diff options
Diffstat (limited to 'drivers/gpu')
106 files changed, 6825 insertions, 2788 deletions
diff --git a/drivers/gpu/drm/drm_crtc.c b/drivers/gpu/drm/drm_crtc.c index fa19c2b9820f..d91fb8c0b7b3 100644 --- a/drivers/gpu/drm/drm_crtc.c +++ b/drivers/gpu/drm/drm_crtc.c @@ -283,7 +283,7 @@ EXPORT_SYMBOL(drm_mode_object_find); * functions & device file and adds it to the master fd list. * * RETURNS: - * Zero on success, error code on falure. + * Zero on success, error code on failure. */ int drm_framebuffer_init(struct drm_device *dev, struct drm_framebuffer *fb, const struct drm_framebuffer_funcs *funcs) @@ -2434,7 +2434,7 @@ int drm_mode_connector_property_set_ioctl(struct drm_device *dev, } else if (connector->funcs->set_property) ret = connector->funcs->set_property(connector, property, out_resp->value); - /* store the property value if succesful */ + /* store the property value if successful */ if (!ret) drm_connector_property_set_value(connector, property, out_resp->value); out: diff --git a/drivers/gpu/drm/drm_crtc_helper.c b/drivers/gpu/drm/drm_crtc_helper.c index 7d0f00a935fa..f2aaf39be398 100644 --- a/drivers/gpu/drm/drm_crtc_helper.c +++ b/drivers/gpu/drm/drm_crtc_helper.c @@ -836,11 +836,7 @@ int drm_crtc_helper_set_config(struct drm_mode_set *set) mode_changed = true; } else if (set->fb == NULL) { mode_changed = true; - } else if ((set->fb->bits_per_pixel != - set->crtc->fb->bits_per_pixel) || - set->fb->depth != set->crtc->fb->depth) - fb_changed = true; - else + } else fb_changed = true; } diff --git a/drivers/gpu/drm/drm_drv.c b/drivers/gpu/drm/drm_drv.c index 766c46875a20..f3c58e2bd75c 100644 --- a/drivers/gpu/drm/drm_drv.c +++ b/drivers/gpu/drm/drm_drv.c @@ -125,28 +125,28 @@ static struct drm_ioctl_desc drm_ioctls[] = { DRM_IOCTL_DEF(DRM_IOCTL_UPDATE_DRAW, drm_update_drawable_info, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), - DRM_IOCTL_DEF(DRM_IOCTL_GEM_CLOSE, drm_gem_close_ioctl, 0), - DRM_IOCTL_DEF(DRM_IOCTL_GEM_FLINK, drm_gem_flink_ioctl, DRM_AUTH), - DRM_IOCTL_DEF(DRM_IOCTL_GEM_OPEN, drm_gem_open_ioctl, DRM_AUTH), - - DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETRESOURCES, drm_mode_getresources, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETCRTC, drm_mode_getcrtc, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETCRTC, drm_mode_setcrtc, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_CURSOR, drm_mode_cursor_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETGAMMA, drm_mode_gamma_get_ioctl, DRM_MASTER), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETGAMMA, drm_mode_gamma_set_ioctl, DRM_MASTER), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETENCODER, drm_mode_getencoder, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETCONNECTOR, drm_mode_getconnector, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_ATTACHMODE, drm_mode_attachmode_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_DETACHMODE, drm_mode_detachmode_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPERTY, drm_mode_getproperty_ioctl, DRM_MASTER | DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETPROPERTY, drm_mode_connector_property_set_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPBLOB, drm_mode_getblob_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETFB, drm_mode_getfb, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_ADDFB, drm_mode_addfb, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_RMFB, drm_mode_rmfb, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_PAGE_FLIP, drm_mode_page_flip_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW), - DRM_IOCTL_DEF(DRM_IOCTL_MODE_DIRTYFB, drm_mode_dirtyfb_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW) + DRM_IOCTL_DEF(DRM_IOCTL_GEM_CLOSE, drm_gem_close_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_GEM_FLINK, drm_gem_flink_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_GEM_OPEN, drm_gem_open_ioctl, DRM_AUTH|DRM_UNLOCKED), + + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETRESOURCES, drm_mode_getresources, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETCRTC, drm_mode_getcrtc, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETCRTC, drm_mode_setcrtc, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_CURSOR, drm_mode_cursor_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETGAMMA, drm_mode_gamma_get_ioctl, DRM_MASTER|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETGAMMA, drm_mode_gamma_set_ioctl, DRM_MASTER|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETENCODER, drm_mode_getencoder, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETCONNECTOR, drm_mode_getconnector, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_ATTACHMODE, drm_mode_attachmode_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_DETACHMODE, drm_mode_detachmode_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPERTY, drm_mode_getproperty_ioctl, DRM_MASTER | DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETPROPERTY, drm_mode_connector_property_set_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPBLOB, drm_mode_getblob_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETFB, drm_mode_getfb, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_ADDFB, drm_mode_addfb, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_RMFB, drm_mode_rmfb, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_PAGE_FLIP, drm_mode_page_flip_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_DIRTYFB, drm_mode_dirtyfb_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED) }; #define DRM_CORE_IOCTL_COUNT ARRAY_SIZE( drm_ioctls ) diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c index f41e91ceaea6..f97e7c42ac8e 100644 --- a/drivers/gpu/drm/drm_edid.c +++ b/drivers/gpu/drm/drm_edid.c @@ -598,6 +598,50 @@ struct drm_display_mode *drm_mode_std(struct drm_device *dev, return mode; } +/* + * EDID is delightfully ambiguous about how interlaced modes are to be + * encoded. Our internal representation is of frame height, but some + * HDTV detailed timings are encoded as field height. + * + * The format list here is from CEA, in frame size. Technically we + * should be checking refresh rate too. Whatever. + */ +static void +drm_mode_do_interlace_quirk(struct drm_display_mode *mode, + struct detailed_pixel_timing *pt) +{ + int i; + static const struct { + int w, h; + } cea_interlaced[] = { + { 1920, 1080 }, + { 720, 480 }, + { 1440, 480 }, + { 2880, 480 }, + { 720, 576 }, + { 1440, 576 }, + { 2880, 576 }, + }; + static const int n_sizes = + sizeof(cea_interlaced)/sizeof(cea_interlaced[0]); + + if (!(pt->misc & DRM_EDID_PT_INTERLACED)) + return; + + for (i = 0; i < n_sizes; i++) { + if ((mode->hdisplay == cea_interlaced[i].w) && + (mode->vdisplay == cea_interlaced[i].h / 2)) { + mode->vdisplay *= 2; + mode->vsync_start *= 2; + mode->vsync_end *= 2; + mode->vtotal *= 2; + mode->vtotal |= 1; + } + } + + mode->flags |= DRM_MODE_FLAG_INTERLACE; +} + /** * drm_mode_detailed - create a new mode from an EDID detailed timing section * @dev: DRM device (needed to create new mode) @@ -680,8 +724,7 @@ static struct drm_display_mode *drm_mode_detailed(struct drm_device *dev, drm_mode_set_name(mode); - if (pt->misc & DRM_EDID_PT_INTERLACED) - mode->flags |= DRM_MODE_FLAG_INTERLACE; + drm_mode_do_interlace_quirk(mode, pt); if (quirks & EDID_QUIRK_DETAILED_SYNC_PP) { pt->misc |= DRM_EDID_PT_HSYNC_POSITIVE | DRM_EDID_PT_VSYNC_POSITIVE; diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c index 0f9e90552dc4..50549703584f 100644 --- a/drivers/gpu/drm/drm_fb_helper.c +++ b/drivers/gpu/drm/drm_fb_helper.c @@ -27,6 +27,7 @@ * Dave Airlie <airlied@linux.ie> * Jesse Barnes <jesse.barnes@intel.com> */ +#include <linux/kernel.h> #include <linux/sysrq.h> #include <linux/fb.h> #include "drmP.h" @@ -50,21 +51,6 @@ int drm_fb_helper_add_connector(struct drm_connector *connector) } EXPORT_SYMBOL(drm_fb_helper_add_connector); -static int my_atoi(const char *name) -{ - int val = 0; - - for (;; name++) { - switch (*name) { - case '0' ... '9': - val = 10*val+(*name-'0'); - break; - default: - return val; - } - } -} - /** * drm_fb_helper_connector_parse_command_line - parse command line for connector * @connector - connector to parse line for @@ -111,7 +97,7 @@ static bool drm_fb_helper_connector_parse_command_line(struct drm_connector *con namelen = i; if (!refresh_specified && !bpp_specified && !yres_specified) { - refresh = my_atoi(&name[i+1]); + refresh = simple_strtol(&name[i+1], NULL, 10); refresh_specified = 1; if (cvt || rb) cvt = 0; @@ -121,7 +107,7 @@ static bool drm_fb_helper_connector_parse_command_line(struct drm_connector *con case '-': namelen = i; if (!bpp_specified && !yres_specified) { - bpp = my_atoi(&name[i+1]); + bpp = simple_strtol(&name[i+1], NULL, 10); bpp_specified = 1; if (cvt || rb) cvt = 0; @@ -130,7 +116,7 @@ static bool drm_fb_helper_connector_parse_command_line(struct drm_connector *con break; case 'x': if (!yres_specified) { - yres = my_atoi(&name[i+1]); + yres = simple_strtol(&name[i+1], NULL, 10); yres_specified = 1; } else goto done; @@ -170,7 +156,7 @@ static bool drm_fb_helper_connector_parse_command_line(struct drm_connector *con } } if (i < 0 && yres_specified) { - xres = my_atoi(name); + xres = simple_strtol(name, NULL, 10); res_specified = 1; } done: @@ -694,7 +680,7 @@ int drm_fb_helper_set_par(struct fb_info *info) int i; if (var->pixclock != 0) { - DRM_ERROR("PIXEL CLCOK SET\n"); + DRM_ERROR("PIXEL CLOCK SET\n"); return -EINVAL; } diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c index e9dbb481c469..aa89d4b0b4c4 100644 --- a/drivers/gpu/drm/drm_gem.c +++ b/drivers/gpu/drm/drm_gem.c @@ -142,19 +142,6 @@ drm_gem_object_alloc(struct drm_device *dev, size_t size) if (IS_ERR(obj->filp)) goto free; - /* Basically we want to disable the OOM killer and handle ENOMEM - * ourselves by sacrificing pages from cached buffers. - * XXX shmem_file_[gs]et_gfp_mask() - */ - mapping_set_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping, - GFP_HIGHUSER | - __GFP_COLD | - __GFP_FS | - __GFP_RECLAIMABLE | - __GFP_NORETRY | - __GFP_NOWARN | - __GFP_NOMEMALLOC); - kref_init(&obj->refcount); kref_init(&obj->handlecount); obj->size = size; @@ -205,9 +192,7 @@ drm_gem_handle_delete(struct drm_file *filp, u32 handle) idr_remove(&filp->object_idr, handle); spin_unlock(&filp->table_lock); - mutex_lock(&dev->struct_mutex); - drm_gem_object_handle_unreference(obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_handle_unreference_unlocked(obj); return 0; } @@ -338,9 +323,7 @@ again: } err: - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(obj); return ret; } @@ -371,9 +354,7 @@ drm_gem_open_ioctl(struct drm_device *dev, void *data, return -ENOENT; ret = drm_gem_handle_create(file_priv, obj, &handle); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(obj); if (ret) return ret; @@ -403,7 +384,7 @@ drm_gem_object_release_handle(int id, void *ptr, void *data) { struct drm_gem_object *obj = ptr; - drm_gem_object_handle_unreference(obj); + drm_gem_object_handle_unreference_unlocked(obj); return 0; } @@ -416,16 +397,25 @@ drm_gem_object_release_handle(int id, void *ptr, void *data) void drm_gem_release(struct drm_device *dev, struct drm_file *file_private) { - mutex_lock(&dev->struct_mutex); idr_for_each(&file_private->object_idr, &drm_gem_object_release_handle, NULL); idr_destroy(&file_private->object_idr); - mutex_unlock(&dev->struct_mutex); +} + +static void +drm_gem_object_free_common(struct drm_gem_object *obj) +{ + struct drm_device *dev = obj->dev; + fput(obj->filp); + atomic_dec(&dev->object_count); + atomic_sub(obj->size, &dev->object_memory); + kfree(obj); } /** * Called after the last reference to the object has been lost. + * Must be called holding struct_ mutex * * Frees the object */ @@ -440,14 +430,40 @@ drm_gem_object_free(struct kref *kref) if (dev->driver->gem_free_object != NULL) dev->driver->gem_free_object(obj); - fput(obj->filp); - atomic_dec(&dev->object_count); - atomic_sub(obj->size, &dev->object_memory); - kfree(obj); + drm_gem_object_free_common(obj); } EXPORT_SYMBOL(drm_gem_object_free); /** + * Called after the last reference to the object has been lost. + * Must be called without holding struct_mutex + * + * Frees the object + */ +void +drm_gem_object_free_unlocked(struct kref *kref) +{ + struct drm_gem_object *obj = (struct drm_gem_object *) kref; + struct drm_device *dev = obj->dev; + + if (dev->driver->gem_free_object_unlocked != NULL) + dev->driver->gem_free_object_unlocked(obj); + else if (dev->driver->gem_free_object != NULL) { + mutex_lock(&dev->struct_mutex); + dev->driver->gem_free_object(obj); + mutex_unlock(&dev->struct_mutex); + } + + drm_gem_object_free_common(obj); +} +EXPORT_SYMBOL(drm_gem_object_free_unlocked); + +static void drm_gem_object_ref_bug(struct kref *list_kref) +{ + BUG(); +} + +/** * Called after the last handle to the object has been closed * * Removes any name for the object. Note that this must be @@ -471,8 +487,10 @@ drm_gem_object_handle_free(struct kref *kref) /* * The object name held a reference to this object, drop * that now. + * + * This cannot be the last reference, since the handle holds one too. */ - drm_gem_object_unreference(obj); + kref_put(&obj->refcount, drm_gem_object_ref_bug); } else spin_unlock(&dev->object_name_lock); @@ -490,11 +508,8 @@ EXPORT_SYMBOL(drm_gem_vm_open); void drm_gem_vm_close(struct vm_area_struct *vma) { struct drm_gem_object *obj = vma->vm_private_data; - struct drm_device *dev = obj->dev; - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(obj); } EXPORT_SYMBOL(drm_gem_vm_close); diff --git a/drivers/gpu/drm/drm_mm.c b/drivers/gpu/drm/drm_mm.c index cdec32977129..2ac074c8f5d2 100644 --- a/drivers/gpu/drm/drm_mm.c +++ b/drivers/gpu/drm/drm_mm.c @@ -405,7 +405,8 @@ struct drm_mm_node *drm_mm_search_free_in_range(const struct drm_mm *mm, wasted += alignment - tmp; } - if (entry->size >= size + wasted) { + if (entry->size >= size + wasted && + (entry->start + wasted + size) <= end) { if (!best_match) return entry; if (entry->size < best_size) { diff --git a/drivers/gpu/drm/drm_modes.c b/drivers/gpu/drm/drm_modes.c index 6d81a02463a3..76d63394c776 100644 --- a/drivers/gpu/drm/drm_modes.c +++ b/drivers/gpu/drm/drm_modes.c @@ -1,9 +1,4 @@ /* - * The list_sort function is (presumably) licensed under the GPL (see the - * top level "COPYING" file for details). - * - * The remainder of this file is: - * * Copyright © 1997-2003 by The XFree86 Project, Inc. * Copyright © 2007 Dave Airlie * Copyright © 2007-2008 Intel Corporation @@ -36,6 +31,7 @@ */ #include <linux/list.h> +#include <linux/list_sort.h> #include "drmP.h" #include "drm.h" #include "drm_crtc.h" @@ -855,6 +851,7 @@ EXPORT_SYMBOL(drm_mode_prune_invalid); /** * drm_mode_compare - compare modes for favorability + * @priv: unused * @lh_a: list_head for first mode * @lh_b: list_head for second mode * @@ -868,7 +865,7 @@ EXPORT_SYMBOL(drm_mode_prune_invalid); * Negative if @lh_a is better than @lh_b, zero if they're equivalent, or * positive if @lh_b is better than @lh_a. */ -static int drm_mode_compare(struct list_head *lh_a, struct list_head *lh_b) +static int drm_mode_compare(void *priv, struct list_head *lh_a, struct list_head *lh_b) { struct drm_display_mode *a = list_entry(lh_a, struct drm_display_mode, head); struct drm_display_mode *b = list_entry(lh_b, struct drm_display_mode, head); @@ -885,85 +882,6 @@ static int drm_mode_compare(struct list_head *lh_a, struct list_head *lh_b) return diff; } -/* FIXME: what we don't have a list sort function? */ -/* list sort from Mark J Roberts (mjr@znex.org) */ -void list_sort(struct list_head *head, - int (*cmp)(struct list_head *a, struct list_head *b)) -{ - struct list_head *p, *q, *e, *list, *tail, *oldhead; - int insize, nmerges, psize, qsize, i; - - list = head->next; - list_del(head); - insize = 1; - for (;;) { - p = oldhead = list; - list = tail = NULL; - nmerges = 0; - - while (p) { - nmerges++; - q = p; - psize = 0; - for (i = 0; i < insize; i++) { - psize++; - q = q->next == oldhead ? NULL : q->next; - if (!q) - break; - } - - qsize = insize; - while (psize > 0 || (qsize > 0 && q)) { - if (!psize) { - e = q; - q = q->next; - qsize--; - if (q == oldhead) - q = NULL; - } else if (!qsize || !q) { - e = p; - p = p->next; - psize--; - if (p == oldhead) - p = NULL; - } else if (cmp(p, q) <= 0) { - e = p; - p = p->next; - psize--; - if (p == oldhead) - p = NULL; - } else { - e = q; - q = q->next; - qsize--; - if (q == oldhead) - q = NULL; - } - if (tail) - tail->next = e; - else - list = e; - e->prev = tail; - tail = e; - } - p = q; - } - - tail->next = list; - list->prev = tail; - - if (nmerges <= 1) - break; - - insize *= 2; - } - - head->next = list; - head->prev = list->prev; - list->prev->next = head; - list->prev = head; -} - /** * drm_mode_sort - sort mode list * @mode_list: list to sort @@ -975,7 +893,7 @@ void list_sort(struct list_head *head, */ void drm_mode_sort(struct list_head *mode_list) { - list_sort(mode_list, drm_mode_compare); + list_sort(NULL, mode_list, drm_mode_compare); } EXPORT_SYMBOL(drm_mode_sort); diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c index 18476bf0b580..1376dfe44c95 100644 --- a/drivers/gpu/drm/i915/i915_debugfs.c +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -162,7 +162,7 @@ static int i915_interrupt_info(struct seq_file *m, void *data) struct drm_device *dev = node->minor->dev; drm_i915_private_t *dev_priv = dev->dev_private; - if (!IS_IRONLAKE(dev)) { + if (!HAS_PCH_SPLIT(dev)) { seq_printf(m, "Interrupt enable: %08x\n", I915_READ(IER)); seq_printf(m, "Interrupt identity: %08x\n", @@ -272,7 +272,7 @@ static void i915_dump_pages(struct seq_file *m, struct page **pages, int page_co mem = kmap_atomic(pages[page], KM_USER0); for (i = 0; i < PAGE_SIZE; i += 4) seq_printf(m, "%08x : %08x\n", i, mem[i / 4]); - kunmap_atomic(pages[page], KM_USER0); + kunmap_atomic(mem, KM_USER0); } } @@ -290,7 +290,7 @@ static int i915_batchbuffer_info(struct seq_file *m, void *data) list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) { obj = obj_priv->obj; if (obj->read_domains & I915_GEM_DOMAIN_COMMAND) { - ret = i915_gem_object_get_pages(obj); + ret = i915_gem_object_get_pages(obj, 0); if (ret) { DRM_ERROR("Failed to get pages: %d\n", ret); spin_unlock(&dev_priv->mm.active_list_lock); @@ -350,6 +350,36 @@ static int i915_ringbuffer_info(struct seq_file *m, void *data) return 0; } +static const char *pin_flag(int pinned) +{ + if (pinned > 0) + return " P"; + else if (pinned < 0) + return " p"; + else + return ""; +} + +static const char *tiling_flag(int tiling) +{ + switch (tiling) { + default: + case I915_TILING_NONE: return ""; + case I915_TILING_X: return " X"; + case I915_TILING_Y: return " Y"; + } +} + +static const char *dirty_flag(int dirty) +{ + return dirty ? " dirty" : ""; +} + +static const char *purgeable_flag(int purgeable) +{ + return purgeable ? " purgeable" : ""; +} + static int i915_error_state(struct seq_file *m, void *unused) { struct drm_info_node *node = (struct drm_info_node *) m->private; @@ -357,6 +387,7 @@ static int i915_error_state(struct seq_file *m, void *unused) drm_i915_private_t *dev_priv = dev->dev_private; struct drm_i915_error_state *error; unsigned long flags; + int i, page, offset, elt; spin_lock_irqsave(&dev_priv->error_lock, flags); if (!dev_priv->first_error) { @@ -368,6 +399,7 @@ static int i915_error_state(struct seq_file *m, void *unused) seq_printf(m, "Time: %ld s %ld us\n", error->time.tv_sec, error->time.tv_usec); + seq_printf(m, "PCI ID: 0x%04x\n", dev->pci_device); seq_printf(m, "EIR: 0x%08x\n", error->eir); seq_printf(m, " PGTBL_ER: 0x%08x\n", error->pgtbl_er); seq_printf(m, " INSTPM: 0x%08x\n", error->instpm); @@ -379,6 +411,59 @@ static int i915_error_state(struct seq_file *m, void *unused) seq_printf(m, " INSTPS: 0x%08x\n", error->instps); seq_printf(m, " INSTDONE1: 0x%08x\n", error->instdone1); } + seq_printf(m, "seqno: 0x%08x\n", error->seqno); + + if (error->active_bo_count) { + seq_printf(m, "Buffers [%d]:\n", error->active_bo_count); + + for (i = 0; i < error->active_bo_count; i++) { + seq_printf(m, " %08x %8zd %08x %08x %08x%s%s%s%s", + error->active_bo[i].gtt_offset, + error->active_bo[i].size, + error->active_bo[i].read_domains, + error->active_bo[i].write_domain, + error->active_bo[i].seqno, + pin_flag(error->active_bo[i].pinned), + tiling_flag(error->active_bo[i].tiling), + dirty_flag(error->active_bo[i].dirty), + purgeable_flag(error->active_bo[i].purgeable)); + + if (error->active_bo[i].name) + seq_printf(m, " (name: %d)", error->active_bo[i].name); + if (error->active_bo[i].fence_reg != I915_FENCE_REG_NONE) + seq_printf(m, " (fence: %d)", error->active_bo[i].fence_reg); + + seq_printf(m, "\n"); + } + } + + for (i = 0; i < ARRAY_SIZE(error->batchbuffer); i++) { + if (error->batchbuffer[i]) { + struct drm_i915_error_object *obj = error->batchbuffer[i]; + + seq_printf(m, "--- gtt_offset = 0x%08x\n", obj->gtt_offset); + offset = 0; + for (page = 0; page < obj->page_count; page++) { + for (elt = 0; elt < PAGE_SIZE/4; elt++) { + seq_printf(m, "%08x : %08x\n", offset, obj->pages[page][elt]); + offset += 4; + } + } + } + } + + if (error->ringbuffer) { + struct drm_i915_error_object *obj = error->ringbuffer; + + seq_printf(m, "--- ringbuffer = 0x%08x\n", obj->gtt_offset); + offset = 0; + for (page = 0; page < obj->page_count; page++) { + for (elt = 0; elt < PAGE_SIZE/4; elt++) { + seq_printf(m, "%08x : %08x\n", offset, obj->pages[page][elt]); + offset += 4; + } + } + } out: spin_unlock_irqrestore(&dev_priv->error_lock, flags); @@ -386,30 +471,161 @@ out: return 0; } -static int i915_registers_info(struct seq_file *m, void *data) { +static int i915_rstdby_delays(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u16 crstanddelay = I915_READ16(CRSTANDVID); + + seq_printf(m, "w/ctx: %d, w/o ctx: %d\n", (crstanddelay >> 8) & 0x3f, (crstanddelay & 0x3f)); + + return 0; +} + +static int i915_cur_delayinfo(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u16 rgvswctl = I915_READ16(MEMSWCTL); + + seq_printf(m, "Last command: 0x%01x\n", (rgvswctl >> 13) & 0x3); + seq_printf(m, "Command status: %d\n", (rgvswctl >> 12) & 1); + seq_printf(m, "P%d DELAY 0x%02x\n", (rgvswctl >> 8) & 0xf, + rgvswctl & 0x3f); + + return 0; +} + +static int i915_delayfreq_table(struct seq_file *m, void *unused) +{ struct drm_info_node *node = (struct drm_info_node *) m->private; struct drm_device *dev = node->minor->dev; drm_i915_private_t *dev_priv = dev->dev_private; - uint32_t reg; - -#define DUMP_RANGE(start, end) \ - for (reg=start; reg < end; reg += 4) \ - seq_printf(m, "%08x\t%08x\n", reg, I915_READ(reg)); - - DUMP_RANGE(0x00000, 0x00fff); /* VGA registers */ - DUMP_RANGE(0x02000, 0x02fff); /* instruction, memory, interrupt control registers */ - DUMP_RANGE(0x03000, 0x031ff); /* FENCE and PPGTT control registers */ - DUMP_RANGE(0x03200, 0x03fff); /* frame buffer compression registers */ - DUMP_RANGE(0x05000, 0x05fff); /* I/O control registers */ - DUMP_RANGE(0x06000, 0x06fff); /* clock control registers */ - DUMP_RANGE(0x07000, 0x07fff); /* 3D internal debug registers */ - DUMP_RANGE(0x07400, 0x088ff); /* GPE debug registers */ - DUMP_RANGE(0x0a000, 0x0afff); /* display palette registers */ - DUMP_RANGE(0x10000, 0x13fff); /* MMIO MCHBAR */ - DUMP_RANGE(0x30000, 0x3ffff); /* overlay registers */ - DUMP_RANGE(0x60000, 0x6ffff); /* display engine pipeline registers */ - DUMP_RANGE(0x70000, 0x72fff); /* display and cursor registers */ - DUMP_RANGE(0x73000, 0x73fff); /* performance counters */ + u32 delayfreq; + int i; + + for (i = 0; i < 16; i++) { + delayfreq = I915_READ(PXVFREQ_BASE + i * 4); + seq_printf(m, "P%02dVIDFREQ: 0x%08x\n", i, delayfreq); + } + + return 0; +} + +static inline int MAP_TO_MV(int map) +{ + return 1250 - (map * 25); +} + +static int i915_inttoext_table(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 inttoext; + int i; + + for (i = 1; i <= 32; i++) { + inttoext = I915_READ(INTTOEXT_BASE_ILK + i * 4); + seq_printf(m, "INTTOEXT%02d: 0x%08x\n", i, inttoext); + } + + return 0; +} + +static int i915_drpc_info(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 rgvmodectl = I915_READ(MEMMODECTL); + + seq_printf(m, "HD boost: %s\n", (rgvmodectl & MEMMODE_BOOST_EN) ? + "yes" : "no"); + seq_printf(m, "Boost freq: %d\n", + (rgvmodectl & MEMMODE_BOOST_FREQ_MASK) >> + MEMMODE_BOOST_FREQ_SHIFT); + seq_printf(m, "HW control enabled: %s\n", + rgvmodectl & MEMMODE_HWIDLE_EN ? "yes" : "no"); + seq_printf(m, "SW control enabled: %s\n", + rgvmodectl & MEMMODE_SWMODE_EN ? "yes" : "no"); + seq_printf(m, "Gated voltage change: %s\n", + rgvmodectl & MEMMODE_RCLK_GATE ? "yes" : "no"); + seq_printf(m, "Starting frequency: P%d\n", + (rgvmodectl & MEMMODE_FSTART_MASK) >> MEMMODE_FSTART_SHIFT); + seq_printf(m, "Max frequency: P%d\n", + (rgvmodectl & MEMMODE_FMAX_MASK) >> MEMMODE_FMAX_SHIFT); + seq_printf(m, "Min frequency: P%d\n", (rgvmodectl & MEMMODE_FMIN_MASK)); + + return 0; +} + +static int i915_fbc_status(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_crtc *crtc; + drm_i915_private_t *dev_priv = dev->dev_private; + bool fbc_enabled = false; + + if (!dev_priv->display.fbc_enabled) { + seq_printf(m, "FBC unsupported on this chipset\n"); + return 0; + } + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + if (!crtc->enabled) + continue; + if (dev_priv->display.fbc_enabled(crtc)) + fbc_enabled = true; + } + + if (fbc_enabled) { + seq_printf(m, "FBC enabled\n"); + } else { + seq_printf(m, "FBC disabled: "); + switch (dev_priv->no_fbc_reason) { + case FBC_STOLEN_TOO_SMALL: + seq_printf(m, "not enough stolen memory"); + break; + case FBC_UNSUPPORTED_MODE: + seq_printf(m, "mode not supported"); + break; + case FBC_MODE_TOO_LARGE: + seq_printf(m, "mode too large"); + break; + case FBC_BAD_PLANE: + seq_printf(m, "FBC unsupported on plane"); + break; + case FBC_NOT_TILED: + seq_printf(m, "scanout buffer not tiled"); + break; + default: + seq_printf(m, "unknown reason"); + } + seq_printf(m, "\n"); + } + return 0; +} + +static int i915_sr_status(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + bool sr_enabled = false; + + if (IS_I965G(dev) || IS_I945G(dev) || IS_I945GM(dev)) + sr_enabled = I915_READ(FW_BLC_SELF) & FW_BLC_SELF_EN; + else if (IS_I915GM(dev)) + sr_enabled = I915_READ(INSTPM) & INSTPM_SELF_EN; + else if (IS_PINEVIEW(dev)) + sr_enabled = I915_READ(DSPFW3) & PINEVIEW_SELF_REFRESH_EN; + + seq_printf(m, "self-refresh: %s\n", sr_enabled ? "enabled" : + "disabled"); return 0; } @@ -519,7 +735,6 @@ static int i915_wedged_create(struct dentry *root, struct drm_minor *minor) } static struct drm_info_list i915_debugfs_list[] = { - {"i915_regs", i915_registers_info, 0}, {"i915_gem_active", i915_gem_object_list_info, 0, (void *) ACTIVE_LIST}, {"i915_gem_flushing", i915_gem_object_list_info, 0, (void *) FLUSHING_LIST}, {"i915_gem_inactive", i915_gem_object_list_info, 0, (void *) INACTIVE_LIST}, @@ -532,6 +747,13 @@ static struct drm_info_list i915_debugfs_list[] = { {"i915_ringbuffer_info", i915_ringbuffer_info, 0}, {"i915_batchbuffers", i915_batchbuffer_info, 0}, {"i915_error_state", i915_error_state, 0}, + {"i915_rstdby_delays", i915_rstdby_delays, 0}, + {"i915_cur_delayinfo", i915_cur_delayinfo, 0}, + {"i915_delayfreq_table", i915_delayfreq_table, 0}, + {"i915_inttoext_table", i915_inttoext_table, 0}, + {"i915_drpc_info", i915_drpc_info, 0}, + {"i915_fbc_status", i915_fbc_status, 0}, + {"i915_sr_status", i915_sr_status, 0}, }; #define I915_DEBUGFS_ENTRIES ARRAY_SIZE(i915_debugfs_list) diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c index 02607ed61399..3e658d6a6b7d 100644 --- a/drivers/gpu/drm/i915/i915_dma.c +++ b/drivers/gpu/drm/i915/i915_dma.c @@ -35,6 +35,8 @@ #include "i915_drv.h" #include "i915_trace.h" #include <linux/vgaarb.h> +#include <linux/acpi.h> +#include <linux/pnp.h> /* Really want an OS-independent resettable timer. Would like to have * this loop run for (eg) 3 sec, but have the timer reset every time @@ -134,6 +136,10 @@ static int i915_init_phys_hws(struct drm_device *dev) memset(dev_priv->hw_status_page, 0, PAGE_SIZE); + if (IS_I965G(dev)) + dev_priv->dma_status_page |= (dev_priv->dma_status_page >> 28) & + 0xf0; + I915_WRITE(HWS_PGA, dev_priv->dma_status_page); DRM_DEBUG_DRIVER("Enabled hardware status page\n"); return 0; @@ -731,8 +737,10 @@ static int i915_cmdbuffer(struct drm_device *dev, void *data, if (cmdbuf->num_cliprects) { cliprects = kcalloc(cmdbuf->num_cliprects, sizeof(struct drm_clip_rect), GFP_KERNEL); - if (cliprects == NULL) + if (cliprects == NULL) { + ret = -ENOMEM; goto fail_batch_free; + } ret = copy_from_user(cliprects, cmdbuf->cliprects, cmdbuf->num_cliprects * @@ -813,9 +821,13 @@ static int i915_getparam(struct drm_device *dev, void *data, case I915_PARAM_HAS_PAGEFLIPPING: value = 1; break; + case I915_PARAM_HAS_EXECBUF2: + /* depends on GEM */ + value = dev_priv->has_gem; + break; default: DRM_DEBUG_DRIVER("Unknown parameter %d\n", - param->param); + param->param); return -EINVAL; } @@ -923,6 +935,120 @@ static int i915_get_bridge_dev(struct drm_device *dev) return 0; } +#define MCHBAR_I915 0x44 +#define MCHBAR_I965 0x48 +#define MCHBAR_SIZE (4*4096) + +#define DEVEN_REG 0x54 +#define DEVEN_MCHBAR_EN (1 << 28) + +/* Allocate space for the MCH regs if needed, return nonzero on error */ +static int +intel_alloc_mchbar_resource(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915; + u32 temp_lo, temp_hi = 0; + u64 mchbar_addr; + int ret = 0; + + if (IS_I965G(dev)) + pci_read_config_dword(dev_priv->bridge_dev, reg + 4, &temp_hi); + pci_read_config_dword(dev_priv->bridge_dev, reg, &temp_lo); + mchbar_addr = ((u64)temp_hi << 32) | temp_lo; + + /* If ACPI doesn't have it, assume we need to allocate it ourselves */ +#ifdef CONFIG_PNP + if (mchbar_addr && + pnp_range_reserved(mchbar_addr, mchbar_addr + MCHBAR_SIZE)) { + ret = 0; + goto out; + } +#endif + + /* Get some space for it */ + ret = pci_bus_alloc_resource(dev_priv->bridge_dev->bus, &dev_priv->mch_res, + MCHBAR_SIZE, MCHBAR_SIZE, + PCIBIOS_MIN_MEM, + 0, pcibios_align_resource, + dev_priv->bridge_dev); + if (ret) { + DRM_DEBUG_DRIVER("failed bus alloc: %d\n", ret); + dev_priv->mch_res.start = 0; + goto out; + } + + if (IS_I965G(dev)) + pci_write_config_dword(dev_priv->bridge_dev, reg + 4, + upper_32_bits(dev_priv->mch_res.start)); + + pci_write_config_dword(dev_priv->bridge_dev, reg, + lower_32_bits(dev_priv->mch_res.start)); +out: + return ret; +} + +/* Setup MCHBAR if possible, return true if we should disable it again */ +static void +intel_setup_mchbar(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int mchbar_reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915; + u32 temp; + bool enabled; + + dev_priv->mchbar_need_disable = false; + + if (IS_I915G(dev) || IS_I915GM(dev)) { + pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp); + enabled = !!(temp & DEVEN_MCHBAR_EN); + } else { + pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); + enabled = temp & 1; + } + + /* If it's already enabled, don't have to do anything */ + if (enabled) + return; + + if (intel_alloc_mchbar_resource(dev)) + return; + + dev_priv->mchbar_need_disable = true; + + /* Space is allocated or reserved, so enable it. */ + if (IS_I915G(dev) || IS_I915GM(dev)) { + pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG, + temp | DEVEN_MCHBAR_EN); + } else { + pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); + pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp | 1); + } +} + +static void +intel_teardown_mchbar(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int mchbar_reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915; + u32 temp; + + if (dev_priv->mchbar_need_disable) { + if (IS_I915G(dev) || IS_I915GM(dev)) { + pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp); + temp &= ~DEVEN_MCHBAR_EN; + pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG, temp); + } else { + pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); + temp &= ~1; + pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp); + } + } + + if (dev_priv->mch_res.start) + release_resource(&dev_priv->mch_res); +} + /** * i915_probe_agp - get AGP bootup configuration * @pdev: PCI device @@ -968,59 +1094,123 @@ static int i915_probe_agp(struct drm_device *dev, uint32_t *aperture_size, * Some of the preallocated space is taken by the GTT * and popup. GTT is 1K per MB of aperture size, and popup is 4K. */ - if (IS_G4X(dev) || IS_PINEVIEW(dev) || IS_IRONLAKE(dev)) + if (IS_G4X(dev) || IS_PINEVIEW(dev) || IS_IRONLAKE(dev) || IS_GEN6(dev)) overhead = 4096; else overhead = (*aperture_size / 1024) + 4096; - switch (tmp & INTEL_GMCH_GMS_MASK) { - case INTEL_855_GMCH_GMS_DISABLED: - DRM_ERROR("video memory is disabled\n"); - return -1; - case INTEL_855_GMCH_GMS_STOLEN_1M: - stolen = 1 * 1024 * 1024; - break; - case INTEL_855_GMCH_GMS_STOLEN_4M: - stolen = 4 * 1024 * 1024; - break; - case INTEL_855_GMCH_GMS_STOLEN_8M: - stolen = 8 * 1024 * 1024; - break; - case INTEL_855_GMCH_GMS_STOLEN_16M: - stolen = 16 * 1024 * 1024; - break; - case INTEL_855_GMCH_GMS_STOLEN_32M: - stolen = 32 * 1024 * 1024; - break; - case INTEL_915G_GMCH_GMS_STOLEN_48M: - stolen = 48 * 1024 * 1024; - break; - case INTEL_915G_GMCH_GMS_STOLEN_64M: - stolen = 64 * 1024 * 1024; - break; - case INTEL_GMCH_GMS_STOLEN_128M: - stolen = 128 * 1024 * 1024; - break; - case INTEL_GMCH_GMS_STOLEN_256M: - stolen = 256 * 1024 * 1024; - break; - case INTEL_GMCH_GMS_STOLEN_96M: - stolen = 96 * 1024 * 1024; - break; - case INTEL_GMCH_GMS_STOLEN_160M: - stolen = 160 * 1024 * 1024; - break; - case INTEL_GMCH_GMS_STOLEN_224M: - stolen = 224 * 1024 * 1024; - break; - case INTEL_GMCH_GMS_STOLEN_352M: - stolen = 352 * 1024 * 1024; - break; - default: - DRM_ERROR("unexpected GMCH_GMS value: 0x%02x\n", - tmp & INTEL_GMCH_GMS_MASK); - return -1; + if (IS_GEN6(dev)) { + /* SNB has memory control reg at 0x50.w */ + pci_read_config_word(dev->pdev, SNB_GMCH_CTRL, &tmp); + + switch (tmp & SNB_GMCH_GMS_STOLEN_MASK) { + case INTEL_855_GMCH_GMS_DISABLED: + DRM_ERROR("video memory is disabled\n"); + return -1; + case SNB_GMCH_GMS_STOLEN_32M: + stolen = 32 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_64M: + stolen = 64 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_96M: + stolen = 96 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_128M: + stolen = 128 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_160M: + stolen = 160 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_192M: + stolen = 192 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_224M: + stolen = 224 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_256M: + stolen = 256 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_288M: + stolen = 288 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_320M: + stolen = 320 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_352M: + stolen = 352 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_384M: + stolen = 384 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_416M: + stolen = 416 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_448M: + stolen = 448 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_480M: + stolen = 480 * 1024 * 1024; + break; + case SNB_GMCH_GMS_STOLEN_512M: + stolen = 512 * 1024 * 1024; + break; + default: + DRM_ERROR("unexpected GMCH_GMS value: 0x%02x\n", + tmp & SNB_GMCH_GMS_STOLEN_MASK); + return -1; + } + } else { + switch (tmp & INTEL_GMCH_GMS_MASK) { + case INTEL_855_GMCH_GMS_DISABLED: + DRM_ERROR("video memory is disabled\n"); + return -1; + case INTEL_855_GMCH_GMS_STOLEN_1M: + stolen = 1 * 1024 * 1024; + break; + case INTEL_855_GMCH_GMS_STOLEN_4M: + stolen = 4 * 1024 * 1024; + break; + case INTEL_855_GMCH_GMS_STOLEN_8M: + stolen = 8 * 1024 * 1024; + break; + case INTEL_855_GMCH_GMS_STOLEN_16M: + stolen = 16 * 1024 * 1024; + break; + case INTEL_855_GMCH_GMS_STOLEN_32M: + stolen = 32 * 1024 * 1024; + break; + case INTEL_915G_GMCH_GMS_STOLEN_48M: + stolen = 48 * 1024 * 1024; + break; + case INTEL_915G_GMCH_GMS_STOLEN_64M: + stolen = 64 * 1024 * 1024; + break; + case INTEL_GMCH_GMS_STOLEN_128M: + stolen = 128 * 1024 * 1024; + break; + case INTEL_GMCH_GMS_STOLEN_256M: + stolen = 256 * 1024 * 1024; + break; + case INTEL_GMCH_GMS_STOLEN_96M: + stolen = 96 * 1024 * 1024; + break; + case INTEL_GMCH_GMS_STOLEN_160M: + stolen = 160 * 1024 * 1024; + break; + case INTEL_GMCH_GMS_STOLEN_224M: + stolen = 224 * 1024 * 1024; + break; + case INTEL_GMCH_GMS_STOLEN_352M: + stolen = 352 * 1024 * 1024; + break; + default: + DRM_ERROR("unexpected GMCH_GMS value: 0x%02x\n", + tmp & INTEL_GMCH_GMS_MASK); + return -1; + } } + *preallocated_size = stolen - overhead; *start = overhead; @@ -1054,7 +1244,7 @@ static unsigned long i915_gtt_to_phys(struct drm_device *dev, int gtt_offset, gtt_size; if (IS_I965G(dev)) { - if (IS_G4X(dev) || IS_IRONLAKE(dev)) { + if (IS_G4X(dev) || IS_IRONLAKE(dev) || IS_GEN6(dev)) { gtt_offset = 2*1024*1024; gtt_size = 2*1024*1024; } else { @@ -1117,11 +1307,13 @@ static void i915_setup_compression(struct drm_device *dev, int size) { struct drm_i915_private *dev_priv = dev->dev_private; struct drm_mm_node *compressed_fb, *compressed_llb; - unsigned long cfb_base, ll_base; + unsigned long cfb_base; + unsigned long ll_base = 0; /* Leave 1M for line length buffer & misc. */ compressed_fb = drm_mm_search_free(&dev_priv->vram, size, 4096, 0); if (!compressed_fb) { + dev_priv->no_fbc_reason = FBC_STOLEN_TOO_SMALL; i915_warn_stolen(dev); return; } @@ -1129,6 +1321,7 @@ static void i915_setup_compression(struct drm_device *dev, int size) compressed_fb = drm_mm_get_block(compressed_fb, size, 4096); if (!compressed_fb) { i915_warn_stolen(dev); + dev_priv->no_fbc_reason = FBC_STOLEN_TOO_SMALL; return; } @@ -1200,14 +1393,6 @@ static int i915_load_modeset_init(struct drm_device *dev, dev->mode_config.fb_base = drm_get_resource_start(dev, fb_bar) & 0xff000000; - if (IS_MOBILE(dev) || IS_I9XX(dev)) - dev_priv->cursor_needs_physical = true; - else - dev_priv->cursor_needs_physical = false; - - if (IS_I965G(dev) || IS_G33(dev)) - dev_priv->cursor_needs_physical = false; - /* Basic memrange allocator for stolen space (aka vram) */ drm_mm_init(&dev_priv->vram, 0, prealloc_size); DRM_INFO("set up %ldM of stolen space\n", prealloc_size / (1024*1024)); @@ -1257,6 +1442,8 @@ static int i915_load_modeset_init(struct drm_device *dev, if (ret) goto destroy_ringbuffer; + intel_modeset_init(dev); + ret = drm_irq_install(dev); if (ret) goto destroy_ringbuffer; @@ -1271,14 +1458,14 @@ static int i915_load_modeset_init(struct drm_device *dev, I915_WRITE(INSTPM, (1 << 5) | (1 << 21)); - intel_modeset_init(dev); - drm_helper_initial_config(dev); return 0; destroy_ringbuffer: + mutex_lock(&dev->struct_mutex); i915_gem_cleanup_ringbuffer(dev); + mutex_unlock(&dev->struct_mutex); out: return ret; } @@ -1360,7 +1547,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) { struct drm_i915_private *dev_priv = dev->dev_private; resource_size_t base, size; - int ret = 0, mmio_bar = IS_I9XX(dev) ? 0 : 1; + int ret = 0, mmio_bar; uint32_t agp_size, prealloc_size, prealloc_start; /* i915 has 4 more counters */ @@ -1376,8 +1563,10 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) dev->dev_private = (void *)dev_priv; dev_priv->dev = dev; + dev_priv->info = (struct intel_device_info *) flags; /* Add register map (needed for suspend/resume) */ + mmio_bar = IS_I9XX(dev) ? 0 : 1; base = drm_get_resource_start(dev, mmio_bar); size = drm_get_resource_len(dev, mmio_bar); @@ -1440,11 +1629,14 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) dev->driver->get_vblank_counter = i915_get_vblank_counter; dev->max_vblank_count = 0xffffff; /* only 24 bits of frame count */ - if (IS_G4X(dev) || IS_IRONLAKE(dev)) { + if (IS_G4X(dev) || IS_IRONLAKE(dev) || IS_GEN6(dev)) { dev->max_vblank_count = 0xffffffff; /* full 32 bit counter */ dev->driver->get_vblank_counter = gm45_get_vblank_counter; } + /* Try to make sure MCHBAR is enabled before poking at it */ + intel_setup_mchbar(dev); + i915_gem_load(dev); /* Init HWS */ @@ -1518,6 +1710,8 @@ int i915_driver_unload(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; + i915_destroy_error_state(dev); + destroy_workqueue(dev_priv->wq); del_timer_sync(&dev_priv->hangcheck_timer); @@ -1564,6 +1758,8 @@ int i915_driver_unload(struct drm_device *dev) intel_cleanup_overlay(dev); } + intel_teardown_mchbar(dev); + pci_dev_put(dev_priv->bridge_dev); kfree(dev->dev_private); @@ -1652,6 +1848,7 @@ struct drm_ioctl_desc i915_ioctls[] = { DRM_IOCTL_DEF(DRM_I915_HWS_ADDR, i915_set_status_page, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), DRM_IOCTL_DEF(DRM_I915_GEM_INIT, i915_gem_init_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), DRM_IOCTL_DEF(DRM_I915_GEM_EXECBUFFER, i915_gem_execbuffer, DRM_AUTH), + DRM_IOCTL_DEF(DRM_I915_GEM_EXECBUFFER2, i915_gem_execbuffer2, DRM_AUTH), DRM_IOCTL_DEF(DRM_I915_GEM_PIN, i915_gem_pin_ioctl, DRM_AUTH|DRM_ROOT_ONLY), DRM_IOCTL_DEF(DRM_I915_GEM_UNPIN, i915_gem_unpin_ioctl, DRM_AUTH|DRM_ROOT_ONLY), DRM_IOCTL_DEF(DRM_I915_GEM_BUSY, i915_gem_busy_ioctl, DRM_AUTH), diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index 24286ca168fc..85ad020125c8 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -33,7 +33,6 @@ #include "i915_drm.h" #include "i915_drv.h" -#include "drm_pciids.h" #include <linux/console.h> #include "drm_crtc_helper.h" @@ -46,36 +45,162 @@ module_param_named(fbpercrtc, i915_fbpercrtc, int, 0400); unsigned int i915_powersave = 1; module_param_named(powersave, i915_powersave, int, 0400); +unsigned int i915_lvds_downclock = 0; +module_param_named(lvds_downclock, i915_lvds_downclock, int, 0400); + static struct drm_driver driver; +extern int intel_agp_enabled; + +#define INTEL_VGA_DEVICE(id, info) { \ + .class = PCI_CLASS_DISPLAY_VGA << 8, \ + .class_mask = 0xffff00, \ + .vendor = 0x8086, \ + .device = id, \ + .subvendor = PCI_ANY_ID, \ + .subdevice = PCI_ANY_ID, \ + .driver_data = (unsigned long) info } + +const static struct intel_device_info intel_i830_info = { + .is_i8xx = 1, .is_mobile = 1, .cursor_needs_physical = 1, +}; + +const static struct intel_device_info intel_845g_info = { + .is_i8xx = 1, +}; + +const static struct intel_device_info intel_i85x_info = { + .is_i8xx = 1, .is_mobile = 1, .cursor_needs_physical = 1, +}; + +const static struct intel_device_info intel_i865g_info = { + .is_i8xx = 1, +}; + +const static struct intel_device_info intel_i915g_info = { + .is_i915g = 1, .is_i9xx = 1, .cursor_needs_physical = 1, +}; +const static struct intel_device_info intel_i915gm_info = { + .is_i9xx = 1, .is_mobile = 1, .has_fbc = 1, + .cursor_needs_physical = 1, +}; +const static struct intel_device_info intel_i945g_info = { + .is_i9xx = 1, .has_hotplug = 1, .cursor_needs_physical = 1, +}; +const static struct intel_device_info intel_i945gm_info = { + .is_i945gm = 1, .is_i9xx = 1, .is_mobile = 1, .has_fbc = 1, + .has_hotplug = 1, .cursor_needs_physical = 1, +}; + +const static struct intel_device_info intel_i965g_info = { + .is_i965g = 1, .is_i9xx = 1, .has_hotplug = 1, +}; + +const static struct intel_device_info intel_i965gm_info = { + .is_i965g = 1, .is_mobile = 1, .is_i965gm = 1, .is_i9xx = 1, + .is_mobile = 1, .has_fbc = 1, .has_rc6 = 1, + .has_hotplug = 1, +}; + +const static struct intel_device_info intel_g33_info = { + .is_g33 = 1, .is_i9xx = 1, .need_gfx_hws = 1, + .has_hotplug = 1, +}; + +const static struct intel_device_info intel_g45_info = { + .is_i965g = 1, .is_g4x = 1, .is_i9xx = 1, .need_gfx_hws = 1, + .has_pipe_cxsr = 1, + .has_hotplug = 1, +}; -static struct pci_device_id pciidlist[] = { - i915_PCI_IDS +const static struct intel_device_info intel_gm45_info = { + .is_i965g = 1, .is_mobile = 1, .is_g4x = 1, .is_i9xx = 1, + .is_mobile = 1, .need_gfx_hws = 1, .has_fbc = 1, .has_rc6 = 1, + .has_pipe_cxsr = 1, + .has_hotplug = 1, +}; + +const static struct intel_device_info intel_pineview_info = { + .is_g33 = 1, .is_pineview = 1, .is_mobile = 1, .is_i9xx = 1, + .need_gfx_hws = 1, + .has_hotplug = 1, +}; + +const static struct intel_device_info intel_ironlake_d_info = { + .is_ironlake = 1, .is_i965g = 1, .is_i9xx = 1, .need_gfx_hws = 1, + .has_pipe_cxsr = 1, + .has_hotplug = 1, +}; + +const static struct intel_device_info intel_ironlake_m_info = { + .is_ironlake = 1, .is_mobile = 1, .is_i965g = 1, .is_i9xx = 1, + .need_gfx_hws = 1, .has_rc6 = 1, + .has_hotplug = 1, +}; + +const static struct intel_device_info intel_sandybridge_d_info = { + .is_i965g = 1, .is_i9xx = 1, .need_gfx_hws = 1, + .has_hotplug = 1, +}; + +const static struct intel_device_info intel_sandybridge_m_info = { + .is_i965g = 1, .is_mobile = 1, .is_i9xx = 1, .need_gfx_hws = 1, + .has_hotplug = 1, +}; + +const static struct pci_device_id pciidlist[] = { + INTEL_VGA_DEVICE(0x3577, &intel_i830_info), + INTEL_VGA_DEVICE(0x2562, &intel_845g_info), + INTEL_VGA_DEVICE(0x3582, &intel_i85x_info), + INTEL_VGA_DEVICE(0x35e8, &intel_i85x_info), + INTEL_VGA_DEVICE(0x2572, &intel_i865g_info), + INTEL_VGA_DEVICE(0x2582, &intel_i915g_info), + INTEL_VGA_DEVICE(0x258a, &intel_i915g_info), + INTEL_VGA_DEVICE(0x2592, &intel_i915gm_info), + INTEL_VGA_DEVICE(0x2772, &intel_i945g_info), + INTEL_VGA_DEVICE(0x27a2, &intel_i945gm_info), + INTEL_VGA_DEVICE(0x27ae, &intel_i945gm_info), + INTEL_VGA_DEVICE(0x2972, &intel_i965g_info), + INTEL_VGA_DEVICE(0x2982, &intel_i965g_info), + INTEL_VGA_DEVICE(0x2992, &intel_i965g_info), + INTEL_VGA_DEVICE(0x29a2, &intel_i965g_info), + INTEL_VGA_DEVICE(0x29b2, &intel_g33_info), + INTEL_VGA_DEVICE(0x29c2, &intel_g33_info), + INTEL_VGA_DEVICE(0x29d2, &intel_g33_info), + INTEL_VGA_DEVICE(0x2a02, &intel_i965gm_info), + INTEL_VGA_DEVICE(0x2a12, &intel_i965gm_info), + INTEL_VGA_DEVICE(0x2a42, &intel_gm45_info), + INTEL_VGA_DEVICE(0x2e02, &intel_g45_info), + INTEL_VGA_DEVICE(0x2e12, &intel_g45_info), + INTEL_VGA_DEVICE(0x2e22, &intel_g45_info), + INTEL_VGA_DEVICE(0x2e32, &intel_g45_info), + INTEL_VGA_DEVICE(0x2e42, &intel_g45_info), + INTEL_VGA_DEVICE(0xa001, &intel_pineview_info), + INTEL_VGA_DEVICE(0xa011, &intel_pineview_info), + INTEL_VGA_DEVICE(0x0042, &intel_ironlake_d_info), + INTEL_VGA_DEVICE(0x0046, &intel_ironlake_m_info), + INTEL_VGA_DEVICE(0x0102, &intel_sandybridge_d_info), + INTEL_VGA_DEVICE(0x0106, &intel_sandybridge_m_info), + {0, 0, 0} }; #if defined(CONFIG_DRM_I915_KMS) MODULE_DEVICE_TABLE(pci, pciidlist); #endif -static int i915_suspend(struct drm_device *dev, pm_message_t state) +static int i915_drm_freeze(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - if (!dev || !dev_priv) { - DRM_ERROR("dev: %p, dev_priv: %p\n", dev, dev_priv); - DRM_ERROR("DRM not initialized, aborting suspend.\n"); - return -ENODEV; - } - - if (state.event == PM_EVENT_PRETHAW) - return 0; - pci_save_state(dev->pdev); /* If KMS is active, we do the leavevt stuff here */ if (drm_core_check_feature(dev, DRIVER_MODESET)) { - if (i915_gem_idle(dev)) + int error = i915_gem_idle(dev); + if (error) { dev_err(&dev->pdev->dev, - "GEM idle failed, resume may fail\n"); + "GEM idle failed, resume might fail\n"); + return error; + } drm_irq_uninstall(dev); } @@ -83,26 +208,42 @@ static int i915_suspend(struct drm_device *dev, pm_message_t state) intel_opregion_free(dev, 1); + /* Modeset on resume, not lid events */ + dev_priv->modeset_on_lid = 0; + + return 0; +} + +static int i915_suspend(struct drm_device *dev, pm_message_t state) +{ + int error; + + if (!dev || !dev->dev_private) { + DRM_ERROR("dev: %p\n", dev); + DRM_ERROR("DRM not initialized, aborting suspend.\n"); + return -ENODEV; + } + + if (state.event == PM_EVENT_PRETHAW) + return 0; + + error = i915_drm_freeze(dev); + if (error) + return error; + if (state.event == PM_EVENT_SUSPEND) { /* Shut down the device */ pci_disable_device(dev->pdev); pci_set_power_state(dev->pdev, PCI_D3hot); } - /* Modeset on resume, not lid events */ - dev_priv->modeset_on_lid = 0; - return 0; } -static int i915_resume(struct drm_device *dev) +static int i915_drm_thaw(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - int ret = 0; - - if (pci_enable_device(dev->pdev)) - return -1; - pci_set_master(dev->pdev); + int error = 0; i915_restore_state(dev); @@ -113,21 +254,28 @@ static int i915_resume(struct drm_device *dev) mutex_lock(&dev->struct_mutex); dev_priv->mm.suspended = 0; - ret = i915_gem_init_ringbuffer(dev); - if (ret != 0) - ret = -1; + error = i915_gem_init_ringbuffer(dev); mutex_unlock(&dev->struct_mutex); drm_irq_install(dev); - } - if (drm_core_check_feature(dev, DRIVER_MODESET)) { + /* Resume the modeset for every activated CRTC */ drm_helper_resume_force_mode(dev); } dev_priv->modeset_on_lid = 0; - return ret; + return error; +} + +static int i915_resume(struct drm_device *dev) +{ + if (pci_enable_device(dev->pdev)) + return -EIO; + + pci_set_master(dev->pdev); + + return i915_drm_thaw(dev); } /** @@ -268,22 +416,73 @@ i915_pci_remove(struct pci_dev *pdev) drm_put_dev(dev); } -static int -i915_pci_suspend(struct pci_dev *pdev, pm_message_t state) +static int i915_pm_suspend(struct device *dev) { - struct drm_device *dev = pci_get_drvdata(pdev); + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + int error; + + if (!drm_dev || !drm_dev->dev_private) { + dev_err(dev, "DRM not initialized, aborting suspend.\n"); + return -ENODEV; + } + + error = i915_drm_freeze(drm_dev); + if (error) + return error; + + pci_disable_device(pdev); + pci_set_power_state(pdev, PCI_D3hot); - return i915_suspend(dev, state); + return 0; } -static int -i915_pci_resume(struct pci_dev *pdev) +static int i915_pm_resume(struct device *dev) { - struct drm_device *dev = pci_get_drvdata(pdev); + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + + return i915_resume(drm_dev); +} + +static int i915_pm_freeze(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + + if (!drm_dev || !drm_dev->dev_private) { + dev_err(dev, "DRM not initialized, aborting suspend.\n"); + return -ENODEV; + } - return i915_resume(dev); + return i915_drm_freeze(drm_dev); } +static int i915_pm_thaw(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + + return i915_drm_thaw(drm_dev); +} + +static int i915_pm_poweroff(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + + return i915_drm_freeze(drm_dev); +} + +const struct dev_pm_ops i915_pm_ops = { + .suspend = i915_pm_suspend, + .resume = i915_pm_resume, + .freeze = i915_pm_freeze, + .thaw = i915_pm_thaw, + .poweroff = i915_pm_poweroff, + .restore = i915_pm_resume, +}; + static struct vm_operations_struct i915_gem_vm_ops = { .fault = i915_gem_fault, .open = drm_gem_vm_open, @@ -303,8 +502,11 @@ static struct drm_driver driver = { .lastclose = i915_driver_lastclose, .preclose = i915_driver_preclose, .postclose = i915_driver_postclose, + + /* Used in place of i915_pm_ops for non-DRIVER_MODESET */ .suspend = i915_suspend, .resume = i915_resume, + .device_is_agp = i915_driver_device_is_agp, .enable_vblank = i915_enable_vblank, .disable_vblank = i915_disable_vblank, @@ -344,10 +546,7 @@ static struct drm_driver driver = { .id_table = pciidlist, .probe = i915_pci_probe, .remove = i915_pci_remove, -#ifdef CONFIG_PM - .resume = i915_pci_resume, - .suspend = i915_pci_suspend, -#endif + .driver.pm = &i915_pm_ops, }, .name = DRIVER_NAME, @@ -360,6 +559,11 @@ static struct drm_driver driver = { static int __init i915_init(void) { + if (!intel_agp_enabled) { + DRM_ERROR("drm/i915 can't work without intel_agp module!\n"); + return -ENODEV; + } + driver.num_ioctls = i915_max_ioctl; i915_gem_shrinker_init(); @@ -385,6 +589,11 @@ static int __init i915_init(void) driver.driver_features &= ~DRIVER_MODESET; #endif + if (!(driver.driver_features & DRIVER_MODESET)) { + driver.suspend = i915_suspend; + driver.resume = i915_resume; + } + return drm_init(&driver); } diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index fbecac72f5bb..f97592609da4 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -150,7 +150,27 @@ struct drm_i915_error_state { u32 instps; u32 instdone1; u32 seqno; + u64 bbaddr; struct timeval time; + struct drm_i915_error_object { + int page_count; + u32 gtt_offset; + u32 *pages[0]; + } *ringbuffer, *batchbuffer[2]; + struct drm_i915_error_buffer { + size_t size; + u32 name; + u32 seqno; + u32 gtt_offset; + u32 read_domains; + u32 write_domain; + u32 fence_reg; + s32 pinned:2; + u32 tiling:2; + u32 dirty:1; + u32 purgeable:1; + } *active_bo; + u32 active_bo_count; }; struct drm_i915_display_funcs { @@ -172,9 +192,39 @@ struct drm_i915_display_funcs { struct intel_overlay; +struct intel_device_info { + u8 is_mobile : 1; + u8 is_i8xx : 1; + u8 is_i915g : 1; + u8 is_i9xx : 1; + u8 is_i945gm : 1; + u8 is_i965g : 1; + u8 is_i965gm : 1; + u8 is_g33 : 1; + u8 need_gfx_hws : 1; + u8 is_g4x : 1; + u8 is_pineview : 1; + u8 is_ironlake : 1; + u8 has_fbc : 1; + u8 has_rc6 : 1; + u8 has_pipe_cxsr : 1; + u8 has_hotplug : 1; + u8 cursor_needs_physical : 1; +}; + +enum no_fbc_reason { + FBC_STOLEN_TOO_SMALL, /* not enough space to hold compressed buffers */ + FBC_UNSUPPORTED_MODE, /* interlace or doublescanned mode */ + FBC_MODE_TOO_LARGE, /* mode too large for compression */ + FBC_BAD_PLANE, /* fbc not supported on plane */ + FBC_NOT_TILED, /* buffer not tiled */ +}; + typedef struct drm_i915_private { struct drm_device *dev; + const struct intel_device_info *info; + int has_gem; void __iomem *regs; @@ -232,8 +282,6 @@ typedef struct drm_i915_private { int hangcheck_count; uint32_t last_acthd; - bool cursor_needs_physical; - struct drm_mm vram; unsigned long cfb_size; @@ -263,6 +311,7 @@ typedef struct drm_i915_private { unsigned int lvds_use_ssc:1; unsigned int edp_support:1; int lvds_ssc_freq; + int edp_bpp; struct notifier_block lid_notifier; @@ -287,8 +336,6 @@ typedef struct drm_i915_private { u32 saveDSPACNTR; u32 saveDSPBCNTR; u32 saveDSPARB; - u32 saveRENDERSTANDBY; - u32 savePWRCTXA; u32 saveHWS; u32 savePIPEACONF; u32 savePIPEBCONF; @@ -433,6 +480,7 @@ typedef struct drm_i915_private { u32 savePIPEB_DATA_N1; u32 savePIPEB_LINK_M1; u32 savePIPEB_LINK_N1; + u32 saveMCHBAR_RENDER_STANDBY; struct { struct drm_mm gtt_space; @@ -474,6 +522,15 @@ typedef struct drm_i915_private { struct list_head flushing_list; /** + * List of objects currently pending a GPU write flush. + * + * All elements on this list will belong to either the + * active_list or flushing_list, last_rendering_seqno can + * be used to differentiate between the two elements. + */ + struct list_head gpu_write_list; + + /** * LRU list of objects which are not in the ringbuffer and * are ready to unbind, but are still in the GTT. * @@ -561,6 +618,15 @@ typedef struct drm_i915_private { u16 orig_clock; int child_dev_num; struct child_device_config *child_dev; + struct drm_connector *int_lvds_connector; + + bool mchbar_need_disable; + + u8 cur_delay; + u8 min_delay; + u8 max_delay; + + enum no_fbc_reason no_fbc_reason; } drm_i915_private_t; /** driver private structure attached to each drm_gem_object */ @@ -572,6 +638,8 @@ struct drm_i915_gem_object { /** This object's place on the active/flushing/inactive lists */ struct list_head list; + /** This object's place on GPU write list */ + struct list_head gpu_write_list; /** This object's place on the fenced object LRU */ struct list_head fence_list; @@ -703,6 +771,7 @@ extern struct drm_ioctl_desc i915_ioctls[]; extern int i915_max_ioctl; extern unsigned int i915_fbpercrtc; extern unsigned int i915_powersave; +extern unsigned int i915_lvds_downclock; extern void i915_save_display(struct drm_device *dev); extern void i915_restore_display(struct drm_device *dev); @@ -729,6 +798,7 @@ extern int i965_reset(struct drm_device *dev, u8 flags); /* i915_irq.c */ void i915_hangcheck_elapsed(unsigned long data); +void i915_destroy_error_state(struct drm_device *dev); extern int i915_irq_emit(struct drm_device *dev, void *data, struct drm_file *file_priv); extern int i915_irq_wait(struct drm_device *dev, void *data, @@ -794,6 +864,8 @@ int i915_gem_sw_finish_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv); int i915_gem_execbuffer(struct drm_device *dev, void *data, struct drm_file *file_priv); +int i915_gem_execbuffer2(struct drm_device *dev, void *data, + struct drm_file *file_priv); int i915_gem_pin_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv); int i915_gem_unpin_ioctl(struct drm_device *dev, void *data, @@ -843,12 +915,13 @@ int i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptib int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf); int i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write); +int i915_gem_object_set_to_display_plane(struct drm_gem_object *obj); int i915_gem_attach_phys_object(struct drm_device *dev, struct drm_gem_object *obj, int id); void i915_gem_detach_phys_object(struct drm_device *dev, struct drm_gem_object *obj); void i915_gem_free_all_phys_object(struct drm_device *dev); -int i915_gem_object_get_pages(struct drm_gem_object *obj); +int i915_gem_object_get_pages(struct drm_gem_object *obj, gfp_t gfpmask); void i915_gem_object_put_pages(struct drm_gem_object *obj); void i915_gem_release(struct drm_device * dev, struct drm_file *file_priv); void i915_gem_object_flush_write_domain(struct drm_gem_object *obj); @@ -860,6 +933,10 @@ void i915_gem_shrinker_exit(void); void i915_gem_detect_bit_6_swizzle(struct drm_device *dev); void i915_gem_object_do_bit_17_swizzle(struct drm_gem_object *obj); void i915_gem_object_save_bit_17_swizzle(struct drm_gem_object *obj); +bool i915_tiling_ok(struct drm_device *dev, int stride, int size, + int tiling_mode); +bool i915_gem_object_fence_offset_ok(struct drm_gem_object *obj, + int tiling_mode); /* i915_gem_debug.c */ void i915_gem_dump_object(struct drm_gem_object *obj, int len, @@ -982,67 +1059,54 @@ extern void g4x_disable_fbc(struct drm_device *dev); extern int i915_wrap_ring(struct drm_device * dev); extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller); -#define IS_I830(dev) ((dev)->pci_device == 0x3577) -#define IS_845G(dev) ((dev)->pci_device == 0x2562) -#define IS_I85X(dev) ((dev)->pci_device == 0x3582) -#define IS_I865G(dev) ((dev)->pci_device == 0x2572) -#define IS_I8XX(dev) (IS_I830(dev) || IS_845G(dev) || IS_I85X(dev) || IS_I865G(dev)) - -#define IS_I915G(dev) ((dev)->pci_device == 0x2582 || (dev)->pci_device == 0x258a) -#define IS_I915GM(dev) ((dev)->pci_device == 0x2592) -#define IS_I945G(dev) ((dev)->pci_device == 0x2772) -#define IS_I945GM(dev) ((dev)->pci_device == 0x27A2 ||\ - (dev)->pci_device == 0x27AE) -#define IS_I965G(dev) ((dev)->pci_device == 0x2972 || \ - (dev)->pci_device == 0x2982 || \ - (dev)->pci_device == 0x2992 || \ - (dev)->pci_device == 0x29A2 || \ - (dev)->pci_device == 0x2A02 || \ - (dev)->pci_device == 0x2A12 || \ - (dev)->pci_device == 0x2A42 || \ - (dev)->pci_device == 0x2E02 || \ - (dev)->pci_device == 0x2E12 || \ - (dev)->pci_device == 0x2E22 || \ - (dev)->pci_device == 0x2E32 || \ - (dev)->pci_device == 0x2E42 || \ - (dev)->pci_device == 0x0042 || \ - (dev)->pci_device == 0x0046) - -#define IS_I965GM(dev) ((dev)->pci_device == 0x2A02 || \ - (dev)->pci_device == 0x2A12) - -#define IS_GM45(dev) ((dev)->pci_device == 0x2A42) - -#define IS_G4X(dev) ((dev)->pci_device == 0x2E02 || \ - (dev)->pci_device == 0x2E12 || \ - (dev)->pci_device == 0x2E22 || \ - (dev)->pci_device == 0x2E32 || \ - (dev)->pci_device == 0x2E42 || \ - IS_GM45(dev)) - -#define IS_PINEVIEW_G(dev) ((dev)->pci_device == 0xa001) -#define IS_PINEVIEW_M(dev) ((dev)->pci_device == 0xa011) -#define IS_PINEVIEW(dev) (IS_PINEVIEW_G(dev) || IS_PINEVIEW_M(dev)) - -#define IS_G33(dev) ((dev)->pci_device == 0x29C2 || \ - (dev)->pci_device == 0x29B2 || \ - (dev)->pci_device == 0x29D2 || \ - (IS_PINEVIEW(dev))) - +#define INTEL_INFO(dev) (((struct drm_i915_private *) (dev)->dev_private)->info) + +#define IS_I830(dev) ((dev)->pci_device == 0x3577) +#define IS_845G(dev) ((dev)->pci_device == 0x2562) +#define IS_I85X(dev) ((dev)->pci_device == 0x3582) +#define IS_I865G(dev) ((dev)->pci_device == 0x2572) +#define IS_GEN2(dev) (INTEL_INFO(dev)->is_i8xx) +#define IS_I915G(dev) (INTEL_INFO(dev)->is_i915g) +#define IS_I915GM(dev) ((dev)->pci_device == 0x2592) +#define IS_I945G(dev) ((dev)->pci_device == 0x2772) +#define IS_I945GM(dev) (INTEL_INFO(dev)->is_i945gm) +#define IS_I965G(dev) (INTEL_INFO(dev)->is_i965g) +#define IS_I965GM(dev) (INTEL_INFO(dev)->is_i965gm) +#define IS_GM45(dev) ((dev)->pci_device == 0x2A42) +#define IS_G4X(dev) (INTEL_INFO(dev)->is_g4x) +#define IS_PINEVIEW_G(dev) ((dev)->pci_device == 0xa001) +#define IS_PINEVIEW_M(dev) ((dev)->pci_device == 0xa011) +#define IS_PINEVIEW(dev) (INTEL_INFO(dev)->is_pineview) +#define IS_G33(dev) (INTEL_INFO(dev)->is_g33) #define IS_IRONLAKE_D(dev) ((dev)->pci_device == 0x0042) #define IS_IRONLAKE_M(dev) ((dev)->pci_device == 0x0046) -#define IS_IRONLAKE(dev) (IS_IRONLAKE_D(dev) || IS_IRONLAKE_M(dev)) - -#define IS_I9XX(dev) (IS_I915G(dev) || IS_I915GM(dev) || IS_I945G(dev) || \ - IS_I945GM(dev) || IS_I965G(dev) || IS_G33(dev) || \ - IS_IRONLAKE(dev)) +#define IS_IRONLAKE(dev) (INTEL_INFO(dev)->is_ironlake) +#define IS_I9XX(dev) (INTEL_INFO(dev)->is_i9xx) +#define IS_MOBILE(dev) (INTEL_INFO(dev)->is_mobile) + +#define IS_GEN3(dev) (IS_I915G(dev) || \ + IS_I915GM(dev) || \ + IS_I945G(dev) || \ + IS_I945GM(dev) || \ + IS_G33(dev) || \ + IS_PINEVIEW(dev)) +#define IS_GEN4(dev) ((dev)->pci_device == 0x2972 || \ + (dev)->pci_device == 0x2982 || \ + (dev)->pci_device == 0x2992 || \ + (dev)->pci_device == 0x29A2 || \ + (dev)->pci_device == 0x2A02 || \ + (dev)->pci_device == 0x2A12 || \ + (dev)->pci_device == 0x2E02 || \ + (dev)->pci_device == 0x2E12 || \ + (dev)->pci_device == 0x2E22 || \ + (dev)->pci_device == 0x2E32 || \ + (dev)->pci_device == 0x2A42 || \ + (dev)->pci_device == 0x2E42) + +#define I915_NEED_GFX_HWS(dev) (INTEL_INFO(dev)->need_gfx_hws) + +#define IS_GEN6(dev) ((dev)->pci_device == 0x0102) -#define IS_MOBILE(dev) (IS_I830(dev) || IS_I85X(dev) || IS_I915GM(dev) || \ - IS_I945GM(dev) || IS_I965GM(dev) || IS_GM45(dev) || \ - IS_PINEVIEW(dev) || IS_IRONLAKE_M(dev)) - -#define I915_NEED_GFX_HWS(dev) (IS_G33(dev) || IS_GM45(dev) || IS_G4X(dev) || \ - IS_IRONLAKE(dev)) /* With the 945 and later, Y tiling got adjusted so that it was 32 128-byte * rows, which changed the alignment requirements and fence programming. */ @@ -1054,17 +1118,17 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller); #define SUPPORTS_EDP(dev) (IS_IRONLAKE_M(dev)) #define SUPPORTS_TV(dev) (IS_I9XX(dev) && IS_MOBILE(dev) && \ !IS_IRONLAKE(dev) && !IS_PINEVIEW(dev)) -#define I915_HAS_HOTPLUG(dev) (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev) || IS_I965G(dev)) +#define I915_HAS_HOTPLUG(dev) (INTEL_INFO(dev)->has_hotplug) /* dsparb controlled by hw only */ #define DSPARB_HWCONTROL(dev) (IS_G4X(dev) || IS_IRONLAKE(dev)) #define HAS_FW_BLC(dev) (IS_I9XX(dev) || IS_G4X(dev) || IS_IRONLAKE(dev)) -#define HAS_PIPE_CXSR(dev) (IS_G4X(dev) || IS_IRONLAKE(dev)) -#define I915_HAS_FBC(dev) (IS_MOBILE(dev) && \ - (IS_I9XX(dev) || IS_GM45(dev)) && \ - !IS_PINEVIEW(dev) && \ - !IS_IRONLAKE(dev)) -#define I915_HAS_RC6(dev) (IS_I965GM(dev) || IS_GM45(dev) || IS_IRONLAKE_M(dev)) +#define HAS_PIPE_CXSR(dev) (INTEL_INFO(dev)->has_pipe_cxsr) +#define I915_HAS_FBC(dev) (INTEL_INFO(dev)->has_fbc) +#define I915_HAS_RC6(dev) (INTEL_INFO(dev)->has_rc6) + +#define HAS_PCH_SPLIT(dev) (IS_IRONLAKE(dev) || \ + IS_GEN6(dev)) #define PRIMARY_RINGBUFFER_SIZE (128*1024) diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index c7f0cbec4e84..fba37e9f775d 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -128,9 +128,7 @@ i915_gem_create_ioctl(struct drm_device *dev, void *data, return -ENOMEM; ret = drm_gem_handle_create(file_priv, obj, &handle); - mutex_lock(&dev->struct_mutex); - drm_gem_object_handle_unreference(obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_handle_unreference_unlocked(obj); if (ret) return ret; @@ -277,7 +275,7 @@ i915_gem_shmem_pread_fast(struct drm_device *dev, struct drm_gem_object *obj, mutex_lock(&dev->struct_mutex); - ret = i915_gem_object_get_pages(obj); + ret = i915_gem_object_get_pages(obj, 0); if (ret != 0) goto fail_unlock; @@ -321,40 +319,24 @@ fail_unlock: return ret; } -static inline gfp_t -i915_gem_object_get_page_gfp_mask (struct drm_gem_object *obj) -{ - return mapping_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping); -} - -static inline void -i915_gem_object_set_page_gfp_mask (struct drm_gem_object *obj, gfp_t gfp) -{ - mapping_set_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping, gfp); -} - static int i915_gem_object_get_pages_or_evict(struct drm_gem_object *obj) { int ret; - ret = i915_gem_object_get_pages(obj); + ret = i915_gem_object_get_pages(obj, __GFP_NORETRY | __GFP_NOWARN); /* If we've insufficient memory to map in the pages, attempt * to make some space by throwing out some old buffers. */ if (ret == -ENOMEM) { struct drm_device *dev = obj->dev; - gfp_t gfp; ret = i915_gem_evict_something(dev, obj->size); if (ret) return ret; - gfp = i915_gem_object_get_page_gfp_mask(obj); - i915_gem_object_set_page_gfp_mask(obj, gfp & ~__GFP_NORETRY); - ret = i915_gem_object_get_pages(obj); - i915_gem_object_set_page_gfp_mask (obj, gfp); + ret = i915_gem_object_get_pages(obj, 0); } return ret; @@ -504,7 +486,7 @@ i915_gem_pread_ioctl(struct drm_device *dev, void *data, */ if (args->offset > obj->size || args->size > obj->size || args->offset + args->size > obj->size) { - drm_gem_object_unreference(obj); + drm_gem_object_unreference_unlocked(obj); return -EINVAL; } @@ -517,7 +499,7 @@ i915_gem_pread_ioctl(struct drm_device *dev, void *data, file_priv); } - drm_gem_object_unreference(obj); + drm_gem_object_unreference_unlocked(obj); return ret; } @@ -790,7 +772,7 @@ i915_gem_shmem_pwrite_fast(struct drm_device *dev, struct drm_gem_object *obj, mutex_lock(&dev->struct_mutex); - ret = i915_gem_object_get_pages(obj); + ret = i915_gem_object_get_pages(obj, 0); if (ret != 0) goto fail_unlock; @@ -977,7 +959,7 @@ i915_gem_pwrite_ioctl(struct drm_device *dev, void *data, */ if (args->offset > obj->size || args->size > obj->size || args->offset + args->size > obj->size) { - drm_gem_object_unreference(obj); + drm_gem_object_unreference_unlocked(obj); return -EINVAL; } @@ -1011,7 +993,7 @@ i915_gem_pwrite_ioctl(struct drm_device *dev, void *data, DRM_INFO("pwrite failed %d\n", ret); #endif - drm_gem_object_unreference(obj); + drm_gem_object_unreference_unlocked(obj); return ret; } @@ -1154,9 +1136,7 @@ i915_gem_mmap_ioctl(struct drm_device *dev, void *data, PROT_READ | PROT_WRITE, MAP_SHARED, args->offset); up_write(¤t->mm->mmap_sem); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(obj); if (IS_ERR((void *)addr)) return addr; @@ -1310,7 +1290,7 @@ out_free_list: * i915_gem_release_mmap - remove physical page mappings * @obj: obj in question * - * Preserve the reservation of the mmaping with the DRM core code, but + * Preserve the reservation of the mmapping with the DRM core code, but * relinquish ownership of the pages back to the system. * * It is vital that we remove the page mapping if we have mapped a tiled @@ -1568,6 +1548,8 @@ i915_gem_object_move_to_inactive(struct drm_gem_object *obj) else list_move_tail(&obj_priv->list, &dev_priv->mm.inactive_list); + BUG_ON(!list_empty(&obj_priv->gpu_write_list)); + obj_priv->last_rendering_seqno = 0; if (obj_priv->active) { obj_priv->active = 0; @@ -1576,6 +1558,38 @@ i915_gem_object_move_to_inactive(struct drm_gem_object *obj) i915_verify_inactive(dev, __FILE__, __LINE__); } +static void +i915_gem_process_flushing_list(struct drm_device *dev, + uint32_t flush_domains, uint32_t seqno) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj_priv, *next; + + list_for_each_entry_safe(obj_priv, next, + &dev_priv->mm.gpu_write_list, + gpu_write_list) { + struct drm_gem_object *obj = obj_priv->obj; + + if ((obj->write_domain & flush_domains) == + obj->write_domain) { + uint32_t old_write_domain = obj->write_domain; + + obj->write_domain = 0; + list_del_init(&obj_priv->gpu_write_list); + i915_gem_object_move_to_active(obj, seqno); + + /* update the fence lru list */ + if (obj_priv->fence_reg != I915_FENCE_REG_NONE) + list_move_tail(&obj_priv->fence_list, + &dev_priv->mm.fence_list); + + trace_i915_gem_object_change_domain(obj, + obj->read_domains, + old_write_domain); + } + } +} + /** * Creates a new sequence number, emitting a write of it to the status page * plus an interrupt, which will trigger i915_user_interrupt_handler. @@ -1634,27 +1648,8 @@ i915_add_request(struct drm_device *dev, struct drm_file *file_priv, /* Associate any objects on the flushing list matching the write * domain we're flushing with our flush. */ - if (flush_domains != 0) { - struct drm_i915_gem_object *obj_priv, *next; - - list_for_each_entry_safe(obj_priv, next, - &dev_priv->mm.flushing_list, list) { - struct drm_gem_object *obj = obj_priv->obj; - - if ((obj->write_domain & flush_domains) == - obj->write_domain) { - uint32_t old_write_domain = obj->write_domain; - - obj->write_domain = 0; - i915_gem_object_move_to_active(obj, seqno); - - trace_i915_gem_object_change_domain(obj, - obj->read_domains, - old_write_domain); - } - } - - } + if (flush_domains != 0) + i915_gem_process_flushing_list(dev, flush_domains, seqno); if (!dev_priv->mm.suspended) { mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD); @@ -1834,7 +1829,7 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible) return -EIO; if (!i915_seqno_passed(i915_get_gem_seqno(dev), seqno)) { - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) ier = I915_READ(DEIER) | I915_READ(GTIER); else ier = I915_READ(IER); @@ -2003,6 +1998,7 @@ int i915_gem_object_unbind(struct drm_gem_object *obj) { struct drm_device *dev = obj->dev; + drm_i915_private_t *dev_priv = dev->dev_private; struct drm_i915_gem_object *obj_priv = obj->driver_private; int ret = 0; @@ -2021,9 +2017,6 @@ i915_gem_object_unbind(struct drm_gem_object *obj) /* blow away mappings if mapped through GTT */ i915_gem_release_mmap(obj); - if (obj_priv->fence_reg != I915_FENCE_REG_NONE) - i915_gem_clear_fence_reg(obj); - /* Move the object to the CPU domain to ensure that * any possible CPU writes while it's not in the GTT * are flushed when we go to remap it. This will @@ -2039,6 +2032,10 @@ i915_gem_object_unbind(struct drm_gem_object *obj) BUG_ON(obj_priv->active); + /* release the fence reg _after_ flushing */ + if (obj_priv->fence_reg != I915_FENCE_REG_NONE) + i915_gem_clear_fence_reg(obj); + if (obj_priv->agp_mem != NULL) { drm_unbind_agp(obj_priv->agp_mem); drm_free_agp(obj_priv->agp_mem, obj->size / PAGE_SIZE); @@ -2057,8 +2054,10 @@ i915_gem_object_unbind(struct drm_gem_object *obj) } /* Remove ourselves from the LRU list if present. */ + spin_lock(&dev_priv->mm.active_list_lock); if (!list_empty(&obj_priv->list)) list_del_init(&obj_priv->list); + spin_unlock(&dev_priv->mm.active_list_lock); if (i915_gem_object_is_purgeable(obj_priv)) i915_gem_object_truncate(obj); @@ -2096,10 +2095,33 @@ i915_gem_find_inactive_object(struct drm_device *dev, int min_size) } static int -i915_gem_evict_everything(struct drm_device *dev) +i915_gpu_idle(struct drm_device *dev) { drm_i915_private_t *dev_priv = dev->dev_private; + bool lists_empty; uint32_t seqno; + + spin_lock(&dev_priv->mm.active_list_lock); + lists_empty = list_empty(&dev_priv->mm.flushing_list) && + list_empty(&dev_priv->mm.active_list); + spin_unlock(&dev_priv->mm.active_list_lock); + + if (lists_empty) + return 0; + + /* Flush everything onto the inactive list. */ + i915_gem_flush(dev, I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS); + seqno = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS); + if (seqno == 0) + return -ENOMEM; + + return i915_wait_request(dev, seqno); +} + +static int +i915_gem_evict_everything(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; int ret; bool lists_empty; @@ -2113,15 +2135,12 @@ i915_gem_evict_everything(struct drm_device *dev) return -ENOSPC; /* Flush everything (on to the inactive lists) and evict */ - i915_gem_flush(dev, I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS); - seqno = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS); - if (seqno == 0) - return -ENOMEM; - - ret = i915_wait_request(dev, seqno); + ret = i915_gpu_idle(dev); if (ret) return ret; + BUG_ON(!list_empty(&dev_priv->mm.flushing_list)); + ret = i915_gem_evict_from_inactive_list(dev); if (ret) return ret; @@ -2229,7 +2248,8 @@ i915_gem_evict_something(struct drm_device *dev, int min_size) } int -i915_gem_object_get_pages(struct drm_gem_object *obj) +i915_gem_object_get_pages(struct drm_gem_object *obj, + gfp_t gfpmask) { struct drm_i915_gem_object *obj_priv = obj->driver_private; int page_count, i; @@ -2255,7 +2275,10 @@ i915_gem_object_get_pages(struct drm_gem_object *obj) inode = obj->filp->f_path.dentry->d_inode; mapping = inode->i_mapping; for (i = 0; i < page_count; i++) { - page = read_mapping_page(mapping, i, NULL); + page = read_cache_page_gfp(mapping, i, + mapping_gfp_mask (mapping) | + __GFP_COLD | + gfpmask); if (IS_ERR(page)) { ret = PTR_ERR(page); i915_gem_object_put_pages(obj); @@ -2270,6 +2293,28 @@ i915_gem_object_get_pages(struct drm_gem_object *obj) return 0; } +static void sandybridge_write_fence_reg(struct drm_i915_fence_reg *reg) +{ + struct drm_gem_object *obj = reg->obj; + struct drm_device *dev = obj->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj_priv = obj->driver_private; + int regnum = obj_priv->fence_reg; + uint64_t val; + + val = (uint64_t)((obj_priv->gtt_offset + obj->size - 4096) & + 0xfffff000) << 32; + val |= obj_priv->gtt_offset & 0xfffff000; + val |= (uint64_t)((obj_priv->stride / 128) - 1) << + SANDYBRIDGE_FENCE_PITCH_SHIFT; + + if (obj_priv->tiling_mode == I915_TILING_Y) + val |= 1 << I965_FENCE_TILING_Y_SHIFT; + val |= I965_FENCE_REG_VALID; + + I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + (regnum * 8), val); +} + static void i965_write_fence_reg(struct drm_i915_fence_reg *reg) { struct drm_gem_object *obj = reg->obj; @@ -2366,6 +2411,58 @@ static void i830_write_fence_reg(struct drm_i915_fence_reg *reg) I915_WRITE(FENCE_REG_830_0 + (regnum * 4), val); } +static int i915_find_fence_reg(struct drm_device *dev) +{ + struct drm_i915_fence_reg *reg = NULL; + struct drm_i915_gem_object *obj_priv = NULL; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_gem_object *obj = NULL; + int i, avail, ret; + + /* First try to find a free reg */ + avail = 0; + for (i = dev_priv->fence_reg_start; i < dev_priv->num_fence_regs; i++) { + reg = &dev_priv->fence_regs[i]; + if (!reg->obj) + return i; + + obj_priv = reg->obj->driver_private; + if (!obj_priv->pin_count) + avail++; + } + + if (avail == 0) + return -ENOSPC; + + /* None available, try to steal one or wait for a user to finish */ + i = I915_FENCE_REG_NONE; + list_for_each_entry(obj_priv, &dev_priv->mm.fence_list, + fence_list) { + obj = obj_priv->obj; + + if (obj_priv->pin_count) + continue; + + /* found one! */ + i = obj_priv->fence_reg; + break; + } + + BUG_ON(i == I915_FENCE_REG_NONE); + + /* We only have a reference on obj from the active list. put_fence_reg + * might drop that one, causing a use-after-free in it. So hold a + * private reference to obj like the other callers of put_fence_reg + * (set_tiling ioctl) do. */ + drm_gem_object_reference(obj); + ret = i915_gem_object_put_fence_reg(obj); + drm_gem_object_unreference(obj); + if (ret != 0) + return ret; + + return i; +} + /** * i915_gem_object_get_fence_reg - set up a fence reg for an object * @obj: object to map through a fence reg @@ -2386,8 +2483,7 @@ i915_gem_object_get_fence_reg(struct drm_gem_object *obj) struct drm_i915_private *dev_priv = dev->dev_private; struct drm_i915_gem_object *obj_priv = obj->driver_private; struct drm_i915_fence_reg *reg = NULL; - struct drm_i915_gem_object *old_obj_priv = NULL; - int i, ret, avail; + int ret; /* Just update our place in the LRU if our fence is getting used. */ if (obj_priv->fence_reg != I915_FENCE_REG_NONE) { @@ -2415,86 +2511,27 @@ i915_gem_object_get_fence_reg(struct drm_gem_object *obj) break; } - /* First try to find a free reg */ - avail = 0; - for (i = dev_priv->fence_reg_start; i < dev_priv->num_fence_regs; i++) { - reg = &dev_priv->fence_regs[i]; - if (!reg->obj) - break; - - old_obj_priv = reg->obj->driver_private; - if (!old_obj_priv->pin_count) - avail++; - } - - /* None available, try to steal one or wait for a user to finish */ - if (i == dev_priv->num_fence_regs) { - struct drm_gem_object *old_obj = NULL; - - if (avail == 0) - return -ENOSPC; - - list_for_each_entry(old_obj_priv, &dev_priv->mm.fence_list, - fence_list) { - old_obj = old_obj_priv->obj; - - if (old_obj_priv->pin_count) - continue; - - /* Take a reference, as otherwise the wait_rendering - * below may cause the object to get freed out from - * under us. - */ - drm_gem_object_reference(old_obj); - - /* i915 uses fences for GPU access to tiled buffers */ - if (IS_I965G(dev) || !old_obj_priv->active) - break; - - /* This brings the object to the head of the LRU if it - * had been written to. The only way this should - * result in us waiting longer than the expected - * optimal amount of time is if there was a - * fence-using buffer later that was read-only. - */ - i915_gem_object_flush_gpu_write_domain(old_obj); - ret = i915_gem_object_wait_rendering(old_obj); - if (ret != 0) { - drm_gem_object_unreference(old_obj); - return ret; - } - - break; - } - - /* - * Zap this virtual mapping so we can set up a fence again - * for this object next time we need it. - */ - i915_gem_release_mmap(old_obj); - - i = old_obj_priv->fence_reg; - reg = &dev_priv->fence_regs[i]; - - old_obj_priv->fence_reg = I915_FENCE_REG_NONE; - list_del_init(&old_obj_priv->fence_list); - - drm_gem_object_unreference(old_obj); - } + ret = i915_find_fence_reg(dev); + if (ret < 0) + return ret; - obj_priv->fence_reg = i; + obj_priv->fence_reg = ret; + reg = &dev_priv->fence_regs[obj_priv->fence_reg]; list_add_tail(&obj_priv->fence_list, &dev_priv->mm.fence_list); reg->obj = obj; - if (IS_I965G(dev)) + if (IS_GEN6(dev)) + sandybridge_write_fence_reg(reg); + else if (IS_I965G(dev)) i965_write_fence_reg(reg); else if (IS_I9XX(dev)) i915_write_fence_reg(reg); else i830_write_fence_reg(reg); - trace_i915_gem_object_get_fence(obj, i, obj_priv->tiling_mode); + trace_i915_gem_object_get_fence(obj, obj_priv->fence_reg, + obj_priv->tiling_mode); return 0; } @@ -2513,9 +2550,12 @@ i915_gem_clear_fence_reg(struct drm_gem_object *obj) drm_i915_private_t *dev_priv = dev->dev_private; struct drm_i915_gem_object *obj_priv = obj->driver_private; - if (IS_I965G(dev)) + if (IS_GEN6(dev)) { + I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + + (obj_priv->fence_reg * 8), 0); + } else if (IS_I965G(dev)) { I915_WRITE64(FENCE_REG_965_0 + (obj_priv->fence_reg * 8), 0); - else { + } else { uint32_t fence_reg; if (obj_priv->fence_reg < 8) @@ -2549,6 +2589,12 @@ i915_gem_object_put_fence_reg(struct drm_gem_object *obj) if (obj_priv->fence_reg == I915_FENCE_REG_NONE) return 0; + /* If we've changed tiling, GTT-mappings of the object + * need to re-fault to ensure that the correct fence register + * setup is in place. + */ + i915_gem_release_mmap(obj); + /* On the i915, GPU access to tiled buffers is via a fence, * therefore we must wait for any outstanding access to complete * before clearing the fence. @@ -2557,12 +2603,12 @@ i915_gem_object_put_fence_reg(struct drm_gem_object *obj) int ret; i915_gem_object_flush_gpu_write_domain(obj); - i915_gem_object_flush_gtt_write_domain(obj); ret = i915_gem_object_wait_rendering(obj); if (ret != 0) return ret; } + i915_gem_object_flush_gtt_write_domain(obj); i915_gem_clear_fence_reg (obj); return 0; @@ -2578,12 +2624,9 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment) drm_i915_private_t *dev_priv = dev->dev_private; struct drm_i915_gem_object *obj_priv = obj->driver_private; struct drm_mm_node *free_space; - bool retry_alloc = false; + gfp_t gfpmask = __GFP_NORETRY | __GFP_NOWARN; int ret; - if (dev_priv->mm.suspended) - return -EBUSY; - if (obj_priv->madv != I915_MADV_WILLNEED) { DRM_ERROR("Attempting to bind a purgeable object\n"); return -EINVAL; @@ -2625,15 +2668,7 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment) DRM_INFO("Binding object of size %zd at 0x%08x\n", obj->size, obj_priv->gtt_offset); #endif - if (retry_alloc) { - i915_gem_object_set_page_gfp_mask (obj, - i915_gem_object_get_page_gfp_mask (obj) & ~__GFP_NORETRY); - } - ret = i915_gem_object_get_pages(obj); - if (retry_alloc) { - i915_gem_object_set_page_gfp_mask (obj, - i915_gem_object_get_page_gfp_mask (obj) | __GFP_NORETRY); - } + ret = i915_gem_object_get_pages(obj, gfpmask); if (ret) { drm_mm_put_block(obj_priv->gtt_space); obj_priv->gtt_space = NULL; @@ -2643,9 +2678,9 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment) ret = i915_gem_evict_something(dev, obj->size); if (ret) { /* now try to shrink everyone else */ - if (! retry_alloc) { - retry_alloc = true; - goto search_free; + if (gfpmask) { + gfpmask = 0; + goto search_free; } return ret; @@ -2713,7 +2748,6 @@ static void i915_gem_object_flush_gpu_write_domain(struct drm_gem_object *obj) { struct drm_device *dev = obj->dev; - uint32_t seqno; uint32_t old_write_domain; if ((obj->write_domain & I915_GEM_GPU_DOMAINS) == 0) @@ -2722,9 +2756,8 @@ i915_gem_object_flush_gpu_write_domain(struct drm_gem_object *obj) /* Queue the GPU write cache flushing we need. */ old_write_domain = obj->write_domain; i915_gem_flush(dev, 0, obj->write_domain); - seqno = i915_add_request(dev, NULL, obj->write_domain); - obj->write_domain = 0; - i915_gem_object_move_to_active(obj, seqno); + (void) i915_add_request(dev, NULL, obj->write_domain); + BUG_ON(obj->write_domain); trace_i915_gem_object_change_domain(obj, obj->read_domains, @@ -2839,6 +2872,57 @@ i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write) return 0; } +/* + * Prepare buffer for display plane. Use uninterruptible for possible flush + * wait, as in modesetting process we're not supposed to be interrupted. + */ +int +i915_gem_object_set_to_display_plane(struct drm_gem_object *obj) +{ + struct drm_device *dev = obj->dev; + struct drm_i915_gem_object *obj_priv = obj->driver_private; + uint32_t old_write_domain, old_read_domains; + int ret; + + /* Not valid to be called on unbound objects. */ + if (obj_priv->gtt_space == NULL) + return -EINVAL; + + i915_gem_object_flush_gpu_write_domain(obj); + + /* Wait on any GPU rendering and flushing to occur. */ + if (obj_priv->active) { +#if WATCH_BUF + DRM_INFO("%s: object %p wait for seqno %08x\n", + __func__, obj, obj_priv->last_rendering_seqno); +#endif + ret = i915_do_wait_request(dev, obj_priv->last_rendering_seqno, 0); + if (ret != 0) + return ret; + } + + old_write_domain = obj->write_domain; + old_read_domains = obj->read_domains; + + obj->read_domains &= I915_GEM_DOMAIN_GTT; + + i915_gem_object_flush_cpu_write_domain(obj); + + /* It should now be out of any other write domains, and we can update + * the domain values for our changes. + */ + BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_GTT) != 0); + obj->read_domains |= I915_GEM_DOMAIN_GTT; + obj->write_domain = I915_GEM_DOMAIN_GTT; + obj_priv->dirty = 1; + + trace_i915_gem_object_change_domain(obj, + old_read_domains, + old_write_domain); + + return 0; +} + /** * Moves a single object to the CPU read, and possibly write domain. * @@ -3198,7 +3282,7 @@ i915_gem_object_set_cpu_read_domain_range(struct drm_gem_object *obj, static int i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, struct drm_file *file_priv, - struct drm_i915_gem_exec_object *entry, + struct drm_i915_gem_exec_object2 *entry, struct drm_i915_gem_relocation_entry *relocs) { struct drm_device *dev = obj->dev; @@ -3206,12 +3290,36 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, struct drm_i915_gem_object *obj_priv = obj->driver_private; int i, ret; void __iomem *reloc_page; + bool need_fence; + + need_fence = entry->flags & EXEC_OBJECT_NEEDS_FENCE && + obj_priv->tiling_mode != I915_TILING_NONE; + + /* Check fence reg constraints and rebind if necessary */ + if (need_fence && !i915_gem_object_fence_offset_ok(obj, + obj_priv->tiling_mode)) + i915_gem_object_unbind(obj); /* Choose the GTT offset for our buffer and put it there. */ ret = i915_gem_object_pin(obj, (uint32_t) entry->alignment); if (ret) return ret; + /* + * Pre-965 chips need a fence register set up in order to + * properly handle blits to/from tiled surfaces. + */ + if (need_fence) { + ret = i915_gem_object_get_fence_reg(obj); + if (ret != 0) { + if (ret != -EBUSY && ret != -ERESTARTSYS) + DRM_ERROR("Failure to install fence: %d\n", + ret); + i915_gem_object_unpin(obj); + return ret; + } + } + entry->offset = obj_priv->gtt_offset; /* Apply the relocations, using the GTT aperture to avoid cache @@ -3259,6 +3367,16 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, } /* Validate that the target is in a valid r/w GPU domain */ + if (reloc->write_domain & (reloc->write_domain - 1)) { + DRM_ERROR("reloc with multiple write domains: " + "obj %p target %d offset %d " + "read %08x write %08x", + obj, reloc->target_handle, + (int) reloc->offset, + reloc->read_domains, + reloc->write_domain); + return -EINVAL; + } if (reloc->write_domain & I915_GEM_DOMAIN_CPU || reloc->read_domains & I915_GEM_DOMAIN_CPU) { DRM_ERROR("reloc with read/write CPU domains: " @@ -3373,7 +3491,7 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, */ static int i915_dispatch_gem_execbuffer(struct drm_device *dev, - struct drm_i915_gem_execbuffer *exec, + struct drm_i915_gem_execbuffer2 *exec, struct drm_clip_rect *cliprects, uint64_t exec_offset) { @@ -3463,7 +3581,7 @@ i915_gem_ring_throttle(struct drm_device *dev, struct drm_file *file_priv) } static int -i915_gem_get_relocs_from_user(struct drm_i915_gem_exec_object *exec_list, +i915_gem_get_relocs_from_user(struct drm_i915_gem_exec_object2 *exec_list, uint32_t buffer_count, struct drm_i915_gem_relocation_entry **relocs) { @@ -3478,8 +3596,10 @@ i915_gem_get_relocs_from_user(struct drm_i915_gem_exec_object *exec_list, } *relocs = drm_calloc_large(reloc_count, sizeof(**relocs)); - if (*relocs == NULL) + if (*relocs == NULL) { + DRM_ERROR("failed to alloc relocs, count %d\n", reloc_count); return -ENOMEM; + } for (i = 0; i < buffer_count; i++) { struct drm_i915_gem_relocation_entry __user *user_relocs; @@ -3503,13 +3623,16 @@ i915_gem_get_relocs_from_user(struct drm_i915_gem_exec_object *exec_list, } static int -i915_gem_put_relocs_to_user(struct drm_i915_gem_exec_object *exec_list, +i915_gem_put_relocs_to_user(struct drm_i915_gem_exec_object2 *exec_list, uint32_t buffer_count, struct drm_i915_gem_relocation_entry *relocs) { uint32_t reloc_count = 0, i; int ret = 0; + if (relocs == NULL) + return 0; + for (i = 0; i < buffer_count; i++) { struct drm_i915_gem_relocation_entry __user *user_relocs; int unwritten; @@ -3536,7 +3659,7 @@ err: } static int -i915_gem_check_execbuffer (struct drm_i915_gem_execbuffer *exec, +i915_gem_check_execbuffer (struct drm_i915_gem_execbuffer2 *exec, uint64_t exec_offset) { uint32_t exec_start, exec_len; @@ -3589,18 +3712,18 @@ i915_gem_wait_for_pending_flip(struct drm_device *dev, } int -i915_gem_execbuffer(struct drm_device *dev, void *data, - struct drm_file *file_priv) +i915_gem_do_execbuffer(struct drm_device *dev, void *data, + struct drm_file *file_priv, + struct drm_i915_gem_execbuffer2 *args, + struct drm_i915_gem_exec_object2 *exec_list) { drm_i915_private_t *dev_priv = dev->dev_private; - struct drm_i915_gem_execbuffer *args = data; - struct drm_i915_gem_exec_object *exec_list = NULL; struct drm_gem_object **object_list = NULL; struct drm_gem_object *batch_obj; struct drm_i915_gem_object *obj_priv; struct drm_clip_rect *cliprects = NULL; - struct drm_i915_gem_relocation_entry *relocs; - int ret, ret2, i, pinned = 0; + struct drm_i915_gem_relocation_entry *relocs = NULL; + int ret = 0, ret2, i, pinned = 0; uint64_t exec_offset; uint32_t seqno, flush_domains, reloc_index; int pin_tries, flips; @@ -3614,31 +3737,21 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, DRM_ERROR("execbuf with %d buffers\n", args->buffer_count); return -EINVAL; } - /* Copy in the exec list from userland */ - exec_list = drm_malloc_ab(sizeof(*exec_list), args->buffer_count); object_list = drm_malloc_ab(sizeof(*object_list), args->buffer_count); - if (exec_list == NULL || object_list == NULL) { - DRM_ERROR("Failed to allocate exec or object list " - "for %d buffers\n", + if (object_list == NULL) { + DRM_ERROR("Failed to allocate object list for %d buffers\n", args->buffer_count); ret = -ENOMEM; goto pre_mutex_err; } - ret = copy_from_user(exec_list, - (struct drm_i915_relocation_entry __user *) - (uintptr_t) args->buffers_ptr, - sizeof(*exec_list) * args->buffer_count); - if (ret != 0) { - DRM_ERROR("copy %d exec entries failed %d\n", - args->buffer_count, ret); - goto pre_mutex_err; - } if (args->num_cliprects != 0) { cliprects = kcalloc(args->num_cliprects, sizeof(*cliprects), GFP_KERNEL); - if (cliprects == NULL) + if (cliprects == NULL) { + ret = -ENOMEM; goto pre_mutex_err; + } ret = copy_from_user(cliprects, (struct drm_clip_rect __user *) @@ -3680,6 +3793,8 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, if (object_list[i] == NULL) { DRM_ERROR("Invalid object handle %d at index %d\n", exec_list[i].handle, i); + /* prevent error path from reading uninitialized data */ + args->buffer_count = i + 1; ret = -EBADF; goto err; } @@ -3688,6 +3803,8 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, if (obj_priv->in_execbuffer) { DRM_ERROR("Object %p appears more than once in object list\n", object_list[i]); + /* prevent error path from reading uninitialized data */ + args->buffer_count = i + 1; ret = -EBADF; goto err; } @@ -3801,16 +3918,23 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, i915_gem_flush(dev, dev->invalidate_domains, dev->flush_domains); - if (dev->flush_domains) + if (dev->flush_domains & I915_GEM_GPU_DOMAINS) (void)i915_add_request(dev, file_priv, dev->flush_domains); } for (i = 0; i < args->buffer_count; i++) { struct drm_gem_object *obj = object_list[i]; + struct drm_i915_gem_object *obj_priv = obj->driver_private; uint32_t old_write_domain = obj->write_domain; obj->write_domain = obj->pending_write_domain; + if (obj->write_domain) + list_move_tail(&obj_priv->gpu_write_list, + &dev_priv->mm.gpu_write_list); + else + list_del_init(&obj_priv->gpu_write_list); + trace_i915_gem_object_change_domain(obj, obj->read_domains, old_write_domain); @@ -3884,8 +4008,101 @@ err: mutex_unlock(&dev->struct_mutex); +pre_mutex_err: + /* Copy the updated relocations out regardless of current error + * state. Failure to update the relocs would mean that the next + * time userland calls execbuf, it would do so with presumed offset + * state that didn't match the actual object state. + */ + ret2 = i915_gem_put_relocs_to_user(exec_list, args->buffer_count, + relocs); + if (ret2 != 0) { + DRM_ERROR("Failed to copy relocations back out: %d\n", ret2); + + if (ret == 0) + ret = ret2; + } + + drm_free_large(object_list); + kfree(cliprects); + + return ret; +} + +/* + * Legacy execbuffer just creates an exec2 list from the original exec object + * list array and passes it to the real function. + */ +int +i915_gem_execbuffer(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_i915_gem_execbuffer *args = data; + struct drm_i915_gem_execbuffer2 exec2; + struct drm_i915_gem_exec_object *exec_list = NULL; + struct drm_i915_gem_exec_object2 *exec2_list = NULL; + int ret, i; + +#if WATCH_EXEC + DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n", + (int) args->buffers_ptr, args->buffer_count, args->batch_len); +#endif + + if (args->buffer_count < 1) { + DRM_ERROR("execbuf with %d buffers\n", args->buffer_count); + return -EINVAL; + } + + /* Copy in the exec list from userland */ + exec_list = drm_malloc_ab(sizeof(*exec_list), args->buffer_count); + exec2_list = drm_malloc_ab(sizeof(*exec2_list), args->buffer_count); + if (exec_list == NULL || exec2_list == NULL) { + DRM_ERROR("Failed to allocate exec list for %d buffers\n", + args->buffer_count); + drm_free_large(exec_list); + drm_free_large(exec2_list); + return -ENOMEM; + } + ret = copy_from_user(exec_list, + (struct drm_i915_relocation_entry __user *) + (uintptr_t) args->buffers_ptr, + sizeof(*exec_list) * args->buffer_count); + if (ret != 0) { + DRM_ERROR("copy %d exec entries failed %d\n", + args->buffer_count, ret); + drm_free_large(exec_list); + drm_free_large(exec2_list); + return -EFAULT; + } + + for (i = 0; i < args->buffer_count; i++) { + exec2_list[i].handle = exec_list[i].handle; + exec2_list[i].relocation_count = exec_list[i].relocation_count; + exec2_list[i].relocs_ptr = exec_list[i].relocs_ptr; + exec2_list[i].alignment = exec_list[i].alignment; + exec2_list[i].offset = exec_list[i].offset; + if (!IS_I965G(dev)) + exec2_list[i].flags = EXEC_OBJECT_NEEDS_FENCE; + else + exec2_list[i].flags = 0; + } + + exec2.buffers_ptr = args->buffers_ptr; + exec2.buffer_count = args->buffer_count; + exec2.batch_start_offset = args->batch_start_offset; + exec2.batch_len = args->batch_len; + exec2.DR1 = args->DR1; + exec2.DR4 = args->DR4; + exec2.num_cliprects = args->num_cliprects; + exec2.cliprects_ptr = args->cliprects_ptr; + exec2.flags = 0; + + ret = i915_gem_do_execbuffer(dev, data, file_priv, &exec2, exec2_list); if (!ret) { /* Copy the new buffer offsets back to the user's exec list. */ + for (i = 0; i < args->buffer_count; i++) + exec_list[i].offset = exec2_list[i].offset; + /* ... and back out to userspace */ ret = copy_to_user((struct drm_i915_relocation_entry __user *) (uintptr_t) args->buffers_ptr, exec_list, @@ -3898,25 +4115,62 @@ err: } } - /* Copy the updated relocations out regardless of current error - * state. Failure to update the relocs would mean that the next - * time userland calls execbuf, it would do so with presumed offset - * state that didn't match the actual object state. - */ - ret2 = i915_gem_put_relocs_to_user(exec_list, args->buffer_count, - relocs); - if (ret2 != 0) { - DRM_ERROR("Failed to copy relocations back out: %d\n", ret2); + drm_free_large(exec_list); + drm_free_large(exec2_list); + return ret; +} - if (ret == 0) - ret = ret2; +int +i915_gem_execbuffer2(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_i915_gem_execbuffer2 *args = data; + struct drm_i915_gem_exec_object2 *exec2_list = NULL; + int ret; + +#if WATCH_EXEC + DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n", + (int) args->buffers_ptr, args->buffer_count, args->batch_len); +#endif + + if (args->buffer_count < 1) { + DRM_ERROR("execbuf2 with %d buffers\n", args->buffer_count); + return -EINVAL; } -pre_mutex_err: - drm_free_large(object_list); - drm_free_large(exec_list); - kfree(cliprects); + exec2_list = drm_malloc_ab(sizeof(*exec2_list), args->buffer_count); + if (exec2_list == NULL) { + DRM_ERROR("Failed to allocate exec list for %d buffers\n", + args->buffer_count); + return -ENOMEM; + } + ret = copy_from_user(exec2_list, + (struct drm_i915_relocation_entry __user *) + (uintptr_t) args->buffers_ptr, + sizeof(*exec2_list) * args->buffer_count); + if (ret != 0) { + DRM_ERROR("copy %d exec entries failed %d\n", + args->buffer_count, ret); + drm_free_large(exec2_list); + return -EFAULT; + } + + ret = i915_gem_do_execbuffer(dev, data, file_priv, args, exec2_list); + if (!ret) { + /* Copy the new buffer offsets back to the user's exec list. */ + ret = copy_to_user((struct drm_i915_relocation_entry __user *) + (uintptr_t) args->buffers_ptr, + exec2_list, + sizeof(*exec2_list) * args->buffer_count); + if (ret) { + ret = -EFAULT; + DRM_ERROR("failed to copy %d exec entries " + "back to user (%d)\n", + args->buffer_count, ret); + } + } + drm_free_large(exec2_list); return ret; } @@ -3933,19 +4187,7 @@ i915_gem_object_pin(struct drm_gem_object *obj, uint32_t alignment) if (ret) return ret; } - /* - * Pre-965 chips need a fence register set up in order to - * properly handle tiled surfaces. - */ - if (!IS_I965G(dev) && obj_priv->tiling_mode != I915_TILING_NONE) { - ret = i915_gem_object_get_fence_reg(obj); - if (ret != 0) { - if (ret != -EBUSY && ret != -ERESTARTSYS) - DRM_ERROR("Failure to install fence: %d\n", - ret); - return ret; - } - } + obj_priv->pin_count++; /* If the object is not active and not pending a flush, @@ -4203,6 +4445,7 @@ int i915_gem_init_object(struct drm_gem_object *obj) obj_priv->obj = obj; obj_priv->fence_reg = I915_FENCE_REG_NONE; INIT_LIST_HEAD(&obj_priv->list); + INIT_LIST_HEAD(&obj_priv->gpu_write_list); INIT_LIST_HEAD(&obj_priv->fence_list); obj_priv->madv = I915_MADV_WILLNEED; @@ -4262,8 +4505,7 @@ int i915_gem_idle(struct drm_device *dev) { drm_i915_private_t *dev_priv = dev->dev_private; - uint32_t seqno, cur_seqno, last_seqno; - int stuck, ret; + int ret; mutex_lock(&dev->struct_mutex); @@ -4272,115 +4514,36 @@ i915_gem_idle(struct drm_device *dev) return 0; } - /* Hack! Don't let anybody do execbuf while we don't control the chip. - * We need to replace this with a semaphore, or something. - */ - dev_priv->mm.suspended = 1; - del_timer(&dev_priv->hangcheck_timer); - - /* Cancel the retire work handler, wait for it to finish if running - */ - mutex_unlock(&dev->struct_mutex); - cancel_delayed_work_sync(&dev_priv->mm.retire_work); - mutex_lock(&dev->struct_mutex); - - i915_kernel_lost_context(dev); - - /* Flush the GPU along with all non-CPU write domains - */ - i915_gem_flush(dev, I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS); - seqno = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS); - - if (seqno == 0) { + ret = i915_gpu_idle(dev); + if (ret) { mutex_unlock(&dev->struct_mutex); - return -ENOMEM; + return ret; } - dev_priv->mm.waiting_gem_seqno = seqno; - last_seqno = 0; - stuck = 0; - for (;;) { - cur_seqno = i915_get_gem_seqno(dev); - if (i915_seqno_passed(cur_seqno, seqno)) - break; - if (last_seqno == cur_seqno) { - if (stuck++ > 100) { - DRM_ERROR("hardware wedged\n"); - atomic_set(&dev_priv->mm.wedged, 1); - DRM_WAKEUP(&dev_priv->irq_queue); - break; - } + /* Under UMS, be paranoid and evict. */ + if (!drm_core_check_feature(dev, DRIVER_MODESET)) { + ret = i915_gem_evict_from_inactive_list(dev); + if (ret) { + mutex_unlock(&dev->struct_mutex); + return ret; } - msleep(10); - last_seqno = cur_seqno; } - dev_priv->mm.waiting_gem_seqno = 0; - i915_gem_retire_requests(dev); - - spin_lock(&dev_priv->mm.active_list_lock); - if (!atomic_read(&dev_priv->mm.wedged)) { - /* Active and flushing should now be empty as we've - * waited for a sequence higher than any pending execbuffer - */ - WARN_ON(!list_empty(&dev_priv->mm.active_list)); - WARN_ON(!list_empty(&dev_priv->mm.flushing_list)); - /* Request should now be empty as we've also waited - * for the last request in the list - */ - WARN_ON(!list_empty(&dev_priv->mm.request_list)); - } - - /* Empty the active and flushing lists to inactive. If there's - * anything left at this point, it means that we're wedged and - * nothing good's going to happen by leaving them there. So strip - * the GPU domains and just stuff them onto inactive. + /* Hack! Don't let anybody do execbuf while we don't control the chip. + * We need to replace this with a semaphore, or something. + * And not confound mm.suspended! */ - while (!list_empty(&dev_priv->mm.active_list)) { - struct drm_gem_object *obj; - uint32_t old_write_domain; - - obj = list_first_entry(&dev_priv->mm.active_list, - struct drm_i915_gem_object, - list)->obj; - old_write_domain = obj->write_domain; - obj->write_domain &= ~I915_GEM_GPU_DOMAINS; - i915_gem_object_move_to_inactive(obj); - - trace_i915_gem_object_change_domain(obj, - obj->read_domains, - old_write_domain); - } - spin_unlock(&dev_priv->mm.active_list_lock); - - while (!list_empty(&dev_priv->mm.flushing_list)) { - struct drm_gem_object *obj; - uint32_t old_write_domain; - - obj = list_first_entry(&dev_priv->mm.flushing_list, - struct drm_i915_gem_object, - list)->obj; - old_write_domain = obj->write_domain; - obj->write_domain &= ~I915_GEM_GPU_DOMAINS; - i915_gem_object_move_to_inactive(obj); - - trace_i915_gem_object_change_domain(obj, - obj->read_domains, - old_write_domain); - } - - - /* Move all inactive buffers out of the GTT. */ - ret = i915_gem_evict_from_inactive_list(dev); - WARN_ON(!list_empty(&dev_priv->mm.inactive_list)); - if (ret) { - mutex_unlock(&dev->struct_mutex); - return ret; - } + dev_priv->mm.suspended = 1; + del_timer(&dev_priv->hangcheck_timer); + i915_kernel_lost_context(dev); i915_gem_cleanup_ringbuffer(dev); + mutex_unlock(&dev->struct_mutex); + /* Cancel the retire work handler, which should be idle now. */ + cancel_delayed_work_sync(&dev_priv->mm.retire_work); + return 0; } @@ -4424,8 +4587,13 @@ i915_gem_init_hws(struct drm_device *dev) } dev_priv->hws_obj = obj; memset(dev_priv->hw_status_page, 0, PAGE_SIZE); - I915_WRITE(HWS_PGA, dev_priv->status_gfx_addr); - I915_READ(HWS_PGA); /* posting read */ + if (IS_GEN6(dev)) { + I915_WRITE(HWS_PGA_GEN6, dev_priv->status_gfx_addr); + I915_READ(HWS_PGA_GEN6); /* posting read */ + } else { + I915_WRITE(HWS_PGA, dev_priv->status_gfx_addr); + I915_READ(HWS_PGA); /* posting read */ + } DRM_DEBUG_DRIVER("hws offset: 0x%08x\n", dev_priv->status_gfx_addr); return 0; @@ -4654,6 +4822,7 @@ i915_gem_load(struct drm_device *dev) spin_lock_init(&dev_priv->mm.active_list_lock); INIT_LIST_HEAD(&dev_priv->mm.active_list); INIT_LIST_HEAD(&dev_priv->mm.flushing_list); + INIT_LIST_HEAD(&dev_priv->mm.gpu_write_list); INIT_LIST_HEAD(&dev_priv->mm.inactive_list); INIT_LIST_HEAD(&dev_priv->mm.request_list); INIT_LIST_HEAD(&dev_priv->mm.fence_list); @@ -4666,7 +4835,8 @@ i915_gem_load(struct drm_device *dev) spin_unlock(&shrink_list_lock); /* Old X drivers will take 0-2 for front, back, depth buffers */ - dev_priv->fence_reg_start = 3; + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + dev_priv->fence_reg_start = 3; if (IS_I965G(dev) || IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) dev_priv->num_fence_regs = 16; @@ -4766,7 +4936,7 @@ void i915_gem_detach_phys_object(struct drm_device *dev, if (!obj_priv->phys_obj) return; - ret = i915_gem_object_get_pages(obj); + ret = i915_gem_object_get_pages(obj, 0); if (ret) goto out; @@ -4824,7 +4994,7 @@ i915_gem_attach_phys_object(struct drm_device *dev, obj_priv->phys_obj = dev_priv->mm.phys_objs[id - 1]; obj_priv->phys_obj->cur_obj = obj; - ret = i915_gem_object_get_pages(obj); + ret = i915_gem_object_get_pages(obj, 0); if (ret) { DRM_ERROR("failed to get page list\n"); goto out; diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c index 30d6af6c09bb..b5c55d88ff76 100644 --- a/drivers/gpu/drm/i915/i915_gem_tiling.c +++ b/drivers/gpu/drm/i915/i915_gem_tiling.c @@ -25,8 +25,6 @@ * */ -#include <linux/acpi.h> -#include <linux/pnp.h> #include "linux/string.h" #include "linux/bitops.h" #include "drmP.h" @@ -83,120 +81,6 @@ * to match what the GPU expects. */ -#define MCHBAR_I915 0x44 -#define MCHBAR_I965 0x48 -#define MCHBAR_SIZE (4*4096) - -#define DEVEN_REG 0x54 -#define DEVEN_MCHBAR_EN (1 << 28) - -/* Allocate space for the MCH regs if needed, return nonzero on error */ -static int -intel_alloc_mchbar_resource(struct drm_device *dev) -{ - drm_i915_private_t *dev_priv = dev->dev_private; - int reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915; - u32 temp_lo, temp_hi = 0; - u64 mchbar_addr; - int ret = 0; - - if (IS_I965G(dev)) - pci_read_config_dword(dev_priv->bridge_dev, reg + 4, &temp_hi); - pci_read_config_dword(dev_priv->bridge_dev, reg, &temp_lo); - mchbar_addr = ((u64)temp_hi << 32) | temp_lo; - - /* If ACPI doesn't have it, assume we need to allocate it ourselves */ -#ifdef CONFIG_PNP - if (mchbar_addr && - pnp_range_reserved(mchbar_addr, mchbar_addr + MCHBAR_SIZE)) { - ret = 0; - goto out; - } -#endif - - /* Get some space for it */ - ret = pci_bus_alloc_resource(dev_priv->bridge_dev->bus, &dev_priv->mch_res, - MCHBAR_SIZE, MCHBAR_SIZE, - PCIBIOS_MIN_MEM, - 0, pcibios_align_resource, - dev_priv->bridge_dev); - if (ret) { - DRM_DEBUG_DRIVER("failed bus alloc: %d\n", ret); - dev_priv->mch_res.start = 0; - goto out; - } - - if (IS_I965G(dev)) - pci_write_config_dword(dev_priv->bridge_dev, reg + 4, - upper_32_bits(dev_priv->mch_res.start)); - - pci_write_config_dword(dev_priv->bridge_dev, reg, - lower_32_bits(dev_priv->mch_res.start)); -out: - return ret; -} - -/* Setup MCHBAR if possible, return true if we should disable it again */ -static bool -intel_setup_mchbar(struct drm_device *dev) -{ - drm_i915_private_t *dev_priv = dev->dev_private; - int mchbar_reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915; - u32 temp; - bool need_disable = false, enabled; - - if (IS_I915G(dev) || IS_I915GM(dev)) { - pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp); - enabled = !!(temp & DEVEN_MCHBAR_EN); - } else { - pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); - enabled = temp & 1; - } - - /* If it's already enabled, don't have to do anything */ - if (enabled) - goto out; - - if (intel_alloc_mchbar_resource(dev)) - goto out; - - need_disable = true; - - /* Space is allocated or reserved, so enable it. */ - if (IS_I915G(dev) || IS_I915GM(dev)) { - pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG, - temp | DEVEN_MCHBAR_EN); - } else { - pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); - pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp | 1); - } -out: - return need_disable; -} - -static void -intel_teardown_mchbar(struct drm_device *dev, bool disable) -{ - drm_i915_private_t *dev_priv = dev->dev_private; - int mchbar_reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915; - u32 temp; - - if (disable) { - if (IS_I915G(dev) || IS_I915GM(dev)) { - pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp); - temp &= ~DEVEN_MCHBAR_EN; - pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG, temp); - } else { - pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); - temp &= ~1; - pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp); - } - } - - if (dev_priv->mch_res.start) - release_resource(&dev_priv->mch_res); -} - /** * Detects bit 6 swizzling of address lookup between IGD access and CPU * access through main memory. @@ -207,9 +91,8 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev) drm_i915_private_t *dev_priv = dev->dev_private; uint32_t swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN; uint32_t swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN; - bool need_disable; - if (IS_IRONLAKE(dev)) { + if (IS_IRONLAKE(dev) || IS_GEN6(dev)) { /* On Ironlake whatever DRAM config, GPU always do * same swizzling setup. */ @@ -224,9 +107,6 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev) } else if (IS_MOBILE(dev)) { uint32_t dcc; - /* Try to make sure MCHBAR is enabled before poking at it */ - need_disable = intel_setup_mchbar(dev); - /* On mobile 9xx chipsets, channel interleave by the CPU is * determined by DCC. For single-channel, neither the CPU * nor the GPU do swizzling. For dual channel interleaved, @@ -266,8 +146,6 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev) swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN; swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN; } - - intel_teardown_mchbar(dev, need_disable); } else { /* The 965, G33, and newer, have a very flexible memory * configuration. It will enable dual-channel mode @@ -302,37 +180,8 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev) dev_priv->mm.bit_6_swizzle_y = swizzle_y; } - -/** - * Returns the size of the fence for a tiled object of the given size. - */ -static int -i915_get_fence_size(struct drm_device *dev, int size) -{ - int i; - int start; - - if (IS_I965G(dev)) { - /* The 965 can have fences at any page boundary. */ - return ALIGN(size, 4096); - } else { - /* Align the size to a power of two greater than the smallest - * fence size. - */ - if (IS_I9XX(dev)) - start = 1024 * 1024; - else - start = 512 * 1024; - - for (i = start; i < size; i <<= 1) - ; - - return i; - } -} - /* Check pitch constriants for all chips & tiling formats */ -static bool +bool i915_tiling_ok(struct drm_device *dev, int stride, int size, int tiling_mode) { int tile_width; @@ -384,16 +233,10 @@ i915_tiling_ok(struct drm_device *dev, int stride, int size, int tiling_mode) if (stride & (stride - 1)) return false; - /* We don't 0handle the aperture area covered by the fence being bigger - * than the object size. - */ - if (i915_get_fence_size(dev, size) != size) - return false; - return true; } -static bool +bool i915_gem_object_fence_offset_ok(struct drm_gem_object *obj, int tiling_mode) { struct drm_device *dev = obj->dev; @@ -440,9 +283,7 @@ i915_gem_set_tiling(struct drm_device *dev, void *data, obj_priv = obj->driver_private; if (!i915_tiling_ok(dev, args->stride, obj->size, args->tiling_mode)) { - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(obj); return -EINVAL; } @@ -495,12 +336,6 @@ i915_gem_set_tiling(struct drm_device *dev, void *data, goto err; } - /* If we've changed tiling, GTT-mappings of the object - * need to re-fault to ensure that the correct fence register - * setup is in place. - */ - i915_gem_release_mmap(obj); - obj_priv->tiling_mode = args->tiling_mode; obj_priv->stride = args->stride; } diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c index 85f4c5de97e2..5388354da0d1 100644 --- a/drivers/gpu/drm/i915/i915_irq.c +++ b/drivers/gpu/drm/i915/i915_irq.c @@ -166,7 +166,7 @@ void intel_enable_asle (struct drm_device *dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) ironlake_enable_display_irq(dev_priv, DE_GSE); else i915_enable_pipestat(dev_priv, 1, @@ -269,12 +269,62 @@ static void i915_hotplug_work_func(struct work_struct *work) drm_sysfs_hotplug_event(dev); } +static void i915_handle_rps_change(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + u32 busy_up, busy_down, max_avg, min_avg; + u16 rgvswctl; + u8 new_delay = dev_priv->cur_delay; + + I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS) & ~MEMINT_EVAL_CHG); + busy_up = I915_READ(RCPREVBSYTUPAVG); + busy_down = I915_READ(RCPREVBSYTDNAVG); + max_avg = I915_READ(RCBMAXAVG); + min_avg = I915_READ(RCBMINAVG); + + /* Handle RCS change request from hw */ + if (busy_up > max_avg) { + if (dev_priv->cur_delay != dev_priv->max_delay) + new_delay = dev_priv->cur_delay - 1; + if (new_delay < dev_priv->max_delay) + new_delay = dev_priv->max_delay; + } else if (busy_down < min_avg) { + if (dev_priv->cur_delay != dev_priv->min_delay) + new_delay = dev_priv->cur_delay + 1; + if (new_delay > dev_priv->min_delay) + new_delay = dev_priv->min_delay; + } + + DRM_DEBUG("rps change requested: %d -> %d\n", + dev_priv->cur_delay, new_delay); + + rgvswctl = I915_READ(MEMSWCTL); + if (rgvswctl & MEMCTL_CMD_STS) { + DRM_ERROR("gpu busy, RCS change rejected\n"); + return; /* still busy with another command */ + } + + /* Program the new state */ + rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) | + (new_delay << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM; + I915_WRITE(MEMSWCTL, rgvswctl); + POSTING_READ(MEMSWCTL); + + rgvswctl |= MEMCTL_CMD_STS; + I915_WRITE(MEMSWCTL, rgvswctl); + + dev_priv->cur_delay = new_delay; + + DRM_DEBUG("rps changed\n"); + + return; +} + irqreturn_t ironlake_irq_handler(struct drm_device *dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; int ret = IRQ_NONE; u32 de_iir, gt_iir, de_ier, pch_iir; - u32 new_de_iir, new_gt_iir, new_pch_iir; struct drm_i915_master_private *master_priv; /* disable master interrupt before clearing iir */ @@ -286,49 +336,63 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev) gt_iir = I915_READ(GTIIR); pch_iir = I915_READ(SDEIIR); - for (;;) { - if (de_iir == 0 && gt_iir == 0 && pch_iir == 0) - break; + if (de_iir == 0 && gt_iir == 0 && pch_iir == 0) + goto done; - ret = IRQ_HANDLED; + ret = IRQ_HANDLED; - /* should clear PCH hotplug event before clear CPU irq */ - I915_WRITE(SDEIIR, pch_iir); - new_pch_iir = I915_READ(SDEIIR); + if (dev->primary->master) { + master_priv = dev->primary->master->driver_priv; + if (master_priv->sarea_priv) + master_priv->sarea_priv->last_dispatch = + READ_BREADCRUMB(dev_priv); + } - I915_WRITE(DEIIR, de_iir); - new_de_iir = I915_READ(DEIIR); - I915_WRITE(GTIIR, gt_iir); - new_gt_iir = I915_READ(GTIIR); + if (gt_iir & GT_USER_INTERRUPT) { + u32 seqno = i915_get_gem_seqno(dev); + dev_priv->mm.irq_gem_seqno = seqno; + trace_i915_gem_request_complete(dev, seqno); + DRM_WAKEUP(&dev_priv->irq_queue); + dev_priv->hangcheck_count = 0; + mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD); + } - if (dev->primary->master) { - master_priv = dev->primary->master->driver_priv; - if (master_priv->sarea_priv) - master_priv->sarea_priv->last_dispatch = - READ_BREADCRUMB(dev_priv); - } + if (de_iir & DE_GSE) + ironlake_opregion_gse_intr(dev); - if (gt_iir & GT_USER_INTERRUPT) { - u32 seqno = i915_get_gem_seqno(dev); - dev_priv->mm.irq_gem_seqno = seqno; - trace_i915_gem_request_complete(dev, seqno); - DRM_WAKEUP(&dev_priv->irq_queue); - } + if (de_iir & DE_PLANEA_FLIP_DONE) { + intel_prepare_page_flip(dev, 0); + intel_finish_page_flip(dev, 0); + } - if (de_iir & DE_GSE) - ironlake_opregion_gse_intr(dev); + if (de_iir & DE_PLANEB_FLIP_DONE) { + intel_prepare_page_flip(dev, 1); + intel_finish_page_flip(dev, 1); + } - /* check event from PCH */ - if ((de_iir & DE_PCH_EVENT) && - (pch_iir & SDE_HOTPLUG_MASK)) { - queue_work(dev_priv->wq, &dev_priv->hotplug_work); - } + if (de_iir & DE_PIPEA_VBLANK) + drm_handle_vblank(dev, 0); - de_iir = new_de_iir; - gt_iir = new_gt_iir; - pch_iir = new_pch_iir; + if (de_iir & DE_PIPEB_VBLANK) + drm_handle_vblank(dev, 1); + + /* check event from PCH */ + if ((de_iir & DE_PCH_EVENT) && + (pch_iir & SDE_HOTPLUG_MASK)) { + queue_work(dev_priv->wq, &dev_priv->hotplug_work); + } + + if (de_iir & DE_PCU_EVENT) { + I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS)); + i915_handle_rps_change(dev); } + /* should clear PCH hotplug event before clear CPU irq */ + I915_WRITE(SDEIIR, pch_iir); + I915_WRITE(GTIIR, gt_iir); + I915_WRITE(DEIIR, de_iir); + +done: I915_WRITE(DEIER, de_ier); (void)I915_READ(DEIER); @@ -368,6 +432,121 @@ static void i915_error_work_func(struct work_struct *work) } } +static struct drm_i915_error_object * +i915_error_object_create(struct drm_device *dev, + struct drm_gem_object *src) +{ + struct drm_i915_error_object *dst; + struct drm_i915_gem_object *src_priv; + int page, page_count; + + if (src == NULL) + return NULL; + + src_priv = src->driver_private; + if (src_priv->pages == NULL) + return NULL; + + page_count = src->size / PAGE_SIZE; + + dst = kmalloc(sizeof(*dst) + page_count * sizeof (u32 *), GFP_ATOMIC); + if (dst == NULL) + return NULL; + + for (page = 0; page < page_count; page++) { + void *s, *d = kmalloc(PAGE_SIZE, GFP_ATOMIC); + if (d == NULL) + goto unwind; + s = kmap_atomic(src_priv->pages[page], KM_USER0); + memcpy(d, s, PAGE_SIZE); + kunmap_atomic(s, KM_USER0); + dst->pages[page] = d; + } + dst->page_count = page_count; + dst->gtt_offset = src_priv->gtt_offset; + + return dst; + +unwind: + while (page--) + kfree(dst->pages[page]); + kfree(dst); + return NULL; +} + +static void +i915_error_object_free(struct drm_i915_error_object *obj) +{ + int page; + + if (obj == NULL) + return; + + for (page = 0; page < obj->page_count; page++) + kfree(obj->pages[page]); + + kfree(obj); +} + +static void +i915_error_state_free(struct drm_device *dev, + struct drm_i915_error_state *error) +{ + i915_error_object_free(error->batchbuffer[0]); + i915_error_object_free(error->batchbuffer[1]); + i915_error_object_free(error->ringbuffer); + kfree(error->active_bo); + kfree(error); +} + +static u32 +i915_get_bbaddr(struct drm_device *dev, u32 *ring) +{ + u32 cmd; + + if (IS_I830(dev) || IS_845G(dev)) + cmd = MI_BATCH_BUFFER; + else if (IS_I965G(dev)) + cmd = (MI_BATCH_BUFFER_START | (2 << 6) | + MI_BATCH_NON_SECURE_I965); + else + cmd = (MI_BATCH_BUFFER_START | (2 << 6)); + + return ring[0] == cmd ? ring[1] : 0; +} + +static u32 +i915_ringbuffer_last_batch(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 head, bbaddr; + u32 *ring; + + /* Locate the current position in the ringbuffer and walk back + * to find the most recently dispatched batch buffer. + */ + bbaddr = 0; + head = I915_READ(PRB0_HEAD) & HEAD_ADDR; + ring = (u32 *)(dev_priv->ring.virtual_start + head); + + while (--ring >= (u32 *)dev_priv->ring.virtual_start) { + bbaddr = i915_get_bbaddr(dev, ring); + if (bbaddr) + break; + } + + if (bbaddr == 0) { + ring = (u32 *)(dev_priv->ring.virtual_start + dev_priv->ring.Size); + while (--ring >= (u32 *)dev_priv->ring.virtual_start) { + bbaddr = i915_get_bbaddr(dev, ring); + if (bbaddr) + break; + } + } + + return bbaddr; +} + /** * i915_capture_error_state - capture an error record for later analysis * @dev: drm device @@ -380,19 +559,26 @@ static void i915_error_work_func(struct work_struct *work) static void i915_capture_error_state(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj_priv; struct drm_i915_error_state *error; + struct drm_gem_object *batchbuffer[2]; unsigned long flags; + u32 bbaddr; + int count; spin_lock_irqsave(&dev_priv->error_lock, flags); - if (dev_priv->first_error) - goto out; + error = dev_priv->first_error; + spin_unlock_irqrestore(&dev_priv->error_lock, flags); + if (error) + return; error = kmalloc(sizeof(*error), GFP_ATOMIC); if (!error) { - DRM_DEBUG_DRIVER("out ot memory, not capturing error state\n"); - goto out; + DRM_DEBUG_DRIVER("out of memory, not capturing error state\n"); + return; } + error->seqno = i915_get_gem_seqno(dev); error->eir = I915_READ(EIR); error->pgtbl_er = I915_READ(PGTBL_ER); error->pipeastat = I915_READ(PIPEASTAT); @@ -403,6 +589,7 @@ static void i915_capture_error_state(struct drm_device *dev) error->ipehr = I915_READ(IPEHR); error->instdone = I915_READ(INSTDONE); error->acthd = I915_READ(ACTHD); + error->bbaddr = 0; } else { error->ipeir = I915_READ(IPEIR_I965); error->ipehr = I915_READ(IPEHR_I965); @@ -410,14 +597,101 @@ static void i915_capture_error_state(struct drm_device *dev) error->instps = I915_READ(INSTPS); error->instdone1 = I915_READ(INSTDONE1); error->acthd = I915_READ(ACTHD_I965); + error->bbaddr = I915_READ64(BB_ADDR); } - do_gettimeofday(&error->time); + bbaddr = i915_ringbuffer_last_batch(dev); + + /* Grab the current batchbuffer, most likely to have crashed. */ + batchbuffer[0] = NULL; + batchbuffer[1] = NULL; + count = 0; + list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) { + struct drm_gem_object *obj = obj_priv->obj; + + if (batchbuffer[0] == NULL && + bbaddr >= obj_priv->gtt_offset && + bbaddr < obj_priv->gtt_offset + obj->size) + batchbuffer[0] = obj; + + if (batchbuffer[1] == NULL && + error->acthd >= obj_priv->gtt_offset && + error->acthd < obj_priv->gtt_offset + obj->size && + batchbuffer[0] != obj) + batchbuffer[1] = obj; + + count++; + } + + /* We need to copy these to an anonymous buffer as the simplest + * method to avoid being overwritten by userpace. + */ + error->batchbuffer[0] = i915_error_object_create(dev, batchbuffer[0]); + error->batchbuffer[1] = i915_error_object_create(dev, batchbuffer[1]); + + /* Record the ringbuffer */ + error->ringbuffer = i915_error_object_create(dev, dev_priv->ring.ring_obj); + + /* Record buffers on the active list. */ + error->active_bo = NULL; + error->active_bo_count = 0; + + if (count) + error->active_bo = kmalloc(sizeof(*error->active_bo)*count, + GFP_ATOMIC); + + if (error->active_bo) { + int i = 0; + list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) { + struct drm_gem_object *obj = obj_priv->obj; + + error->active_bo[i].size = obj->size; + error->active_bo[i].name = obj->name; + error->active_bo[i].seqno = obj_priv->last_rendering_seqno; + error->active_bo[i].gtt_offset = obj_priv->gtt_offset; + error->active_bo[i].read_domains = obj->read_domains; + error->active_bo[i].write_domain = obj->write_domain; + error->active_bo[i].fence_reg = obj_priv->fence_reg; + error->active_bo[i].pinned = 0; + if (obj_priv->pin_count > 0) + error->active_bo[i].pinned = 1; + if (obj_priv->user_pin_count > 0) + error->active_bo[i].pinned = -1; + error->active_bo[i].tiling = obj_priv->tiling_mode; + error->active_bo[i].dirty = obj_priv->dirty; + error->active_bo[i].purgeable = obj_priv->madv != I915_MADV_WILLNEED; + + if (++i == count) + break; + } + error->active_bo_count = i; + } - dev_priv->first_error = error; + do_gettimeofday(&error->time); -out: + spin_lock_irqsave(&dev_priv->error_lock, flags); + if (dev_priv->first_error == NULL) { + dev_priv->first_error = error; + error = NULL; + } spin_unlock_irqrestore(&dev_priv->error_lock, flags); + + if (error) + i915_error_state_free(dev, error); +} + +void i915_destroy_error_state(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_error_state *error; + + spin_lock(&dev_priv->error_lock); + error = dev_priv->first_error; + dev_priv->first_error = NULL; + spin_unlock(&dev_priv->error_lock); + + if (error) + i915_error_state_free(dev, error); } /** @@ -568,7 +842,7 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS) atomic_inc(&dev_priv->irq_received); - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) return ironlake_irq_handler(dev); iir = I915_READ(IIR); @@ -729,7 +1003,7 @@ void i915_user_irq_get(struct drm_device *dev) spin_lock_irqsave(&dev_priv->user_irq_lock, irqflags); if (dev->irq_enabled && (++dev_priv->user_irq_refcount == 1)) { - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) ironlake_enable_graphics_irq(dev_priv, GT_USER_INTERRUPT); else i915_enable_irq(dev_priv, I915_USER_INTERRUPT); @@ -745,7 +1019,7 @@ void i915_user_irq_put(struct drm_device *dev) spin_lock_irqsave(&dev_priv->user_irq_lock, irqflags); BUG_ON(dev->irq_enabled && dev_priv->user_irq_refcount <= 0); if (dev->irq_enabled && (--dev_priv->user_irq_refcount == 0)) { - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) ironlake_disable_graphics_irq(dev_priv, GT_USER_INTERRUPT); else i915_disable_irq(dev_priv, I915_USER_INTERRUPT); @@ -852,11 +1126,11 @@ int i915_enable_vblank(struct drm_device *dev, int pipe) if (!(pipeconf & PIPEACONF_ENABLE)) return -EINVAL; - if (IS_IRONLAKE(dev)) - return 0; - spin_lock_irqsave(&dev_priv->user_irq_lock, irqflags); - if (IS_I965G(dev)) + if (HAS_PCH_SPLIT(dev)) + ironlake_enable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK: DE_PIPEB_VBLANK); + else if (IS_I965G(dev)) i915_enable_pipestat(dev_priv, pipe, PIPE_START_VBLANK_INTERRUPT_ENABLE); else @@ -874,13 +1148,14 @@ void i915_disable_vblank(struct drm_device *dev, int pipe) drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; unsigned long irqflags; - if (IS_IRONLAKE(dev)) - return; - spin_lock_irqsave(&dev_priv->user_irq_lock, irqflags); - i915_disable_pipestat(dev_priv, pipe, - PIPE_VBLANK_INTERRUPT_ENABLE | - PIPE_START_VBLANK_INTERRUPT_ENABLE); + if (HAS_PCH_SPLIT(dev)) + ironlake_disable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK: DE_PIPEB_VBLANK); + else + i915_disable_pipestat(dev_priv, pipe, + PIPE_VBLANK_INTERRUPT_ENABLE | + PIPE_START_VBLANK_INTERRUPT_ENABLE); spin_unlock_irqrestore(&dev_priv->user_irq_lock, irqflags); } @@ -888,7 +1163,7 @@ void i915_enable_interrupt (struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - if (!IS_IRONLAKE(dev)) + if (!HAS_PCH_SPLIT(dev)) opregion_enable_asle(dev); dev_priv->irq_enabled = 1; } @@ -964,7 +1239,11 @@ void i915_hangcheck_elapsed(unsigned long data) struct drm_device *dev = (struct drm_device *)data; drm_i915_private_t *dev_priv = dev->dev_private; uint32_t acthd; - + + /* No reset support on this chip yet. */ + if (IS_GEN6(dev)) + return; + if (!IS_I965G(dev)) acthd = I915_READ(ACTHD); else @@ -1023,13 +1302,14 @@ static int ironlake_irq_postinstall(struct drm_device *dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; /* enable kind of interrupts always enabled */ - u32 display_mask = DE_MASTER_IRQ_CONTROL | DE_GSE | DE_PCH_EVENT; + u32 display_mask = DE_MASTER_IRQ_CONTROL | DE_GSE | DE_PCH_EVENT | + DE_PLANEA_FLIP_DONE | DE_PLANEB_FLIP_DONE; u32 render_mask = GT_USER_INTERRUPT; u32 hotplug_mask = SDE_CRT_HOTPLUG | SDE_PORTB_HOTPLUG | SDE_PORTC_HOTPLUG | SDE_PORTD_HOTPLUG; dev_priv->irq_mask_reg = ~display_mask; - dev_priv->de_irq_enable_reg = display_mask; + dev_priv->de_irq_enable_reg = display_mask | DE_PIPEA_VBLANK | DE_PIPEB_VBLANK; /* should always can generate irq */ I915_WRITE(DEIIR, I915_READ(DEIIR)); @@ -1054,6 +1334,13 @@ static int ironlake_irq_postinstall(struct drm_device *dev) I915_WRITE(SDEIER, dev_priv->pch_irq_enable_reg); (void) I915_READ(SDEIER); + if (IS_IRONLAKE_M(dev)) { + /* Clear & enable PCU event interrupts */ + I915_WRITE(DEIIR, DE_PCU_EVENT); + I915_WRITE(DEIER, I915_READ(DEIER) | DE_PCU_EVENT); + ironlake_enable_display_irq(dev_priv, DE_PCU_EVENT); + } + return 0; } @@ -1066,7 +1353,7 @@ void i915_driver_irq_preinstall(struct drm_device * dev) INIT_WORK(&dev_priv->hotplug_work, i915_hotplug_work_func); INIT_WORK(&dev_priv->error_work, i915_error_work_func); - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { ironlake_irq_preinstall(dev); return; } @@ -1084,6 +1371,10 @@ void i915_driver_irq_preinstall(struct drm_device * dev) (void) I915_READ(IER); } +/* + * Must be called after intel_modeset_init or hotplug interrupts won't be + * enabled correctly. + */ int i915_driver_irq_postinstall(struct drm_device *dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; @@ -1094,7 +1385,7 @@ int i915_driver_irq_postinstall(struct drm_device *dev) dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) return ironlake_irq_postinstall(dev); /* Unmask the interrupts that we always want on. */ @@ -1106,19 +1397,23 @@ int i915_driver_irq_postinstall(struct drm_device *dev) if (I915_HAS_HOTPLUG(dev)) { u32 hotplug_en = I915_READ(PORT_HOTPLUG_EN); - /* Leave other bits alone */ - hotplug_en |= HOTPLUG_EN_MASK; + /* Note HDMI and DP share bits */ + if (dev_priv->hotplug_supported_mask & HDMIB_HOTPLUG_INT_STATUS) + hotplug_en |= HDMIB_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & HDMIC_HOTPLUG_INT_STATUS) + hotplug_en |= HDMIC_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & HDMID_HOTPLUG_INT_STATUS) + hotplug_en |= HDMID_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & SDVOC_HOTPLUG_INT_STATUS) + hotplug_en |= SDVOC_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & SDVOB_HOTPLUG_INT_STATUS) + hotplug_en |= SDVOB_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & CRT_HOTPLUG_INT_STATUS) + hotplug_en |= CRT_HOTPLUG_INT_EN; + /* Ignore TV since it's buggy */ + I915_WRITE(PORT_HOTPLUG_EN, hotplug_en); - dev_priv->hotplug_supported_mask = CRT_HOTPLUG_INT_STATUS | - TV_HOTPLUG_INT_STATUS | SDVOC_HOTPLUG_INT_STATUS | - SDVOB_HOTPLUG_INT_STATUS; - if (IS_G4X(dev)) { - dev_priv->hotplug_supported_mask |= - HDMIB_HOTPLUG_INT_STATUS | - HDMIC_HOTPLUG_INT_STATUS | - HDMID_HOTPLUG_INT_STATUS; - } /* Enable in IER... */ enable_mask |= I915_DISPLAY_PORT_INTERRUPT; /* and unmask in IMR */ @@ -1178,7 +1473,7 @@ void i915_driver_irq_uninstall(struct drm_device * dev) dev_priv->vblank_pipe = 0; - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { ironlake_irq_uninstall(dev); return; } diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h index 974b3cf70618..3d59862c7ccd 100644 --- a/drivers/gpu/drm/i915/i915_reg.h +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -53,6 +53,25 @@ #define INTEL_GMCH_GMS_STOLEN_224M (0xc << 4) #define INTEL_GMCH_GMS_STOLEN_352M (0xd << 4) +#define SNB_GMCH_CTRL 0x50 +#define SNB_GMCH_GMS_STOLEN_MASK 0xF8 +#define SNB_GMCH_GMS_STOLEN_32M (1 << 3) +#define SNB_GMCH_GMS_STOLEN_64M (2 << 3) +#define SNB_GMCH_GMS_STOLEN_96M (3 << 3) +#define SNB_GMCH_GMS_STOLEN_128M (4 << 3) +#define SNB_GMCH_GMS_STOLEN_160M (5 << 3) +#define SNB_GMCH_GMS_STOLEN_192M (6 << 3) +#define SNB_GMCH_GMS_STOLEN_224M (7 << 3) +#define SNB_GMCH_GMS_STOLEN_256M (8 << 3) +#define SNB_GMCH_GMS_STOLEN_288M (9 << 3) +#define SNB_GMCH_GMS_STOLEN_320M (0xa << 3) +#define SNB_GMCH_GMS_STOLEN_352M (0xb << 3) +#define SNB_GMCH_GMS_STOLEN_384M (0xc << 3) +#define SNB_GMCH_GMS_STOLEN_416M (0xd << 3) +#define SNB_GMCH_GMS_STOLEN_448M (0xe << 3) +#define SNB_GMCH_GMS_STOLEN_480M (0xf << 3) +#define SNB_GMCH_GMS_STOLEN_512M (0x10 << 3) + /* PCI config space */ #define HPLLCC 0xc0 /* 855 only */ @@ -61,6 +80,7 @@ #define GC_CLOCK_100_200 (1 << 0) #define GC_CLOCK_100_133 (2 << 0) #define GC_CLOCK_166_250 (3 << 0) +#define GCFGC2 0xda #define GCFGC 0xf0 /* 915+ only */ #define GC_LOW_FREQUENCY_ENABLE (1 << 7) #define GC_DISPLAY_CLOCK_190_200_MHZ (0 << 4) @@ -234,6 +254,9 @@ #define I965_FENCE_REG_VALID (1<<0) #define I965_FENCE_MAX_PITCH_VAL 0x0400 +#define FENCE_REG_SANDYBRIDGE_0 0x100000 +#define SANDYBRIDGE_FENCE_PITCH_SHIFT 32 + /* * Instruction and interrupt control regs */ @@ -265,6 +288,7 @@ #define INSTDONE1 0x0207c /* 965+ only */ #define ACTHD_I965 0x02074 #define HWS_PGA 0x02080 +#define HWS_PGA_GEN6 0x04080 #define HWS_ADDRESS_MASK 0xfffff000 #define HWS_START_ADDRESS_SHIFT 4 #define PWRCTXA 0x2088 /* 965GM+ only */ @@ -282,7 +306,7 @@ #define I915_PIPE_CONTROL_NOTIFY_INTERRUPT (1<<18) #define I915_DISPLAY_PORT_INTERRUPT (1<<17) #define I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT (1<<15) -#define I915_GMCH_THERMAL_SENSOR_EVENT_INTERRUPT (1<<14) +#define I915_GMCH_THERMAL_SENSOR_EVENT_INTERRUPT (1<<14) /* p-state */ #define I915_HWB_OOM_INTERRUPT (1<<13) #define I915_SYNC_STATUS_INTERRUPT (1<<12) #define I915_DISPLAY_PLANE_A_FLIP_PENDING_INTERRUPT (1<<11) @@ -306,11 +330,14 @@ #define I915_ERROR_MEMORY_REFRESH (1<<1) #define I915_ERROR_INSTRUCTION (1<<0) #define INSTPM 0x020c0 +#define INSTPM_SELF_EN (1<<12) /* 915GM only */ #define ACTHD 0x020c8 #define FW_BLC 0x020d8 #define FW_BLC2 0x020dc #define FW_BLC_SELF 0x020e0 /* 915+ only */ -#define FW_BLC_SELF_EN (1<<15) +#define FW_BLC_SELF_EN_MASK (1<<31) +#define FW_BLC_SELF_FIFO_MASK (1<<16) /* 945 only */ +#define FW_BLC_SELF_EN (1<<15) /* 945 only */ #define MM_BURST_LENGTH 0x00700000 #define MM_FIFO_WATERMARK 0x0001F000 #define LM_BURST_LENGTH 0x00000700 @@ -324,6 +351,7 @@ #define CM0_COLOR_EVICT_DISABLE (1<<3) #define CM0_DEPTH_WRITE_DISABLE (1<<1) #define CM0_RC_OP_FLUSH_DISABLE (1<<0) +#define BB_ADDR 0x02140 /* 8 bytes */ #define GFX_FLSH_CNTL 0x02170 /* 915+ only */ @@ -338,6 +366,7 @@ #define FBC_CTL_PERIODIC (1<<30) #define FBC_CTL_INTERVAL_SHIFT (16) #define FBC_CTL_UNCOMPRESSIBLE (1<<14) +#define FBC_C3_IDLE (1<<13) #define FBC_CTL_STRIDE_SHIFT (5) #define FBC_CTL_FENCENO (1<<0) #define FBC_COMMAND 0x0320c @@ -783,10 +812,144 @@ #define CLKCFG_MEM_800 (3 << 4) #define CLKCFG_MEM_MASK (7 << 4) -/** GM965 GM45 render standby register */ -#define MCHBAR_RENDER_STANDBY 0x111B8 +#define CRSTANDVID 0x11100 +#define PXVFREQ_BASE 0x11110 /* P[0-15]VIDFREQ (0x1114c) (Ironlake) */ +#define PXVFREQ_PX_MASK 0x7f000000 +#define PXVFREQ_PX_SHIFT 24 +#define VIDFREQ_BASE 0x11110 +#define VIDFREQ1 0x11110 /* VIDFREQ1-4 (0x1111c) (Cantiga) */ +#define VIDFREQ2 0x11114 +#define VIDFREQ3 0x11118 +#define VIDFREQ4 0x1111c +#define VIDFREQ_P0_MASK 0x1f000000 +#define VIDFREQ_P0_SHIFT 24 +#define VIDFREQ_P0_CSCLK_MASK 0x00f00000 +#define VIDFREQ_P0_CSCLK_SHIFT 20 +#define VIDFREQ_P0_CRCLK_MASK 0x000f0000 +#define VIDFREQ_P0_CRCLK_SHIFT 16 +#define VIDFREQ_P1_MASK 0x00001f00 +#define VIDFREQ_P1_SHIFT 8 +#define VIDFREQ_P1_CSCLK_MASK 0x000000f0 +#define VIDFREQ_P1_CSCLK_SHIFT 4 +#define VIDFREQ_P1_CRCLK_MASK 0x0000000f +#define INTTOEXT_BASE_ILK 0x11300 +#define INTTOEXT_BASE 0x11120 /* INTTOEXT1-8 (0x1113c) */ +#define INTTOEXT_MAP3_SHIFT 24 +#define INTTOEXT_MAP3_MASK (0x1f << INTTOEXT_MAP3_SHIFT) +#define INTTOEXT_MAP2_SHIFT 16 +#define INTTOEXT_MAP2_MASK (0x1f << INTTOEXT_MAP2_SHIFT) +#define INTTOEXT_MAP1_SHIFT 8 +#define INTTOEXT_MAP1_MASK (0x1f << INTTOEXT_MAP1_SHIFT) +#define INTTOEXT_MAP0_SHIFT 0 +#define INTTOEXT_MAP0_MASK (0x1f << INTTOEXT_MAP0_SHIFT) +#define MEMSWCTL 0x11170 /* Ironlake only */ +#define MEMCTL_CMD_MASK 0xe000 +#define MEMCTL_CMD_SHIFT 13 +#define MEMCTL_CMD_RCLK_OFF 0 +#define MEMCTL_CMD_RCLK_ON 1 +#define MEMCTL_CMD_CHFREQ 2 +#define MEMCTL_CMD_CHVID 3 +#define MEMCTL_CMD_VMMOFF 4 +#define MEMCTL_CMD_VMMON 5 +#define MEMCTL_CMD_STS (1<<12) /* write 1 triggers command, clears + when command complete */ +#define MEMCTL_FREQ_MASK 0x0f00 /* jitter, from 0-15 */ +#define MEMCTL_FREQ_SHIFT 8 +#define MEMCTL_SFCAVM (1<<7) +#define MEMCTL_TGT_VID_MASK 0x007f +#define MEMIHYST 0x1117c +#define MEMINTREN 0x11180 /* 16 bits */ +#define MEMINT_RSEXIT_EN (1<<8) +#define MEMINT_CX_SUPR_EN (1<<7) +#define MEMINT_CONT_BUSY_EN (1<<6) +#define MEMINT_AVG_BUSY_EN (1<<5) +#define MEMINT_EVAL_CHG_EN (1<<4) +#define MEMINT_MON_IDLE_EN (1<<3) +#define MEMINT_UP_EVAL_EN (1<<2) +#define MEMINT_DOWN_EVAL_EN (1<<1) +#define MEMINT_SW_CMD_EN (1<<0) +#define MEMINTRSTR 0x11182 /* 16 bits */ +#define MEM_RSEXIT_MASK 0xc000 +#define MEM_RSEXIT_SHIFT 14 +#define MEM_CONT_BUSY_MASK 0x3000 +#define MEM_CONT_BUSY_SHIFT 12 +#define MEM_AVG_BUSY_MASK 0x0c00 +#define MEM_AVG_BUSY_SHIFT 10 +#define MEM_EVAL_CHG_MASK 0x0300 +#define MEM_EVAL_BUSY_SHIFT 8 +#define MEM_MON_IDLE_MASK 0x00c0 +#define MEM_MON_IDLE_SHIFT 6 +#define MEM_UP_EVAL_MASK 0x0030 +#define MEM_UP_EVAL_SHIFT 4 +#define MEM_DOWN_EVAL_MASK 0x000c +#define MEM_DOWN_EVAL_SHIFT 2 +#define MEM_SW_CMD_MASK 0x0003 +#define MEM_INT_STEER_GFX 0 +#define MEM_INT_STEER_CMR 1 +#define MEM_INT_STEER_SMI 2 +#define MEM_INT_STEER_SCI 3 +#define MEMINTRSTS 0x11184 +#define MEMINT_RSEXIT (1<<7) +#define MEMINT_CONT_BUSY (1<<6) +#define MEMINT_AVG_BUSY (1<<5) +#define MEMINT_EVAL_CHG (1<<4) +#define MEMINT_MON_IDLE (1<<3) +#define MEMINT_UP_EVAL (1<<2) +#define MEMINT_DOWN_EVAL (1<<1) +#define MEMINT_SW_CMD (1<<0) +#define MEMMODECTL 0x11190 +#define MEMMODE_BOOST_EN (1<<31) +#define MEMMODE_BOOST_FREQ_MASK 0x0f000000 /* jitter for boost, 0-15 */ +#define MEMMODE_BOOST_FREQ_SHIFT 24 +#define MEMMODE_IDLE_MODE_MASK 0x00030000 +#define MEMMODE_IDLE_MODE_SHIFT 16 +#define MEMMODE_IDLE_MODE_EVAL 0 +#define MEMMODE_IDLE_MODE_CONT 1 +#define MEMMODE_HWIDLE_EN (1<<15) +#define MEMMODE_SWMODE_EN (1<<14) +#define MEMMODE_RCLK_GATE (1<<13) +#define MEMMODE_HW_UPDATE (1<<12) +#define MEMMODE_FSTART_MASK 0x00000f00 /* starting jitter, 0-15 */ +#define MEMMODE_FSTART_SHIFT 8 +#define MEMMODE_FMAX_MASK 0x000000f0 /* max jitter, 0-15 */ +#define MEMMODE_FMAX_SHIFT 4 +#define MEMMODE_FMIN_MASK 0x0000000f /* min jitter, 0-15 */ +#define RCBMAXAVG 0x1119c +#define MEMSWCTL2 0x1119e /* Cantiga only */ +#define SWMEMCMD_RENDER_OFF (0 << 13) +#define SWMEMCMD_RENDER_ON (1 << 13) +#define SWMEMCMD_SWFREQ (2 << 13) +#define SWMEMCMD_TARVID (3 << 13) +#define SWMEMCMD_VRM_OFF (4 << 13) +#define SWMEMCMD_VRM_ON (5 << 13) +#define CMDSTS (1<<12) +#define SFCAVM (1<<11) +#define SWFREQ_MASK 0x0380 /* P0-7 */ +#define SWFREQ_SHIFT 7 +#define TARVID_MASK 0x001f +#define MEMSTAT_CTG 0x111a0 +#define RCBMINAVG 0x111a0 +#define RCUPEI 0x111b0 +#define RCDNEI 0x111b4 +#define MCHBAR_RENDER_STANDBY 0x111b8 #define RCX_SW_EXIT (1<<23) #define RSX_STATUS_MASK 0x00700000 +#define VIDCTL 0x111c0 +#define VIDSTS 0x111c8 +#define VIDSTART 0x111cc /* 8 bits */ +#define MEMSTAT_ILK 0x111f8 +#define MEMSTAT_VID_MASK 0x7f00 +#define MEMSTAT_VID_SHIFT 8 +#define MEMSTAT_PSTATE_MASK 0x00f8 +#define MEMSTAT_PSTATE_SHIFT 3 +#define MEMSTAT_MON_ACTV (1<<2) +#define MEMSTAT_SRC_CTL_MASK 0x0003 +#define MEMSTAT_SRC_CTL_CORE 0 +#define MEMSTAT_SRC_CTL_TRB 1 +#define MEMSTAT_SRC_CTL_THM 2 +#define MEMSTAT_SRC_CTL_STDBY 3 +#define RCPREVBSYTUPAVG 0x113b8 +#define RCPREVBSYTDNAVG 0x113bc #define PEG_BAND_GAP_DATA 0x14d68 /* @@ -879,13 +1042,6 @@ #define CRT_HOTPLUG_DETECT_VOLTAGE_475MV (1 << 2) #define CRT_HOTPLUG_MASK (0x3fc) /* Bits 9-2 */ #define CRT_FORCE_HOTPLUG_MASK 0xfffffe1f -#define HOTPLUG_EN_MASK (HDMIB_HOTPLUG_INT_EN | \ - HDMIC_HOTPLUG_INT_EN | \ - HDMID_HOTPLUG_INT_EN | \ - SDVOB_HOTPLUG_INT_EN | \ - SDVOC_HOTPLUG_INT_EN | \ - CRT_HOTPLUG_INT_EN) - #define PORT_HOTPLUG_STAT 0x61114 #define HDMIB_HOTPLUG_INT_STATUS (1 << 29) @@ -982,6 +1138,8 @@ #define LVDS_PORT_EN (1 << 31) /* Selects pipe B for LVDS data. Must be set on pre-965. */ #define LVDS_PIPEB_SELECT (1 << 30) +/* LVDS dithering flag on 965/g4x platform */ +#define LVDS_ENABLE_DITHER (1 << 25) /* Enable border for unscaled (or aspect-scaled) display */ #define LVDS_BORDER_ENABLE (1 << 15) /* @@ -1751,6 +1909,8 @@ /* Display & cursor control */ +/* dithering flag on Ironlake */ +#define PIPE_ENABLE_DITHER (1 << 4) /* Pipe A */ #define PIPEADSL 0x70000 #define PIPEACONF 0x70008 @@ -1818,7 +1978,7 @@ #define DSPFW_PLANEB_SHIFT 8 #define DSPFW2 0x70038 #define DSPFW_CURSORA_MASK 0x00003f00 -#define DSPFW_CURSORA_SHIFT 16 +#define DSPFW_CURSORA_SHIFT 8 #define DSPFW3 0x7003c #define DSPFW_HPLL_SR_EN (1<<31) #define DSPFW_CURSOR_SR_SHIFT 24 diff --git a/drivers/gpu/drm/i915/i915_suspend.c b/drivers/gpu/drm/i915/i915_suspend.c index d5ebb00a9d49..ac0d1a73ac22 100644 --- a/drivers/gpu/drm/i915/i915_suspend.c +++ b/drivers/gpu/drm/i915/i915_suspend.c @@ -682,6 +682,8 @@ void i915_restore_display(struct drm_device *dev) I915_WRITE(PCH_PP_OFF_DELAYS, dev_priv->savePP_OFF_DELAYS); I915_WRITE(PCH_PP_DIVISOR, dev_priv->savePP_DIVISOR); I915_WRITE(PCH_PP_CONTROL, dev_priv->savePP_CONTROL); + I915_WRITE(MCHBAR_RENDER_STANDBY, + dev_priv->saveMCHBAR_RENDER_STANDBY); } else { I915_WRITE(PFIT_PGM_RATIOS, dev_priv->savePFIT_PGM_RATIOS); I915_WRITE(BLC_PWM_CTL, dev_priv->saveBLC_PWM_CTL); @@ -732,12 +734,6 @@ int i915_save_state(struct drm_device *dev) pci_read_config_byte(dev->pdev, LBB, &dev_priv->saveLBB); - /* Render Standby */ - if (I915_HAS_RC6(dev)) { - dev_priv->saveRENDERSTANDBY = I915_READ(MCHBAR_RENDER_STANDBY); - dev_priv->savePWRCTXA = I915_READ(PWRCTXA); - } - /* Hardware status page */ dev_priv->saveHWS = I915_READ(HWS_PGA); @@ -751,11 +747,16 @@ int i915_save_state(struct drm_device *dev) dev_priv->saveGTIMR = I915_READ(GTIMR); dev_priv->saveFDI_RXA_IMR = I915_READ(FDI_RXA_IMR); dev_priv->saveFDI_RXB_IMR = I915_READ(FDI_RXB_IMR); + dev_priv->saveMCHBAR_RENDER_STANDBY = + I915_READ(MCHBAR_RENDER_STANDBY); } else { dev_priv->saveIER = I915_READ(IER); dev_priv->saveIMR = I915_READ(IMR); } + if (IS_IRONLAKE_M(dev)) + ironlake_disable_drps(dev); + /* Cache mode state */ dev_priv->saveCACHE_MODE_0 = I915_READ(CACHE_MODE_0); @@ -793,12 +794,6 @@ int i915_restore_state(struct drm_device *dev) pci_write_config_byte(dev->pdev, LBB, dev_priv->saveLBB); - /* Render Standby */ - if (I915_HAS_RC6(dev)) { - I915_WRITE(MCHBAR_RENDER_STANDBY, dev_priv->saveRENDERSTANDBY); - I915_WRITE(PWRCTXA, dev_priv->savePWRCTXA); - } - /* Hardware status page */ I915_WRITE(HWS_PGA, dev_priv->saveHWS); @@ -832,6 +827,9 @@ int i915_restore_state(struct drm_device *dev) /* Clock gating state */ intel_init_clock_gating(dev); + if (IS_IRONLAKE_M(dev)) + ironlake_enable_drps(dev); + /* Cache mode state */ I915_WRITE (CACHE_MODE_0, dev_priv->saveCACHE_MODE_0 | 0xffff0000); diff --git a/drivers/gpu/drm/i915/intel_bios.c b/drivers/gpu/drm/i915/intel_bios.c index f27567747580..70c9d4ba7042 100644 --- a/drivers/gpu/drm/i915/intel_bios.c +++ b/drivers/gpu/drm/i915/intel_bios.c @@ -33,6 +33,8 @@ #define SLAVE_ADDR1 0x70 #define SLAVE_ADDR2 0x72 +static int panel_type; + static void * find_section(struct bdb_header *bdb, int section_id) { @@ -128,6 +130,7 @@ parse_lfp_panel_data(struct drm_i915_private *dev_priv, dev_priv->lvds_dither = lvds_options->pixel_dither; if (lvds_options->panel_type == 0xff) return; + panel_type = lvds_options->panel_type; lvds_lfp_data = find_section(bdb, BDB_LVDS_LFP_DATA); if (!lvds_lfp_data) @@ -197,7 +200,8 @@ parse_lfp_panel_data(struct drm_i915_private *dev_priv, memset(temp_mode, 0, sizeof(*temp_mode)); } kfree(temp_mode); - if (temp_downclock < panel_fixed_mode->clock) { + if (temp_downclock < panel_fixed_mode->clock && + i915_lvds_downclock) { dev_priv->lvds_downclock_avail = 1; dev_priv->lvds_downclock = temp_downclock; DRM_DEBUG_KMS("LVDS downclock is found in VBT. ", @@ -243,6 +247,7 @@ static void parse_general_features(struct drm_i915_private *dev_priv, struct bdb_header *bdb) { + struct drm_device *dev = dev_priv->dev; struct bdb_general_features *general; /* Set sensible defaults in case we can't find the general block */ @@ -259,7 +264,7 @@ parse_general_features(struct drm_i915_private *dev_priv, if (IS_I85X(dev_priv->dev)) dev_priv->lvds_ssc_freq = general->ssc_freq ? 66 : 48; - else if (IS_IRONLAKE(dev_priv->dev)) + else if (IS_IRONLAKE(dev_priv->dev) || IS_GEN6(dev)) dev_priv->lvds_ssc_freq = general->ssc_freq ? 100 : 120; else @@ -405,6 +410,34 @@ parse_driver_features(struct drm_i915_private *dev_priv, } static void +parse_edp(struct drm_i915_private *dev_priv, struct bdb_header *bdb) +{ + struct bdb_edp *edp; + + edp = find_section(bdb, BDB_EDP); + if (!edp) { + if (SUPPORTS_EDP(dev_priv->dev) && dev_priv->edp_support) { + DRM_DEBUG_KMS("No eDP BDB found but eDP panel supported,\ + assume 18bpp panel color depth.\n"); + dev_priv->edp_bpp = 18; + } + return; + } + + switch ((edp->color_depth >> (panel_type * 2)) & 3) { + case EDP_18BPP: + dev_priv->edp_bpp = 18; + break; + case EDP_24BPP: + dev_priv->edp_bpp = 24; + break; + case EDP_30BPP: + dev_priv->edp_bpp = 30; + break; + } +} + +static void parse_device_mapping(struct drm_i915_private *dev_priv, struct bdb_header *bdb) { @@ -521,6 +554,7 @@ intel_init_bios(struct drm_device *dev) parse_sdvo_device_mapping(dev_priv, bdb); parse_device_mapping(dev_priv, bdb); parse_driver_features(dev_priv, bdb); + parse_edp(dev_priv, bdb); pci_unmap_rom(pdev, bios); diff --git a/drivers/gpu/drm/i915/intel_bios.h b/drivers/gpu/drm/i915/intel_bios.h index 425ac9d7f724..4c18514f6f80 100644 --- a/drivers/gpu/drm/i915/intel_bios.h +++ b/drivers/gpu/drm/i915/intel_bios.h @@ -98,6 +98,7 @@ struct vbios_data { #define BDB_SDVO_LVDS_PNP_IDS 24 #define BDB_SDVO_LVDS_POWER_SEQ 25 #define BDB_TV_OPTIONS 26 +#define BDB_EDP 27 #define BDB_LVDS_OPTIONS 40 #define BDB_LVDS_LFP_DATA_PTRS 41 #define BDB_LVDS_LFP_DATA 42 @@ -426,6 +427,45 @@ struct bdb_driver_features { u8 custom_vbt_version; } __attribute__((packed)); +#define EDP_18BPP 0 +#define EDP_24BPP 1 +#define EDP_30BPP 2 +#define EDP_RATE_1_62 0 +#define EDP_RATE_2_7 1 +#define EDP_LANE_1 0 +#define EDP_LANE_2 1 +#define EDP_LANE_4 3 +#define EDP_PREEMPHASIS_NONE 0 +#define EDP_PREEMPHASIS_3_5dB 1 +#define EDP_PREEMPHASIS_6dB 2 +#define EDP_PREEMPHASIS_9_5dB 3 +#define EDP_VSWING_0_4V 0 +#define EDP_VSWING_0_6V 1 +#define EDP_VSWING_0_8V 2 +#define EDP_VSWING_1_2V 3 + +struct edp_power_seq { + u16 t3; + u16 t7; + u16 t9; + u16 t10; + u16 t12; +} __attribute__ ((packed)); + +struct edp_link_params { + u8 rate:4; + u8 lanes:4; + u8 preemphasis:4; + u8 vswing:4; +} __attribute__ ((packed)); + +struct bdb_edp { + struct edp_power_seq power_seqs[16]; + u32 color_depth; + u32 sdrrs_msa_timing_delay; + struct edp_link_params link_params[16]; +} __attribute__ ((packed)); + bool intel_init_bios(struct drm_device *dev); /* diff --git a/drivers/gpu/drm/i915/intel_crt.c b/drivers/gpu/drm/i915/intel_crt.c index 9f3d3e563414..fccf07470c8f 100644 --- a/drivers/gpu/drm/i915/intel_crt.c +++ b/drivers/gpu/drm/i915/intel_crt.c @@ -39,7 +39,7 @@ static void intel_crt_dpms(struct drm_encoder *encoder, int mode) struct drm_i915_private *dev_priv = dev->dev_private; u32 temp, reg; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) reg = PCH_ADPA; else reg = ADPA; @@ -113,7 +113,7 @@ static void intel_crt_mode_set(struct drm_encoder *encoder, else dpll_md_reg = DPLL_B_MD; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) adpa_reg = PCH_ADPA; else adpa_reg = ADPA; @@ -122,7 +122,7 @@ static void intel_crt_mode_set(struct drm_encoder *encoder, * Disable separate mode multiplier used when cloning SDVO to CRT * XXX this needs to be adjusted when we really are cloning */ - if (IS_I965G(dev) && !IS_IRONLAKE(dev)) { + if (IS_I965G(dev) && !HAS_PCH_SPLIT(dev)) { dpll_md = I915_READ(dpll_md_reg); I915_WRITE(dpll_md_reg, dpll_md & ~DPLL_MD_UDI_MULTIPLIER_MASK); @@ -136,11 +136,11 @@ static void intel_crt_mode_set(struct drm_encoder *encoder, if (intel_crtc->pipe == 0) { adpa |= ADPA_PIPE_A_SELECT; - if (!IS_IRONLAKE(dev)) + if (!HAS_PCH_SPLIT(dev)) I915_WRITE(BCLRPAT_A, 0); } else { adpa |= ADPA_PIPE_B_SELECT; - if (!IS_IRONLAKE(dev)) + if (!HAS_PCH_SPLIT(dev)) I915_WRITE(BCLRPAT_B, 0); } @@ -157,6 +157,9 @@ static bool intel_ironlake_crt_detect_hotplug(struct drm_connector *connector) adpa = I915_READ(PCH_ADPA); adpa &= ~ADPA_CRT_HOTPLUG_MASK; + /* disable HPD first */ + I915_WRITE(PCH_ADPA, adpa); + (void)I915_READ(PCH_ADPA); adpa |= (ADPA_CRT_HOTPLUG_PERIOD_128 | ADPA_CRT_HOTPLUG_WARMUP_10MS | @@ -199,7 +202,7 @@ static bool intel_crt_detect_hotplug(struct drm_connector *connector) u32 hotplug_en; int i, tries = 0; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) return intel_ironlake_crt_detect_hotplug(connector); /* @@ -521,7 +524,7 @@ void intel_crt_init(struct drm_device *dev) &intel_output->enc); /* Set up the DDC bus. */ - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) i2c_reg = PCH_GPIOA; else { i2c_reg = GPIOA; @@ -548,4 +551,6 @@ void intel_crt_init(struct drm_device *dev) drm_connector_helper_add(connector, &intel_crt_connector_helper_funcs); drm_sysfs_connector_add(connector); + + dev_priv->hotplug_supported_mask |= CRT_HOTPLUG_INT_STATUS; } diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index 52cd9b006da2..9cd6de5f9906 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -70,8 +70,6 @@ struct intel_limit { intel_p2_t p2; bool (* find_pll)(const intel_limit_t *, struct drm_crtc *, int, int, intel_clock_t *); - bool (* find_reduced_pll)(const intel_limit_t *, struct drm_crtc *, - int, int, intel_clock_t *); }; #define I8XX_DOT_MIN 25000 @@ -234,7 +232,7 @@ struct intel_limit { #define G4X_P2_DISPLAY_PORT_FAST 10 #define G4X_P2_DISPLAY_PORT_LIMIT 0 -/* Ironlake */ +/* Ironlake / Sandybridge */ /* as we calculate clock using (register_value + 2) for N/M1/M2, so here the range value for them is (actual_value-2). */ @@ -242,38 +240,93 @@ struct intel_limit { #define IRONLAKE_DOT_MAX 350000 #define IRONLAKE_VCO_MIN 1760000 #define IRONLAKE_VCO_MAX 3510000 -#define IRONLAKE_N_MIN 1 -#define IRONLAKE_N_MAX 5 -#define IRONLAKE_M_MIN 79 -#define IRONLAKE_M_MAX 118 #define IRONLAKE_M1_MIN 12 -#define IRONLAKE_M1_MAX 23 +#define IRONLAKE_M1_MAX 22 #define IRONLAKE_M2_MIN 5 #define IRONLAKE_M2_MAX 9 -#define IRONLAKE_P_SDVO_DAC_MIN 5 -#define IRONLAKE_P_SDVO_DAC_MAX 80 -#define IRONLAKE_P_LVDS_MIN 28 -#define IRONLAKE_P_LVDS_MAX 112 -#define IRONLAKE_P1_MIN 1 -#define IRONLAKE_P1_MAX 8 -#define IRONLAKE_P2_SDVO_DAC_SLOW 10 -#define IRONLAKE_P2_SDVO_DAC_FAST 5 -#define IRONLAKE_P2_LVDS_SLOW 14 /* single channel */ -#define IRONLAKE_P2_LVDS_FAST 7 /* double channel */ #define IRONLAKE_P2_DOT_LIMIT 225000 /* 225Mhz */ +/* We have parameter ranges for different type of outputs. */ + +/* DAC & HDMI Refclk 120Mhz */ +#define IRONLAKE_DAC_N_MIN 1 +#define IRONLAKE_DAC_N_MAX 5 +#define IRONLAKE_DAC_M_MIN 79 +#define IRONLAKE_DAC_M_MAX 127 +#define IRONLAKE_DAC_P_MIN 5 +#define IRONLAKE_DAC_P_MAX 80 +#define IRONLAKE_DAC_P1_MIN 1 +#define IRONLAKE_DAC_P1_MAX 8 +#define IRONLAKE_DAC_P2_SLOW 10 +#define IRONLAKE_DAC_P2_FAST 5 + +/* LVDS single-channel 120Mhz refclk */ +#define IRONLAKE_LVDS_S_N_MIN 1 +#define IRONLAKE_LVDS_S_N_MAX 3 +#define IRONLAKE_LVDS_S_M_MIN 79 +#define IRONLAKE_LVDS_S_M_MAX 118 +#define IRONLAKE_LVDS_S_P_MIN 28 +#define IRONLAKE_LVDS_S_P_MAX 112 +#define IRONLAKE_LVDS_S_P1_MIN 2 +#define IRONLAKE_LVDS_S_P1_MAX 8 +#define IRONLAKE_LVDS_S_P2_SLOW 14 +#define IRONLAKE_LVDS_S_P2_FAST 14 + +/* LVDS dual-channel 120Mhz refclk */ +#define IRONLAKE_LVDS_D_N_MIN 1 +#define IRONLAKE_LVDS_D_N_MAX 3 +#define IRONLAKE_LVDS_D_M_MIN 79 +#define IRONLAKE_LVDS_D_M_MAX 127 +#define IRONLAKE_LVDS_D_P_MIN 14 +#define IRONLAKE_LVDS_D_P_MAX 56 +#define IRONLAKE_LVDS_D_P1_MIN 2 +#define IRONLAKE_LVDS_D_P1_MAX 8 +#define IRONLAKE_LVDS_D_P2_SLOW 7 +#define IRONLAKE_LVDS_D_P2_FAST 7 + +/* LVDS single-channel 100Mhz refclk */ +#define IRONLAKE_LVDS_S_SSC_N_MIN 1 +#define IRONLAKE_LVDS_S_SSC_N_MAX 2 +#define IRONLAKE_LVDS_S_SSC_M_MIN 79 +#define IRONLAKE_LVDS_S_SSC_M_MAX 126 +#define IRONLAKE_LVDS_S_SSC_P_MIN 28 +#define IRONLAKE_LVDS_S_SSC_P_MAX 112 +#define IRONLAKE_LVDS_S_SSC_P1_MIN 2 +#define IRONLAKE_LVDS_S_SSC_P1_MAX 8 +#define IRONLAKE_LVDS_S_SSC_P2_SLOW 14 +#define IRONLAKE_LVDS_S_SSC_P2_FAST 14 + +/* LVDS dual-channel 100Mhz refclk */ +#define IRONLAKE_LVDS_D_SSC_N_MIN 1 +#define IRONLAKE_LVDS_D_SSC_N_MAX 3 +#define IRONLAKE_LVDS_D_SSC_M_MIN 79 +#define IRONLAKE_LVDS_D_SSC_M_MAX 126 +#define IRONLAKE_LVDS_D_SSC_P_MIN 14 +#define IRONLAKE_LVDS_D_SSC_P_MAX 42 +#define IRONLAKE_LVDS_D_SSC_P1_MIN 2 +#define IRONLAKE_LVDS_D_SSC_P1_MAX 6 +#define IRONLAKE_LVDS_D_SSC_P2_SLOW 7 +#define IRONLAKE_LVDS_D_SSC_P2_FAST 7 + +/* DisplayPort */ +#define IRONLAKE_DP_N_MIN 1 +#define IRONLAKE_DP_N_MAX 2 +#define IRONLAKE_DP_M_MIN 81 +#define IRONLAKE_DP_M_MAX 90 +#define IRONLAKE_DP_P_MIN 10 +#define IRONLAKE_DP_P_MAX 20 +#define IRONLAKE_DP_P2_FAST 10 +#define IRONLAKE_DP_P2_SLOW 10 +#define IRONLAKE_DP_P2_LIMIT 0 +#define IRONLAKE_DP_P1_MIN 1 +#define IRONLAKE_DP_P1_MAX 2 + static bool intel_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, int target, int refclk, intel_clock_t *best_clock); static bool -intel_find_best_reduced_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, - int target, int refclk, intel_clock_t *best_clock); -static bool intel_g4x_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, int target, int refclk, intel_clock_t *best_clock); -static bool -intel_ironlake_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, - int target, int refclk, intel_clock_t *best_clock); static bool intel_find_pll_g4x_dp(const intel_limit_t *, struct drm_crtc *crtc, @@ -294,7 +347,6 @@ static const intel_limit_t intel_limits_i8xx_dvo = { .p2 = { .dot_limit = I8XX_P2_SLOW_LIMIT, .p2_slow = I8XX_P2_SLOW, .p2_fast = I8XX_P2_FAST }, .find_pll = intel_find_best_PLL, - .find_reduced_pll = intel_find_best_reduced_PLL, }; static const intel_limit_t intel_limits_i8xx_lvds = { @@ -309,7 +361,6 @@ static const intel_limit_t intel_limits_i8xx_lvds = { .p2 = { .dot_limit = I8XX_P2_SLOW_LIMIT, .p2_slow = I8XX_P2_LVDS_SLOW, .p2_fast = I8XX_P2_LVDS_FAST }, .find_pll = intel_find_best_PLL, - .find_reduced_pll = intel_find_best_reduced_PLL, }; static const intel_limit_t intel_limits_i9xx_sdvo = { @@ -324,7 +375,6 @@ static const intel_limit_t intel_limits_i9xx_sdvo = { .p2 = { .dot_limit = I9XX_P2_SDVO_DAC_SLOW_LIMIT, .p2_slow = I9XX_P2_SDVO_DAC_SLOW, .p2_fast = I9XX_P2_SDVO_DAC_FAST }, .find_pll = intel_find_best_PLL, - .find_reduced_pll = intel_find_best_reduced_PLL, }; static const intel_limit_t intel_limits_i9xx_lvds = { @@ -342,7 +392,6 @@ static const intel_limit_t intel_limits_i9xx_lvds = { .p2 = { .dot_limit = I9XX_P2_LVDS_SLOW_LIMIT, .p2_slow = I9XX_P2_LVDS_SLOW, .p2_fast = I9XX_P2_LVDS_FAST }, .find_pll = intel_find_best_PLL, - .find_reduced_pll = intel_find_best_reduced_PLL, }; /* below parameter and function is for G4X Chipset Family*/ @@ -360,7 +409,6 @@ static const intel_limit_t intel_limits_g4x_sdvo = { .p2_fast = G4X_P2_SDVO_FAST }, .find_pll = intel_g4x_find_best_PLL, - .find_reduced_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_g4x_hdmi = { @@ -377,7 +425,6 @@ static const intel_limit_t intel_limits_g4x_hdmi = { .p2_fast = G4X_P2_HDMI_DAC_FAST }, .find_pll = intel_g4x_find_best_PLL, - .find_reduced_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_g4x_single_channel_lvds = { @@ -402,7 +449,6 @@ static const intel_limit_t intel_limits_g4x_single_channel_lvds = { .p2_fast = G4X_P2_SINGLE_CHANNEL_LVDS_FAST }, .find_pll = intel_g4x_find_best_PLL, - .find_reduced_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_g4x_dual_channel_lvds = { @@ -427,7 +473,6 @@ static const intel_limit_t intel_limits_g4x_dual_channel_lvds = { .p2_fast = G4X_P2_DUAL_CHANNEL_LVDS_FAST }, .find_pll = intel_g4x_find_best_PLL, - .find_reduced_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_g4x_display_port = { @@ -465,7 +510,6 @@ static const intel_limit_t intel_limits_pineview_sdvo = { .p2 = { .dot_limit = I9XX_P2_SDVO_DAC_SLOW_LIMIT, .p2_slow = I9XX_P2_SDVO_DAC_SLOW, .p2_fast = I9XX_P2_SDVO_DAC_FAST }, .find_pll = intel_find_best_PLL, - .find_reduced_pll = intel_find_best_reduced_PLL, }; static const intel_limit_t intel_limits_pineview_lvds = { @@ -481,46 +525,135 @@ static const intel_limit_t intel_limits_pineview_lvds = { .p2 = { .dot_limit = I9XX_P2_LVDS_SLOW_LIMIT, .p2_slow = I9XX_P2_LVDS_SLOW, .p2_fast = I9XX_P2_LVDS_SLOW }, .find_pll = intel_find_best_PLL, - .find_reduced_pll = intel_find_best_reduced_PLL, }; -static const intel_limit_t intel_limits_ironlake_sdvo = { +static const intel_limit_t intel_limits_ironlake_dac = { + .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, + .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, + .n = { .min = IRONLAKE_DAC_N_MIN, .max = IRONLAKE_DAC_N_MAX }, + .m = { .min = IRONLAKE_DAC_M_MIN, .max = IRONLAKE_DAC_M_MAX }, + .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, + .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, + .p = { .min = IRONLAKE_DAC_P_MIN, .max = IRONLAKE_DAC_P_MAX }, + .p1 = { .min = IRONLAKE_DAC_P1_MIN, .max = IRONLAKE_DAC_P1_MAX }, + .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, + .p2_slow = IRONLAKE_DAC_P2_SLOW, + .p2_fast = IRONLAKE_DAC_P2_FAST }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_ironlake_single_lvds = { + .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, + .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, + .n = { .min = IRONLAKE_LVDS_S_N_MIN, .max = IRONLAKE_LVDS_S_N_MAX }, + .m = { .min = IRONLAKE_LVDS_S_M_MIN, .max = IRONLAKE_LVDS_S_M_MAX }, + .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, + .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, + .p = { .min = IRONLAKE_LVDS_S_P_MIN, .max = IRONLAKE_LVDS_S_P_MAX }, + .p1 = { .min = IRONLAKE_LVDS_S_P1_MIN, .max = IRONLAKE_LVDS_S_P1_MAX }, + .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, + .p2_slow = IRONLAKE_LVDS_S_P2_SLOW, + .p2_fast = IRONLAKE_LVDS_S_P2_FAST }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_ironlake_dual_lvds = { + .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, + .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, + .n = { .min = IRONLAKE_LVDS_D_N_MIN, .max = IRONLAKE_LVDS_D_N_MAX }, + .m = { .min = IRONLAKE_LVDS_D_M_MIN, .max = IRONLAKE_LVDS_D_M_MAX }, + .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, + .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, + .p = { .min = IRONLAKE_LVDS_D_P_MIN, .max = IRONLAKE_LVDS_D_P_MAX }, + .p1 = { .min = IRONLAKE_LVDS_D_P1_MIN, .max = IRONLAKE_LVDS_D_P1_MAX }, + .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, + .p2_slow = IRONLAKE_LVDS_D_P2_SLOW, + .p2_fast = IRONLAKE_LVDS_D_P2_FAST }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_ironlake_single_lvds_100m = { .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, - .n = { .min = IRONLAKE_N_MIN, .max = IRONLAKE_N_MAX }, - .m = { .min = IRONLAKE_M_MIN, .max = IRONLAKE_M_MAX }, + .n = { .min = IRONLAKE_LVDS_S_SSC_N_MIN, .max = IRONLAKE_LVDS_S_SSC_N_MAX }, + .m = { .min = IRONLAKE_LVDS_S_SSC_M_MIN, .max = IRONLAKE_LVDS_S_SSC_M_MAX }, .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, - .p = { .min = IRONLAKE_P_SDVO_DAC_MIN, .max = IRONLAKE_P_SDVO_DAC_MAX }, - .p1 = { .min = IRONLAKE_P1_MIN, .max = IRONLAKE_P1_MAX }, + .p = { .min = IRONLAKE_LVDS_S_SSC_P_MIN, .max = IRONLAKE_LVDS_S_SSC_P_MAX }, + .p1 = { .min = IRONLAKE_LVDS_S_SSC_P1_MIN,.max = IRONLAKE_LVDS_S_SSC_P1_MAX }, .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, - .p2_slow = IRONLAKE_P2_SDVO_DAC_SLOW, - .p2_fast = IRONLAKE_P2_SDVO_DAC_FAST }, - .find_pll = intel_ironlake_find_best_PLL, + .p2_slow = IRONLAKE_LVDS_S_SSC_P2_SLOW, + .p2_fast = IRONLAKE_LVDS_S_SSC_P2_FAST }, + .find_pll = intel_g4x_find_best_PLL, }; -static const intel_limit_t intel_limits_ironlake_lvds = { +static const intel_limit_t intel_limits_ironlake_dual_lvds_100m = { .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, - .n = { .min = IRONLAKE_N_MIN, .max = IRONLAKE_N_MAX }, - .m = { .min = IRONLAKE_M_MIN, .max = IRONLAKE_M_MAX }, + .n = { .min = IRONLAKE_LVDS_D_SSC_N_MIN, .max = IRONLAKE_LVDS_D_SSC_N_MAX }, + .m = { .min = IRONLAKE_LVDS_D_SSC_M_MIN, .max = IRONLAKE_LVDS_D_SSC_M_MAX }, .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, - .p = { .min = IRONLAKE_P_LVDS_MIN, .max = IRONLAKE_P_LVDS_MAX }, - .p1 = { .min = IRONLAKE_P1_MIN, .max = IRONLAKE_P1_MAX }, + .p = { .min = IRONLAKE_LVDS_D_SSC_P_MIN, .max = IRONLAKE_LVDS_D_SSC_P_MAX }, + .p1 = { .min = IRONLAKE_LVDS_D_SSC_P1_MIN,.max = IRONLAKE_LVDS_D_SSC_P1_MAX }, .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, - .p2_slow = IRONLAKE_P2_LVDS_SLOW, - .p2_fast = IRONLAKE_P2_LVDS_FAST }, - .find_pll = intel_ironlake_find_best_PLL, + .p2_slow = IRONLAKE_LVDS_D_SSC_P2_SLOW, + .p2_fast = IRONLAKE_LVDS_D_SSC_P2_FAST }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_ironlake_display_port = { + .dot = { .min = IRONLAKE_DOT_MIN, + .max = IRONLAKE_DOT_MAX }, + .vco = { .min = IRONLAKE_VCO_MIN, + .max = IRONLAKE_VCO_MAX}, + .n = { .min = IRONLAKE_DP_N_MIN, + .max = IRONLAKE_DP_N_MAX }, + .m = { .min = IRONLAKE_DP_M_MIN, + .max = IRONLAKE_DP_M_MAX }, + .m1 = { .min = IRONLAKE_M1_MIN, + .max = IRONLAKE_M1_MAX }, + .m2 = { .min = IRONLAKE_M2_MIN, + .max = IRONLAKE_M2_MAX }, + .p = { .min = IRONLAKE_DP_P_MIN, + .max = IRONLAKE_DP_P_MAX }, + .p1 = { .min = IRONLAKE_DP_P1_MIN, + .max = IRONLAKE_DP_P1_MAX}, + .p2 = { .dot_limit = IRONLAKE_DP_P2_LIMIT, + .p2_slow = IRONLAKE_DP_P2_SLOW, + .p2_fast = IRONLAKE_DP_P2_FAST }, + .find_pll = intel_find_pll_ironlake_dp, }; static const intel_limit_t *intel_ironlake_limit(struct drm_crtc *crtc) { + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; const intel_limit_t *limit; - if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) - limit = &intel_limits_ironlake_lvds; + int refclk = 120; + + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { + if (dev_priv->lvds_use_ssc && dev_priv->lvds_ssc_freq == 100) + refclk = 100; + + if ((I915_READ(PCH_LVDS) & LVDS_CLKB_POWER_MASK) == + LVDS_CLKB_POWER_UP) { + /* LVDS dual channel */ + if (refclk == 100) + limit = &intel_limits_ironlake_dual_lvds_100m; + else + limit = &intel_limits_ironlake_dual_lvds; + } else { + if (refclk == 100) + limit = &intel_limits_ironlake_single_lvds_100m; + else + limit = &intel_limits_ironlake_single_lvds; + } + } else if (intel_pipe_has_type(crtc, INTEL_OUTPUT_DISPLAYPORT) || + HAS_eDP) + limit = &intel_limits_ironlake_display_port; else - limit = &intel_limits_ironlake_sdvo; + limit = &intel_limits_ironlake_dac; return limit; } @@ -557,7 +690,7 @@ static const intel_limit_t *intel_limit(struct drm_crtc *crtc) struct drm_device *dev = crtc->dev; const intel_limit_t *limit; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) limit = intel_ironlake_limit(crtc); else if (IS_G4X(dev)) { limit = intel_g4x_limit(crtc); @@ -737,46 +870,6 @@ intel_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, return (err != target); } - -static bool -intel_find_best_reduced_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, - int target, int refclk, intel_clock_t *best_clock) - -{ - struct drm_device *dev = crtc->dev; - intel_clock_t clock; - int err = target; - bool found = false; - - memcpy(&clock, best_clock, sizeof(intel_clock_t)); - - for (clock.m1 = limit->m1.min; clock.m1 <= limit->m1.max; clock.m1++) { - for (clock.m2 = limit->m2.min; clock.m2 <= limit->m2.max; clock.m2++) { - /* m1 is always 0 in Pineview */ - if (clock.m2 >= clock.m1 && !IS_PINEVIEW(dev)) - break; - for (clock.n = limit->n.min; clock.n <= limit->n.max; - clock.n++) { - int this_err; - - intel_clock(dev, refclk, &clock); - - if (!intel_PLL_is_valid(crtc, &clock)) - continue; - - this_err = abs(clock.dot - target); - if (this_err < err) { - *best_clock = clock; - err = this_err; - found = true; - } - } - } - } - - return found; -} - static bool intel_g4x_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, int target, int refclk, intel_clock_t *best_clock) @@ -791,7 +884,13 @@ intel_g4x_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, found = false; if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { - if ((I915_READ(LVDS) & LVDS_CLKB_POWER_MASK) == + int lvds_reg; + + if (HAS_PCH_SPLIT(dev)) + lvds_reg = PCH_LVDS; + else + lvds_reg = LVDS; + if ((I915_READ(lvds_reg) & LVDS_CLKB_POWER_MASK) == LVDS_CLKB_POWER_UP) clock.p2 = limit->p2.p2_fast; else @@ -839,6 +938,11 @@ intel_find_pll_ironlake_dp(const intel_limit_t *limit, struct drm_crtc *crtc, { struct drm_device *dev = crtc->dev; intel_clock_t clock; + + /* return directly when it is eDP */ + if (HAS_eDP) + return true; + if (target < 200000) { clock.n = 1; clock.p1 = 2; @@ -857,68 +961,6 @@ intel_find_pll_ironlake_dp(const intel_limit_t *limit, struct drm_crtc *crtc, return true; } -static bool -intel_ironlake_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, - int target, int refclk, intel_clock_t *best_clock) -{ - struct drm_device *dev = crtc->dev; - struct drm_i915_private *dev_priv = dev->dev_private; - intel_clock_t clock; - int err_most = 47; - int err_min = 10000; - - /* eDP has only 2 clock choice, no n/m/p setting */ - if (HAS_eDP) - return true; - - if (intel_pipe_has_type(crtc, INTEL_OUTPUT_DISPLAYPORT)) - return intel_find_pll_ironlake_dp(limit, crtc, target, - refclk, best_clock); - - if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { - if ((I915_READ(PCH_LVDS) & LVDS_CLKB_POWER_MASK) == - LVDS_CLKB_POWER_UP) - clock.p2 = limit->p2.p2_fast; - else - clock.p2 = limit->p2.p2_slow; - } else { - if (target < limit->p2.dot_limit) - clock.p2 = limit->p2.p2_slow; - else - clock.p2 = limit->p2.p2_fast; - } - - memset(best_clock, 0, sizeof(*best_clock)); - for (clock.p1 = limit->p1.max; clock.p1 >= limit->p1.min; clock.p1--) { - /* based on hardware requriment prefer smaller n to precision */ - for (clock.n = limit->n.min; clock.n <= limit->n.max; clock.n++) { - /* based on hardware requirment prefere larger m1,m2 */ - for (clock.m1 = limit->m1.max; - clock.m1 >= limit->m1.min; clock.m1--) { - for (clock.m2 = limit->m2.max; - clock.m2 >= limit->m2.min; clock.m2--) { - int this_err; - - intel_clock(dev, refclk, &clock); - if (!intel_PLL_is_valid(crtc, &clock)) - continue; - this_err = abs((10000 - (target*10000/clock.dot))); - if (this_err < err_most) { - *best_clock = clock; - /* found on first matching */ - goto out; - } else if (this_err < err_min) { - *best_clock = clock; - err_min = this_err; - } - } - } - } - } -out: - return true; -} - /* DisplayPort has only two frequencies, 162MHz and 270MHz */ static bool intel_find_pll_g4x_dp(const intel_limit_t *limit, struct drm_crtc *crtc, @@ -989,6 +1031,8 @@ static void i8xx_enable_fbc(struct drm_crtc *crtc, unsigned long interval) /* enable it... */ fbc_ctl = FBC_CTL_EN | FBC_CTL_PERIODIC; + if (IS_I945GM(dev)) + fbc_ctl |= FBC_C3_IDLE; /* 945 needs special SR handling */ fbc_ctl |= (dev_priv->cfb_pitch & 0xff) << FBC_CTL_STRIDE_SHIFT; fbc_ctl |= (interval & 0x2fff) << FBC_CTL_INTERVAL_SHIFT; if (obj_priv->tiling_mode != I915_TILING_NONE) @@ -1144,25 +1188,30 @@ static void intel_update_fbc(struct drm_crtc *crtc, if (intel_fb->obj->size > dev_priv->cfb_size) { DRM_DEBUG_KMS("framebuffer too large, disabling " "compression\n"); + dev_priv->no_fbc_reason = FBC_STOLEN_TOO_SMALL; goto out_disable; } if ((mode->flags & DRM_MODE_FLAG_INTERLACE) || (mode->flags & DRM_MODE_FLAG_DBLSCAN)) { DRM_DEBUG_KMS("mode incompatible with compression, " "disabling\n"); + dev_priv->no_fbc_reason = FBC_UNSUPPORTED_MODE; goto out_disable; } if ((mode->hdisplay > 2048) || (mode->vdisplay > 1536)) { DRM_DEBUG_KMS("mode too large for compression, disabling\n"); + dev_priv->no_fbc_reason = FBC_MODE_TOO_LARGE; goto out_disable; } if ((IS_I915GM(dev) || IS_I945GM(dev)) && plane != 0) { DRM_DEBUG_KMS("plane not 0, disabling compression\n"); + dev_priv->no_fbc_reason = FBC_BAD_PLANE; goto out_disable; } if (obj_priv->tiling_mode != I915_TILING_X) { DRM_DEBUG_KMS("framebuffer not tiled, disabling compression\n"); + dev_priv->no_fbc_reason = FBC_NOT_TILED; goto out_disable; } @@ -1282,7 +1331,7 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, return ret; } - ret = i915_gem_object_set_to_gtt_domain(obj, 1); + ret = i915_gem_object_set_to_display_plane(obj); if (ret != 0) { i915_gem_object_unpin(obj); mutex_unlock(&dev->struct_mutex); @@ -1322,7 +1371,7 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, dspcntr &= ~DISPPLANE_TILED; } - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) /* must disable */ dspcntr |= DISPPLANE_TRICKLE_FEED_DISABLE; @@ -1383,7 +1432,7 @@ static void i915_disable_vga (struct drm_device *dev) u8 sr1; u32 vga_reg; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) vga_reg = CPU_VGACNTRL; else vga_reg = VGACNTRL; @@ -1493,6 +1542,10 @@ static void ironlake_crtc_dpms(struct drm_crtc *crtc, int mode) int trans_vsync_reg = (pipe == 0) ? TRANS_VSYNC_A : TRANS_VSYNC_B; u32 temp; int tries = 5, j, n; + u32 pipe_bpc; + + temp = I915_READ(pipeconf_reg); + pipe_bpc = temp & PIPE_BPC_MASK; /* XXX: When our outputs are all unaware of DPMS modes other than off * and on, we should map those modes to DRM_MODE_DPMS_OFF in the CRTC. @@ -1524,6 +1577,12 @@ static void ironlake_crtc_dpms(struct drm_crtc *crtc, int mode) /* enable PCH FDI RX PLL, wait warmup plus DMI latency */ temp = I915_READ(fdi_rx_reg); + /* + * make the BPC in FDI Rx be consistent with that in + * pipeconf reg. + */ + temp &= ~(0x7 << 16); + temp |= (pipe_bpc << 11); I915_WRITE(fdi_rx_reg, temp | FDI_RX_PLL_ENABLE | FDI_SEL_PCDCLK | FDI_DP_PORT_WIDTH_X4); /* default 4 lanes */ @@ -1666,6 +1725,12 @@ static void ironlake_crtc_dpms(struct drm_crtc *crtc, int mode) /* enable PCH transcoder */ temp = I915_READ(transconf_reg); + /* + * make the BPC in transcoder be consistent with + * that in pipeconf reg. + */ + temp &= ~PIPE_BPC_MASK; + temp |= pipe_bpc; I915_WRITE(transconf_reg, temp | TRANS_ENABLE); I915_READ(transconf_reg); @@ -1697,6 +1762,7 @@ static void ironlake_crtc_dpms(struct drm_crtc *crtc, int mode) case DRM_MODE_DPMS_OFF: DRM_DEBUG_KMS("crtc %d dpms off\n", pipe); + drm_vblank_off(dev, pipe); /* Disable display plane */ temp = I915_READ(dspcntr_reg); if ((temp & DISPLAY_PLANE_ENABLE) != 0) { @@ -1745,6 +1811,9 @@ static void ironlake_crtc_dpms(struct drm_crtc *crtc, int mode) I915_READ(fdi_tx_reg); temp = I915_READ(fdi_rx_reg); + /* BPC in FDI rx is consistent with that in pipeconf */ + temp &= ~(0x07 << 16); + temp |= (pipe_bpc << 11); I915_WRITE(fdi_rx_reg, temp & ~FDI_RX_ENABLE); I915_READ(fdi_rx_reg); @@ -1789,7 +1858,12 @@ static void ironlake_crtc_dpms(struct drm_crtc *crtc, int mode) } } } - + temp = I915_READ(transconf_reg); + /* BPC in transcoder is consistent with that in pipeconf */ + temp &= ~PIPE_BPC_MASK; + temp |= pipe_bpc; + I915_WRITE(transconf_reg, temp); + I915_READ(transconf_reg); udelay(100); /* disable PCH DPLL */ @@ -2042,7 +2116,7 @@ static bool intel_crtc_mode_fixup(struct drm_crtc *crtc, struct drm_display_mode *adjusted_mode) { struct drm_device *dev = crtc->dev; - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { /* FDI link clock is fixed at 2.7G */ if (mode->clock * 3 > 27000 * 4) return MODE_CLOCK_HIGH; @@ -2448,7 +2522,7 @@ static void pineview_enable_cxsr(struct drm_device *dev, unsigned long clock, * A value of 5us seems to be a good balance; safe for very low end * platforms but not overly aggressive on lower latency configs. */ -const static int latency_ns = 5000; +static const int latency_ns = 5000; static int i9xx_get_fifo_size(struct drm_device *dev, int plane) { @@ -2559,7 +2633,7 @@ static void g4x_update_wm(struct drm_device *dev, int planea_clock, /* Calc sr entries for one plane configs */ if (sr_hdisplay && (!planea_clock || !planeb_clock)) { /* self-refresh has much higher latency */ - const static int sr_latency_ns = 12000; + static const int sr_latency_ns = 12000; sr_clock = planea_clock ? planea_clock : planeb_clock; line_time_us = ((sr_hdisplay * 1000) / sr_clock); @@ -2570,6 +2644,10 @@ static void g4x_update_wm(struct drm_device *dev, int planea_clock, sr_entries = roundup(sr_entries / cacheline_size, 1); DRM_DEBUG("self-refresh entries: %d\n", sr_entries); I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN); + } else { + /* Turn off self refresh if both pipes are enabled */ + I915_WRITE(FW_BLC_SELF, I915_READ(FW_BLC_SELF) + & ~FW_BLC_SELF_EN); } DRM_DEBUG("Setting FIFO watermarks - A: %d, B: %d, SR %d\n", @@ -2598,7 +2676,7 @@ static void i965_update_wm(struct drm_device *dev, int planea_clock, /* Calc sr entries for one plane configs */ if (sr_hdisplay && (!planea_clock || !planeb_clock)) { /* self-refresh has much higher latency */ - const static int sr_latency_ns = 12000; + static const int sr_latency_ns = 12000; sr_clock = planea_clock ? planea_clock : planeb_clock; line_time_us = ((sr_hdisplay * 1000) / sr_clock); @@ -2613,6 +2691,10 @@ static void i965_update_wm(struct drm_device *dev, int planea_clock, srwm = 1; srwm &= 0x3f; I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN); + } else { + /* Turn off self refresh if both pipes are enabled */ + I915_WRITE(FW_BLC_SELF, I915_READ(FW_BLC_SELF) + & ~FW_BLC_SELF_EN); } DRM_DEBUG_KMS("Setting FIFO watermarks - A: 8, B: 8, C: 8, SR %d\n", @@ -2667,7 +2749,7 @@ static void i9xx_update_wm(struct drm_device *dev, int planea_clock, if (HAS_FW_BLC(dev) && sr_hdisplay && (!planea_clock || !planeb_clock)) { /* self-refresh has much higher latency */ - const static int sr_latency_ns = 6000; + static const int sr_latency_ns = 6000; sr_clock = planea_clock ? planea_clock : planeb_clock; line_time_us = ((sr_hdisplay * 1000) / sr_clock); @@ -2680,7 +2762,22 @@ static void i9xx_update_wm(struct drm_device *dev, int planea_clock, srwm = total_size - sr_entries; if (srwm < 0) srwm = 1; - I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN | (srwm & 0x3f)); + + if (IS_I945G(dev) || IS_I945GM(dev)) + I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_FIFO_MASK | (srwm & 0xff)); + else if (IS_I915GM(dev)) { + /* 915M has a smaller SRWM field */ + I915_WRITE(FW_BLC_SELF, srwm & 0x3f); + I915_WRITE(INSTPM, I915_READ(INSTPM) | INSTPM_SELF_EN); + } + } else { + /* Turn off self refresh if both pipes are enabled */ + if (IS_I945G(dev) || IS_I945GM(dev)) { + I915_WRITE(FW_BLC_SELF, I915_READ(FW_BLC_SELF) + & ~FW_BLC_SELF_EN); + } else if (IS_I915GM(dev)) { + I915_WRITE(INSTPM, I915_READ(INSTPM) & ~INSTPM_SELF_EN); + } } DRM_DEBUG_KMS("Setting FIFO watermarks - A: %d, B: %d, C: %d, SR %d\n", @@ -2886,7 +2983,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, refclk / 1000); } else if (IS_I9XX(dev)) { refclk = 96000; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) refclk = 120000; /* 120Mhz refclk */ } else { refclk = 48000; @@ -2906,10 +3003,8 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, return -EINVAL; } - if (is_lvds && limit->find_reduced_pll && - dev_priv->lvds_downclock_avail) { - memcpy(&reduced_clock, &clock, sizeof(intel_clock_t)); - has_reduced_clock = limit->find_reduced_pll(limit, crtc, + if (is_lvds && dev_priv->lvds_downclock_avail) { + has_reduced_clock = limit->find_pll(limit, crtc, dev_priv->lvds_downclock, refclk, &reduced_clock); @@ -2946,7 +3041,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, } /* FDI link */ - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { int lane, link_bw, bpp; /* eDP doesn't require FDI link, so just set DP M/N according to current link config */ @@ -2969,6 +3064,33 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, /* determine panel color depth */ temp = I915_READ(pipeconf_reg); + temp &= ~PIPE_BPC_MASK; + if (is_lvds) { + int lvds_reg = I915_READ(PCH_LVDS); + /* the BPC will be 6 if it is 18-bit LVDS panel */ + if ((lvds_reg & LVDS_A3_POWER_MASK) == LVDS_A3_POWER_UP) + temp |= PIPE_8BPC; + else + temp |= PIPE_6BPC; + } else if (is_edp) { + switch (dev_priv->edp_bpp/3) { + case 8: + temp |= PIPE_8BPC; + break; + case 10: + temp |= PIPE_10BPC; + break; + case 6: + temp |= PIPE_6BPC; + break; + case 12: + temp |= PIPE_12BPC; + break; + } + } else + temp |= PIPE_8BPC; + I915_WRITE(pipeconf_reg, temp); + I915_READ(pipeconf_reg); switch (temp & PIPE_BPC_MASK) { case PIPE_8BPC: @@ -2996,7 +3118,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, * PCH B stepping, previous chipset stepping should be * ignoring this setting. */ - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { temp = I915_READ(PCH_DREF_CONTROL); /* Always enable nonspread source */ temp &= ~DREF_NONSPREAD_SOURCE_MASK; @@ -3043,7 +3165,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, reduced_clock.m2; } - if (!IS_IRONLAKE(dev)) + if (!HAS_PCH_SPLIT(dev)) dpll = DPLL_VGA_MODE_DIS; if (IS_I9XX(dev)) { @@ -3056,7 +3178,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, sdvo_pixel_multiply = adjusted_mode->clock / mode->clock; if (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) dpll |= (sdvo_pixel_multiply - 1) << SDVO_MULTIPLIER_SHIFT_HIRES; - else if (IS_IRONLAKE(dev)) + else if (HAS_PCH_SPLIT(dev)) dpll |= (sdvo_pixel_multiply - 1) << PLL_REF_SDVO_HDMI_MULTIPLIER_SHIFT; } if (is_dp) @@ -3068,7 +3190,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, else { dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA01_P1_POST_DIV_SHIFT; /* also FPA1 */ - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA1_P1_POST_DIV_SHIFT; if (IS_G4X(dev) && has_reduced_clock) dpll |= (1 << (reduced_clock.p1 - 1)) << DPLL_FPA1_P1_POST_DIV_SHIFT; @@ -3087,7 +3209,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, dpll |= DPLLB_LVDS_P2_CLOCK_DIV_14; break; } - if (IS_I965G(dev) && !IS_IRONLAKE(dev)) + if (IS_I965G(dev) && !HAS_PCH_SPLIT(dev)) dpll |= (6 << PLL_LOAD_PULSE_PHASE_SHIFT); } else { if (is_lvds) { @@ -3121,7 +3243,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, /* Ironlake's plane is forced to pipe, bit 24 is to enable color space conversion */ - if (!IS_IRONLAKE(dev)) { + if (!HAS_PCH_SPLIT(dev)) { if (pipe == 0) dspcntr &= ~DISPPLANE_SEL_PIPE_MASK; else @@ -3148,14 +3270,14 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, /* Disable the panel fitter if it was on our pipe */ - if (!IS_IRONLAKE(dev) && intel_panel_fitter_pipe(dev) == pipe) + if (!HAS_PCH_SPLIT(dev) && intel_panel_fitter_pipe(dev) == pipe) I915_WRITE(PFIT_CONTROL, 0); DRM_DEBUG_KMS("Mode for pipe %c:\n", pipe == 0 ? 'A' : 'B'); drm_mode_debug_printmodeline(mode); /* assign to Ironlake registers */ - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { fp_reg = pch_fp_reg; dpll_reg = pch_dpll_reg; } @@ -3176,7 +3298,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, if (is_lvds) { u32 lvds; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) lvds_reg = PCH_LVDS; lvds = I915_READ(lvds_reg); @@ -3195,7 +3317,20 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, * appropriately here, but we need to look more thoroughly into how * panels behave in the two modes. */ - + /* set the dithering flag */ + if (IS_I965G(dev)) { + if (dev_priv->lvds_dither) { + if (HAS_PCH_SPLIT(dev)) + pipeconf |= PIPE_ENABLE_DITHER; + else + lvds |= LVDS_ENABLE_DITHER; + } else { + if (HAS_PCH_SPLIT(dev)) + pipeconf &= ~PIPE_ENABLE_DITHER; + else + lvds &= ~LVDS_ENABLE_DITHER; + } + } I915_WRITE(lvds_reg, lvds); I915_READ(lvds_reg); } @@ -3209,7 +3344,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, /* Wait for the clocks to stabilize. */ udelay(150); - if (IS_I965G(dev) && !IS_IRONLAKE(dev)) { + if (IS_I965G(dev) && !HAS_PCH_SPLIT(dev)) { if (is_sdvo) { sdvo_pixel_multiply = adjusted_mode->clock / mode->clock; I915_WRITE(dpll_md_reg, (0 << DPLL_MD_UDI_DIVIDER_SHIFT) | @@ -3256,14 +3391,14 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, /* pipesrc and dspsize control the size that is scaled from, which should * always be the user's requested size. */ - if (!IS_IRONLAKE(dev)) { + if (!HAS_PCH_SPLIT(dev)) { I915_WRITE(dspsize_reg, ((mode->vdisplay - 1) << 16) | (mode->hdisplay - 1)); I915_WRITE(dsppos_reg, 0); } I915_WRITE(pipesrc_reg, ((mode->hdisplay - 1) << 16) | (mode->vdisplay - 1)); - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { I915_WRITE(data_m1_reg, TU_SIZE(m_n.tu) | m_n.gmch_m); I915_WRITE(data_n1_reg, TU_SIZE(m_n.tu) | m_n.gmch_n); I915_WRITE(link_m1_reg, m_n.link_m); @@ -3319,7 +3454,7 @@ void intel_crtc_load_lut(struct drm_crtc *crtc) return; /* use legacy palette for Ironlake */ - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) palreg = (intel_crtc->pipe == 0) ? LGC_PALETTE_A : LGC_PALETTE_B; @@ -3385,7 +3520,7 @@ static int intel_crtc_cursor_set(struct drm_crtc *crtc, /* we only need to pin inside GTT if cursor is non-phy */ mutex_lock(&dev->struct_mutex); - if (!dev_priv->cursor_needs_physical) { + if (!dev_priv->info->cursor_needs_physical) { ret = i915_gem_object_pin(bo, PAGE_SIZE); if (ret) { DRM_ERROR("failed to pin cursor bo\n"); @@ -3420,7 +3555,7 @@ static int intel_crtc_cursor_set(struct drm_crtc *crtc, I915_WRITE(base, addr); if (intel_crtc->cursor_bo) { - if (dev_priv->cursor_needs_physical) { + if (dev_priv->info->cursor_needs_physical) { if (intel_crtc->cursor_bo != bo) i915_gem_detach_phys_object(dev, intel_crtc->cursor_bo); } else @@ -3434,11 +3569,10 @@ static int intel_crtc_cursor_set(struct drm_crtc *crtc, intel_crtc->cursor_bo = bo; return 0; -fail: - mutex_lock(&dev->struct_mutex); fail_locked: - drm_gem_object_unreference(bo); mutex_unlock(&dev->struct_mutex); +fail: + drm_gem_object_unreference_unlocked(bo); return ret; } @@ -3779,125 +3913,6 @@ static void intel_gpu_idle_timer(unsigned long arg) queue_work(dev_priv->wq, &dev_priv->idle_work); } -void intel_increase_renderclock(struct drm_device *dev, bool schedule) -{ - drm_i915_private_t *dev_priv = dev->dev_private; - - if (IS_IRONLAKE(dev)) - return; - - if (!dev_priv->render_reclock_avail) { - DRM_DEBUG_DRIVER("not reclocking render clock\n"); - return; - } - - /* Restore render clock frequency to original value */ - if (IS_G4X(dev) || IS_I9XX(dev)) - pci_write_config_word(dev->pdev, GCFGC, dev_priv->orig_clock); - else if (IS_I85X(dev)) - pci_write_config_word(dev->pdev, HPLLCC, dev_priv->orig_clock); - DRM_DEBUG_DRIVER("increasing render clock frequency\n"); - - /* Schedule downclock */ - if (schedule) - mod_timer(&dev_priv->idle_timer, jiffies + - msecs_to_jiffies(GPU_IDLE_TIMEOUT)); -} - -void intel_decrease_renderclock(struct drm_device *dev) -{ - drm_i915_private_t *dev_priv = dev->dev_private; - - if (IS_IRONLAKE(dev)) - return; - - if (!dev_priv->render_reclock_avail) { - DRM_DEBUG_DRIVER("not reclocking render clock\n"); - return; - } - - if (IS_G4X(dev)) { - u16 gcfgc; - - /* Adjust render clock... */ - pci_read_config_word(dev->pdev, GCFGC, &gcfgc); - - /* Down to minimum... */ - gcfgc &= ~GM45_GC_RENDER_CLOCK_MASK; - gcfgc |= GM45_GC_RENDER_CLOCK_266_MHZ; - - pci_write_config_word(dev->pdev, GCFGC, gcfgc); - } else if (IS_I965G(dev)) { - u16 gcfgc; - - /* Adjust render clock... */ - pci_read_config_word(dev->pdev, GCFGC, &gcfgc); - - /* Down to minimum... */ - gcfgc &= ~I965_GC_RENDER_CLOCK_MASK; - gcfgc |= I965_GC_RENDER_CLOCK_267_MHZ; - - pci_write_config_word(dev->pdev, GCFGC, gcfgc); - } else if (IS_I945G(dev) || IS_I945GM(dev)) { - u16 gcfgc; - - /* Adjust render clock... */ - pci_read_config_word(dev->pdev, GCFGC, &gcfgc); - - /* Down to minimum... */ - gcfgc &= ~I945_GC_RENDER_CLOCK_MASK; - gcfgc |= I945_GC_RENDER_CLOCK_166_MHZ; - - pci_write_config_word(dev->pdev, GCFGC, gcfgc); - } else if (IS_I915G(dev)) { - u16 gcfgc; - - /* Adjust render clock... */ - pci_read_config_word(dev->pdev, GCFGC, &gcfgc); - - /* Down to minimum... */ - gcfgc &= ~I915_GC_RENDER_CLOCK_MASK; - gcfgc |= I915_GC_RENDER_CLOCK_166_MHZ; - - pci_write_config_word(dev->pdev, GCFGC, gcfgc); - } else if (IS_I85X(dev)) { - u16 hpllcc; - - /* Adjust render clock... */ - pci_read_config_word(dev->pdev, HPLLCC, &hpllcc); - - /* Up to maximum... */ - hpllcc &= ~GC_CLOCK_CONTROL_MASK; - hpllcc |= GC_CLOCK_133_200; - - pci_write_config_word(dev->pdev, HPLLCC, hpllcc); - } - DRM_DEBUG_DRIVER("decreasing render clock frequency\n"); -} - -/* Note that no increase function is needed for this - increase_renderclock() - * will also rewrite these bits - */ -void intel_decrease_displayclock(struct drm_device *dev) -{ - if (IS_IRONLAKE(dev)) - return; - - if (IS_I945G(dev) || IS_I945GM(dev) || IS_I915G(dev) || - IS_I915GM(dev)) { - u16 gcfgc; - - /* Adjust render clock... */ - pci_read_config_word(dev->pdev, GCFGC, &gcfgc); - - /* Down to minimum... */ - gcfgc &= ~0xf0; - gcfgc |= 0x80; - - pci_write_config_word(dev->pdev, GCFGC, gcfgc); - } -} - #define CRTC_IDLE_TIMEOUT 1000 /* ms */ static void intel_crtc_idle_timer(unsigned long arg) @@ -3922,7 +3937,7 @@ static void intel_increase_pllclock(struct drm_crtc *crtc, bool schedule) int dpll_reg = (pipe == 0) ? DPLL_A : DPLL_B; int dpll = I915_READ(dpll_reg); - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) return; if (!dev_priv->lvds_downclock_avail) @@ -3961,7 +3976,7 @@ static void intel_decrease_pllclock(struct drm_crtc *crtc) int dpll_reg = (pipe == 0) ? DPLL_A : DPLL_B; int dpll = I915_READ(dpll_reg); - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) return; if (!dev_priv->lvds_downclock_avail) @@ -4011,10 +4026,9 @@ static void intel_idle_update(struct work_struct *work) mutex_lock(&dev->struct_mutex); - /* GPU isn't processing, downclock it. */ - if (!dev_priv->busy) { - intel_decrease_renderclock(dev); - intel_decrease_displayclock(dev); + if (IS_I945G(dev) || IS_I945GM(dev)) { + DRM_DEBUG_DRIVER("enable memory self refresh on 945\n"); + I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN_MASK | FW_BLC_SELF_EN); } list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { @@ -4051,12 +4065,18 @@ void intel_mark_busy(struct drm_device *dev, struct drm_gem_object *obj) return; if (!dev_priv->busy) { + if (IS_I945G(dev) || IS_I945GM(dev)) { + u32 fw_blc_self; + + DRM_DEBUG_DRIVER("disable memory self refresh on 945\n"); + fw_blc_self = I915_READ(FW_BLC_SELF); + fw_blc_self &= ~FW_BLC_SELF_EN; + I915_WRITE(FW_BLC_SELF, fw_blc_self | FW_BLC_SELF_EN_MASK); + } dev_priv->busy = true; - intel_increase_renderclock(dev, true); - } else { + } else mod_timer(&dev_priv->idle_timer, jiffies + msecs_to_jiffies(GPU_IDLE_TIMEOUT)); - } list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { if (!crtc->fb) @@ -4066,6 +4086,14 @@ void intel_mark_busy(struct drm_device *dev, struct drm_gem_object *obj) intel_fb = to_intel_framebuffer(crtc->fb); if (intel_fb->obj == obj) { if (!intel_crtc->busy) { + if (IS_I945G(dev) || IS_I945GM(dev)) { + u32 fw_blc_self; + + DRM_DEBUG_DRIVER("disable memory self refresh on 945\n"); + fw_blc_self = I915_READ(FW_BLC_SELF); + fw_blc_self &= ~FW_BLC_SELF_EN; + I915_WRITE(FW_BLC_SELF, fw_blc_self | FW_BLC_SELF_EN_MASK); + } /* Non-busy -> busy, upclock */ intel_increase_pllclock(crtc, true); intel_crtc->busy = true; @@ -4089,7 +4117,8 @@ static void intel_crtc_destroy(struct drm_crtc *crtc) struct intel_unpin_work { struct work_struct work; struct drm_device *dev; - struct drm_gem_object *obj; + struct drm_gem_object *old_fb_obj; + struct drm_gem_object *pending_flip_obj; struct drm_pending_vblank_event *event; int pending; }; @@ -4100,8 +4129,9 @@ static void intel_unpin_work_fn(struct work_struct *__work) container_of(__work, struct intel_unpin_work, work); mutex_lock(&work->dev->struct_mutex); - i915_gem_object_unpin(work->obj); - drm_gem_object_unreference(work->obj); + i915_gem_object_unpin(work->old_fb_obj); + drm_gem_object_unreference(work->pending_flip_obj); + drm_gem_object_unreference(work->old_fb_obj); mutex_unlock(&work->dev->struct_mutex); kfree(work); } @@ -4124,6 +4154,12 @@ void intel_finish_page_flip(struct drm_device *dev, int pipe) spin_lock_irqsave(&dev->event_lock, flags); work = intel_crtc->unpin_work; if (work == NULL || !work->pending) { + if (work && !work->pending) { + obj_priv = work->pending_flip_obj->driver_private; + DRM_DEBUG_DRIVER("flip finish: %p (%d) not pending?\n", + obj_priv, + atomic_read(&obj_priv->pending_flip)); + } spin_unlock_irqrestore(&dev->event_lock, flags); return; } @@ -4144,8 +4180,11 @@ void intel_finish_page_flip(struct drm_device *dev, int pipe) spin_unlock_irqrestore(&dev->event_lock, flags); - obj_priv = work->obj->driver_private; - if (atomic_dec_and_test(&obj_priv->pending_flip)) + obj_priv = work->pending_flip_obj->driver_private; + + /* Initial scanout buffer will have a 0 pending flip count */ + if ((atomic_read(&obj_priv->pending_flip) == 0) || + atomic_dec_and_test(&obj_priv->pending_flip)) DRM_WAKEUP(&dev_priv->pending_flip_queue); schedule_work(&work->work); } @@ -4158,8 +4197,11 @@ void intel_prepare_page_flip(struct drm_device *dev, int plane) unsigned long flags; spin_lock_irqsave(&dev->event_lock, flags); - if (intel_crtc->unpin_work) + if (intel_crtc->unpin_work) { intel_crtc->unpin_work->pending = 1; + } else { + DRM_DEBUG_DRIVER("preparing flip with no unpin work?\n"); + } spin_unlock_irqrestore(&dev->event_lock, flags); } @@ -4175,7 +4217,8 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc, struct intel_crtc *intel_crtc = to_intel_crtc(crtc); struct intel_unpin_work *work; unsigned long flags; - int ret; + int pipesrc_reg = (intel_crtc->pipe == 0) ? PIPEASRC : PIPEBSRC; + int ret, pipesrc; RING_LOCALS; work = kzalloc(sizeof *work, GFP_KERNEL); @@ -4187,12 +4230,13 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc, work->event = event; work->dev = crtc->dev; intel_fb = to_intel_framebuffer(crtc->fb); - work->obj = intel_fb->obj; + work->old_fb_obj = intel_fb->obj; INIT_WORK(&work->work, intel_unpin_work_fn); /* We borrow the event spin lock for protecting unpin_work */ spin_lock_irqsave(&dev->event_lock, flags); if (intel_crtc->unpin_work) { + DRM_DEBUG_DRIVER("flip queue: crtc already busy\n"); spin_unlock_irqrestore(&dev->event_lock, flags); kfree(work); mutex_unlock(&dev->struct_mutex); @@ -4206,19 +4250,24 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc, ret = intel_pin_and_fence_fb_obj(dev, obj); if (ret != 0) { + DRM_DEBUG_DRIVER("flip queue: %p pin & fence failed\n", + obj->driver_private); kfree(work); + intel_crtc->unpin_work = NULL; mutex_unlock(&dev->struct_mutex); return ret; } - /* Reference the old fb object for the scheduled work. */ - drm_gem_object_reference(work->obj); + /* Reference the objects for the scheduled work. */ + drm_gem_object_reference(work->old_fb_obj); + drm_gem_object_reference(obj); crtc->fb = fb; i915_gem_object_flush_write_domain(obj); drm_vblank_get(dev, intel_crtc->pipe); obj_priv = obj->driver_private; atomic_inc(&obj_priv->pending_flip); + work->pending_flip_obj = obj; BEGIN_LP_RING(4); OUT_RING(MI_DISPLAY_FLIP | @@ -4226,7 +4275,8 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc, OUT_RING(fb->pitch); if (IS_I965G(dev)) { OUT_RING(obj_priv->gtt_offset | obj_priv->tiling_mode); - OUT_RING((fb->width << 16) | fb->height); + pipesrc = I915_READ(pipesrc_reg); + OUT_RING(pipesrc & 0x0fff0fff); } else { OUT_RING(obj_priv->gtt_offset); OUT_RING(MI_NOOP); @@ -4368,7 +4418,7 @@ static void intel_setup_outputs(struct drm_device *dev) if (IS_MOBILE(dev) && !IS_I830(dev)) intel_lvds_init(dev); - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { int found; if (IS_MOBILE(dev) && (I915_READ(DP_A) & DP_DETECTED)) @@ -4400,30 +4450,44 @@ static void intel_setup_outputs(struct drm_device *dev) bool found = false; if (I915_READ(SDVOB) & SDVO_DETECTED) { + DRM_DEBUG_KMS("probing SDVOB\n"); found = intel_sdvo_init(dev, SDVOB); - if (!found && SUPPORTS_INTEGRATED_HDMI(dev)) + if (!found && SUPPORTS_INTEGRATED_HDMI(dev)) { + DRM_DEBUG_KMS("probing HDMI on SDVOB\n"); intel_hdmi_init(dev, SDVOB); + } - if (!found && SUPPORTS_INTEGRATED_DP(dev)) + if (!found && SUPPORTS_INTEGRATED_DP(dev)) { + DRM_DEBUG_KMS("probing DP_B\n"); intel_dp_init(dev, DP_B); + } } /* Before G4X SDVOC doesn't have its own detect register */ - if (I915_READ(SDVOB) & SDVO_DETECTED) + if (I915_READ(SDVOB) & SDVO_DETECTED) { + DRM_DEBUG_KMS("probing SDVOC\n"); found = intel_sdvo_init(dev, SDVOC); + } if (!found && (I915_READ(SDVOC) & SDVO_DETECTED)) { - if (SUPPORTS_INTEGRATED_HDMI(dev)) + if (SUPPORTS_INTEGRATED_HDMI(dev)) { + DRM_DEBUG_KMS("probing HDMI on SDVOC\n"); intel_hdmi_init(dev, SDVOC); - if (SUPPORTS_INTEGRATED_DP(dev)) + } + if (SUPPORTS_INTEGRATED_DP(dev)) { + DRM_DEBUG_KMS("probing DP_C\n"); intel_dp_init(dev, DP_C); + } } - if (SUPPORTS_INTEGRATED_DP(dev) && (I915_READ(DP_D) & DP_DETECTED)) + if (SUPPORTS_INTEGRATED_DP(dev) && + (I915_READ(DP_D) & DP_DETECTED)) { + DRM_DEBUG_KMS("probing DP_D\n"); intel_dp_init(dev, DP_D); - } else if (IS_I8XX(dev)) + } + } else if (IS_GEN2(dev)) intel_dvo_init(dev); if (SUPPORTS_TV(dev)) @@ -4448,9 +4512,7 @@ static void intel_user_framebuffer_destroy(struct drm_framebuffer *fb) intelfb_remove(dev, fb); drm_framebuffer_cleanup(fb); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(intel_fb->obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(intel_fb->obj); kfree(intel_fb); } @@ -4513,9 +4575,7 @@ intel_user_framebuffer_create(struct drm_device *dev, ret = intel_framebuffer_create(dev, mode_cmd, &fb, obj); if (ret) { - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(obj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(obj); return NULL; } @@ -4527,6 +4587,127 @@ static const struct drm_mode_config_funcs intel_mode_funcs = { .fb_changed = intelfb_probe, }; +static struct drm_gem_object * +intel_alloc_power_context(struct drm_device *dev) +{ + struct drm_gem_object *pwrctx; + int ret; + + pwrctx = drm_gem_object_alloc(dev, 4096); + if (!pwrctx) { + DRM_DEBUG("failed to alloc power context, RC6 disabled\n"); + return NULL; + } + + mutex_lock(&dev->struct_mutex); + ret = i915_gem_object_pin(pwrctx, 4096); + if (ret) { + DRM_ERROR("failed to pin power context: %d\n", ret); + goto err_unref; + } + + ret = i915_gem_object_set_to_gtt_domain(pwrctx, 1); + if (ret) { + DRM_ERROR("failed to set-domain on power context: %d\n", ret); + goto err_unpin; + } + mutex_unlock(&dev->struct_mutex); + + return pwrctx; + +err_unpin: + i915_gem_object_unpin(pwrctx); +err_unref: + drm_gem_object_unreference(pwrctx); + mutex_unlock(&dev->struct_mutex); + return NULL; +} + +void ironlake_enable_drps(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 rgvmodectl = I915_READ(MEMMODECTL), rgvswctl; + u8 fmax, fmin, fstart, vstart; + int i = 0; + + /* 100ms RC evaluation intervals */ + I915_WRITE(RCUPEI, 100000); + I915_WRITE(RCDNEI, 100000); + + /* Set max/min thresholds to 90ms and 80ms respectively */ + I915_WRITE(RCBMAXAVG, 90000); + I915_WRITE(RCBMINAVG, 80000); + + I915_WRITE(MEMIHYST, 1); + + /* Set up min, max, and cur for interrupt handling */ + fmax = (rgvmodectl & MEMMODE_FMAX_MASK) >> MEMMODE_FMAX_SHIFT; + fmin = (rgvmodectl & MEMMODE_FMIN_MASK); + fstart = (rgvmodectl & MEMMODE_FSTART_MASK) >> + MEMMODE_FSTART_SHIFT; + vstart = (I915_READ(PXVFREQ_BASE + (fstart * 4)) & PXVFREQ_PX_MASK) >> + PXVFREQ_PX_SHIFT; + + dev_priv->max_delay = fstart; /* can't go to fmax w/o IPS */ + dev_priv->min_delay = fmin; + dev_priv->cur_delay = fstart; + + I915_WRITE(MEMINTREN, MEMINT_CX_SUPR_EN | MEMINT_EVAL_CHG_EN); + + /* + * Interrupts will be enabled in ironlake_irq_postinstall + */ + + I915_WRITE(VIDSTART, vstart); + POSTING_READ(VIDSTART); + + rgvmodectl |= MEMMODE_SWMODE_EN; + I915_WRITE(MEMMODECTL, rgvmodectl); + + while (I915_READ(MEMSWCTL) & MEMCTL_CMD_STS) { + if (i++ > 100) { + DRM_ERROR("stuck trying to change perf mode\n"); + break; + } + msleep(1); + } + msleep(1); + + rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) | + (fstart << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM; + I915_WRITE(MEMSWCTL, rgvswctl); + POSTING_READ(MEMSWCTL); + + rgvswctl |= MEMCTL_CMD_STS; + I915_WRITE(MEMSWCTL, rgvswctl); +} + +void ironlake_disable_drps(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 rgvswctl; + u8 fstart; + + /* Ack interrupts, disable EFC interrupt */ + I915_WRITE(MEMINTREN, I915_READ(MEMINTREN) & ~MEMINT_EVAL_CHG_EN); + I915_WRITE(MEMINTRSTS, MEMINT_EVAL_CHG); + I915_WRITE(DEIER, I915_READ(DEIER) & ~DE_PCU_EVENT); + I915_WRITE(DEIIR, DE_PCU_EVENT); + I915_WRITE(DEIMR, I915_READ(DEIMR) | DE_PCU_EVENT); + + /* Go back to the starting frequency */ + fstart = (I915_READ(MEMMODECTL) & MEMMODE_FSTART_MASK) >> + MEMMODE_FSTART_SHIFT; + rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) | + (fstart << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM; + I915_WRITE(MEMSWCTL, rgvswctl); + msleep(1); + rgvswctl |= MEMCTL_CMD_STS; + I915_WRITE(MEMSWCTL, rgvswctl); + msleep(1); + +} + void intel_init_clock_gating(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -4535,7 +4716,7 @@ void intel_init_clock_gating(struct drm_device *dev) * Disable clock gating reported to work incorrectly according to the * specs, but enable as much else as we can. */ - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { return; } else if (IS_G4X(dev)) { uint32_t dspclk_gate; @@ -4579,42 +4760,27 @@ void intel_init_clock_gating(struct drm_device *dev) * GPU can automatically power down the render unit if given a page * to save state. */ - if (I915_HAS_RC6(dev)) { - struct drm_gem_object *pwrctx; - struct drm_i915_gem_object *obj_priv; - int ret; + if (I915_HAS_RC6(dev) && drm_core_check_feature(dev, DRIVER_MODESET)) { + struct drm_i915_gem_object *obj_priv = NULL; if (dev_priv->pwrctx) { obj_priv = dev_priv->pwrctx->driver_private; } else { - pwrctx = drm_gem_object_alloc(dev, 4096); - if (!pwrctx) { - DRM_DEBUG("failed to alloc power context, " - "RC6 disabled\n"); - goto out; - } + struct drm_gem_object *pwrctx; - ret = i915_gem_object_pin(pwrctx, 4096); - if (ret) { - DRM_ERROR("failed to pin power context: %d\n", - ret); - drm_gem_object_unreference(pwrctx); - goto out; + pwrctx = intel_alloc_power_context(dev); + if (pwrctx) { + dev_priv->pwrctx = pwrctx; + obj_priv = pwrctx->driver_private; } - - i915_gem_object_set_to_gtt_domain(pwrctx, 1); - - dev_priv->pwrctx = pwrctx; - obj_priv = pwrctx->driver_private; } - I915_WRITE(PWRCTXA, obj_priv->gtt_offset | PWRCTX_EN); - I915_WRITE(MCHBAR_RENDER_STANDBY, - I915_READ(MCHBAR_RENDER_STANDBY) & ~RCX_SW_EXIT); + if (obj_priv) { + I915_WRITE(PWRCTXA, obj_priv->gtt_offset | PWRCTX_EN); + I915_WRITE(MCHBAR_RENDER_STANDBY, + I915_READ(MCHBAR_RENDER_STANDBY) & ~RCX_SW_EXIT); + } } - -out: - return; } /* Set up chip specific display functions */ @@ -4623,7 +4789,7 @@ static void intel_init_display(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; /* We always want a DPMS function */ - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) dev_priv->display.dpms = ironlake_crtc_dpms; else dev_priv->display.dpms = i9xx_crtc_dpms; @@ -4666,7 +4832,7 @@ static void intel_init_display(struct drm_device *dev) i830_get_display_clock_speed; /* For FIFO watermark updates */ - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) dev_priv->display.update_wm = NULL; else if (IS_G4X(dev)) dev_priv->display.update_wm = g4x_update_wm; @@ -4725,11 +4891,6 @@ void intel_modeset_init(struct drm_device *dev) DRM_DEBUG_KMS("%d display pipe%s available.\n", num_pipe, num_pipe > 1 ? "s" : ""); - if (IS_I85X(dev)) - pci_read_config_word(dev->pdev, HPLLCC, &dev_priv->orig_clock); - else if (IS_I9XX(dev) || IS_G4X(dev)) - pci_read_config_word(dev->pdev, GCFGC, &dev_priv->orig_clock); - for (i = 0; i < num_pipe; i++) { intel_crtc_init(dev, i); } @@ -4738,6 +4899,9 @@ void intel_modeset_init(struct drm_device *dev) intel_init_clock_gating(dev); + if (IS_IRONLAKE_M(dev)) + ironlake_enable_drps(dev); + INIT_WORK(&dev_priv->idle_work, intel_idle_update); setup_timer(&dev_priv->idle_timer, intel_gpu_idle_timer, (unsigned long)dev); @@ -4770,7 +4934,6 @@ void intel_modeset_cleanup(struct drm_device *dev) del_timer_sync(&intel_crtc->idle_timer); } - intel_increase_renderclock(dev, false); del_timer_sync(&dev_priv->idle_timer); if (dev_priv->display.disable_fbc) @@ -4786,6 +4949,9 @@ void intel_modeset_cleanup(struct drm_device *dev) drm_gem_object_unreference(dev_priv->pwrctx); } + if (IS_IRONLAKE_M(dev)) + ironlake_disable_drps(dev); + mutex_unlock(&dev->struct_mutex); drm_mode_config_cleanup(dev); diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c index 4e7aa8b7b938..3ef3a0d0edd0 100644 --- a/drivers/gpu/drm/i915/intel_dp.c +++ b/drivers/gpu/drm/i915/intel_dp.c @@ -125,9 +125,15 @@ intel_dp_link_clock(uint8_t link_bw) /* I think this is a fiction */ static int -intel_dp_link_required(int pixel_clock) +intel_dp_link_required(struct drm_device *dev, + struct intel_output *intel_output, int pixel_clock) { - return pixel_clock * 3; + struct drm_i915_private *dev_priv = dev->dev_private; + + if (IS_eDP(intel_output)) + return (pixel_clock * dev_priv->edp_bpp) / 8; + else + return pixel_clock * 3; } static int @@ -138,7 +144,8 @@ intel_dp_mode_valid(struct drm_connector *connector, int max_link_clock = intel_dp_link_clock(intel_dp_max_link_bw(intel_output)); int max_lanes = intel_dp_max_lane_count(intel_output); - if (intel_dp_link_required(mode->clock) > max_link_clock * max_lanes) + if (intel_dp_link_required(connector->dev, intel_output, mode->clock) + > max_link_clock * max_lanes) return MODE_CLOCK_HIGH; if (mode->clock < 10000) @@ -224,7 +231,7 @@ intel_dp_aux_ch(struct intel_output *intel_output, */ if (IS_eDP(intel_output)) aux_clock_divider = 225; /* eDP input clock at 450Mhz */ - else if (IS_IRONLAKE(dev)) + else if (HAS_PCH_SPLIT(dev)) aux_clock_divider = 62; /* IRL input clock fixed at 125Mhz */ else aux_clock_divider = intel_hrawclk(dev) / 2; @@ -492,7 +499,8 @@ intel_dp_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, for (clock = 0; clock <= max_clock; clock++) { int link_avail = intel_dp_link_clock(bws[clock]) * lane_count; - if (intel_dp_link_required(mode->clock) <= link_avail) { + if (intel_dp_link_required(encoder->dev, intel_output, mode->clock) + <= link_avail) { dp_priv->link_bw = bws[clock]; dp_priv->lane_count = lane_count; adjusted_mode->clock = intel_dp_link_clock(dp_priv->link_bw); @@ -576,7 +584,7 @@ intel_dp_set_m_n(struct drm_crtc *crtc, struct drm_display_mode *mode, intel_dp_compute_m_n(3, lane_count, mode->clock, adjusted_mode->clock, &m_n); - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { if (intel_crtc->pipe == 0) { I915_WRITE(TRANSA_DATA_M1, ((m_n.tu - 1) << PIPE_GMCH_DATA_M_TU_SIZE_SHIFT) | @@ -1168,7 +1176,7 @@ intel_dp_detect(struct drm_connector *connector) dp_priv->has_audio = false; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) return ironlake_dp_detect(connector); temp = I915_READ(PORT_HOTPLUG_EN); @@ -1289,53 +1297,7 @@ intel_dp_hot_plug(struct intel_output *intel_output) if (dp_priv->dpms_mode == DRM_MODE_DPMS_ON) intel_dp_check_link_status(intel_output); } -/* - * Enumerate the child dev array parsed from VBT to check whether - * the given DP is present. - * If it is present, return 1. - * If it is not present, return false. - * If no child dev is parsed from VBT, it is assumed that the given - * DP is present. - */ -static int dp_is_present_in_vbt(struct drm_device *dev, int dp_reg) -{ - struct drm_i915_private *dev_priv = dev->dev_private; - struct child_device_config *p_child; - int i, dp_port, ret; - - if (!dev_priv->child_dev_num) - return 1; - - dp_port = 0; - if (dp_reg == DP_B || dp_reg == PCH_DP_B) - dp_port = PORT_IDPB; - else if (dp_reg == DP_C || dp_reg == PCH_DP_C) - dp_port = PORT_IDPC; - else if (dp_reg == DP_D || dp_reg == PCH_DP_D) - dp_port = PORT_IDPD; - - ret = 0; - for (i = 0; i < dev_priv->child_dev_num; i++) { - p_child = dev_priv->child_dev + i; - /* - * If the device type is not DP, continue. - */ - if (p_child->device_type != DEVICE_TYPE_DP && - p_child->device_type != DEVICE_TYPE_eDP) - continue; - /* Find the eDP port */ - if (dp_reg == DP_A && p_child->device_type == DEVICE_TYPE_eDP) { - ret = 1; - break; - } - /* Find the DP port */ - if (p_child->dvo_port == dp_port) { - ret = 1; - break; - } - } - return ret; -} + void intel_dp_init(struct drm_device *dev, int output_reg) { @@ -1345,10 +1307,6 @@ intel_dp_init(struct drm_device *dev, int output_reg) struct intel_dp_priv *dp_priv; const char *name = NULL; - if (!dp_is_present_in_vbt(dev, output_reg)) { - DRM_DEBUG_KMS("DP is not present. Ignore it\n"); - return; - } intel_output = kcalloc(sizeof(struct intel_output) + sizeof(struct intel_dp_priv), 1, GFP_KERNEL); if (!intel_output) @@ -1373,11 +1331,10 @@ intel_dp_init(struct drm_device *dev, int output_reg) else if (output_reg == DP_D || output_reg == PCH_DP_D) intel_output->clone_mask = (1 << INTEL_DP_D_CLONE_BIT); - if (IS_eDP(intel_output)) { - intel_output->crtc_mask = (1 << 1); + if (IS_eDP(intel_output)) intel_output->clone_mask = (1 << INTEL_EDP_CLONE_BIT); - } else - intel_output->crtc_mask = (1 << 0) | (1 << 1); + + intel_output->crtc_mask = (1 << 0) | (1 << 1); connector->interlace_allowed = true; connector->doublescan_allowed = 0; @@ -1402,14 +1359,20 @@ intel_dp_init(struct drm_device *dev, int output_reg) break; case DP_B: case PCH_DP_B: + dev_priv->hotplug_supported_mask |= + HDMIB_HOTPLUG_INT_STATUS; name = "DPDDC-B"; break; case DP_C: case PCH_DP_C: + dev_priv->hotplug_supported_mask |= + HDMIC_HOTPLUG_INT_STATUS; name = "DPDDC-C"; break; case DP_D: case PCH_DP_D: + dev_priv->hotplug_supported_mask |= + HDMID_HOTPLUG_INT_STATUS; name = "DPDDC-D"; break; } diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h index a51573da1ff6..3a467ca57857 100644 --- a/drivers/gpu/drm/i915/intel_drv.h +++ b/drivers/gpu/drm/i915/intel_drv.h @@ -209,6 +209,8 @@ extern void intel_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, extern void intel_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, u16 *blue, int regno); extern void intel_init_clock_gating(struct drm_device *dev); +extern void ironlake_enable_drps(struct drm_device *dev); +extern void ironlake_disable_drps(struct drm_device *dev); extern int intel_framebuffer_create(struct drm_device *dev, struct drm_mode_fb_cmd *mode_cmd, diff --git a/drivers/gpu/drm/i915/intel_fb.c b/drivers/gpu/drm/i915/intel_fb.c index d4823cc87895..aaabbcbe5905 100644 --- a/drivers/gpu/drm/i915/intel_fb.c +++ b/drivers/gpu/drm/i915/intel_fb.c @@ -70,7 +70,7 @@ static struct drm_fb_helper_funcs intel_fb_helper_funcs = { /** - * Curretly it is assumed that the old framebuffer is reused. + * Currently it is assumed that the old framebuffer is reused. * * LOCKING * caller should hold the mode config lock. @@ -148,7 +148,7 @@ static int intelfb_create(struct drm_device *dev, uint32_t fb_width, mutex_lock(&dev->struct_mutex); - ret = i915_gem_object_pin(fbo, PAGE_SIZE); + ret = i915_gem_object_pin(fbo, 64*1024); if (ret) { DRM_ERROR("failed to pin fb: %d\n", ret); goto out_unref; diff --git a/drivers/gpu/drm/i915/intel_hdmi.c b/drivers/gpu/drm/i915/intel_hdmi.c index f04dbbe7d400..a30f8bfc1985 100644 --- a/drivers/gpu/drm/i915/intel_hdmi.c +++ b/drivers/gpu/drm/i915/intel_hdmi.c @@ -82,7 +82,7 @@ static void intel_hdmi_dpms(struct drm_encoder *encoder, int mode) /* HW workaround, need to toggle enable bit off and on for 12bpc, but * we do this anyway which shows more stable in testing. */ - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { I915_WRITE(hdmi_priv->sdvox_reg, temp & ~SDVO_ENABLE); POSTING_READ(hdmi_priv->sdvox_reg); } @@ -99,7 +99,7 @@ static void intel_hdmi_dpms(struct drm_encoder *encoder, int mode) /* HW workaround, need to write this twice for issue that may result * in first write getting masked. */ - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { I915_WRITE(hdmi_priv->sdvox_reg, temp); POSTING_READ(hdmi_priv->sdvox_reg); } @@ -225,52 +225,6 @@ static const struct drm_encoder_funcs intel_hdmi_enc_funcs = { .destroy = intel_hdmi_enc_destroy, }; -/* - * Enumerate the child dev array parsed from VBT to check whether - * the given HDMI is present. - * If it is present, return 1. - * If it is not present, return false. - * If no child dev is parsed from VBT, it assumes that the given - * HDMI is present. - */ -static int hdmi_is_present_in_vbt(struct drm_device *dev, int hdmi_reg) -{ - struct drm_i915_private *dev_priv = dev->dev_private; - struct child_device_config *p_child; - int i, hdmi_port, ret; - - if (!dev_priv->child_dev_num) - return 1; - - if (hdmi_reg == SDVOB) - hdmi_port = DVO_B; - else if (hdmi_reg == SDVOC) - hdmi_port = DVO_C; - else if (hdmi_reg == HDMIB) - hdmi_port = DVO_B; - else if (hdmi_reg == HDMIC) - hdmi_port = DVO_C; - else if (hdmi_reg == HDMID) - hdmi_port = DVO_D; - else - return 0; - - ret = 0; - for (i = 0; i < dev_priv->child_dev_num; i++) { - p_child = dev_priv->child_dev + i; - /* - * If the device type is not HDMI, continue. - */ - if (p_child->device_type != DEVICE_TYPE_HDMI) - continue; - /* Find the HDMI port */ - if (p_child->dvo_port == hdmi_port) { - ret = 1; - break; - } - } - return ret; -} void intel_hdmi_init(struct drm_device *dev, int sdvox_reg) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -278,10 +232,6 @@ void intel_hdmi_init(struct drm_device *dev, int sdvox_reg) struct intel_output *intel_output; struct intel_hdmi_priv *hdmi_priv; - if (!hdmi_is_present_in_vbt(dev, sdvox_reg)) { - DRM_DEBUG_KMS("HDMI is not present. Ignored it \n"); - return; - } intel_output = kcalloc(sizeof(struct intel_output) + sizeof(struct intel_hdmi_priv), 1, GFP_KERNEL); if (!intel_output) @@ -303,21 +253,26 @@ void intel_hdmi_init(struct drm_device *dev, int sdvox_reg) if (sdvox_reg == SDVOB) { intel_output->clone_mask = (1 << INTEL_HDMIB_CLONE_BIT); intel_output->ddc_bus = intel_i2c_create(dev, GPIOE, "HDMIB"); + dev_priv->hotplug_supported_mask |= HDMIB_HOTPLUG_INT_STATUS; } else if (sdvox_reg == SDVOC) { intel_output->clone_mask = (1 << INTEL_HDMIC_CLONE_BIT); intel_output->ddc_bus = intel_i2c_create(dev, GPIOD, "HDMIC"); + dev_priv->hotplug_supported_mask |= HDMIC_HOTPLUG_INT_STATUS; } else if (sdvox_reg == HDMIB) { intel_output->clone_mask = (1 << INTEL_HDMID_CLONE_BIT); intel_output->ddc_bus = intel_i2c_create(dev, PCH_GPIOE, "HDMIB"); + dev_priv->hotplug_supported_mask |= HDMIB_HOTPLUG_INT_STATUS; } else if (sdvox_reg == HDMIC) { intel_output->clone_mask = (1 << INTEL_HDMIE_CLONE_BIT); intel_output->ddc_bus = intel_i2c_create(dev, PCH_GPIOD, "HDMIC"); + dev_priv->hotplug_supported_mask |= HDMIC_HOTPLUG_INT_STATUS; } else if (sdvox_reg == HDMID) { intel_output->clone_mask = (1 << INTEL_HDMIF_CLONE_BIT); intel_output->ddc_bus = intel_i2c_create(dev, PCH_GPIOF, "HDMID"); + dev_priv->hotplug_supported_mask |= HDMID_HOTPLUG_INT_STATUS; } if (!intel_output->ddc_bus) goto err_connector; diff --git a/drivers/gpu/drm/i915/intel_i2c.c b/drivers/gpu/drm/i915/intel_i2c.c index 8673c735b8ab..fcc753ca5d94 100644 --- a/drivers/gpu/drm/i915/intel_i2c.c +++ b/drivers/gpu/drm/i915/intel_i2c.c @@ -128,7 +128,7 @@ intel_i2c_reset_gmbus(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { I915_WRITE(PCH_GMBUS0, 0); } else { I915_WRITE(GMBUS0, 0); diff --git a/drivers/gpu/drm/i915/intel_lvds.c b/drivers/gpu/drm/i915/intel_lvds.c index b04d1e63d439..14e516fdc2dd 100644 --- a/drivers/gpu/drm/i915/intel_lvds.c +++ b/drivers/gpu/drm/i915/intel_lvds.c @@ -56,7 +56,7 @@ static void intel_lvds_set_backlight(struct drm_device *dev, int level) struct drm_i915_private *dev_priv = dev->dev_private; u32 blc_pwm_ctl, reg; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) reg = BLC_PWM_CPU_CTL; else reg = BLC_PWM_CTL; @@ -74,7 +74,7 @@ static u32 intel_lvds_get_max_backlight(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; u32 reg; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) reg = BLC_PWM_PCH_CTL2; else reg = BLC_PWM_CTL; @@ -89,17 +89,22 @@ static u32 intel_lvds_get_max_backlight(struct drm_device *dev) static void intel_lvds_set_power(struct drm_device *dev, bool on) { struct drm_i915_private *dev_priv = dev->dev_private; - u32 pp_status, ctl_reg, status_reg; + u32 pp_status, ctl_reg, status_reg, lvds_reg; - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { ctl_reg = PCH_PP_CONTROL; status_reg = PCH_PP_STATUS; + lvds_reg = PCH_LVDS; } else { ctl_reg = PP_CONTROL; status_reg = PP_STATUS; + lvds_reg = LVDS; } if (on) { + I915_WRITE(lvds_reg, I915_READ(lvds_reg) | LVDS_PORT_EN); + POSTING_READ(lvds_reg); + I915_WRITE(ctl_reg, I915_READ(ctl_reg) | POWER_TARGET_ON); do { @@ -115,6 +120,9 @@ static void intel_lvds_set_power(struct drm_device *dev, bool on) do { pp_status = I915_READ(status_reg); } while (pp_status & PP_ON); + + I915_WRITE(lvds_reg, I915_READ(lvds_reg) & ~LVDS_PORT_EN); + POSTING_READ(lvds_reg); } } @@ -137,7 +145,7 @@ static void intel_lvds_save(struct drm_connector *connector) u32 pp_on_reg, pp_off_reg, pp_ctl_reg, pp_div_reg; u32 pwm_ctl_reg; - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { pp_on_reg = PCH_PP_ON_DELAYS; pp_off_reg = PCH_PP_OFF_DELAYS; pp_ctl_reg = PCH_PP_CONTROL; @@ -174,7 +182,7 @@ static void intel_lvds_restore(struct drm_connector *connector) u32 pp_on_reg, pp_off_reg, pp_ctl_reg, pp_div_reg; u32 pwm_ctl_reg; - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { pp_on_reg = PCH_PP_ON_DELAYS; pp_off_reg = PCH_PP_OFF_DELAYS; pp_ctl_reg = PCH_PP_CONTROL; @@ -297,7 +305,7 @@ static bool intel_lvds_mode_fixup(struct drm_encoder *encoder, } /* full screen scale for now */ - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) goto out; /* 965+ wants fuzzy fitting */ @@ -327,7 +335,7 @@ static bool intel_lvds_mode_fixup(struct drm_encoder *encoder, * to register description and PRM. * Change the value here to see the borders for debugging */ - if (!IS_IRONLAKE(dev)) { + if (!HAS_PCH_SPLIT(dev)) { I915_WRITE(BCLRPAT_A, 0); I915_WRITE(BCLRPAT_B, 0); } @@ -548,7 +556,7 @@ static void intel_lvds_prepare(struct drm_encoder *encoder) struct drm_i915_private *dev_priv = dev->dev_private; u32 reg; - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) reg = BLC_PWM_CPU_CTL; else reg = BLC_PWM_CTL; @@ -587,7 +595,7 @@ static void intel_lvds_mode_set(struct drm_encoder *encoder, * settings. */ - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) return; /* @@ -602,12 +610,47 @@ static void intel_lvds_mode_set(struct drm_encoder *encoder, /* Some lid devices report incorrect lid status, assume they're connected */ static const struct dmi_system_id bad_lid_status[] = { { + .ident = "Compaq nx9020", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"), + DMI_MATCH(DMI_BOARD_NAME, "3084"), + }, + }, + { + .ident = "Samsung SX20S", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Samsung Electronics"), + DMI_MATCH(DMI_BOARD_NAME, "SX20S"), + }, + }, + { .ident = "Aspire One", .matches = { DMI_MATCH(DMI_SYS_VENDOR, "Acer"), DMI_MATCH(DMI_PRODUCT_NAME, "Aspire one"), }, }, + { + .ident = "Aspire 1810T", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Acer"), + DMI_MATCH(DMI_PRODUCT_NAME, "Aspire 1810T"), + }, + }, + { + .ident = "PC-81005", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "MALATA"), + DMI_MATCH(DMI_PRODUCT_NAME, "PC-81005"), + }, + }, + { + .ident = "Clevo M5x0N", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "CLEVO Co."), + DMI_MATCH(DMI_BOARD_NAME, "M5x0N"), + }, + }, { } }; @@ -620,9 +663,16 @@ static const struct dmi_system_id bad_lid_status[] = { */ static enum drm_connector_status intel_lvds_detect(struct drm_connector *connector) { + struct drm_device *dev = connector->dev; enum drm_connector_status status = connector_status_connected; - if (!acpi_lid_open() && !dmi_check_system(bad_lid_status)) + /* ACPI lid methods were generally unreliable in this generation, so + * don't even bother. + */ + if (IS_GEN2(dev)) + return connector_status_connected; + + if (!dmi_check_system(bad_lid_status) && !acpi_lid_open()) status = connector_status_disconnected; return status; @@ -679,7 +729,14 @@ static int intel_lid_notify(struct notifier_block *nb, unsigned long val, struct drm_i915_private *dev_priv = container_of(nb, struct drm_i915_private, lid_notifier); struct drm_device *dev = dev_priv->dev; + struct drm_connector *connector = dev_priv->int_lvds_connector; + /* + * check and update the status of LVDS connector after receiving + * the LID nofication event. + */ + if (connector) + connector->status = connector->funcs->detect(connector); if (!acpi_lid_open()) { dev_priv->modeset_on_lid = 1; return NOTIFY_OK; @@ -854,65 +911,6 @@ static const struct dmi_system_id intel_no_lvds[] = { { } /* terminating entry */ }; -#ifdef CONFIG_ACPI -/* - * check_lid_device -- check whether @handle is an ACPI LID device. - * @handle: ACPI device handle - * @level : depth in the ACPI namespace tree - * @context: the number of LID device when we find the device - * @rv: a return value to fill if desired (Not use) - */ -static acpi_status -check_lid_device(acpi_handle handle, u32 level, void *context, - void **return_value) -{ - struct acpi_device *acpi_dev; - int *lid_present = context; - - acpi_dev = NULL; - /* Get the acpi device for device handle */ - if (acpi_bus_get_device(handle, &acpi_dev) || !acpi_dev) { - /* If there is no ACPI device for handle, return */ - return AE_OK; - } - - if (!strncmp(acpi_device_hid(acpi_dev), "PNP0C0D", 7)) - *lid_present = 1; - - return AE_OK; -} - -/** - * check whether there exists the ACPI LID device by enumerating the ACPI - * device tree. - */ -static int intel_lid_present(void) -{ - int lid_present = 0; - - if (acpi_disabled) { - /* If ACPI is disabled, there is no ACPI device tree to - * check, so assume the LID device would have been present. - */ - return 1; - } - - acpi_walk_namespace(ACPI_TYPE_DEVICE, ACPI_ROOT_OBJECT, - ACPI_UINT32_MAX, - check_lid_device, &lid_present, NULL); - - return lid_present; -} -#else -static int intel_lid_present(void) -{ - /* In the absence of ACPI built in, assume that the LID device would - * have been present. - */ - return 1; -} -#endif - /** * intel_find_lvds_downclock - find the reduced downclock for LVDS in EDID * @dev: drm device @@ -957,7 +955,8 @@ static void intel_find_lvds_downclock(struct drm_device *dev, } } mutex_unlock(&dev->mode_config.mutex); - if (temp_downclock < panel_fixed_mode->clock) { + if (temp_downclock < panel_fixed_mode->clock && + i915_lvds_downclock) { /* We found the downclock for LVDS. */ dev_priv->lvds_downclock_avail = 1; dev_priv->lvds_downclock = temp_downclock; @@ -1031,16 +1030,12 @@ void intel_lvds_init(struct drm_device *dev) if (dmi_check_system(intel_no_lvds)) return; - /* - * Assume LVDS is present if there's an ACPI lid device or if the - * device is present in the VBT. - */ - if (!lvds_is_present_in_vbt(dev) && !intel_lid_present()) { - DRM_DEBUG_KMS("LVDS is not present in VBT and no lid detected\n"); + if (!lvds_is_present_in_vbt(dev)) { + DRM_DEBUG_KMS("LVDS is not present in VBT\n"); return; } - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { if ((I915_READ(PCH_LVDS) & LVDS_DETECTED) == 0) return; if (dev_priv->edp_support) { @@ -1143,7 +1138,7 @@ void intel_lvds_init(struct drm_device *dev) */ /* Ironlake: FIXME if still fail, not try pipe mode now */ - if (IS_IRONLAKE(dev)) + if (HAS_PCH_SPLIT(dev)) goto failed; lvds = I915_READ(LVDS); @@ -1164,7 +1159,7 @@ void intel_lvds_init(struct drm_device *dev) goto failed; out: - if (IS_IRONLAKE(dev)) { + if (HAS_PCH_SPLIT(dev)) { u32 pwm; /* make sure PWM is enabled */ pwm = I915_READ(BLC_PWM_CPU_CTL2); @@ -1180,6 +1175,8 @@ out: DRM_DEBUG_KMS("lid notifier registration failed\n"); dev_priv->lid_notifier.notifier_call = NULL; } + /* keep the LVDS connector */ + dev_priv->int_lvds_connector = connector; drm_sysfs_connector_add(connector); return; diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c index 2639591c72e9..d355d1d527e7 100644 --- a/drivers/gpu/drm/i915/intel_overlay.c +++ b/drivers/gpu/drm/i915/intel_overlay.c @@ -172,7 +172,7 @@ struct overlay_registers { #define OFC_UPDATE 0x1 #define OVERLAY_NONPHYSICAL(dev) (IS_G33(dev) || IS_I965G(dev)) -#define OVERLAY_EXISTS(dev) (!IS_G4X(dev) && !IS_IRONLAKE(dev)) +#define OVERLAY_EXISTS(dev) (!IS_G4X(dev) && !IS_IRONLAKE(dev) && !IS_GEN6(dev)) static struct overlay_registers *intel_overlay_map_regs_atomic(struct intel_overlay *overlay) @@ -199,16 +199,11 @@ static struct overlay_registers *intel_overlay_map_regs_atomic(struct intel_over static void intel_overlay_unmap_regs_atomic(struct intel_overlay *overlay) { - struct drm_device *dev = overlay->dev; - drm_i915_private_t *dev_priv = dev->dev_private; - if (OVERLAY_NONPHYSICAL(overlay->dev)) io_mapping_unmap_atomic(overlay->virt_addr); overlay->virt_addr = NULL; - I915_READ(OVADD); /* flush wc cashes */ - return; } @@ -225,9 +220,7 @@ static int intel_overlay_on(struct intel_overlay *overlay) overlay->active = 1; overlay->hw_wedged = NEEDS_WAIT_FOR_FLIP; - BEGIN_LP_RING(6); - OUT_RING(MI_FLUSH); - OUT_RING(MI_NOOP); + BEGIN_LP_RING(4); OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_ON); OUT_RING(overlay->flip_addr | OFC_UPDATE); OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); @@ -267,9 +260,7 @@ static void intel_overlay_continue(struct intel_overlay *overlay, if (tmp & (1 << 17)) DRM_DEBUG("overlay underrun, DOVSTA: %x\n", tmp); - BEGIN_LP_RING(4); - OUT_RING(MI_FLUSH); - OUT_RING(MI_NOOP); + BEGIN_LP_RING(2); OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE); OUT_RING(flip_addr); ADVANCE_LP_RING(); @@ -338,9 +329,7 @@ static int intel_overlay_off(struct intel_overlay *overlay) /* wait for overlay to go idle */ overlay->hw_wedged = SWITCH_OFF_STAGE_1; - BEGIN_LP_RING(6); - OUT_RING(MI_FLUSH); - OUT_RING(MI_NOOP); + BEGIN_LP_RING(4); OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE); OUT_RING(flip_addr); OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); @@ -358,9 +347,7 @@ static int intel_overlay_off(struct intel_overlay *overlay) /* turn overlay off */ overlay->hw_wedged = SWITCH_OFF_STAGE_2; - BEGIN_LP_RING(6); - OUT_RING(MI_FLUSH); - OUT_RING(MI_NOOP); + BEGIN_LP_RING(4); OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_OFF); OUT_RING(flip_addr); OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); @@ -435,9 +422,7 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay, overlay->hw_wedged = SWITCH_OFF_STAGE_2; - BEGIN_LP_RING(6); - OUT_RING(MI_FLUSH); - OUT_RING(MI_NOOP); + BEGIN_LP_RING(4); OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_OFF); OUT_RING(flip_addr); OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); @@ -1179,7 +1164,7 @@ int intel_overlay_put_image(struct drm_device *dev, void *data, out_unlock: mutex_unlock(&dev->struct_mutex); mutex_unlock(&dev->mode_config.mutex); - drm_gem_object_unreference(new_bo); + drm_gem_object_unreference_unlocked(new_bo); kfree(params); return ret; diff --git a/drivers/gpu/drm/i915/intel_sdvo.c b/drivers/gpu/drm/i915/intel_sdvo.c index dba5147f4064..48daee5c9c63 100644 --- a/drivers/gpu/drm/i915/intel_sdvo.c +++ b/drivers/gpu/drm/i915/intel_sdvo.c @@ -35,6 +35,7 @@ #include "i915_drm.h" #include "i915_drv.h" #include "intel_sdvo_regs.h" +#include <linux/dmi.h> static char *tv_format_names[] = { "NTSC_M" , "NTSC_J" , "NTSC_443", @@ -462,14 +463,63 @@ static int intel_sdvo_get_pixel_multiplier(struct drm_display_mode *mode) } /** - * Don't check status code from this as it switches the bus back to the - * SDVO chips which defeats the purpose of doing a bus switch in the first - * place. + * Try to read the response after issuie the DDC switch command. But it + * is noted that we must do the action of reading response and issuing DDC + * switch command in one I2C transaction. Otherwise when we try to start + * another I2C transaction after issuing the DDC bus switch, it will be + * switched to the internal SDVO register. */ static void intel_sdvo_set_control_bus_switch(struct intel_output *intel_output, u8 target) { - intel_sdvo_write_cmd(intel_output, SDVO_CMD_SET_CONTROL_BUS_SWITCH, &target, 1); + struct intel_sdvo_priv *sdvo_priv = intel_output->dev_priv; + u8 out_buf[2], cmd_buf[2], ret_value[2], ret; + struct i2c_msg msgs[] = { + { + .addr = sdvo_priv->slave_addr >> 1, + .flags = 0, + .len = 2, + .buf = out_buf, + }, + /* the following two are to read the response */ + { + .addr = sdvo_priv->slave_addr >> 1, + .flags = 0, + .len = 1, + .buf = cmd_buf, + }, + { + .addr = sdvo_priv->slave_addr >> 1, + .flags = I2C_M_RD, + .len = 1, + .buf = ret_value, + }, + }; + + intel_sdvo_debug_write(intel_output, SDVO_CMD_SET_CONTROL_BUS_SWITCH, + &target, 1); + /* write the DDC switch command argument */ + intel_sdvo_write_byte(intel_output, SDVO_I2C_ARG_0, target); + + out_buf[0] = SDVO_I2C_OPCODE; + out_buf[1] = SDVO_CMD_SET_CONTROL_BUS_SWITCH; + cmd_buf[0] = SDVO_I2C_CMD_STATUS; + cmd_buf[1] = 0; + ret_value[0] = 0; + ret_value[1] = 0; + + ret = i2c_transfer(intel_output->i2c_bus, msgs, 3); + if (ret != 3) { + /* failure in I2C transfer */ + DRM_DEBUG_KMS("I2c transfer returned %d\n", ret); + return; + } + if (ret_value[0] != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("DDC switch command returns response %d\n", + ret_value[0]); + return; + } + return; } static bool intel_sdvo_set_target_input(struct intel_output *intel_output, bool target_0, bool target_1) @@ -1579,6 +1629,32 @@ intel_sdvo_hdmi_sink_detect(struct drm_connector *connector, u16 response) edid = drm_get_edid(&intel_output->base, intel_output->ddc_bus); + /* This is only applied to SDVO cards with multiple outputs */ + if (edid == NULL && intel_sdvo_multifunc_encoder(intel_output)) { + uint8_t saved_ddc, temp_ddc; + saved_ddc = sdvo_priv->ddc_bus; + temp_ddc = sdvo_priv->ddc_bus >> 1; + /* + * Don't use the 1 as the argument of DDC bus switch to get + * the EDID. It is used for SDVO SPD ROM. + */ + while(temp_ddc > 1) { + sdvo_priv->ddc_bus = temp_ddc; + edid = drm_get_edid(&intel_output->base, + intel_output->ddc_bus); + if (edid) { + /* + * When we can get the EDID, maybe it is the + * correct DDC bus. Update it. + */ + sdvo_priv->ddc_bus = temp_ddc; + break; + } + temp_ddc >>= 1; + } + if (edid == NULL) + sdvo_priv->ddc_bus = saved_ddc; + } /* when there is no edid and no monitor is connected with VGA * port, try to use the CRT ddc to read the EDID for DVI-connector */ @@ -2208,6 +2284,25 @@ intel_sdvo_get_slave_addr(struct drm_device *dev, int output_device) return 0x72; } +static int intel_sdvo_bad_tv_callback(const struct dmi_system_id *id) +{ + DRM_DEBUG_KMS("Ignoring bad SDVO TV connector for %s\n", id->ident); + return 1; +} + +static struct dmi_system_id intel_sdvo_bad_tv[] = { + { + .callback = intel_sdvo_bad_tv_callback, + .ident = "IntelG45/ICH10R/DME1737", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "IBM CORPORATION"), + DMI_MATCH(DMI_PRODUCT_NAME, "4800784"), + }, + }, + + { } /* terminating entry */ +}; + static bool intel_sdvo_output_setup(struct intel_output *intel_output, uint16_t flags) { @@ -2248,7 +2343,8 @@ intel_sdvo_output_setup(struct intel_output *intel_output, uint16_t flags) (1 << INTEL_SDVO_NON_TV_CLONE_BIT) | (1 << INTEL_ANALOG_CLONE_BIT); } - } else if (flags & SDVO_OUTPUT_SVID0) { + } else if ((flags & SDVO_OUTPUT_SVID0) && + !dmi_check_system(intel_sdvo_bad_tv)) { sdvo_priv->controlled_output = SDVO_OUTPUT_SVID0; encoder->encoder_type = DRM_MODE_ENCODER_TVDAC; @@ -2270,6 +2366,14 @@ intel_sdvo_output_setup(struct intel_output *intel_output, uint16_t flags) connector->connector_type = DRM_MODE_CONNECTOR_VGA; intel_output->clone_mask = (1 << INTEL_SDVO_NON_TV_CLONE_BIT) | (1 << INTEL_ANALOG_CLONE_BIT); + } else if (flags & SDVO_OUTPUT_CVBS0) { + + sdvo_priv->controlled_output = SDVO_OUTPUT_CVBS0; + encoder->encoder_type = DRM_MODE_ENCODER_TVDAC; + connector->connector_type = DRM_MODE_CONNECTOR_SVIDEO; + sdvo_priv->is_tv = true; + intel_output->needs_tv_clock = true; + intel_output->clone_mask = 1 << INTEL_SDVO_TV_CLONE_BIT; } else if (flags & SDVO_OUTPUT_LVDS0) { sdvo_priv->controlled_output = SDVO_OUTPUT_LVDS0; @@ -2662,6 +2766,7 @@ static void intel_sdvo_create_enhance_property(struct drm_connector *connector) bool intel_sdvo_init(struct drm_device *dev, int output_device) { + struct drm_i915_private *dev_priv = dev->dev_private; struct drm_connector *connector; struct intel_output *intel_output; struct intel_sdvo_priv *sdvo_priv; @@ -2708,10 +2813,12 @@ bool intel_sdvo_init(struct drm_device *dev, int output_device) intel_output->ddc_bus = intel_i2c_create(dev, GPIOE, "SDVOB DDC BUS"); sdvo_priv->analog_ddc_bus = intel_i2c_create(dev, GPIOA, "SDVOB/VGA DDC BUS"); + dev_priv->hotplug_supported_mask |= SDVOB_HOTPLUG_INT_STATUS; } else { intel_output->ddc_bus = intel_i2c_create(dev, GPIOE, "SDVOC DDC BUS"); sdvo_priv->analog_ddc_bus = intel_i2c_create(dev, GPIOA, "SDVOC/VGA DDC BUS"); + dev_priv->hotplug_supported_mask |= SDVOC_HOTPLUG_INT_STATUS; } if (intel_output->ddc_bus == NULL) @@ -2720,7 +2827,7 @@ bool intel_sdvo_init(struct drm_device *dev, int output_device) /* Wrap with our custom algo which switches to DDC mode */ intel_output->ddc_bus->algo = &intel_sdvo_i2c_bit_algo; - /* In defaut case sdvo lvds is false */ + /* In default case sdvo lvds is false */ intel_sdvo_get_capabilities(intel_output, &sdvo_priv->caps); if (intel_sdvo_output_setup(intel_output, diff --git a/drivers/gpu/drm/nouveau/Makefile b/drivers/gpu/drm/nouveau/Makefile index 48c290b5da8c..32db806f3b5a 100644 --- a/drivers/gpu/drm/nouveau/Makefile +++ b/drivers/gpu/drm/nouveau/Makefile @@ -16,7 +16,7 @@ nouveau-y := nouveau_drv.o nouveau_state.o nouveau_channel.o nouveau_mem.o \ nv04_fifo.o nv10_fifo.o nv40_fifo.o nv50_fifo.o \ nv04_graph.o nv10_graph.o nv20_graph.o \ nv40_graph.o nv50_graph.o \ - nv40_grctx.o \ + nv40_grctx.o nv50_grctx.o \ nv04_instmem.o nv50_instmem.o \ nv50_crtc.o nv50_dac.o nv50_sor.o \ nv50_cursor.o nv50_display.o nv50_fbcon.o \ diff --git a/drivers/gpu/drm/nouveau/nouveau_acpi.c b/drivers/gpu/drm/nouveau/nouveau_acpi.c index 1cf488247a16..48227e744753 100644 --- a/drivers/gpu/drm/nouveau/nouveau_acpi.c +++ b/drivers/gpu/drm/nouveau/nouveau_acpi.c @@ -90,21 +90,21 @@ int nouveau_hybrid_setup(struct drm_device *dev) { int result; - if (nouveau_dsm(dev, NOUVEAU_DSM_ACTIVE, NOUVEAU_DSM_ACTIVE_QUERY, + if (nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_STATE, &result)) return -ENODEV; NV_INFO(dev, "_DSM hardware status gave 0x%x\n", result); - if (result & 0x1) { /* Stamina mode - disable the external GPU */ + if (result) { /* Ensure that the external GPU is enabled */ + nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_SPEED, NULL); + nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_SPEED, + NULL); + } else { /* Stamina mode - disable the external GPU */ nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_STAMINA, NULL); nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_STAMINA, NULL); - } else { /* Ensure that the external GPU is enabled */ - nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_SPEED, NULL); - nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_SPEED, - NULL); } return 0; diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.c b/drivers/gpu/drm/nouveau/nouveau_bios.c index d7f8d8b4a4b8..71247da17da5 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bios.c +++ b/drivers/gpu/drm/nouveau/nouveau_bios.c @@ -311,11 +311,11 @@ valid_reg(struct nvbios *bios, uint32_t reg) /* C51 has misaligned regs on purpose. Marvellous */ if (reg & 0x2 || - (reg & 0x1 && dev_priv->VBIOS.pub.chip_version != 0x51)) + (reg & 0x1 && dev_priv->vbios.chip_version != 0x51)) NV_ERROR(dev, "======= misaligned reg 0x%08X =======\n", reg); /* warn on C51 regs that haven't been verified accessible in tracing */ - if (reg & 0x1 && dev_priv->VBIOS.pub.chip_version == 0x51 && + if (reg & 0x1 && dev_priv->vbios.chip_version == 0x51 && reg != 0x130d && reg != 0x1311 && reg != 0x60081d) NV_WARN(dev, "=== C51 misaligned reg 0x%08X not verified ===\n", reg); @@ -420,7 +420,7 @@ bios_wr32(struct nvbios *bios, uint32_t reg, uint32_t data) LOG_OLD_VALUE(bios_rd32(bios, reg)); BIOSLOG(bios, " Write: Reg: 0x%08X, Data: 0x%08X\n", reg, data); - if (dev_priv->VBIOS.execute) { + if (dev_priv->vbios.execute) { still_alive(); nv_wr32(bios->dev, reg, data); } @@ -647,7 +647,7 @@ nv50_pll_set(struct drm_device *dev, uint32_t reg, uint32_t clk) reg0 = (reg0 & 0xfff8ffff) | (pll.log2P << 16); reg1 = (reg1 & 0xffff0000) | (pll.N1 << 8) | pll.M1; - if (dev_priv->VBIOS.execute) { + if (dev_priv->vbios.execute) { still_alive(); nv_wr32(dev, reg + 4, reg1); nv_wr32(dev, reg + 0, reg0); @@ -689,7 +689,7 @@ setPLL(struct nvbios *bios, uint32_t reg, uint32_t clk) static int dcb_entry_idx_from_crtchead(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; /* * For the results of this function to be correct, CR44 must have been @@ -700,7 +700,7 @@ static int dcb_entry_idx_from_crtchead(struct drm_device *dev) uint8_t dcb_entry = NVReadVgaCrtc5758(dev, bios->state.crtchead, 0); - if (dcb_entry > bios->bdcb.dcb.entries) { + if (dcb_entry > bios->dcb.entries) { NV_ERROR(dev, "CR58 doesn't have a valid DCB entry currently " "(%02X)\n", dcb_entry); dcb_entry = 0x7f; /* unused / invalid marker */ @@ -713,25 +713,26 @@ static struct nouveau_i2c_chan * init_i2c_device_find(struct drm_device *dev, int i2c_index) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct bios_parsed_dcb *bdcb = &dev_priv->VBIOS.bdcb; + struct dcb_table *dcb = &dev_priv->vbios.dcb; if (i2c_index == 0xff) { /* note: dcb_entry_idx_from_crtchead needs pre-script set-up */ int idx = dcb_entry_idx_from_crtchead(dev), shift = 0; - int default_indices = bdcb->i2c_default_indices; + int default_indices = dcb->i2c_default_indices; - if (idx != 0x7f && bdcb->dcb.entry[idx].i2c_upper_default) + if (idx != 0x7f && dcb->entry[idx].i2c_upper_default) shift = 4; i2c_index = (default_indices >> shift) & 0xf; } if (i2c_index == 0x80) /* g80+ */ - i2c_index = bdcb->i2c_default_indices & 0xf; + i2c_index = dcb->i2c_default_indices & 0xf; return nouveau_i2c_find(dev, i2c_index); } -static uint32_t get_tmds_index_reg(struct drm_device *dev, uint8_t mlv) +static uint32_t +get_tmds_index_reg(struct drm_device *dev, uint8_t mlv) { /* * For mlv < 0x80, it is an index into a table of TMDS base addresses. @@ -744,6 +745,7 @@ static uint32_t get_tmds_index_reg(struct drm_device *dev, uint8_t mlv) */ struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; const int pramdac_offset[13] = { 0, 0, 0x8, 0, 0x2000, 0, 0, 0, 0x2008, 0, 0, 0, 0x2000 }; const uint32_t pramdac_table[4] = { @@ -756,13 +758,12 @@ static uint32_t get_tmds_index_reg(struct drm_device *dev, uint8_t mlv) dcb_entry = dcb_entry_idx_from_crtchead(dev); if (dcb_entry == 0x7f) return 0; - dacoffset = pramdac_offset[ - dev_priv->VBIOS.bdcb.dcb.entry[dcb_entry].or]; + dacoffset = pramdac_offset[bios->dcb.entry[dcb_entry].or]; if (mlv == 0x81) dacoffset ^= 8; return 0x6808b0 + dacoffset; } else { - if (mlv > ARRAY_SIZE(pramdac_table)) { + if (mlv >= ARRAY_SIZE(pramdac_table)) { NV_ERROR(dev, "Magic Lookup Value too big (%02X)\n", mlv); return 0; @@ -1865,7 +1866,7 @@ init_compute_mem(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) struct drm_nouveau_private *dev_priv = bios->dev->dev_private; - if (dev_priv->card_type >= NV_50) + if (dev_priv->card_type >= NV_40) return 1; /* @@ -2574,19 +2575,19 @@ init_gpio(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) const uint32_t nv50_gpio_reg[4] = { 0xe104, 0xe108, 0xe280, 0xe284 }; const uint32_t nv50_gpio_ctl[2] = { 0xe100, 0xe28c }; - const uint8_t *gpio_table = &bios->data[bios->bdcb.gpio_table_ptr]; + const uint8_t *gpio_table = &bios->data[bios->dcb.gpio_table_ptr]; const uint8_t *gpio_entry; int i; if (!iexec->execute) return 1; - if (bios->bdcb.version != 0x40) { + if (bios->dcb.version != 0x40) { NV_ERROR(bios->dev, "DCB table not version 4.0\n"); return 0; } - if (!bios->bdcb.gpio_table_ptr) { + if (!bios->dcb.gpio_table_ptr) { NV_WARN(bios->dev, "Invalid pointer to INIT_8E table\n"); return 0; } @@ -3123,7 +3124,7 @@ run_digital_op_script(struct drm_device *dev, uint16_t scriptptr, struct dcb_entry *dcbent, int head, bool dl) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; struct init_exec iexec = {true, false}; NV_TRACE(dev, "0x%04X: Parsing digital output script table\n", @@ -3140,7 +3141,7 @@ run_digital_op_script(struct drm_device *dev, uint16_t scriptptr, static int call_lvds_manufacturer_script(struct drm_device *dev, struct dcb_entry *dcbent, int head, enum LVDS_script script) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; uint8_t sub = bios->data[bios->fp.xlated_entry + script] + (bios->fp.link_c_increment && dcbent->or & OUTPUT_C ? 1 : 0); uint16_t scriptofs = ROM16(bios->data[bios->init_script_tbls_ptr + sub * 2]); @@ -3194,7 +3195,7 @@ static int run_lvds_table(struct drm_device *dev, struct dcb_entry *dcbent, int * of a list of pxclks and script pointers. */ struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; unsigned int outputset = (dcbent->or == 4) ? 1 : 0; uint16_t scriptptr = 0, clktable; uint8_t clktableptr = 0; @@ -3261,7 +3262,7 @@ int call_lvds_script(struct drm_device *dev, struct dcb_entry *dcbent, int head, */ struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; uint8_t lvds_ver = bios->data[bios->fp.lvdsmanufacturerpointer]; uint32_t sel_clk_binding, sel_clk; int ret; @@ -3395,7 +3396,7 @@ static int parse_fp_mode_table(struct drm_device *dev, struct nvbios *bios) #ifndef __powerpc__ NV_ERROR(dev, "Pointer to flat panel table invalid\n"); #endif - bios->pub.digital_min_front_porch = 0x4b; + bios->digital_min_front_porch = 0x4b; return 0; } @@ -3428,7 +3429,7 @@ static int parse_fp_mode_table(struct drm_device *dev, struct nvbios *bios) * fptable[4] is the minimum * RAMDAC_FP_HCRTC -> RAMDAC_FP_HSYNC_START gap */ - bios->pub.digital_min_front_porch = fptable[4]; + bios->digital_min_front_porch = fptable[4]; ofs = -7; break; default: @@ -3467,7 +3468,7 @@ static int parse_fp_mode_table(struct drm_device *dev, struct nvbios *bios) /* nv4x cards need both a strap value and fpindex of 0xf to use DDC */ if (lth.lvds_ver > 0x10) - bios->pub.fp_no_ddc = fpstrapping != 0xf || fpindex != 0xf; + bios->fp_no_ddc = fpstrapping != 0xf || fpindex != 0xf; /* * If either the strap or xlated fpindex value are 0xf there is no @@ -3491,7 +3492,7 @@ static int parse_fp_mode_table(struct drm_device *dev, struct nvbios *bios) bool nouveau_bios_fp_mode(struct drm_device *dev, struct drm_display_mode *mode) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; uint8_t *mode_entry = &bios->data[bios->fp.mode_ptr]; if (!mode) /* just checking whether we can produce a mode */ @@ -3562,11 +3563,11 @@ int nouveau_bios_parse_lvds_table(struct drm_device *dev, int pxclk, bool *dl, b * until later, when this function should be called with non-zero pxclk */ struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; int fpstrapping = get_fp_strap(dev, bios), lvdsmanufacturerindex = 0; struct lvdstableheader lth; uint16_t lvdsofs; - int ret, chip_version = bios->pub.chip_version; + int ret, chip_version = bios->chip_version; ret = parse_lvds_manufacturer_table_header(dev, bios, <h); if (ret) @@ -3682,7 +3683,7 @@ bios_output_config_match(struct drm_device *dev, struct dcb_entry *dcbent, uint16_t record, int record_len, int record_nr) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; uint32_t entry; uint16_t table; int i, v; @@ -3716,7 +3717,7 @@ nouveau_bios_dp_table(struct drm_device *dev, struct dcb_entry *dcbent, int *length) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; uint8_t *table; if (!bios->display.dp_table_ptr) { @@ -3725,7 +3726,7 @@ nouveau_bios_dp_table(struct drm_device *dev, struct dcb_entry *dcbent, } table = &bios->data[bios->display.dp_table_ptr]; - if (table[0] != 0x21) { + if (table[0] != 0x20 && table[0] != 0x21) { NV_ERROR(dev, "DisplayPort table version 0x%02x unknown\n", table[0]); return NULL; @@ -3765,8 +3766,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent, */ struct drm_nouveau_private *dev_priv = dev->dev_private; - struct init_exec iexec = {true, false}; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; uint8_t *table = &bios->data[bios->display.script_table_ptr]; uint8_t *otable = NULL; uint16_t script; @@ -3845,8 +3845,6 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent, } } - bios->display.output = dcbent; - if (pxclk == 0) { script = ROM16(otable[6]); if (!script) { @@ -3855,7 +3853,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent, } NV_TRACE(dev, "0x%04X: parsing output script 0\n", script); - parse_init_table(bios, script, &iexec); + nouveau_bios_run_init_table(dev, script, dcbent); } else if (pxclk == -1) { script = ROM16(otable[8]); @@ -3865,7 +3863,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent, } NV_TRACE(dev, "0x%04X: parsing output script 1\n", script); - parse_init_table(bios, script, &iexec); + nouveau_bios_run_init_table(dev, script, dcbent); } else if (pxclk == -2) { if (table[4] >= 12) @@ -3878,7 +3876,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent, } NV_TRACE(dev, "0x%04X: parsing output script 2\n", script); - parse_init_table(bios, script, &iexec); + nouveau_bios_run_init_table(dev, script, dcbent); } else if (pxclk > 0) { script = ROM16(otable[table[4] + i*6 + 2]); @@ -3890,7 +3888,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent, } NV_TRACE(dev, "0x%04X: parsing clock script 0\n", script); - parse_init_table(bios, script, &iexec); + nouveau_bios_run_init_table(dev, script, dcbent); } else if (pxclk < 0) { script = ROM16(otable[table[4] + i*6 + 4]); @@ -3902,7 +3900,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent, } NV_TRACE(dev, "0x%04X: parsing clock script 1\n", script); - parse_init_table(bios, script, &iexec); + nouveau_bios_run_init_table(dev, script, dcbent); } return 0; @@ -3921,8 +3919,8 @@ int run_tmds_table(struct drm_device *dev, struct dcb_entry *dcbent, int head, i */ struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; - int cv = bios->pub.chip_version; + struct nvbios *bios = &dev_priv->vbios; + int cv = bios->chip_version; uint16_t clktable = 0, scriptptr; uint32_t sel_clk_binding, sel_clk; @@ -3981,8 +3979,8 @@ int get_pll_limits(struct drm_device *dev, uint32_t limit_match, struct pll_lims */ struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; - int cv = bios->pub.chip_version, pllindex = 0; + struct nvbios *bios = &dev_priv->vbios; + int cv = bios->chip_version, pllindex = 0; uint8_t pll_lim_ver = 0, headerlen = 0, recordlen = 0, entries = 0; uint32_t crystal_strap_mask, crystal_straps; @@ -4335,7 +4333,7 @@ static void parse_bios_version(struct drm_device *dev, struct nvbios *bios, uint */ bios->major_version = bios->data[offset + 3]; - bios->pub.chip_version = bios->data[offset + 2]; + bios->chip_version = bios->data[offset + 2]; NV_TRACE(dev, "Bios version %02x.%02x.%02x.%02x\n", bios->data[offset + 3], bios->data[offset + 2], bios->data[offset + 1], bios->data[offset]); @@ -4405,7 +4403,7 @@ static int parse_bit_A_tbl_entry(struct drm_device *dev, struct nvbios *bios, st } /* First entry is normal dac, 2nd tv-out perhaps? */ - bios->pub.dactestval = ROM32(bios->data[load_table_ptr + headerlen]) & 0x3ff; + bios->dactestval = ROM32(bios->data[load_table_ptr + headerlen]) & 0x3ff; return 0; } @@ -4529,8 +4527,8 @@ static int parse_bit_i_tbl_entry(struct drm_device *dev, struct nvbios *bios, st return -ENOSYS; } - bios->pub.dactestval = ROM32(bios->data[daccmpoffset + dacheaderlen]); - bios->pub.tvdactestval = ROM32(bios->data[daccmpoffset + dacheaderlen + 4]); + bios->dactestval = ROM32(bios->data[daccmpoffset + dacheaderlen]); + bios->tvdactestval = ROM32(bios->data[daccmpoffset + dacheaderlen + 4]); return 0; } @@ -4799,11 +4797,11 @@ static int parse_bmp_structure(struct drm_device *dev, struct nvbios *bios, unsi uint16_t legacy_scripts_offset, legacy_i2c_offset; /* load needed defaults in case we can't parse this info */ - bios->bdcb.dcb.i2c[0].write = NV_CIO_CRE_DDC_WR__INDEX; - bios->bdcb.dcb.i2c[0].read = NV_CIO_CRE_DDC_STATUS__INDEX; - bios->bdcb.dcb.i2c[1].write = NV_CIO_CRE_DDC0_WR__INDEX; - bios->bdcb.dcb.i2c[1].read = NV_CIO_CRE_DDC0_STATUS__INDEX; - bios->pub.digital_min_front_porch = 0x4b; + bios->dcb.i2c[0].write = NV_CIO_CRE_DDC_WR__INDEX; + bios->dcb.i2c[0].read = NV_CIO_CRE_DDC_STATUS__INDEX; + bios->dcb.i2c[1].write = NV_CIO_CRE_DDC0_WR__INDEX; + bios->dcb.i2c[1].read = NV_CIO_CRE_DDC0_STATUS__INDEX; + bios->digital_min_front_porch = 0x4b; bios->fmaxvco = 256000; bios->fminvco = 128000; bios->fp.duallink_transition_clk = 90000; @@ -4910,10 +4908,10 @@ static int parse_bmp_structure(struct drm_device *dev, struct nvbios *bios, unsi bios->legacy.i2c_indices.crt = bios->data[legacy_i2c_offset]; bios->legacy.i2c_indices.tv = bios->data[legacy_i2c_offset + 1]; bios->legacy.i2c_indices.panel = bios->data[legacy_i2c_offset + 2]; - bios->bdcb.dcb.i2c[0].write = bios->data[legacy_i2c_offset + 4]; - bios->bdcb.dcb.i2c[0].read = bios->data[legacy_i2c_offset + 5]; - bios->bdcb.dcb.i2c[1].write = bios->data[legacy_i2c_offset + 6]; - bios->bdcb.dcb.i2c[1].read = bios->data[legacy_i2c_offset + 7]; + bios->dcb.i2c[0].write = bios->data[legacy_i2c_offset + 4]; + bios->dcb.i2c[0].read = bios->data[legacy_i2c_offset + 5]; + bios->dcb.i2c[1].write = bios->data[legacy_i2c_offset + 6]; + bios->dcb.i2c[1].read = bios->data[legacy_i2c_offset + 7]; if (bmplength > 74) { bios->fmaxvco = ROM32(bmp[67]); @@ -4987,7 +4985,8 @@ read_dcb_i2c_entry(struct drm_device *dev, int dcb_version, uint8_t *i2ctable, i else NV_WARN(dev, "DCB I2C table has more entries than indexable " - "(%d entries, max index 15)\n", i2ctable[2]); + "(%d entries, max %d)\n", i2ctable[2], + DCB_MAX_NUM_I2C_ENTRIES); entry_len = i2ctable[3]; /* [4] is i2c_default_indices, read in parse_dcb_table() */ } @@ -5003,8 +5002,8 @@ read_dcb_i2c_entry(struct drm_device *dev, int dcb_version, uint8_t *i2ctable, i if (index == 0xf) return 0; - if (index > i2c_entries) { - NV_ERROR(dev, "DCB I2C index too big (%d > %d)\n", + if (index >= i2c_entries) { + NV_ERROR(dev, "DCB I2C index too big (%d >= %d)\n", index, i2ctable[2]); return -ENOENT; } @@ -5039,7 +5038,7 @@ read_dcb_i2c_entry(struct drm_device *dev, int dcb_version, uint8_t *i2ctable, i static struct dcb_gpio_entry * new_gpio_entry(struct nvbios *bios) { - struct parsed_dcb_gpio *gpio = &bios->bdcb.gpio; + struct dcb_gpio_table *gpio = &bios->dcb.gpio; return &gpio->entry[gpio->entries++]; } @@ -5048,14 +5047,14 @@ struct dcb_gpio_entry * nouveau_bios_gpio_entry(struct drm_device *dev, enum dcb_gpio_tag tag) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; int i; - for (i = 0; i < bios->bdcb.gpio.entries; i++) { - if (bios->bdcb.gpio.entry[i].tag != tag) + for (i = 0; i < bios->dcb.gpio.entries; i++) { + if (bios->dcb.gpio.entry[i].tag != tag) continue; - return &bios->bdcb.gpio.entry[i]; + return &bios->dcb.gpio.entry[i]; } return NULL; @@ -5103,7 +5102,7 @@ static void parse_dcb_gpio_table(struct nvbios *bios) { struct drm_device *dev = bios->dev; - uint16_t gpio_table_ptr = bios->bdcb.gpio_table_ptr; + uint16_t gpio_table_ptr = bios->dcb.gpio_table_ptr; uint8_t *gpio_table = &bios->data[gpio_table_ptr]; int header_len = gpio_table[1], entries = gpio_table[2], @@ -5111,7 +5110,7 @@ parse_dcb_gpio_table(struct nvbios *bios) void (*parse_entry)(struct nvbios *, uint16_t) = NULL; int i; - if (bios->bdcb.version >= 0x40) { + if (bios->dcb.version >= 0x40) { if (gpio_table_ptr && entry_len != 4) { NV_WARN(dev, "Invalid DCB GPIO table entry length.\n"); return; @@ -5119,7 +5118,7 @@ parse_dcb_gpio_table(struct nvbios *bios) parse_entry = parse_dcb40_gpio_entry; - } else if (bios->bdcb.version >= 0x30) { + } else if (bios->dcb.version >= 0x30) { if (gpio_table_ptr && entry_len != 2) { NV_WARN(dev, "Invalid DCB GPIO table entry length.\n"); return; @@ -5127,7 +5126,7 @@ parse_dcb_gpio_table(struct nvbios *bios) parse_entry = parse_dcb30_gpio_entry; - } else if (bios->bdcb.version >= 0x22) { + } else if (bios->dcb.version >= 0x22) { /* * DCBs older than v3.0 don't really have a GPIO * table, instead they keep some GPIO info at fixed @@ -5161,30 +5160,67 @@ struct dcb_connector_table_entry * nouveau_bios_connector_entry(struct drm_device *dev, int index) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; struct dcb_connector_table_entry *cte; - if (index >= bios->bdcb.connector.entries) + if (index >= bios->dcb.connector.entries) return NULL; - cte = &bios->bdcb.connector.entry[index]; + cte = &bios->dcb.connector.entry[index]; if (cte->type == 0xff) return NULL; return cte; } +static enum dcb_connector_type +divine_connector_type(struct nvbios *bios, int index) +{ + struct dcb_table *dcb = &bios->dcb; + unsigned encoders = 0, type = DCB_CONNECTOR_NONE; + int i; + + for (i = 0; i < dcb->entries; i++) { + if (dcb->entry[i].connector == index) + encoders |= (1 << dcb->entry[i].type); + } + + if (encoders & (1 << OUTPUT_DP)) { + if (encoders & (1 << OUTPUT_TMDS)) + type = DCB_CONNECTOR_DP; + else + type = DCB_CONNECTOR_eDP; + } else + if (encoders & (1 << OUTPUT_TMDS)) { + if (encoders & (1 << OUTPUT_ANALOG)) + type = DCB_CONNECTOR_DVI_I; + else + type = DCB_CONNECTOR_DVI_D; + } else + if (encoders & (1 << OUTPUT_ANALOG)) { + type = DCB_CONNECTOR_VGA; + } else + if (encoders & (1 << OUTPUT_LVDS)) { + type = DCB_CONNECTOR_LVDS; + } else + if (encoders & (1 << OUTPUT_TV)) { + type = DCB_CONNECTOR_TV_0; + } + + return type; +} + static void parse_dcb_connector_table(struct nvbios *bios) { struct drm_device *dev = bios->dev; - struct dcb_connector_table *ct = &bios->bdcb.connector; + struct dcb_connector_table *ct = &bios->dcb.connector; struct dcb_connector_table_entry *cte; - uint8_t *conntab = &bios->data[bios->bdcb.connector_table_ptr]; + uint8_t *conntab = &bios->data[bios->dcb.connector_table_ptr]; uint8_t *entry; int i; - if (!bios->bdcb.connector_table_ptr) { + if (!bios->dcb.connector_table_ptr) { NV_DEBUG_KMS(dev, "No DCB connector table present\n"); return; } @@ -5206,6 +5242,7 @@ parse_dcb_connector_table(struct nvbios *bios) cte->entry = ROM16(entry[0]); else cte->entry = ROM32(entry[0]); + cte->type = (cte->entry & 0x000000ff) >> 0; cte->index = (cte->entry & 0x00000f00) >> 8; switch (cte->entry & 0x00033000) { @@ -5231,10 +5268,33 @@ parse_dcb_connector_table(struct nvbios *bios) NV_INFO(dev, " %d: 0x%08x: type 0x%02x idx %d tag 0x%02x\n", i, cte->entry, cte->type, cte->index, cte->gpio_tag); + + /* check for known types, fallback to guessing the type + * from attached encoders if we hit an unknown. + */ + switch (cte->type) { + case DCB_CONNECTOR_VGA: + case DCB_CONNECTOR_TV_0: + case DCB_CONNECTOR_TV_1: + case DCB_CONNECTOR_TV_3: + case DCB_CONNECTOR_DVI_I: + case DCB_CONNECTOR_DVI_D: + case DCB_CONNECTOR_LVDS: + case DCB_CONNECTOR_DP: + case DCB_CONNECTOR_eDP: + case DCB_CONNECTOR_HDMI_0: + case DCB_CONNECTOR_HDMI_1: + break; + default: + cte->type = divine_connector_type(bios, cte->index); + NV_WARN(dev, "unknown type, using 0x%02x", cte->type); + break; + } + } } -static struct dcb_entry *new_dcb_entry(struct parsed_dcb *dcb) +static struct dcb_entry *new_dcb_entry(struct dcb_table *dcb) { struct dcb_entry *entry = &dcb->entry[dcb->entries]; @@ -5244,7 +5304,7 @@ static struct dcb_entry *new_dcb_entry(struct parsed_dcb *dcb) return entry; } -static void fabricate_vga_output(struct parsed_dcb *dcb, int i2c, int heads) +static void fabricate_vga_output(struct dcb_table *dcb, int i2c, int heads) { struct dcb_entry *entry = new_dcb_entry(dcb); @@ -5255,7 +5315,7 @@ static void fabricate_vga_output(struct parsed_dcb *dcb, int i2c, int heads) /* "or" mostly unused in early gen crt modesetting, 0 is fine */ } -static void fabricate_dvi_i_output(struct parsed_dcb *dcb, bool twoHeads) +static void fabricate_dvi_i_output(struct dcb_table *dcb, bool twoHeads) { struct dcb_entry *entry = new_dcb_entry(dcb); @@ -5282,7 +5342,7 @@ static void fabricate_dvi_i_output(struct parsed_dcb *dcb, bool twoHeads) #endif } -static void fabricate_tv_output(struct parsed_dcb *dcb, bool twoHeads) +static void fabricate_tv_output(struct dcb_table *dcb, bool twoHeads) { struct dcb_entry *entry = new_dcb_entry(dcb); @@ -5293,13 +5353,13 @@ static void fabricate_tv_output(struct parsed_dcb *dcb, bool twoHeads) } static bool -parse_dcb20_entry(struct drm_device *dev, struct bios_parsed_dcb *bdcb, +parse_dcb20_entry(struct drm_device *dev, struct dcb_table *dcb, uint32_t conn, uint32_t conf, struct dcb_entry *entry) { entry->type = conn & 0xf; entry->i2c_index = (conn >> 4) & 0xf; entry->heads = (conn >> 8) & 0xf; - if (bdcb->version >= 0x40) + if (dcb->version >= 0x40) entry->connector = (conn >> 12) & 0xf; entry->bus = (conn >> 16) & 0xf; entry->location = (conn >> 20) & 0x3; @@ -5317,7 +5377,7 @@ parse_dcb20_entry(struct drm_device *dev, struct bios_parsed_dcb *bdcb, * Although the rest of a CRT conf dword is usually * zeros, mac biosen have stuff there so we must mask */ - entry->crtconf.maxfreq = (bdcb->version < 0x30) ? + entry->crtconf.maxfreq = (dcb->version < 0x30) ? (conf & 0xffff) * 10 : (conf & 0xff) * 10000; break; @@ -5326,7 +5386,7 @@ parse_dcb20_entry(struct drm_device *dev, struct bios_parsed_dcb *bdcb, uint32_t mask; if (conf & 0x1) entry->lvdsconf.use_straps_for_mode = true; - if (bdcb->version < 0x22) { + if (dcb->version < 0x22) { mask = ~0xd; /* * The laptop in bug 14567 lies and claims to not use @@ -5350,7 +5410,7 @@ parse_dcb20_entry(struct drm_device *dev, struct bios_parsed_dcb *bdcb, * Until we even try to use these on G8x, it's * useless reporting unknown bits. They all are. */ - if (bdcb->version >= 0x40) + if (dcb->version >= 0x40) break; NV_ERROR(dev, "Unknown LVDS configuration bits, " @@ -5360,7 +5420,7 @@ parse_dcb20_entry(struct drm_device *dev, struct bios_parsed_dcb *bdcb, } case OUTPUT_TV: { - if (bdcb->version >= 0x30) + if (dcb->version >= 0x30) entry->tvconf.has_component_output = conf & (0x8 << 4); else entry->tvconf.has_component_output = false; @@ -5387,8 +5447,10 @@ parse_dcb20_entry(struct drm_device *dev, struct bios_parsed_dcb *bdcb, break; case 0xe: /* weird g80 mobile type that "nv" treats as a terminator */ - bdcb->dcb.entries--; + dcb->entries--; return false; + default: + break; } /* unsure what DCB version introduces this, 3.0? */ @@ -5399,7 +5461,7 @@ parse_dcb20_entry(struct drm_device *dev, struct bios_parsed_dcb *bdcb, } static bool -parse_dcb15_entry(struct drm_device *dev, struct parsed_dcb *dcb, +parse_dcb15_entry(struct drm_device *dev, struct dcb_table *dcb, uint32_t conn, uint32_t conf, struct dcb_entry *entry) { switch (conn & 0x0000000f) { @@ -5465,27 +5527,27 @@ parse_dcb15_entry(struct drm_device *dev, struct parsed_dcb *dcb, return true; } -static bool parse_dcb_entry(struct drm_device *dev, struct bios_parsed_dcb *bdcb, +static bool parse_dcb_entry(struct drm_device *dev, struct dcb_table *dcb, uint32_t conn, uint32_t conf) { - struct dcb_entry *entry = new_dcb_entry(&bdcb->dcb); + struct dcb_entry *entry = new_dcb_entry(dcb); bool ret; - if (bdcb->version >= 0x20) - ret = parse_dcb20_entry(dev, bdcb, conn, conf, entry); + if (dcb->version >= 0x20) + ret = parse_dcb20_entry(dev, dcb, conn, conf, entry); else - ret = parse_dcb15_entry(dev, &bdcb->dcb, conn, conf, entry); + ret = parse_dcb15_entry(dev, dcb, conn, conf, entry); if (!ret) return ret; - read_dcb_i2c_entry(dev, bdcb->version, bdcb->i2c_table, - entry->i2c_index, &bdcb->dcb.i2c[entry->i2c_index]); + read_dcb_i2c_entry(dev, dcb->version, dcb->i2c_table, + entry->i2c_index, &dcb->i2c[entry->i2c_index]); return true; } static -void merge_like_dcb_entries(struct drm_device *dev, struct parsed_dcb *dcb) +void merge_like_dcb_entries(struct drm_device *dev, struct dcb_table *dcb) { /* * DCB v2.0 lists each output combination separately. @@ -5537,8 +5599,7 @@ static int parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct bios_parsed_dcb *bdcb = &bios->bdcb; - struct parsed_dcb *dcb; + struct dcb_table *dcb = &bios->dcb; uint16_t dcbptr = 0, i2ctabptr = 0; uint8_t *dcbtable; uint8_t headerlen = 0x4, entries = DCB_MAX_NUM_ENTRIES; @@ -5546,9 +5607,6 @@ parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads) int recordlength = 8, confofs = 4; int i; - dcb = bios->pub.dcb = &bdcb->dcb; - dcb->entries = 0; - /* get the offset from 0x36 */ if (dev_priv->card_type > NV_04) { dcbptr = ROM16(bios->data[0x36]); @@ -5570,21 +5628,21 @@ parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads) dcbtable = &bios->data[dcbptr]; /* get DCB version */ - bdcb->version = dcbtable[0]; + dcb->version = dcbtable[0]; NV_TRACE(dev, "Found Display Configuration Block version %d.%d\n", - bdcb->version >> 4, bdcb->version & 0xf); + dcb->version >> 4, dcb->version & 0xf); - if (bdcb->version >= 0x20) { /* NV17+ */ + if (dcb->version >= 0x20) { /* NV17+ */ uint32_t sig; - if (bdcb->version >= 0x30) { /* NV40+ */ + if (dcb->version >= 0x30) { /* NV40+ */ headerlen = dcbtable[1]; entries = dcbtable[2]; recordlength = dcbtable[3]; i2ctabptr = ROM16(dcbtable[4]); sig = ROM32(dcbtable[6]); - bdcb->gpio_table_ptr = ROM16(dcbtable[10]); - bdcb->connector_table_ptr = ROM16(dcbtable[20]); + dcb->gpio_table_ptr = ROM16(dcbtable[10]); + dcb->connector_table_ptr = ROM16(dcbtable[20]); } else { i2ctabptr = ROM16(dcbtable[2]); sig = ROM32(dcbtable[4]); @@ -5596,7 +5654,7 @@ parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads) "signature (%08X)\n", sig); return -EINVAL; } - } else if (bdcb->version >= 0x15) { /* some NV11 and NV20 */ + } else if (dcb->version >= 0x15) { /* some NV11 and NV20 */ char sig[8] = { 0 }; strncpy(sig, (char *)&dcbtable[-7], 7); @@ -5644,14 +5702,11 @@ parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads) if (!i2ctabptr) NV_WARN(dev, "No pointer to DCB I2C port table\n"); else { - bdcb->i2c_table = &bios->data[i2ctabptr]; - if (bdcb->version >= 0x30) - bdcb->i2c_default_indices = bdcb->i2c_table[4]; + dcb->i2c_table = &bios->data[i2ctabptr]; + if (dcb->version >= 0x30) + dcb->i2c_default_indices = dcb->i2c_table[4]; } - parse_dcb_gpio_table(bios); - parse_dcb_connector_table(bios); - if (entries > DCB_MAX_NUM_ENTRIES) entries = DCB_MAX_NUM_ENTRIES; @@ -5676,7 +5731,7 @@ parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads) NV_TRACEWARN(dev, "Raw DCB entry %d: %08x %08x\n", dcb->entries, connection, config); - if (!parse_dcb_entry(dev, bdcb, connection, config)) + if (!parse_dcb_entry(dev, dcb, connection, config)) break; } @@ -5684,18 +5739,22 @@ parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads) * apart for v2.1+ not being known for requiring merging, this * guarantees dcbent->index is the index of the entry in the rom image */ - if (bdcb->version < 0x21) + if (dcb->version < 0x21) merge_like_dcb_entries(dev, dcb); - return dcb->entries ? 0 : -ENXIO; + if (!dcb->entries) + return -ENXIO; + + parse_dcb_gpio_table(bios); + parse_dcb_connector_table(bios); + return 0; } static void fixup_legacy_connector(struct nvbios *bios) { - struct bios_parsed_dcb *bdcb = &bios->bdcb; - struct parsed_dcb *dcb = &bdcb->dcb; - int high = 0, i; + struct dcb_table *dcb = &bios->dcb; + int i, i2c, i2c_conn[DCB_MAX_NUM_I2C_ENTRIES] = { }; /* * DCB 3.0 also has the table in most cases, but there are some cards @@ -5703,9 +5762,11 @@ fixup_legacy_connector(struct nvbios *bios) * indices are all 0. We don't need the connector indices on pre-G80 * chips (yet?) so limit the use to DCB 4.0 and above. */ - if (bdcb->version >= 0x40) + if (dcb->version >= 0x40) return; + dcb->connector.entries = 0; + /* * No known connector info before v3.0, so make it up. the rule here * is: anything on the same i2c bus is considered to be on the same @@ -5713,37 +5774,38 @@ fixup_legacy_connector(struct nvbios *bios) * its own unique connector index. */ for (i = 0; i < dcb->entries; i++) { - if (dcb->entry[i].i2c_index == 0xf) - continue; - /* * Ignore the I2C index for on-chip TV-out, as there * are cards with bogus values (nv31m in bug 23212), * and it's otherwise useless. */ if (dcb->entry[i].type == OUTPUT_TV && - dcb->entry[i].location == DCB_LOC_ON_CHIP) { + dcb->entry[i].location == DCB_LOC_ON_CHIP) dcb->entry[i].i2c_index = 0xf; + i2c = dcb->entry[i].i2c_index; + + if (i2c_conn[i2c]) { + dcb->entry[i].connector = i2c_conn[i2c] - 1; continue; } - dcb->entry[i].connector = dcb->entry[i].i2c_index; - if (dcb->entry[i].connector > high) - high = dcb->entry[i].connector; + dcb->entry[i].connector = dcb->connector.entries++; + if (i2c != 0xf) + i2c_conn[i2c] = dcb->connector.entries; } - for (i = 0; i < dcb->entries; i++) { - if (dcb->entry[i].i2c_index != 0xf) - continue; - - dcb->entry[i].connector = ++high; + /* Fake the connector table as well as just connector indices */ + for (i = 0; i < dcb->connector.entries; i++) { + dcb->connector.entry[i].index = i; + dcb->connector.entry[i].type = divine_connector_type(bios, i); + dcb->connector.entry[i].gpio_tag = 0xff; } } static void fixup_legacy_i2c(struct nvbios *bios) { - struct parsed_dcb *dcb = &bios->bdcb.dcb; + struct dcb_table *dcb = &bios->dcb; int i; for (i = 0; i < dcb->entries; i++) { @@ -5829,7 +5891,7 @@ static int load_nv17_hw_sequencer_ucode(struct drm_device *dev, uint8_t *nouveau_bios_embedded_edid(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; const uint8_t edid_sig[] = { 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00 }; uint16_t offset = 0; @@ -5862,20 +5924,23 @@ nouveau_bios_run_init_table(struct drm_device *dev, uint16_t table, struct dcb_entry *dcbent) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; struct init_exec iexec = { true, false }; + mutex_lock(&bios->lock); bios->display.output = dcbent; parse_init_table(bios, table, &iexec); bios->display.output = NULL; + mutex_unlock(&bios->lock); } static bool NVInitVBIOS(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; memset(bios, 0, sizeof(struct nvbios)); + mutex_init(&bios->lock); bios->dev = dev; if (!NVShadowVBIOS(dev, bios->data)) @@ -5888,7 +5953,7 @@ static bool NVInitVBIOS(struct drm_device *dev) static int nouveau_parse_vbios_struct(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; const uint8_t bit_signature[] = { 0xff, 0xb8, 'B', 'I', 'T' }; const uint8_t bmp_signature[] = { 0xff, 0x7f, 'N', 'V', 0x0 }; int offset; @@ -5915,7 +5980,7 @@ int nouveau_run_vbios_init(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; int i, ret = 0; NVLockVgaCrtcs(dev, false); @@ -5946,9 +6011,9 @@ nouveau_run_vbios_init(struct drm_device *dev) } if (dev_priv->card_type >= NV_50) { - for (i = 0; i < bios->bdcb.dcb.entries; i++) { + for (i = 0; i < bios->dcb.entries; i++) { nouveau_bios_run_display_table(dev, - &bios->bdcb.dcb.entry[i], + &bios->dcb.entry[i], 0, 0); } } @@ -5962,11 +6027,11 @@ static void nouveau_bios_i2c_devices_takedown(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; struct dcb_i2c_entry *entry; int i; - entry = &bios->bdcb.dcb.i2c[0]; + entry = &bios->dcb.i2c[0]; for (i = 0; i < DCB_MAX_NUM_I2C_ENTRIES; i++, entry++) nouveau_i2c_fini(dev, entry); } @@ -5975,13 +6040,11 @@ int nouveau_bios_init(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; uint32_t saved_nv_pextdev_boot_0; bool was_locked; int ret; - dev_priv->vbios = &bios->pub; - if (!NVInitVBIOS(dev)) return -ENODEV; @@ -6023,10 +6086,8 @@ nouveau_bios_init(struct drm_device *dev) bios_wr32(bios, NV_PEXTDEV_BOOT_0, saved_nv_pextdev_boot_0); ret = nouveau_run_vbios_init(dev); - if (ret) { - dev_priv->vbios = NULL; + if (ret) return ret; - } /* feature_byte on BMP is poor, but init always sets CR4B */ was_locked = NVLockVgaCrtcs(dev, false); diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.h b/drivers/gpu/drm/nouveau/nouveau_bios.h index 058e98c76d89..9f688aa9a655 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bios.h +++ b/drivers/gpu/drm/nouveau/nouveau_bios.h @@ -34,9 +34,67 @@ #define DCB_LOC_ON_CHIP 0 +struct dcb_i2c_entry { + uint8_t port_type; + uint8_t read, write; + struct nouveau_i2c_chan *chan; +}; + +enum dcb_gpio_tag { + DCB_GPIO_TVDAC0 = 0xc, + DCB_GPIO_TVDAC1 = 0x2d, +}; + +struct dcb_gpio_entry { + enum dcb_gpio_tag tag; + int line; + bool invert; +}; + +struct dcb_gpio_table { + int entries; + struct dcb_gpio_entry entry[DCB_MAX_NUM_GPIO_ENTRIES]; +}; + +enum dcb_connector_type { + DCB_CONNECTOR_VGA = 0x00, + DCB_CONNECTOR_TV_0 = 0x10, + DCB_CONNECTOR_TV_1 = 0x11, + DCB_CONNECTOR_TV_3 = 0x13, + DCB_CONNECTOR_DVI_I = 0x30, + DCB_CONNECTOR_DVI_D = 0x31, + DCB_CONNECTOR_LVDS = 0x40, + DCB_CONNECTOR_DP = 0x46, + DCB_CONNECTOR_eDP = 0x47, + DCB_CONNECTOR_HDMI_0 = 0x60, + DCB_CONNECTOR_HDMI_1 = 0x61, + DCB_CONNECTOR_NONE = 0xff +}; + +struct dcb_connector_table_entry { + uint32_t entry; + enum dcb_connector_type type; + uint8_t index; + uint8_t gpio_tag; +}; + +struct dcb_connector_table { + int entries; + struct dcb_connector_table_entry entry[DCB_MAX_NUM_CONNECTOR_ENTRIES]; +}; + +enum dcb_type { + OUTPUT_ANALOG = 0, + OUTPUT_TV = 1, + OUTPUT_TMDS = 2, + OUTPUT_LVDS = 3, + OUTPUT_DP = 6, + OUTPUT_ANY = -1 +}; + struct dcb_entry { int index; /* may not be raw dcb index if merging has happened */ - uint8_t type; + enum dcb_type type; uint8_t i2c_index; uint8_t heads; uint8_t connector; @@ -71,69 +129,22 @@ struct dcb_entry { bool i2c_upper_default; }; -struct dcb_i2c_entry { - uint8_t port_type; - uint8_t read, write; - struct nouveau_i2c_chan *chan; -}; +struct dcb_table { + uint8_t version; -struct parsed_dcb { int entries; struct dcb_entry entry[DCB_MAX_NUM_ENTRIES]; - struct dcb_i2c_entry i2c[DCB_MAX_NUM_I2C_ENTRIES]; -}; - -enum dcb_gpio_tag { - DCB_GPIO_TVDAC0 = 0xc, - DCB_GPIO_TVDAC1 = 0x2d, -}; - -struct dcb_gpio_entry { - enum dcb_gpio_tag tag; - int line; - bool invert; -}; - -struct parsed_dcb_gpio { - int entries; - struct dcb_gpio_entry entry[DCB_MAX_NUM_GPIO_ENTRIES]; -}; - -struct dcb_connector_table_entry { - uint32_t entry; - uint8_t type; - uint8_t index; - uint8_t gpio_tag; -}; - -struct dcb_connector_table { - int entries; - struct dcb_connector_table_entry entry[DCB_MAX_NUM_CONNECTOR_ENTRIES]; -}; - -struct bios_parsed_dcb { - uint8_t version; - - struct parsed_dcb dcb; uint8_t *i2c_table; uint8_t i2c_default_indices; + struct dcb_i2c_entry i2c[DCB_MAX_NUM_I2C_ENTRIES]; uint16_t gpio_table_ptr; - struct parsed_dcb_gpio gpio; + struct dcb_gpio_table gpio; uint16_t connector_table_ptr; struct dcb_connector_table connector; }; -enum nouveau_encoder_type { - OUTPUT_ANALOG = 0, - OUTPUT_TV = 1, - OUTPUT_TMDS = 2, - OUTPUT_LVDS = 3, - OUTPUT_DP = 6, - OUTPUT_ANY = -1 -}; - enum nouveau_or { OUTPUT_A = (1 << 0), OUTPUT_B = (1 << 1), @@ -190,8 +201,8 @@ struct pll_lims { int refclk; }; -struct nouveau_bios_info { - struct parsed_dcb *dcb; +struct nvbios { + struct drm_device *dev; uint8_t chip_version; @@ -199,11 +210,8 @@ struct nouveau_bios_info { uint32_t tvdactestval; uint8_t digital_min_front_porch; bool fp_no_ddc; -}; -struct nvbios { - struct drm_device *dev; - struct nouveau_bios_info pub; + struct mutex lock; uint8_t data[NV_PROM_SIZE]; unsigned int length; @@ -232,7 +240,7 @@ struct nvbios { uint16_t some_script_ptr; /* BIT I + 14 */ uint16_t init96_tbl_ptr; /* BIT I + 16 */ - struct bios_parsed_dcb bdcb; + struct dcb_table dcb; struct { int crtchead; diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index db0ed4c13f98..028719fddf76 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -65,8 +65,10 @@ nouveau_bo_fixup_align(struct drm_device *dev, /* * Some of the tile_flags have a periodic structure of N*4096 bytes, - * align to to that as well as the page size. Overallocate memory to - * avoid corruption of other buffer objects. + * align to to that as well as the page size. Align the size to the + * appropriate boundaries. This does imply that sizes are rounded up + * 3-7 pages, so be aware of this and do not waste memory by allocating + * many small buffers. */ if (dev_priv->card_type == NV_50) { uint32_t block_size = nouveau_mem_fb_amount(dev) >> 15; @@ -77,22 +79,20 @@ nouveau_bo_fixup_align(struct drm_device *dev, case 0x2800: case 0x4800: case 0x7a00: - *size = roundup(*size, block_size); if (is_power_of_2(block_size)) { - *size += 3 * block_size; for (i = 1; i < 10; i++) { *align = 12 * i * block_size; if (!(*align % 65536)) break; } } else { - *size += 6 * block_size; for (i = 1; i < 10; i++) { *align = 8 * i * block_size; if (!(*align % 65536)) break; } } + *size = roundup(*size, *align); break; default: break; diff --git a/drivers/gpu/drm/nouveau/nouveau_calc.c b/drivers/gpu/drm/nouveau/nouveau_calc.c index ee2b84504d05..88f9bc0941eb 100644 --- a/drivers/gpu/drm/nouveau/nouveau_calc.c +++ b/drivers/gpu/drm/nouveau/nouveau_calc.c @@ -274,7 +274,7 @@ getMNP_single(struct drm_device *dev, struct pll_lims *pll_lim, int clk, * returns calculated clock */ struct drm_nouveau_private *dev_priv = dev->dev_private; - int cv = dev_priv->vbios->chip_version; + int cv = dev_priv->vbios.chip_version; int minvco = pll_lim->vco1.minfreq, maxvco = pll_lim->vco1.maxfreq; int minM = pll_lim->vco1.min_m, maxM = pll_lim->vco1.max_m; int minN = pll_lim->vco1.min_n, maxN = pll_lim->vco1.max_n; @@ -373,7 +373,7 @@ getMNP_double(struct drm_device *dev, struct pll_lims *pll_lim, int clk, * returns calculated clock */ struct drm_nouveau_private *dev_priv = dev->dev_private; - int chip_version = dev_priv->vbios->chip_version; + int chip_version = dev_priv->vbios.chip_version; int minvco1 = pll_lim->vco1.minfreq, maxvco1 = pll_lim->vco1.maxfreq; int minvco2 = pll_lim->vco2.minfreq, maxvco2 = pll_lim->vco2.maxfreq; int minU1 = pll_lim->vco1.min_inputfreq, minU2 = pll_lim->vco2.min_inputfreq; diff --git a/drivers/gpu/drm/nouveau/nouveau_channel.c b/drivers/gpu/drm/nouveau/nouveau_channel.c index 343d718a9667..6dfb425cbae9 100644 --- a/drivers/gpu/drm/nouveau/nouveau_channel.c +++ b/drivers/gpu/drm/nouveau/nouveau_channel.c @@ -35,22 +35,27 @@ nouveau_channel_pushbuf_ctxdma_init(struct nouveau_channel *chan) struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_bo *pb = chan->pushbuf_bo; struct nouveau_gpuobj *pushbuf = NULL; - uint32_t start = pb->bo.mem.mm_node->start << PAGE_SHIFT; int ret; + if (dev_priv->card_type >= NV_50) { + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0, + dev_priv->vm_end, NV_DMA_ACCESS_RO, + NV_DMA_TARGET_AGP, &pushbuf); + chan->pushbuf_base = pb->bo.offset; + } else if (pb->bo.mem.mem_type == TTM_PL_TT) { ret = nouveau_gpuobj_gart_dma_new(chan, 0, dev_priv->gart_info.aper_size, NV_DMA_ACCESS_RO, &pushbuf, NULL); - chan->pushbuf_base = start; + chan->pushbuf_base = pb->bo.mem.mm_node->start << PAGE_SHIFT; } else if (dev_priv->card_type != NV_04) { ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0, dev_priv->fb_available_size, NV_DMA_ACCESS_RO, NV_DMA_TARGET_VIDMEM, &pushbuf); - chan->pushbuf_base = start; + chan->pushbuf_base = pb->bo.mem.mm_node->start << PAGE_SHIFT; } else { /* NV04 cmdbuf hack, from original ddx.. not sure of it's * exact reason for existing :) PCI access to cmdbuf in @@ -61,7 +66,7 @@ nouveau_channel_pushbuf_ctxdma_init(struct nouveau_channel *chan) dev_priv->fb_available_size, NV_DMA_ACCESS_RO, NV_DMA_TARGET_PCI, &pushbuf); - chan->pushbuf_base = start; + chan->pushbuf_base = pb->bo.mem.mm_node->start << PAGE_SHIFT; } ret = nouveau_gpuobj_ref_add(dev, chan, 0, pushbuf, &chan->pushbuf); @@ -275,15 +280,23 @@ nouveau_channel_free(struct nouveau_channel *chan) */ nouveau_fence_fini(chan); - /* Ensure the channel is no longer active on the GPU */ + /* This will prevent pfifo from switching channels. */ pfifo->reassign(dev, false); - if (pgraph->channel(dev) == chan) { - pgraph->fifo_access(dev, false); + /* We want to give pgraph a chance to idle and get rid of all potential + * errors. We need to do this before the lock, otherwise the irq handler + * is unable to process them. + */ + if (pgraph->channel(dev) == chan) + nouveau_wait_for_idle(dev); + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + + pgraph->fifo_access(dev, false); + if (pgraph->channel(dev) == chan) pgraph->unload_context(dev); - pgraph->fifo_access(dev, true); - } pgraph->destroy_context(chan); + pgraph->fifo_access(dev, true); if (pfifo->channel_id(dev) == chan->id) { pfifo->disable(dev); @@ -294,6 +307,8 @@ nouveau_channel_free(struct nouveau_channel *chan) pfifo->reassign(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + /* Release the channel's resources */ nouveau_gpuobj_ref_del(dev, &chan->pushbuf); if (chan->pushbuf_bo) { @@ -370,6 +385,14 @@ nouveau_ioctl_fifo_alloc(struct drm_device *dev, void *data, return ret; init->channel = chan->id; + if (chan->dma.ib_max) + init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_VRAM | + NOUVEAU_GEM_DOMAIN_GART; + else if (chan->pushbuf_bo->bo.mem.mem_type == TTM_PL_VRAM) + init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_VRAM; + else + init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_GART; + init->subchan[0].handle = NvM2MF; if (dev_priv->card_type < NV_50) init->subchan[0].grclass = 0x0039; @@ -409,7 +432,6 @@ nouveau_ioctl_fifo_free(struct drm_device *dev, void *data, ***********************************/ struct drm_ioctl_desc nouveau_ioctls[] = { - DRM_IOCTL_DEF(DRM_NOUVEAU_CARD_INIT, nouveau_ioctl_card_init, DRM_AUTH), DRM_IOCTL_DEF(DRM_NOUVEAU_GETPARAM, nouveau_ioctl_getparam, DRM_AUTH), DRM_IOCTL_DEF(DRM_NOUVEAU_SETPARAM, nouveau_ioctl_setparam, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), DRM_IOCTL_DEF(DRM_NOUVEAU_CHANNEL_ALLOC, nouveau_ioctl_fifo_alloc, DRM_AUTH), @@ -419,13 +441,9 @@ struct drm_ioctl_desc nouveau_ioctls[] = { DRM_IOCTL_DEF(DRM_NOUVEAU_GPUOBJ_FREE, nouveau_ioctl_gpuobj_free, DRM_AUTH), DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_NEW, nouveau_gem_ioctl_new, DRM_AUTH), DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_PUSHBUF, nouveau_gem_ioctl_pushbuf, DRM_AUTH), - DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_PUSHBUF_CALL, nouveau_gem_ioctl_pushbuf_call, DRM_AUTH), - DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_PIN, nouveau_gem_ioctl_pin, DRM_AUTH), - DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_UNPIN, nouveau_gem_ioctl_unpin, DRM_AUTH), DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_CPU_PREP, nouveau_gem_ioctl_cpu_prep, DRM_AUTH), DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_CPU_FINI, nouveau_gem_ioctl_cpu_fini, DRM_AUTH), DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_INFO, nouveau_gem_ioctl_info, DRM_AUTH), - DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_PUSHBUF_CALL2, nouveau_gem_ioctl_pushbuf_call2, DRM_AUTH), }; int nouveau_max_ioctl = DRM_ARRAY_SIZE(nouveau_ioctls); diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c index 7e6d673f3a23..24327f468c4b 100644 --- a/drivers/gpu/drm/nouveau/nouveau_connector.c +++ b/drivers/gpu/drm/nouveau/nouveau_connector.c @@ -88,13 +88,14 @@ nouveau_connector_destroy(struct drm_connector *drm_connector) { struct nouveau_connector *nv_connector = nouveau_connector(drm_connector); - struct drm_device *dev = nv_connector->base.dev; - - NV_DEBUG_KMS(dev, "\n"); + struct drm_device *dev; if (!nv_connector) return; + dev = nv_connector->base.dev; + NV_DEBUG_KMS(dev, "\n"); + kfree(nv_connector->edid); drm_sysfs_connector_remove(drm_connector); drm_connector_cleanup(drm_connector); @@ -217,7 +218,7 @@ nouveau_connector_set_encoder(struct drm_connector *connector, connector->interlace_allowed = true; } - if (connector->connector_type == DRM_MODE_CONNECTOR_DVII) { + if (nv_connector->dcb->type == DCB_CONNECTOR_DVI_I) { drm_connector_property_set_value(connector, dev->mode_config.dvi_i_subconnector_property, nv_encoder->dcb->type == OUTPUT_TMDS ? @@ -235,15 +236,17 @@ nouveau_connector_detect(struct drm_connector *connector) struct nouveau_i2c_chan *i2c; int type, flags; - if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS) + if (nv_connector->dcb->type == DCB_CONNECTOR_LVDS) nv_encoder = find_encoder_by_type(connector, OUTPUT_LVDS); if (nv_encoder && nv_connector->native_mode) { + unsigned status = connector_status_connected; + #ifdef CONFIG_ACPI if (!nouveau_ignorelid && !acpi_lid_open()) - return connector_status_disconnected; + status = connector_status_unknown; #endif nouveau_connector_set_encoder(connector, nv_encoder); - return connector_status_connected; + return status; } /* Cleanup the previous EDID block. */ @@ -278,7 +281,7 @@ nouveau_connector_detect(struct drm_connector *connector) * same i2c channel so the value returned from ddc_detect * isn't necessarily correct. */ - if (connector->connector_type == DRM_MODE_CONNECTOR_DVII) { + if (nv_connector->dcb->type == DCB_CONNECTOR_DVI_I) { if (nv_connector->edid->input & DRM_EDID_INPUT_DIGITAL) type = OUTPUT_TMDS; else @@ -320,11 +323,11 @@ detect_analog: static void nouveau_connector_force(struct drm_connector *connector) { - struct drm_device *dev = connector->dev; + struct nouveau_connector *nv_connector = nouveau_connector(connector); struct nouveau_encoder *nv_encoder; int type; - if (connector->connector_type == DRM_MODE_CONNECTOR_DVII) { + if (nv_connector->dcb->type == DCB_CONNECTOR_DVI_I) { if (connector->force == DRM_FORCE_ON_DIGITAL) type = OUTPUT_TMDS; else @@ -334,7 +337,7 @@ nouveau_connector_force(struct drm_connector *connector) nv_encoder = find_encoder_by_type(connector, type); if (!nv_encoder) { - NV_ERROR(dev, "can't find encoder to force %s on!\n", + NV_ERROR(connector->dev, "can't find encoder to force %s on!\n", drm_get_connector_name(connector)); connector->status = connector_status_disconnected; return; @@ -368,7 +371,7 @@ nouveau_connector_set_property(struct drm_connector *connector, } /* LVDS always needs gpu scaling */ - if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS && + if (nv_connector->dcb->type == DCB_CONNECTOR_LVDS && value == DRM_MODE_SCALE_NONE) return -EINVAL; @@ -534,7 +537,7 @@ nouveau_connector_get_modes(struct drm_connector *connector) /* If we're not LVDS, destroy the previous native mode, the attached * monitor could have changed. */ - if (connector->connector_type != DRM_MODE_CONNECTOR_LVDS && + if (nv_connector->dcb->type != DCB_CONNECTOR_LVDS && nv_connector->native_mode) { drm_mode_destroy(dev, nv_connector->native_mode); nv_connector->native_mode = NULL; @@ -562,7 +565,7 @@ nouveau_connector_get_modes(struct drm_connector *connector) ret = get_slave_funcs(nv_encoder)-> get_modes(to_drm_encoder(nv_encoder), connector); - if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS) + if (nv_encoder->dcb->type == OUTPUT_LVDS) ret += nouveau_connector_scaler_modes_add(connector); return ret; @@ -612,6 +615,9 @@ nouveau_connector_mode_valid(struct drm_connector *connector, clock *= 3; break; + default: + BUG_ON(1); + return MODE_BAD; } if (clock < min_clock) @@ -679,7 +685,7 @@ nouveau_connector_create_lvds(struct drm_device *dev, /* Firstly try getting EDID over DDC, if allowed and I2C channel * is available. */ - if (!dev_priv->VBIOS.pub.fp_no_ddc && nv_encoder->dcb->i2c_index < 0xf) + if (!dev_priv->vbios.fp_no_ddc && nv_encoder->dcb->i2c_index < 0xf) i2c = nouveau_i2c_find(dev, nv_encoder->dcb->i2c_index); if (i2c) { @@ -694,7 +700,7 @@ nouveau_connector_create_lvds(struct drm_device *dev, */ if (!nv_connector->edid && nouveau_bios_fp_mode(dev, &native) && (nv_encoder->dcb->lvdsconf.use_straps_for_mode || - dev_priv->VBIOS.pub.fp_no_ddc)) { + dev_priv->vbios.fp_no_ddc)) { nv_connector->native_mode = drm_mode_duplicate(dev, &native); goto out; } @@ -703,7 +709,7 @@ nouveau_connector_create_lvds(struct drm_device *dev, * stored for the panel stored in them. */ if (!nv_connector->edid && !nv_connector->native_mode && - !dev_priv->VBIOS.pub.fp_no_ddc) { + !dev_priv->vbios.fp_no_ddc) { struct edid *edid = (struct edid *)nouveau_bios_embedded_edid(dev); if (edid) { @@ -738,46 +744,66 @@ out: } int -nouveau_connector_create(struct drm_device *dev, int index, int type) +nouveau_connector_create(struct drm_device *dev, + struct dcb_connector_table_entry *dcb) { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_connector *nv_connector = NULL; struct drm_connector *connector; struct drm_encoder *encoder; - int ret; + int ret, type; NV_DEBUG_KMS(dev, "\n"); - nv_connector = kzalloc(sizeof(*nv_connector), GFP_KERNEL); - if (!nv_connector) - return -ENOMEM; - nv_connector->dcb = nouveau_bios_connector_entry(dev, index); - connector = &nv_connector->base; - - switch (type) { - case DRM_MODE_CONNECTOR_VGA: + switch (dcb->type) { + case DCB_CONNECTOR_NONE: + return 0; + case DCB_CONNECTOR_VGA: NV_INFO(dev, "Detected a VGA connector\n"); + type = DRM_MODE_CONNECTOR_VGA; break; - case DRM_MODE_CONNECTOR_DVID: - NV_INFO(dev, "Detected a DVI-D connector\n"); + case DCB_CONNECTOR_TV_0: + case DCB_CONNECTOR_TV_1: + case DCB_CONNECTOR_TV_3: + NV_INFO(dev, "Detected a TV connector\n"); + type = DRM_MODE_CONNECTOR_TV; break; - case DRM_MODE_CONNECTOR_DVII: + case DCB_CONNECTOR_DVI_I: NV_INFO(dev, "Detected a DVI-I connector\n"); + type = DRM_MODE_CONNECTOR_DVII; break; - case DRM_MODE_CONNECTOR_LVDS: - NV_INFO(dev, "Detected a LVDS connector\n"); + case DCB_CONNECTOR_DVI_D: + NV_INFO(dev, "Detected a DVI-D connector\n"); + type = DRM_MODE_CONNECTOR_DVID; break; - case DRM_MODE_CONNECTOR_TV: - NV_INFO(dev, "Detected a TV connector\n"); + case DCB_CONNECTOR_HDMI_0: + case DCB_CONNECTOR_HDMI_1: + NV_INFO(dev, "Detected a HDMI connector\n"); + type = DRM_MODE_CONNECTOR_HDMIA; break; - case DRM_MODE_CONNECTOR_DisplayPort: + case DCB_CONNECTOR_LVDS: + NV_INFO(dev, "Detected a LVDS connector\n"); + type = DRM_MODE_CONNECTOR_LVDS; + break; + case DCB_CONNECTOR_DP: NV_INFO(dev, "Detected a DisplayPort connector\n"); + type = DRM_MODE_CONNECTOR_DisplayPort; break; - default: - NV_ERROR(dev, "Unknown connector, this is not good.\n"); + case DCB_CONNECTOR_eDP: + NV_INFO(dev, "Detected an eDP connector\n"); + type = DRM_MODE_CONNECTOR_eDP; break; + default: + NV_ERROR(dev, "unknown connector type: 0x%02x!!\n", dcb->type); + return -EINVAL; } + nv_connector = kzalloc(sizeof(*nv_connector), GFP_KERNEL); + if (!nv_connector) + return -ENOMEM; + nv_connector->dcb = dcb; + connector = &nv_connector->base; + /* defaults, will get overridden in detect() */ connector->interlace_allowed = false; connector->doublescan_allowed = false; @@ -785,55 +811,65 @@ nouveau_connector_create(struct drm_device *dev, int index, int type) drm_connector_init(dev, connector, &nouveau_connector_funcs, type); drm_connector_helper_add(connector, &nouveau_connector_helper_funcs); + /* attach encoders */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (nv_encoder->dcb->connector != dcb->index) + continue; + + if (get_slave_funcs(nv_encoder)) + get_slave_funcs(nv_encoder)->create_resources(encoder, connector); + + drm_mode_connector_attach_encoder(connector, encoder); + } + + if (!connector->encoder_ids[0]) { + NV_WARN(dev, " no encoders, ignoring\n"); + drm_connector_cleanup(connector); + kfree(connector); + return 0; + } + /* Init DVI-I specific properties */ - if (type == DRM_MODE_CONNECTOR_DVII) { + if (dcb->type == DCB_CONNECTOR_DVI_I) { drm_mode_create_dvi_i_properties(dev); drm_connector_attach_property(connector, dev->mode_config.dvi_i_subconnector_property, 0); drm_connector_attach_property(connector, dev->mode_config.dvi_i_select_subconnector_property, 0); } - if (type != DRM_MODE_CONNECTOR_LVDS) + if (dcb->type != DCB_CONNECTOR_LVDS) nv_connector->use_dithering = false; - if (type == DRM_MODE_CONNECTOR_DVID || - type == DRM_MODE_CONNECTOR_DVII || - type == DRM_MODE_CONNECTOR_LVDS || - type == DRM_MODE_CONNECTOR_DisplayPort) { - nv_connector->scaling_mode = DRM_MODE_SCALE_FULLSCREEN; - - drm_connector_attach_property(connector, dev->mode_config.scaling_mode_property, - nv_connector->scaling_mode); - drm_connector_attach_property(connector, dev->mode_config.dithering_mode_property, - nv_connector->use_dithering ? DRM_MODE_DITHERING_ON - : DRM_MODE_DITHERING_OFF); - - } else { - nv_connector->scaling_mode = DRM_MODE_SCALE_NONE; - - if (type == DRM_MODE_CONNECTOR_VGA && - dev_priv->card_type >= NV_50) { + switch (dcb->type) { + case DCB_CONNECTOR_VGA: + if (dev_priv->card_type >= NV_50) { drm_connector_attach_property(connector, dev->mode_config.scaling_mode_property, nv_connector->scaling_mode); } - } - - /* attach encoders */ - list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { - struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); - - if (nv_encoder->dcb->connector != index) - continue; - - if (get_slave_funcs(nv_encoder)) - get_slave_funcs(nv_encoder)->create_resources(encoder, connector); + /* fall-through */ + case DCB_CONNECTOR_TV_0: + case DCB_CONNECTOR_TV_1: + case DCB_CONNECTOR_TV_3: + nv_connector->scaling_mode = DRM_MODE_SCALE_NONE; + break; + default: + nv_connector->scaling_mode = DRM_MODE_SCALE_FULLSCREEN; - drm_mode_connector_attach_encoder(connector, encoder); + drm_connector_attach_property(connector, + dev->mode_config.scaling_mode_property, + nv_connector->scaling_mode); + drm_connector_attach_property(connector, + dev->mode_config.dithering_mode_property, + nv_connector->use_dithering ? + DRM_MODE_DITHERING_ON : DRM_MODE_DITHERING_OFF); + break; } drm_sysfs_connector_add(connector); - if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS) { + if (dcb->type == DCB_CONNECTOR_LVDS) { ret = nouveau_connector_create_lvds(dev, connector); if (ret) { connector->funcs->destroy(connector); diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.h b/drivers/gpu/drm/nouveau/nouveau_connector.h index 728b8090e5ff..4ef38abc2d9c 100644 --- a/drivers/gpu/drm/nouveau/nouveau_connector.h +++ b/drivers/gpu/drm/nouveau/nouveau_connector.h @@ -49,6 +49,7 @@ static inline struct nouveau_connector *nouveau_connector( return container_of(con, struct nouveau_connector, base); } -int nouveau_connector_create(struct drm_device *dev, int i2c_index, int type); +int nouveau_connector_create(struct drm_device *, + struct dcb_connector_table_entry *); #endif /* __NOUVEAU_CONNECTOR_H__ */ diff --git a/drivers/gpu/drm/nouveau/nouveau_debugfs.c b/drivers/gpu/drm/nouveau/nouveau_debugfs.c index d79db3698f16..8ff9ef5d4b47 100644 --- a/drivers/gpu/drm/nouveau/nouveau_debugfs.c +++ b/drivers/gpu/drm/nouveau/nouveau_debugfs.c @@ -47,12 +47,23 @@ nouveau_debugfs_channel_info(struct seq_file *m, void *data) seq_printf(m, " cur: 0x%08x\n", chan->dma.cur << 2); seq_printf(m, " put: 0x%08x\n", chan->dma.put << 2); seq_printf(m, " free: 0x%08x\n", chan->dma.free << 2); + if (chan->dma.ib_max) { + seq_printf(m, " ib max: 0x%08x\n", chan->dma.ib_max); + seq_printf(m, " ib put: 0x%08x\n", chan->dma.ib_put); + seq_printf(m, " ib free: 0x%08x\n", chan->dma.ib_free); + } seq_printf(m, "gpu fifo state:\n"); seq_printf(m, " get: 0x%08x\n", nvchan_rd32(chan, chan->user_get)); seq_printf(m, " put: 0x%08x\n", nvchan_rd32(chan, chan->user_put)); + if (chan->dma.ib_max) { + seq_printf(m, " ib get: 0x%08x\n", + nvchan_rd32(chan, 0x88)); + seq_printf(m, " ib put: 0x%08x\n", + nvchan_rd32(chan, 0x8c)); + } seq_printf(m, "last fence : %d\n", chan->fence.sequence); seq_printf(m, "last signalled: %d\n", chan->fence.sequence_ack); @@ -133,9 +144,22 @@ nouveau_debugfs_memory_info(struct seq_file *m, void *data) return 0; } +static int +nouveau_debugfs_vbios_image(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_nouveau_private *dev_priv = node->minor->dev->dev_private; + int i; + + for (i = 0; i < dev_priv->vbios.length; i++) + seq_printf(m, "%c", dev_priv->vbios.data[i]); + return 0; +} + static struct drm_info_list nouveau_debugfs_list[] = { { "chipset", nouveau_debugfs_chipset_info, 0, NULL }, { "memory", nouveau_debugfs_memory_info, 0, NULL }, + { "vbios.rom", nouveau_debugfs_vbios_image, 0, NULL }, }; #define NOUVEAU_DEBUGFS_ENTRIES ARRAY_SIZE(nouveau_debugfs_list) diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c b/drivers/gpu/drm/nouveau/nouveau_display.c index dfc94391d71e..cf1c5c0a0abe 100644 --- a/drivers/gpu/drm/nouveau/nouveau_display.c +++ b/drivers/gpu/drm/nouveau/nouveau_display.c @@ -39,11 +39,8 @@ nouveau_user_framebuffer_destroy(struct drm_framebuffer *drm_fb) if (drm_fb->fbdev) nouveau_fbcon_remove(dev, drm_fb); - if (fb->nvbo) { - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(fb->nvbo->gem); - mutex_unlock(&dev->struct_mutex); - } + if (fb->nvbo) + drm_gem_object_unreference_unlocked(fb->nvbo->gem); drm_framebuffer_cleanup(drm_fb); kfree(fb); diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.c b/drivers/gpu/drm/nouveau/nouveau_dma.c index 50d9e67745af..c8482a108a78 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dma.c +++ b/drivers/gpu/drm/nouveau/nouveau_dma.c @@ -32,7 +32,22 @@ void nouveau_dma_pre_init(struct nouveau_channel *chan) { - chan->dma.max = (chan->pushbuf_bo->bo.mem.size >> 2) - 2; + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_bo *pushbuf = chan->pushbuf_bo; + + if (dev_priv->card_type == NV_50) { + const int ib_size = pushbuf->bo.mem.size / 2; + + chan->dma.ib_base = (pushbuf->bo.mem.size - ib_size) >> 2; + chan->dma.ib_max = (ib_size / 8) - 1; + chan->dma.ib_put = 0; + chan->dma.ib_free = chan->dma.ib_max - chan->dma.ib_put; + + chan->dma.max = (pushbuf->bo.mem.size - ib_size) >> 2; + } else { + chan->dma.max = (pushbuf->bo.mem.size >> 2) - 2; + } + chan->dma.put = 0; chan->dma.cur = chan->dma.put; chan->dma.free = chan->dma.max - chan->dma.cur; @@ -162,12 +177,101 @@ READ_GET(struct nouveau_channel *chan, uint32_t *prev_get, uint32_t *timeout) return (val - chan->pushbuf_base) >> 2; } +void +nv50_dma_push(struct nouveau_channel *chan, struct nouveau_bo *bo, + int delta, int length) +{ + struct nouveau_bo *pb = chan->pushbuf_bo; + uint64_t offset = bo->bo.offset + delta; + int ip = (chan->dma.ib_put * 2) + chan->dma.ib_base; + + BUG_ON(chan->dma.ib_free < 1); + nouveau_bo_wr32(pb, ip++, lower_32_bits(offset)); + nouveau_bo_wr32(pb, ip++, upper_32_bits(offset) | length << 8); + + chan->dma.ib_put = (chan->dma.ib_put + 1) & chan->dma.ib_max; + nvchan_wr32(chan, 0x8c, chan->dma.ib_put); + chan->dma.ib_free--; +} + +static int +nv50_dma_push_wait(struct nouveau_channel *chan, int count) +{ + uint32_t cnt = 0, prev_get = 0; + + while (chan->dma.ib_free < count) { + uint32_t get = nvchan_rd32(chan, 0x88); + if (get != prev_get) { + prev_get = get; + cnt = 0; + } + + if ((++cnt & 0xff) == 0) { + DRM_UDELAY(1); + if (cnt > 100000) + return -EBUSY; + } + + chan->dma.ib_free = get - chan->dma.ib_put; + if (chan->dma.ib_free <= 0) + chan->dma.ib_free += chan->dma.ib_max + 1; + } + + return 0; +} + +static int +nv50_dma_wait(struct nouveau_channel *chan, int slots, int count) +{ + uint32_t cnt = 0, prev_get = 0; + int ret; + + ret = nv50_dma_push_wait(chan, slots + 1); + if (unlikely(ret)) + return ret; + + while (chan->dma.free < count) { + int get = READ_GET(chan, &prev_get, &cnt); + if (unlikely(get < 0)) { + if (get == -EINVAL) + continue; + + return get; + } + + if (get <= chan->dma.cur) { + chan->dma.free = chan->dma.max - chan->dma.cur; + if (chan->dma.free >= count) + break; + + FIRE_RING(chan); + do { + get = READ_GET(chan, &prev_get, &cnt); + if (unlikely(get < 0)) { + if (get == -EINVAL) + continue; + return get; + } + } while (get == 0); + chan->dma.cur = 0; + chan->dma.put = 0; + } + + chan->dma.free = get - chan->dma.cur - 1; + } + + return 0; +} + int -nouveau_dma_wait(struct nouveau_channel *chan, int size) +nouveau_dma_wait(struct nouveau_channel *chan, int slots, int size) { uint32_t prev_get = 0, cnt = 0; int get; + if (chan->dma.ib_max) + return nv50_dma_wait(chan, slots, size); + while (chan->dma.free < size) { get = READ_GET(chan, &prev_get, &cnt); if (unlikely(get == -EBUSY)) diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.h b/drivers/gpu/drm/nouveau/nouveau_dma.h index dabfd655f93e..8b05c15866d5 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dma.h +++ b/drivers/gpu/drm/nouveau/nouveau_dma.h @@ -31,6 +31,9 @@ #define NOUVEAU_DMA_DEBUG 0 #endif +void nv50_dma_push(struct nouveau_channel *, struct nouveau_bo *, + int delta, int length); + /* * There's a hw race condition where you can't jump to your PUT offset, * to avoid this we jump to offset + SKIPS and fill the difference with @@ -96,13 +99,11 @@ enum { static __must_check inline int RING_SPACE(struct nouveau_channel *chan, int size) { - if (chan->dma.free < size) { - int ret; + int ret; - ret = nouveau_dma_wait(chan, size); - if (ret) - return ret; - } + ret = nouveau_dma_wait(chan, 1, size); + if (ret) + return ret; chan->dma.free -= size; return 0; @@ -146,7 +147,13 @@ FIRE_RING(struct nouveau_channel *chan) return; chan->accel_done = true; - WRITE_PUT(chan->dma.cur); + if (chan->dma.ib_max) { + nv50_dma_push(chan, chan->pushbuf_bo, chan->dma.put << 2, + (chan->dma.cur - chan->dma.put) << 2); + } else { + WRITE_PUT(chan->dma.cur); + } + chan->dma.put = chan->dma.cur; } diff --git a/drivers/gpu/drm/nouveau/nouveau_dp.c b/drivers/gpu/drm/nouveau/nouveau_dp.c index dd4937224220..f954ad93e81f 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dp.c +++ b/drivers/gpu/drm/nouveau/nouveau_dp.c @@ -502,12 +502,12 @@ nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr, break; } - if ((stat & NV50_AUXCH_STAT_COUNT) != data_nr) { - ret = -EREMOTEIO; - goto out; - } - if (cmd & 1) { + if ((stat & NV50_AUXCH_STAT_COUNT) != data_nr) { + ret = -EREMOTEIO; + goto out; + } + for (i = 0; i < 4; i++) { data32[i] = nv_rd32(dev, NV50_AUXCH_DATA_IN(index, i)); NV_DEBUG_KMS(dev, "rd %d: 0x%08x\n", i, data32[i]); diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.c b/drivers/gpu/drm/nouveau/nouveau_drv.c index 343ab7f17ccc..874adf55a43f 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.c +++ b/drivers/gpu/drm/nouveau/nouveau_drv.c @@ -56,7 +56,7 @@ int nouveau_vram_pushbuf; module_param_named(vram_pushbuf, nouveau_vram_pushbuf, int, 0400); MODULE_PARM_DESC(vram_notify, "Force DMA notifiers to be in VRAM"); -int nouveau_vram_notify; +int nouveau_vram_notify = 1; module_param_named(vram_notify, nouveau_vram_notify, int, 0400); MODULE_PARM_DESC(duallink, "Allow dual-link TMDS (>=GeForce 8)"); @@ -75,6 +75,14 @@ MODULE_PARM_DESC(ignorelid, "Ignore ACPI lid status"); int nouveau_ignorelid = 0; module_param_named(ignorelid, nouveau_ignorelid, int, 0400); +MODULE_PARM_DESC(noaccel, "Disable all acceleration"); +int nouveau_noaccel = 0; +module_param_named(noaccel, nouveau_noaccel, int, 0400); + +MODULE_PARM_DESC(nofbaccel, "Disable fbcon acceleration"); +int nouveau_nofbaccel = 0; +module_param_named(nofbaccel, nouveau_nofbaccel, int, 0400); + MODULE_PARM_DESC(tv_norm, "Default TV norm.\n" "\t\tSupported: PAL, PAL-M, PAL-N, PAL-Nc, NTSC-M, NTSC-J,\n" "\t\t\thd480i, hd480p, hd576i, hd576p, hd720p, hd1080i.\n" diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h index 6b9690418bc7..2f8ce42f0725 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.h +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h @@ -34,7 +34,7 @@ #define DRIVER_MAJOR 0 #define DRIVER_MINOR 0 -#define DRIVER_PATCHLEVEL 15 +#define DRIVER_PATCHLEVEL 16 #define NOUVEAU_FAMILY 0x0000FFFF #define NOUVEAU_FLAGS 0xFFFF0000 @@ -83,6 +83,7 @@ struct nouveau_bo { struct drm_file *reserved_by; struct list_head entry; int pbbo_index; + bool validate_mapped; struct nouveau_channel *channel; @@ -239,6 +240,11 @@ struct nouveau_channel { int cur; int put; /* access via pushbuf_bo */ + + int ib_base; + int ib_max; + int ib_free; + int ib_put; } dma; uint32_t sw_subchannel[8]; @@ -533,6 +539,9 @@ struct drm_nouveau_private { struct nouveau_engine engine; struct nouveau_channel *channel; + /* For PFIFO and PGRAPH. */ + spinlock_t context_switch_lock; + /* RAMIN configuration, RAMFC, RAMHT and RAMRO offsets */ struct nouveau_gpuobj *ramht; uint32_t ramin_rsvd_vram; @@ -583,6 +592,7 @@ struct drm_nouveau_private { uint64_t vm_end; struct nouveau_gpuobj *vm_vram_pt[NV50_VM_VRAM_NR]; int vm_vram_pt_nr; + uint64_t vram_sys_base; /* the mtrr covering the FB */ int fb_mtrr; @@ -595,8 +605,7 @@ struct drm_nouveau_private { struct list_head gpuobj_list; - struct nvbios VBIOS; - struct nouveau_bios_info *vbios; + struct nvbios vbios; struct nv04_mode_state mode_reg; struct nv04_mode_state saved_reg; @@ -678,6 +687,8 @@ extern int nouveau_reg_debug; extern char *nouveau_vbios; extern int nouveau_ctxfw; extern int nouveau_ignorelid; +extern int nouveau_nofbaccel; +extern int nouveau_noaccel; /* nouveau_state.c */ extern void nouveau_preclose(struct drm_device *dev, struct drm_file *); @@ -693,12 +704,6 @@ extern bool nouveau_wait_until(struct drm_device *, uint64_t timeout, uint32_t reg, uint32_t mask, uint32_t val); extern bool nouveau_wait_for_idle(struct drm_device *); extern int nouveau_card_init(struct drm_device *); -extern int nouveau_ioctl_card_init(struct drm_device *, void *data, - struct drm_file *); -extern int nouveau_ioctl_suspend(struct drm_device *, void *data, - struct drm_file *); -extern int nouveau_ioctl_resume(struct drm_device *, void *data, - struct drm_file *); /* nouveau_mem.c */ extern int nouveau_mem_init_heap(struct mem_block **, uint64_t start, @@ -842,7 +847,7 @@ nouveau_debugfs_channel_fini(struct nouveau_channel *chan) /* nouveau_dma.c */ extern void nouveau_dma_pre_init(struct nouveau_channel *); extern int nouveau_dma_init(struct nouveau_channel *); -extern int nouveau_dma_wait(struct nouveau_channel *, int size); +extern int nouveau_dma_wait(struct nouveau_channel *, int slots, int size); /* nouveau_acpi.c */ #ifdef CONFIG_ACPI @@ -1024,6 +1029,7 @@ extern void nv50_graph_destroy_context(struct nouveau_channel *); extern int nv50_graph_load_context(struct nouveau_channel *); extern int nv50_graph_unload_context(struct drm_device *); extern void nv50_graph_context_switch(struct drm_device *); +extern int nv50_grctx_init(struct nouveau_grctx *); /* nouveau_grctx.c */ extern int nouveau_grctx_prog_load(struct drm_device *); @@ -1149,16 +1155,6 @@ extern int nouveau_gem_ioctl_new(struct drm_device *, void *, struct drm_file *); extern int nouveau_gem_ioctl_pushbuf(struct drm_device *, void *, struct drm_file *); -extern int nouveau_gem_ioctl_pushbuf_call(struct drm_device *, void *, - struct drm_file *); -extern int nouveau_gem_ioctl_pushbuf_call2(struct drm_device *, void *, - struct drm_file *); -extern int nouveau_gem_ioctl_pin(struct drm_device *, void *, - struct drm_file *); -extern int nouveau_gem_ioctl_unpin(struct drm_device *, void *, - struct drm_file *); -extern int nouveau_gem_ioctl_tile(struct drm_device *, void *, - struct drm_file *); extern int nouveau_gem_ioctl_cpu_prep(struct drm_device *, void *, struct drm_file *); extern int nouveau_gem_ioctl_cpu_fini(struct drm_device *, void *, diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.c b/drivers/gpu/drm/nouveau/nouveau_fbcon.c index 0b05c869e0e7..d48c59cdefe4 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fbcon.c +++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.c @@ -107,6 +107,34 @@ static struct fb_ops nouveau_fbcon_ops = { .fb_setcmap = drm_fb_helper_setcmap, }; +static struct fb_ops nv04_fbcon_ops = { + .owner = THIS_MODULE, + .fb_check_var = drm_fb_helper_check_var, + .fb_set_par = drm_fb_helper_set_par, + .fb_setcolreg = drm_fb_helper_setcolreg, + .fb_fillrect = nv04_fbcon_fillrect, + .fb_copyarea = nv04_fbcon_copyarea, + .fb_imageblit = nv04_fbcon_imageblit, + .fb_sync = nouveau_fbcon_sync, + .fb_pan_display = drm_fb_helper_pan_display, + .fb_blank = drm_fb_helper_blank, + .fb_setcmap = drm_fb_helper_setcmap, +}; + +static struct fb_ops nv50_fbcon_ops = { + .owner = THIS_MODULE, + .fb_check_var = drm_fb_helper_check_var, + .fb_set_par = drm_fb_helper_set_par, + .fb_setcolreg = drm_fb_helper_setcolreg, + .fb_fillrect = nv50_fbcon_fillrect, + .fb_copyarea = nv50_fbcon_copyarea, + .fb_imageblit = nv50_fbcon_imageblit, + .fb_sync = nouveau_fbcon_sync, + .fb_pan_display = drm_fb_helper_pan_display, + .fb_blank = drm_fb_helper_blank, + .fb_setcmap = drm_fb_helper_setcmap, +}; + static void nouveau_fbcon_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, u16 blue, int regno) { @@ -267,8 +295,12 @@ nouveau_fbcon_create(struct drm_device *dev, uint32_t fb_width, dev_priv->fbdev_info = info; strcpy(info->fix.id, "nouveaufb"); - info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_COPYAREA | - FBINFO_HWACCEL_FILLRECT | FBINFO_HWACCEL_IMAGEBLIT; + if (nouveau_nofbaccel) + info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_DISABLED; + else + info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_COPYAREA | + FBINFO_HWACCEL_FILLRECT | + FBINFO_HWACCEL_IMAGEBLIT; info->fbops = &nouveau_fbcon_ops; info->fix.smem_start = dev->mode_config.fb_base + nvbo->bo.offset - dev_priv->vm_vram_base; @@ -316,13 +348,15 @@ nouveau_fbcon_create(struct drm_device *dev, uint32_t fb_width, par->nouveau_fb = nouveau_fb; par->dev = dev; - if (dev_priv->channel) { + if (dev_priv->channel && !nouveau_nofbaccel) { switch (dev_priv->card_type) { case NV_50: nv50_fbcon_accel_init(info); + info->fbops = &nv50_fbcon_ops; break; default: nv04_fbcon_accel_init(info); + info->fbops = &nv04_fbcon_ops; break; }; } @@ -367,10 +401,8 @@ nouveau_fbcon_remove(struct drm_device *dev, struct drm_framebuffer *fb) unregister_framebuffer(info); nouveau_bo_unmap(nouveau_fb->nvbo); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(nouveau_fb->nvbo->gem); + drm_gem_object_unreference_unlocked(nouveau_fb->nvbo->gem); nouveau_fb->nvbo = NULL; - mutex_unlock(&dev->struct_mutex); if (par) drm_fb_helper_free(&par->helper); framebuffer_release(info); diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.h b/drivers/gpu/drm/nouveau/nouveau_fbcon.h index 462e0b87b4bd..f9c34e1a8c11 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fbcon.h +++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.h @@ -40,7 +40,13 @@ int nouveau_fbcon_remove(struct drm_device *dev, struct drm_framebuffer *fb); void nouveau_fbcon_restore(void); void nouveau_fbcon_zfill(struct drm_device *dev); +void nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region); +void nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect); +void nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image); int nv04_fbcon_accel_init(struct fb_info *info); +void nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect); +void nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region); +void nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image); int nv50_fbcon_accel_init(struct fb_info *info); void nouveau_fbcon_gpu_lockup(struct fb_info *info); diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c index 6ac804b0c9f9..0d22f66f1c79 100644 --- a/drivers/gpu/drm/nouveau/nouveau_gem.c +++ b/drivers/gpu/drm/nouveau/nouveau_gem.c @@ -167,12 +167,10 @@ nouveau_gem_ioctl_new(struct drm_device *dev, void *data, ret = drm_gem_handle_create(file_priv, nvbo->gem, &req->info.handle); out: - mutex_lock(&dev->struct_mutex); - drm_gem_object_handle_unreference(nvbo->gem); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_handle_unreference_unlocked(nvbo->gem); if (ret) - drm_gem_object_unreference(nvbo->gem); + drm_gem_object_unreference_unlocked(nvbo->gem); return ret; } @@ -243,6 +241,11 @@ validate_fini_list(struct list_head *list, struct nouveau_fence *fence) nouveau_fence_unref((void *)&prev_fence); } + if (unlikely(nvbo->validate_mapped)) { + ttm_bo_kunmap(&nvbo->kmap); + nvbo->validate_mapped = false; + } + list_del(&nvbo->entry); nvbo->reserved_by = NULL; ttm_bo_unreserve(&nvbo->bo); @@ -302,11 +305,14 @@ retry: if (ret == -EAGAIN) ret = ttm_bo_wait_unreserved(&nvbo->bo, false); drm_gem_object_unreference(gem); - if (ret) + if (ret) { + NV_ERROR(dev, "fail reserve\n"); return ret; + } goto retry; } + b->user_priv = (uint64_t)(unsigned long)nvbo; nvbo->reserved_by = file_priv; nvbo->pbbo_index = i; if ((b->valid_domains & NOUVEAU_GEM_DOMAIN_VRAM) && @@ -336,8 +342,10 @@ retry: } ret = ttm_bo_wait_cpu(&nvbo->bo, false); - if (ret) + if (ret) { + NV_ERROR(dev, "fail wait_cpu\n"); return ret; + } goto retry; } } @@ -351,6 +359,7 @@ validate_list(struct nouveau_channel *chan, struct list_head *list, { struct drm_nouveau_gem_pushbuf_bo __user *upbbo = (void __force __user *)(uintptr_t)user_pbbo_ptr; + struct drm_device *dev = chan->dev; struct nouveau_bo *nvbo; int ret, relocs = 0; @@ -362,39 +371,46 @@ validate_list(struct nouveau_channel *chan, struct list_head *list, spin_lock(&nvbo->bo.lock); ret = ttm_bo_wait(&nvbo->bo, false, false, false); spin_unlock(&nvbo->bo.lock); - if (unlikely(ret)) + if (unlikely(ret)) { + NV_ERROR(dev, "fail wait other chan\n"); return ret; + } } ret = nouveau_gem_set_domain(nvbo->gem, b->read_domains, b->write_domains, b->valid_domains); - if (unlikely(ret)) + if (unlikely(ret)) { + NV_ERROR(dev, "fail set_domain\n"); return ret; + } nvbo->channel = chan; ret = ttm_bo_validate(&nvbo->bo, &nvbo->placement, false, false); nvbo->channel = NULL; - if (unlikely(ret)) + if (unlikely(ret)) { + NV_ERROR(dev, "fail ttm_validate\n"); return ret; + } - if (nvbo->bo.offset == b->presumed_offset && + if (nvbo->bo.offset == b->presumed.offset && ((nvbo->bo.mem.mem_type == TTM_PL_VRAM && - b->presumed_domain & NOUVEAU_GEM_DOMAIN_VRAM) || + b->presumed.domain & NOUVEAU_GEM_DOMAIN_VRAM) || (nvbo->bo.mem.mem_type == TTM_PL_TT && - b->presumed_domain & NOUVEAU_GEM_DOMAIN_GART))) + b->presumed.domain & NOUVEAU_GEM_DOMAIN_GART))) continue; if (nvbo->bo.mem.mem_type == TTM_PL_TT) - b->presumed_domain = NOUVEAU_GEM_DOMAIN_GART; + b->presumed.domain = NOUVEAU_GEM_DOMAIN_GART; else - b->presumed_domain = NOUVEAU_GEM_DOMAIN_VRAM; - b->presumed_offset = nvbo->bo.offset; - b->presumed_ok = 0; + b->presumed.domain = NOUVEAU_GEM_DOMAIN_VRAM; + b->presumed.offset = nvbo->bo.offset; + b->presumed.valid = 0; relocs++; - if (DRM_COPY_TO_USER(&upbbo[nvbo->pbbo_index], b, sizeof(*b))) + if (DRM_COPY_TO_USER(&upbbo[nvbo->pbbo_index].presumed, + &b->presumed, sizeof(b->presumed))) return -EFAULT; } @@ -408,6 +424,7 @@ nouveau_gem_pushbuf_validate(struct nouveau_channel *chan, uint64_t user_buffers, int nr_buffers, struct validate_op *op, int *apply_relocs) { + struct drm_device *dev = chan->dev; int ret, relocs = 0; INIT_LIST_HEAD(&op->vram_list); @@ -418,11 +435,14 @@ nouveau_gem_pushbuf_validate(struct nouveau_channel *chan, return 0; ret = validate_init(chan, file_priv, pbbo, nr_buffers, op); - if (unlikely(ret)) + if (unlikely(ret)) { + NV_ERROR(dev, "validate_init\n"); return ret; + } ret = validate_list(chan, &op->vram_list, pbbo, user_buffers); if (unlikely(ret < 0)) { + NV_ERROR(dev, "validate vram_list\n"); validate_fini(op, NULL); return ret; } @@ -430,6 +450,7 @@ nouveau_gem_pushbuf_validate(struct nouveau_channel *chan, ret = validate_list(chan, &op->gart_list, pbbo, user_buffers); if (unlikely(ret < 0)) { + NV_ERROR(dev, "validate gart_list\n"); validate_fini(op, NULL); return ret; } @@ -437,6 +458,7 @@ nouveau_gem_pushbuf_validate(struct nouveau_channel *chan, ret = validate_list(chan, &op->both_list, pbbo, user_buffers); if (unlikely(ret < 0)) { + NV_ERROR(dev, "validate both_list\n"); validate_fini(op, NULL); return ret; } @@ -465,59 +487,82 @@ u_memcpya(uint64_t user, unsigned nmemb, unsigned size) } static int -nouveau_gem_pushbuf_reloc_apply(struct nouveau_channel *chan, int nr_bo, - struct drm_nouveau_gem_pushbuf_bo *bo, - unsigned nr_relocs, uint64_t ptr_relocs, - unsigned nr_dwords, unsigned first_dword, - uint32_t *pushbuf, bool is_iomem) +nouveau_gem_pushbuf_reloc_apply(struct drm_device *dev, + struct drm_nouveau_gem_pushbuf *req, + struct drm_nouveau_gem_pushbuf_bo *bo) { struct drm_nouveau_gem_pushbuf_reloc *reloc = NULL; - struct drm_device *dev = chan->dev; int ret = 0; unsigned i; - reloc = u_memcpya(ptr_relocs, nr_relocs, sizeof(*reloc)); + reloc = u_memcpya(req->relocs, req->nr_relocs, sizeof(*reloc)); if (IS_ERR(reloc)) return PTR_ERR(reloc); - for (i = 0; i < nr_relocs; i++) { + for (i = 0; i < req->nr_relocs; i++) { struct drm_nouveau_gem_pushbuf_reloc *r = &reloc[i]; struct drm_nouveau_gem_pushbuf_bo *b; + struct nouveau_bo *nvbo; uint32_t data; - if (r->bo_index >= nr_bo || r->reloc_index < first_dword || - r->reloc_index >= first_dword + nr_dwords) { - NV_ERROR(dev, "Bad relocation %d\n", i); - NV_ERROR(dev, " bo: %d max %d\n", r->bo_index, nr_bo); - NV_ERROR(dev, " id: %d max %d\n", r->reloc_index, nr_dwords); + if (unlikely(r->bo_index > req->nr_buffers)) { + NV_ERROR(dev, "reloc bo index invalid\n"); ret = -EINVAL; break; } b = &bo[r->bo_index]; - if (b->presumed_ok) + if (b->presumed.valid) continue; + if (unlikely(r->reloc_bo_index > req->nr_buffers)) { + NV_ERROR(dev, "reloc container bo index invalid\n"); + ret = -EINVAL; + break; + } + nvbo = (void *)(unsigned long)bo[r->reloc_bo_index].user_priv; + + if (unlikely(r->reloc_bo_offset + 4 > + nvbo->bo.mem.num_pages << PAGE_SHIFT)) { + NV_ERROR(dev, "reloc outside of bo\n"); + ret = -EINVAL; + break; + } + + if (!nvbo->kmap.virtual) { + ret = ttm_bo_kmap(&nvbo->bo, 0, nvbo->bo.mem.num_pages, + &nvbo->kmap); + if (ret) { + NV_ERROR(dev, "failed kmap for reloc\n"); + break; + } + nvbo->validate_mapped = true; + } + if (r->flags & NOUVEAU_GEM_RELOC_LOW) - data = b->presumed_offset + r->data; + data = b->presumed.offset + r->data; else if (r->flags & NOUVEAU_GEM_RELOC_HIGH) - data = (b->presumed_offset + r->data) >> 32; + data = (b->presumed.offset + r->data) >> 32; else data = r->data; if (r->flags & NOUVEAU_GEM_RELOC_OR) { - if (b->presumed_domain == NOUVEAU_GEM_DOMAIN_GART) + if (b->presumed.domain == NOUVEAU_GEM_DOMAIN_GART) data |= r->tor; else data |= r->vor; } - if (is_iomem) - iowrite32_native(data, (void __force __iomem *) - &pushbuf[r->reloc_index]); - else - pushbuf[r->reloc_index] = data; + spin_lock(&nvbo->bo.lock); + ret = ttm_bo_wait(&nvbo->bo, false, false, false); + spin_unlock(&nvbo->bo.lock); + if (ret) { + NV_ERROR(dev, "reloc wait_idle failed: %d\n", ret); + break; + } + + nouveau_bo_wr32(nvbo, r->reloc_bo_offset >> 2, data); } kfree(reloc); @@ -528,127 +573,50 @@ int nouveau_gem_ioctl_pushbuf(struct drm_device *dev, void *data, struct drm_file *file_priv) { + struct drm_nouveau_private *dev_priv = dev->dev_private; struct drm_nouveau_gem_pushbuf *req = data; - struct drm_nouveau_gem_pushbuf_bo *bo = NULL; + struct drm_nouveau_gem_pushbuf_push *push; + struct drm_nouveau_gem_pushbuf_bo *bo; struct nouveau_channel *chan; struct validate_op op; - struct nouveau_fence* fence = 0; - uint32_t *pushbuf = NULL; - int ret = 0, do_reloc = 0, i; + struct nouveau_fence *fence = 0; + int i, j, ret = 0, do_reloc = 0; NOUVEAU_CHECK_INITIALISED_WITH_RETURN; NOUVEAU_GET_USER_CHANNEL_WITH_RETURN(req->channel, file_priv, chan); - if (req->nr_dwords >= chan->dma.max || - req->nr_buffers > NOUVEAU_GEM_MAX_BUFFERS || - req->nr_relocs > NOUVEAU_GEM_MAX_RELOCS) { - NV_ERROR(dev, "Pushbuf config exceeds limits:\n"); - NV_ERROR(dev, " dwords : %d max %d\n", req->nr_dwords, - chan->dma.max - 1); - NV_ERROR(dev, " buffers: %d max %d\n", req->nr_buffers, - NOUVEAU_GEM_MAX_BUFFERS); - NV_ERROR(dev, " relocs : %d max %d\n", req->nr_relocs, - NOUVEAU_GEM_MAX_RELOCS); - return -EINVAL; - } - - pushbuf = u_memcpya(req->dwords, req->nr_dwords, sizeof(uint32_t)); - if (IS_ERR(pushbuf)) - return PTR_ERR(pushbuf); - - bo = u_memcpya(req->buffers, req->nr_buffers, sizeof(*bo)); - if (IS_ERR(bo)) { - kfree(pushbuf); - return PTR_ERR(bo); - } - - mutex_lock(&dev->struct_mutex); - - /* Validate buffer list */ - ret = nouveau_gem_pushbuf_validate(chan, file_priv, bo, req->buffers, - req->nr_buffers, &op, &do_reloc); - if (ret) - goto out; - - /* Apply any relocations that are required */ - if (do_reloc) { - ret = nouveau_gem_pushbuf_reloc_apply(chan, req->nr_buffers, - bo, req->nr_relocs, - req->relocs, - req->nr_dwords, 0, - pushbuf, false); - if (ret) - goto out; - } - - /* Emit push buffer to the hw - */ - ret = RING_SPACE(chan, req->nr_dwords); - if (ret) - goto out; - - OUT_RINGp(chan, pushbuf, req->nr_dwords); + req->vram_available = dev_priv->fb_aper_free; + req->gart_available = dev_priv->gart_info.aper_free; + if (unlikely(req->nr_push == 0)) + goto out_next; - ret = nouveau_fence_new(chan, &fence, true); - if (ret) { - NV_ERROR(dev, "error fencing pushbuf: %d\n", ret); - WIND_RING(chan); - goto out; + if (unlikely(req->nr_push > NOUVEAU_GEM_MAX_PUSH)) { + NV_ERROR(dev, "pushbuf push count exceeds limit: %d max %d\n", + req->nr_push, NOUVEAU_GEM_MAX_PUSH); + return -EINVAL; } - if (nouveau_gem_pushbuf_sync(chan)) { - ret = nouveau_fence_wait(fence, NULL, false, false); - if (ret) { - for (i = 0; i < req->nr_dwords; i++) - NV_ERROR(dev, "0x%08x\n", pushbuf[i]); - NV_ERROR(dev, "^^ above push buffer is fail :(\n"); - } + if (unlikely(req->nr_buffers > NOUVEAU_GEM_MAX_BUFFERS)) { + NV_ERROR(dev, "pushbuf bo count exceeds limit: %d max %d\n", + req->nr_buffers, NOUVEAU_GEM_MAX_BUFFERS); + return -EINVAL; } -out: - validate_fini(&op, fence); - nouveau_fence_unref((void**)&fence); - mutex_unlock(&dev->struct_mutex); - kfree(pushbuf); - kfree(bo); - return ret; -} - -#define PUSHBUF_CAL (dev_priv->card_type >= NV_20) - -int -nouveau_gem_ioctl_pushbuf_call(struct drm_device *dev, void *data, - struct drm_file *file_priv) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct drm_nouveau_gem_pushbuf_call *req = data; - struct drm_nouveau_gem_pushbuf_bo *bo = NULL; - struct nouveau_channel *chan; - struct drm_gem_object *gem; - struct nouveau_bo *pbbo; - struct validate_op op; - struct nouveau_fence* fence = 0; - int i, ret = 0, do_reloc = 0; - - NOUVEAU_CHECK_INITIALISED_WITH_RETURN; - NOUVEAU_GET_USER_CHANNEL_WITH_RETURN(req->channel, file_priv, chan); - - if (unlikely(req->handle == 0)) - goto out_next; - - if (req->nr_buffers > NOUVEAU_GEM_MAX_BUFFERS || - req->nr_relocs > NOUVEAU_GEM_MAX_RELOCS) { - NV_ERROR(dev, "Pushbuf config exceeds limits:\n"); - NV_ERROR(dev, " buffers: %d max %d\n", req->nr_buffers, - NOUVEAU_GEM_MAX_BUFFERS); - NV_ERROR(dev, " relocs : %d max %d\n", req->nr_relocs, - NOUVEAU_GEM_MAX_RELOCS); + if (unlikely(req->nr_relocs > NOUVEAU_GEM_MAX_RELOCS)) { + NV_ERROR(dev, "pushbuf reloc count exceeds limit: %d max %d\n", + req->nr_relocs, NOUVEAU_GEM_MAX_RELOCS); return -EINVAL; } + push = u_memcpya(req->push, req->nr_push, sizeof(*push)); + if (IS_ERR(push)) + return PTR_ERR(push); + bo = u_memcpya(req->buffers, req->nr_buffers, sizeof(*bo)); - if (IS_ERR(bo)) + if (IS_ERR(bo)) { + kfree(push); return PTR_ERR(bo); + } mutex_lock(&dev->struct_mutex); @@ -660,122 +628,84 @@ nouveau_gem_ioctl_pushbuf_call(struct drm_device *dev, void *data, goto out; } - /* Validate DMA push buffer */ - gem = drm_gem_object_lookup(dev, file_priv, req->handle); - if (!gem) { - NV_ERROR(dev, "Unknown pb handle 0x%08x\n", req->handle); - ret = -EINVAL; - goto out; - } - pbbo = nouveau_gem_object(gem); - - if ((req->offset & 3) || req->nr_dwords < 2 || - (unsigned long)req->offset > (unsigned long)pbbo->bo.mem.size || - (unsigned long)req->nr_dwords > - ((unsigned long)(pbbo->bo.mem.size - req->offset ) >> 2)) { - NV_ERROR(dev, "pb call misaligned or out of bounds: " - "%d + %d * 4 > %ld\n", - req->offset, req->nr_dwords, pbbo->bo.mem.size); - ret = -EINVAL; - drm_gem_object_unreference(gem); - goto out; - } - - ret = ttm_bo_reserve(&pbbo->bo, false, false, true, - chan->fence.sequence); - if (ret) { - NV_ERROR(dev, "resv pb: %d\n", ret); - drm_gem_object_unreference(gem); - goto out; - } - - nouveau_bo_placement_set(pbbo, 1 << chan->pushbuf_bo->bo.mem.mem_type); - ret = ttm_bo_validate(&pbbo->bo, &pbbo->placement, false, false); - if (ret) { - NV_ERROR(dev, "validate pb: %d\n", ret); - ttm_bo_unreserve(&pbbo->bo); - drm_gem_object_unreference(gem); - goto out; - } - - list_add_tail(&pbbo->entry, &op.both_list); - - /* If presumed return address doesn't match, we need to map the - * push buffer and fix it.. - */ - if (!PUSHBUF_CAL) { - uint32_t retaddy; - - if (chan->dma.free < 4 + NOUVEAU_DMA_SKIPS) { - ret = nouveau_dma_wait(chan, 4 + NOUVEAU_DMA_SKIPS); - if (ret) { - NV_ERROR(dev, "jmp_space: %d\n", ret); - goto out; - } - } - - retaddy = chan->pushbuf_base + ((chan->dma.cur + 2) << 2); - retaddy |= 0x20000000; - if (retaddy != req->suffix0) { - req->suffix0 = retaddy; - do_reloc = 1; - } - } - /* Apply any relocations that are required */ if (do_reloc) { - void *pbvirt; - bool is_iomem; - ret = ttm_bo_kmap(&pbbo->bo, 0, pbbo->bo.mem.num_pages, - &pbbo->kmap); + ret = nouveau_gem_pushbuf_reloc_apply(dev, req, bo); if (ret) { - NV_ERROR(dev, "kmap pb: %d\n", ret); + NV_ERROR(dev, "reloc apply: %d\n", ret); goto out; } + } - pbvirt = ttm_kmap_obj_virtual(&pbbo->kmap, &is_iomem); - ret = nouveau_gem_pushbuf_reloc_apply(chan, req->nr_buffers, bo, - req->nr_relocs, - req->relocs, - req->nr_dwords, - req->offset / 4, - pbvirt, is_iomem); - - if (!PUSHBUF_CAL) { - nouveau_bo_wr32(pbbo, - req->offset / 4 + req->nr_dwords - 2, - req->suffix0); - } - - ttm_bo_kunmap(&pbbo->kmap); + if (chan->dma.ib_max) { + ret = nouveau_dma_wait(chan, req->nr_push + 1, 6); if (ret) { - NV_ERROR(dev, "reloc apply: %d\n", ret); + NV_INFO(dev, "nv50cal_space: %d\n", ret); goto out; } - } - if (PUSHBUF_CAL) { - ret = RING_SPACE(chan, 2); + for (i = 0; i < req->nr_push; i++) { + struct nouveau_bo *nvbo = (void *)(unsigned long) + bo[push[i].bo_index].user_priv; + + nv50_dma_push(chan, nvbo, push[i].offset, + push[i].length); + } + } else + if (dev_priv->card_type >= NV_20) { + ret = RING_SPACE(chan, req->nr_push * 2); if (ret) { NV_ERROR(dev, "cal_space: %d\n", ret); goto out; } - OUT_RING(chan, ((pbbo->bo.mem.mm_node->start << PAGE_SHIFT) + - req->offset) | 2); - OUT_RING(chan, 0); + + for (i = 0; i < req->nr_push; i++) { + struct nouveau_bo *nvbo = (void *)(unsigned long) + bo[push[i].bo_index].user_priv; + struct drm_mm_node *mem = nvbo->bo.mem.mm_node; + + OUT_RING(chan, ((mem->start << PAGE_SHIFT) + + push[i].offset) | 2); + OUT_RING(chan, 0); + } } else { - ret = RING_SPACE(chan, 2 + NOUVEAU_DMA_SKIPS); + ret = RING_SPACE(chan, req->nr_push * (2 + NOUVEAU_DMA_SKIPS)); if (ret) { NV_ERROR(dev, "jmp_space: %d\n", ret); goto out; } - OUT_RING(chan, ((pbbo->bo.mem.mm_node->start << PAGE_SHIFT) + - req->offset) | 0x20000000); - OUT_RING(chan, 0); - /* Space the jumps apart with NOPs. */ - for (i = 0; i < NOUVEAU_DMA_SKIPS; i++) + for (i = 0; i < req->nr_push; i++) { + struct nouveau_bo *nvbo = (void *)(unsigned long) + bo[push[i].bo_index].user_priv; + struct drm_mm_node *mem = nvbo->bo.mem.mm_node; + uint32_t cmd; + + cmd = chan->pushbuf_base + ((chan->dma.cur + 2) << 2); + cmd |= 0x20000000; + if (unlikely(cmd != req->suffix0)) { + if (!nvbo->kmap.virtual) { + ret = ttm_bo_kmap(&nvbo->bo, 0, + nvbo->bo.mem. + num_pages, + &nvbo->kmap); + if (ret) { + WIND_RING(chan); + goto out; + } + nvbo->validate_mapped = true; + } + + nouveau_bo_wr32(nvbo, (push[i].offset + + push[i].length - 8) / 4, cmd); + } + + OUT_RING(chan, ((mem->start << PAGE_SHIFT) + + push[i].offset) | 0x20000000); OUT_RING(chan, 0); + for (j = 0; j < NOUVEAU_DMA_SKIPS; j++) + OUT_RING(chan, 0); + } } ret = nouveau_fence_new(chan, &fence, true); @@ -790,9 +720,14 @@ out: nouveau_fence_unref((void**)&fence); mutex_unlock(&dev->struct_mutex); kfree(bo); + kfree(push); out_next: - if (PUSHBUF_CAL) { + if (chan->dma.ib_max) { + req->suffix0 = 0x00000000; + req->suffix1 = 0x00000000; + } else + if (dev_priv->card_type >= NV_20) { req->suffix0 = 0x00020000; req->suffix1 = 0x00000000; } else { @@ -804,19 +739,6 @@ out_next: return ret; } -int -nouveau_gem_ioctl_pushbuf_call2(struct drm_device *dev, void *data, - struct drm_file *file_priv) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct drm_nouveau_gem_pushbuf_call *req = data; - - req->vram_available = dev_priv->fb_aper_free; - req->gart_available = dev_priv->gart_info.aper_free; - - return nouveau_gem_ioctl_pushbuf_call(dev, data, file_priv); -} - static inline uint32_t domain_to_ttm(struct nouveau_bo *nvbo, uint32_t domain) { @@ -831,74 +753,6 @@ domain_to_ttm(struct nouveau_bo *nvbo, uint32_t domain) } int -nouveau_gem_ioctl_pin(struct drm_device *dev, void *data, - struct drm_file *file_priv) -{ - struct drm_nouveau_gem_pin *req = data; - struct drm_gem_object *gem; - struct nouveau_bo *nvbo; - int ret = 0; - - NOUVEAU_CHECK_INITIALISED_WITH_RETURN; - - if (drm_core_check_feature(dev, DRIVER_MODESET)) { - NV_ERROR(dev, "pin only allowed without kernel modesetting\n"); - return -EINVAL; - } - - if (!DRM_SUSER(DRM_CURPROC)) - return -EPERM; - - gem = drm_gem_object_lookup(dev, file_priv, req->handle); - if (!gem) - return -EINVAL; - nvbo = nouveau_gem_object(gem); - - ret = nouveau_bo_pin(nvbo, domain_to_ttm(nvbo, req->domain)); - if (ret) - goto out; - - req->offset = nvbo->bo.offset; - if (nvbo->bo.mem.mem_type == TTM_PL_TT) - req->domain = NOUVEAU_GEM_DOMAIN_GART; - else - req->domain = NOUVEAU_GEM_DOMAIN_VRAM; - -out: - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gem); - mutex_unlock(&dev->struct_mutex); - - return ret; -} - -int -nouveau_gem_ioctl_unpin(struct drm_device *dev, void *data, - struct drm_file *file_priv) -{ - struct drm_nouveau_gem_pin *req = data; - struct drm_gem_object *gem; - int ret; - - NOUVEAU_CHECK_INITIALISED_WITH_RETURN; - - if (drm_core_check_feature(dev, DRIVER_MODESET)) - return -EINVAL; - - gem = drm_gem_object_lookup(dev, file_priv, req->handle); - if (!gem) - return -EINVAL; - - ret = nouveau_bo_unpin(nouveau_gem_object(gem)); - - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gem); - mutex_unlock(&dev->struct_mutex); - - return ret; -} - -int nouveau_gem_ioctl_cpu_prep(struct drm_device *dev, void *data, struct drm_file *file_priv) { @@ -925,7 +779,9 @@ nouveau_gem_ioctl_cpu_prep(struct drm_device *dev, void *data, } if (req->flags & NOUVEAU_GEM_CPU_PREP_NOBLOCK) { + spin_lock(&nvbo->bo.lock); ret = ttm_bo_wait(&nvbo->bo, false, false, no_wait); + spin_unlock(&nvbo->bo.lock); } else { ret = ttm_bo_synccpu_write_grab(&nvbo->bo, no_wait); if (ret == 0) @@ -933,9 +789,7 @@ nouveau_gem_ioctl_cpu_prep(struct drm_device *dev, void *data, } out: - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gem); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gem); return ret; } @@ -963,9 +817,7 @@ nouveau_gem_ioctl_cpu_fini(struct drm_device *dev, void *data, ret = 0; out: - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gem); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gem); return ret; } @@ -984,9 +836,7 @@ nouveau_gem_ioctl_info(struct drm_device *dev, void *data, return -EINVAL; ret = nouveau_gem_info(gem, req); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gem); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gem); return ret; } diff --git a/drivers/gpu/drm/nouveau/nouveau_grctx.c b/drivers/gpu/drm/nouveau/nouveau_grctx.c index 419f4c2b3b89..c7ebec696747 100644 --- a/drivers/gpu/drm/nouveau/nouveau_grctx.c +++ b/drivers/gpu/drm/nouveau/nouveau_grctx.c @@ -97,8 +97,8 @@ nouveau_grctx_prog_load(struct drm_device *dev) } pgraph->ctxvals = kmalloc(fw->size, GFP_KERNEL); - if (!pgraph->ctxprog) { - NV_ERROR(dev, "OOM copying ctxprog\n"); + if (!pgraph->ctxvals) { + NV_ERROR(dev, "OOM copying ctxvals\n"); release_firmware(fw); nouveau_grctx_fini(dev); return -ENOMEM; diff --git a/drivers/gpu/drm/nouveau/nouveau_hw.c b/drivers/gpu/drm/nouveau/nouveau_hw.c index dc46792a5c96..7855b35effc3 100644 --- a/drivers/gpu/drm/nouveau/nouveau_hw.c +++ b/drivers/gpu/drm/nouveau/nouveau_hw.c @@ -160,7 +160,7 @@ static void setPLL_single(struct drm_device *dev, uint32_t reg, struct nouveau_pll_vals *pv) { struct drm_nouveau_private *dev_priv = dev->dev_private; - int chip_version = dev_priv->vbios->chip_version; + int chip_version = dev_priv->vbios.chip_version; uint32_t oldpll = NVReadRAMDAC(dev, 0, reg); int oldN = (oldpll >> 8) & 0xff, oldM = oldpll & 0xff; uint32_t pll = (oldpll & 0xfff80000) | pv->log2P << 16 | pv->NM1; @@ -216,7 +216,7 @@ setPLL_double_highregs(struct drm_device *dev, uint32_t reg1, struct nouveau_pll_vals *pv) { struct drm_nouveau_private *dev_priv = dev->dev_private; - int chip_version = dev_priv->vbios->chip_version; + int chip_version = dev_priv->vbios.chip_version; bool nv3035 = chip_version == 0x30 || chip_version == 0x35; uint32_t reg2 = reg1 + ((reg1 == NV_RAMDAC_VPLL2) ? 0x5c : 0x70); uint32_t oldpll1 = NVReadRAMDAC(dev, 0, reg1); @@ -374,7 +374,7 @@ nouveau_hw_setpll(struct drm_device *dev, uint32_t reg1, struct nouveau_pll_vals *pv) { struct drm_nouveau_private *dev_priv = dev->dev_private; - int cv = dev_priv->vbios->chip_version; + int cv = dev_priv->vbios.chip_version; if (cv == 0x30 || cv == 0x31 || cv == 0x35 || cv == 0x36 || cv >= 0x40) { diff --git a/drivers/gpu/drm/nouveau/nouveau_i2c.c b/drivers/gpu/drm/nouveau/nouveau_i2c.c index 70e994d28122..88583e7bf651 100644 --- a/drivers/gpu/drm/nouveau/nouveau_i2c.c +++ b/drivers/gpu/drm/nouveau/nouveau_i2c.c @@ -254,16 +254,16 @@ struct nouveau_i2c_chan * nouveau_i2c_find(struct drm_device *dev, int index) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; - if (index > DCB_MAX_NUM_I2C_ENTRIES) + if (index >= DCB_MAX_NUM_I2C_ENTRIES) return NULL; - if (!bios->bdcb.dcb.i2c[index].chan) { - if (nouveau_i2c_init(dev, &bios->bdcb.dcb.i2c[index], index)) + if (!bios->dcb.i2c[index].chan) { + if (nouveau_i2c_init(dev, &bios->dcb.i2c[index], index)) return NULL; } - return bios->bdcb.dcb.i2c[index].chan; + return bios->dcb.i2c[index].chan; } diff --git a/drivers/gpu/drm/nouveau/nouveau_irq.c b/drivers/gpu/drm/nouveau/nouveau_irq.c index 3b9bad66162a..95220ddebb45 100644 --- a/drivers/gpu/drm/nouveau/nouveau_irq.c +++ b/drivers/gpu/drm/nouveau/nouveau_irq.c @@ -211,6 +211,20 @@ nouveau_fifo_irq_handler(struct drm_device *dev) get + 4); } + if (status & NV_PFIFO_INTR_SEMAPHORE) { + uint32_t sem; + + status &= ~NV_PFIFO_INTR_SEMAPHORE; + nv_wr32(dev, NV03_PFIFO_INTR_0, + NV_PFIFO_INTR_SEMAPHORE); + + sem = nv_rd32(dev, NV10_PFIFO_CACHE1_SEMAPHORE); + nv_wr32(dev, NV10_PFIFO_CACHE1_SEMAPHORE, sem | 0x1); + + nv_wr32(dev, NV03_PFIFO_CACHE1_GET, get + 4); + nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, 1); + } + if (status) { NV_INFO(dev, "PFIFO_INTR 0x%08x - Ch %d\n", status, chid); @@ -566,86 +580,99 @@ nouveau_pgraph_irq_handler(struct drm_device *dev) static void nv50_pgraph_irq_handler(struct drm_device *dev) { - uint32_t status, nsource; + uint32_t status; - status = nv_rd32(dev, NV03_PGRAPH_INTR); - nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE); + while ((status = nv_rd32(dev, NV03_PGRAPH_INTR))) { + uint32_t nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE); - if (status & 0x00000001) { - nouveau_pgraph_intr_notify(dev, nsource); - status &= ~0x00000001; - nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000001); - } + if (status & 0x00000001) { + nouveau_pgraph_intr_notify(dev, nsource); + status &= ~0x00000001; + nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000001); + } - if (status & 0x00000010) { - nouveau_pgraph_intr_error(dev, nsource | - NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD); + if (status & 0x00000010) { + nouveau_pgraph_intr_error(dev, nsource | + NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD); - status &= ~0x00000010; - nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000010); - } + status &= ~0x00000010; + nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000010); + } - if (status & 0x00001000) { - nv_wr32(dev, 0x400500, 0x00000000); - nv_wr32(dev, NV03_PGRAPH_INTR, NV_PGRAPH_INTR_CONTEXT_SWITCH); - nv_wr32(dev, NV40_PGRAPH_INTR_EN, nv_rd32(dev, - NV40_PGRAPH_INTR_EN) & ~NV_PGRAPH_INTR_CONTEXT_SWITCH); - nv_wr32(dev, 0x400500, 0x00010001); + if (status & 0x00001000) { + nv_wr32(dev, 0x400500, 0x00000000); + nv_wr32(dev, NV03_PGRAPH_INTR, + NV_PGRAPH_INTR_CONTEXT_SWITCH); + nv_wr32(dev, NV40_PGRAPH_INTR_EN, nv_rd32(dev, + NV40_PGRAPH_INTR_EN) & + ~NV_PGRAPH_INTR_CONTEXT_SWITCH); + nv_wr32(dev, 0x400500, 0x00010001); - nv50_graph_context_switch(dev); + nv50_graph_context_switch(dev); - status &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; - } + status &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; + } - if (status & 0x00100000) { - nouveau_pgraph_intr_error(dev, nsource | - NV03_PGRAPH_NSOURCE_DATA_ERROR); + if (status & 0x00100000) { + nouveau_pgraph_intr_error(dev, nsource | + NV03_PGRAPH_NSOURCE_DATA_ERROR); - status &= ~0x00100000; - nv_wr32(dev, NV03_PGRAPH_INTR, 0x00100000); - } + status &= ~0x00100000; + nv_wr32(dev, NV03_PGRAPH_INTR, 0x00100000); + } - if (status & 0x00200000) { - int r; - - nouveau_pgraph_intr_error(dev, nsource | - NV03_PGRAPH_NSOURCE_PROTECTION_ERROR); - - NV_ERROR(dev, "magic set 1:\n"); - for (r = 0x408900; r <= 0x408910; r += 4) - NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r)); - nv_wr32(dev, 0x408900, nv_rd32(dev, 0x408904) | 0xc0000000); - for (r = 0x408e08; r <= 0x408e24; r += 4) - NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r)); - nv_wr32(dev, 0x408e08, nv_rd32(dev, 0x408e08) | 0xc0000000); - - NV_ERROR(dev, "magic set 2:\n"); - for (r = 0x409900; r <= 0x409910; r += 4) - NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r)); - nv_wr32(dev, 0x409900, nv_rd32(dev, 0x409904) | 0xc0000000); - for (r = 0x409e08; r <= 0x409e24; r += 4) - NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r)); - nv_wr32(dev, 0x409e08, nv_rd32(dev, 0x409e08) | 0xc0000000); - - status &= ~0x00200000; - nv_wr32(dev, NV03_PGRAPH_NSOURCE, nsource); - nv_wr32(dev, NV03_PGRAPH_INTR, 0x00200000); - } + if (status & 0x00200000) { + int r; + + nouveau_pgraph_intr_error(dev, nsource | + NV03_PGRAPH_NSOURCE_PROTECTION_ERROR); + + NV_ERROR(dev, "magic set 1:\n"); + for (r = 0x408900; r <= 0x408910; r += 4) + NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, + nv_rd32(dev, r)); + nv_wr32(dev, 0x408900, + nv_rd32(dev, 0x408904) | 0xc0000000); + for (r = 0x408e08; r <= 0x408e24; r += 4) + NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, + nv_rd32(dev, r)); + nv_wr32(dev, 0x408e08, + nv_rd32(dev, 0x408e08) | 0xc0000000); + + NV_ERROR(dev, "magic set 2:\n"); + for (r = 0x409900; r <= 0x409910; r += 4) + NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, + nv_rd32(dev, r)); + nv_wr32(dev, 0x409900, + nv_rd32(dev, 0x409904) | 0xc0000000); + for (r = 0x409e08; r <= 0x409e24; r += 4) + NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, + nv_rd32(dev, r)); + nv_wr32(dev, 0x409e08, + nv_rd32(dev, 0x409e08) | 0xc0000000); + + status &= ~0x00200000; + nv_wr32(dev, NV03_PGRAPH_NSOURCE, nsource); + nv_wr32(dev, NV03_PGRAPH_INTR, 0x00200000); + } - if (status) { - NV_INFO(dev, "Unhandled PGRAPH_INTR - 0x%08x\n", status); - nv_wr32(dev, NV03_PGRAPH_INTR, status); - } + if (status) { + NV_INFO(dev, "Unhandled PGRAPH_INTR - 0x%08x\n", + status); + nv_wr32(dev, NV03_PGRAPH_INTR, status); + } - { - const int isb = (1 << 16) | (1 << 0); + { + const int isb = (1 << 16) | (1 << 0); - if ((nv_rd32(dev, 0x400500) & isb) != isb) - nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) | isb); - nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) & ~(1 << 31)); + if ((nv_rd32(dev, 0x400500) & isb) != isb) + nv_wr32(dev, 0x400500, + nv_rd32(dev, 0x400500) | isb); + } } nv_wr32(dev, NV03_PMC_INTR_0, NV_PMC_INTR_0_PGRAPH_PENDING); + nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) & ~(1 << 31)); } static void @@ -664,11 +691,14 @@ nouveau_irq_handler(DRM_IRQ_ARGS) struct drm_device *dev = (struct drm_device *)arg; struct drm_nouveau_private *dev_priv = dev->dev_private; uint32_t status, fbdev_flags = 0; + unsigned long flags; status = nv_rd32(dev, NV03_PMC_INTR_0); if (!status) return IRQ_NONE; + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + if (dev_priv->fbdev_info) { fbdev_flags = dev_priv->fbdev_info->flags; dev_priv->fbdev_info->flags |= FBINFO_HWACCEL_DISABLED; @@ -706,5 +736,7 @@ nouveau_irq_handler(DRM_IRQ_ARGS) if (dev_priv->fbdev_info) dev_priv->fbdev_info->flags = fbdev_flags; + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + return IRQ_HANDLED; } diff --git a/drivers/gpu/drm/nouveau/nouveau_mem.c b/drivers/gpu/drm/nouveau/nouveau_mem.c index 8f3a12f614ed..2dc09dbd817d 100644 --- a/drivers/gpu/drm/nouveau/nouveau_mem.c +++ b/drivers/gpu/drm/nouveau/nouveau_mem.c @@ -285,53 +285,50 @@ nv50_mem_vm_bind_linear(struct drm_device *dev, uint64_t virt, uint32_t size, uint32_t flags, uint64_t phys) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_gpuobj **pgt; - unsigned psz, pfl, pages; - - if (virt >= dev_priv->vm_gart_base && - (virt + size) < (dev_priv->vm_gart_base + dev_priv->vm_gart_size)) { - psz = 12; - pgt = &dev_priv->gart_info.sg_ctxdma; - pfl = 0x21; - virt -= dev_priv->vm_gart_base; - } else - if (virt >= dev_priv->vm_vram_base && - (virt + size) < (dev_priv->vm_vram_base + dev_priv->vm_vram_size)) { - psz = 16; - pgt = dev_priv->vm_vram_pt; - pfl = 0x01; - virt -= dev_priv->vm_vram_base; - } else { - NV_ERROR(dev, "Invalid address: 0x%16llx-0x%16llx\n", - virt, virt + size - 1); - return -EINVAL; - } + struct nouveau_gpuobj *pgt; + unsigned block; + int i; - pages = size >> psz; + virt = ((virt - dev_priv->vm_vram_base) >> 16) << 1; + size = (size >> 16) << 1; + + phys |= ((uint64_t)flags << 32); + phys |= 1; + if (dev_priv->vram_sys_base) { + phys += dev_priv->vram_sys_base; + phys |= 0x30; + } dev_priv->engine.instmem.prepare_access(dev, true); - if (flags & 0x80000000) { - while (pages--) { - struct nouveau_gpuobj *pt = pgt[virt >> 29]; - unsigned pte = ((virt & 0x1fffffffULL) >> psz) << 1; + while (size) { + unsigned offset_h = upper_32_bits(phys); + unsigned offset_l = lower_32_bits(phys); + unsigned pte, end; + + for (i = 7; i >= 0; i--) { + block = 1 << (i + 1); + if (size >= block && !(virt & (block - 1))) + break; + } + offset_l |= (i << 7); - nv_wo32(dev, pt, pte++, 0x00000000); - nv_wo32(dev, pt, pte++, 0x00000000); + phys += block << 15; + size -= block; - virt += (1 << psz); - } - } else { - while (pages--) { - struct nouveau_gpuobj *pt = pgt[virt >> 29]; - unsigned pte = ((virt & 0x1fffffffULL) >> psz) << 1; - unsigned offset_h = upper_32_bits(phys) & 0xff; - unsigned offset_l = lower_32_bits(phys); + while (block) { + pgt = dev_priv->vm_vram_pt[virt >> 14]; + pte = virt & 0x3ffe; - nv_wo32(dev, pt, pte++, offset_l | pfl); - nv_wo32(dev, pt, pte++, offset_h | flags); + end = pte + block; + if (end > 16384) + end = 16384; + block -= (end - pte); + virt += (end - pte); - phys += (1 << psz); - virt += (1 << psz); + while (pte < end) { + nv_wo32(dev, pgt, pte++, offset_l); + nv_wo32(dev, pgt, pte++, offset_h); + } } } dev_priv->engine.instmem.finish_access(dev); @@ -356,7 +353,41 @@ nv50_mem_vm_bind_linear(struct drm_device *dev, uint64_t virt, uint32_t size, void nv50_mem_vm_unbind(struct drm_device *dev, uint64_t virt, uint32_t size) { - nv50_mem_vm_bind_linear(dev, virt, size, 0x80000000, 0); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *pgt; + unsigned pages, pte, end; + + virt -= dev_priv->vm_vram_base; + pages = (size >> 16) << 1; + + dev_priv->engine.instmem.prepare_access(dev, true); + while (pages) { + pgt = dev_priv->vm_vram_pt[virt >> 29]; + pte = (virt & 0x1ffe0000ULL) >> 15; + + end = pte + pages; + if (end > 16384) + end = 16384; + pages -= (end - pte); + virt += (end - pte) << 15; + + while (pte < end) + nv_wo32(dev, pgt, pte++, 0); + } + dev_priv->engine.instmem.finish_access(dev); + + nv_wr32(dev, 0x100c80, 0x00050001); + if (!nv_wait(0x100c80, 0x00000001, 0x00000000)) { + NV_ERROR(dev, "timeout: (0x100c80 & 1) == 0 (2)\n"); + NV_ERROR(dev, "0x100c80 = 0x%08x\n", nv_rd32(dev, 0x100c80)); + return; + } + + nv_wr32(dev, 0x100c80, 0x00000001); + if (!nv_wait(0x100c80, 0x00000001, 0x00000000)) { + NV_ERROR(dev, "timeout: (0x100c80 & 1) == 0 (2)\n"); + NV_ERROR(dev, "0x100c80 = 0x%08x\n", nv_rd32(dev, 0x100c80)); + } } /* diff --git a/drivers/gpu/drm/nouveau/nouveau_notifier.c b/drivers/gpu/drm/nouveau/nouveau_notifier.c index 6c66a34b6345..9537f3e30115 100644 --- a/drivers/gpu/drm/nouveau/nouveau_notifier.c +++ b/drivers/gpu/drm/nouveau/nouveau_notifier.c @@ -34,15 +34,20 @@ nouveau_notifier_init_channel(struct nouveau_channel *chan) { struct drm_device *dev = chan->dev; struct nouveau_bo *ntfy = NULL; + uint32_t flags; int ret; - ret = nouveau_gem_new(dev, NULL, PAGE_SIZE, 0, nouveau_vram_notify ? - TTM_PL_FLAG_VRAM : TTM_PL_FLAG_TT, + if (nouveau_vram_notify) + flags = TTM_PL_FLAG_VRAM; + else + flags = TTM_PL_FLAG_TT; + + ret = nouveau_gem_new(dev, NULL, PAGE_SIZE, 0, flags, 0, 0x0000, false, true, &ntfy); if (ret) return ret; - ret = nouveau_bo_pin(ntfy, TTM_PL_FLAG_VRAM); + ret = nouveau_bo_pin(ntfy, flags); if (ret) goto out_err; @@ -56,11 +61,8 @@ nouveau_notifier_init_channel(struct nouveau_channel *chan) chan->notifier_bo = ntfy; out_err: - if (ret) { - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(ntfy->gem); - mutex_unlock(&dev->struct_mutex); - } + if (ret) + drm_gem_object_unreference_unlocked(ntfy->gem); return ret; } @@ -76,8 +78,8 @@ nouveau_notifier_takedown_channel(struct nouveau_channel *chan) nouveau_bo_unmap(chan->notifier_bo); mutex_lock(&dev->struct_mutex); nouveau_bo_unpin(chan->notifier_bo); - drm_gem_object_unreference(chan->notifier_bo->gem); mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(chan->notifier_bo->gem); nouveau_mem_takedown(&chan->notifier_heap); } @@ -128,6 +130,8 @@ nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle, target = NV_DMA_TARGET_PCI; } else { target = NV_DMA_TARGET_AGP; + if (dev_priv->card_type >= NV_50) + offset += dev_priv->vm_gart_base; } } else { NV_ERROR(dev, "Bad DMA target, mem_type %d!\n", diff --git a/drivers/gpu/drm/nouveau/nouveau_object.c b/drivers/gpu/drm/nouveau/nouveau_object.c index 6c2cf81716df..e7c100ba63a1 100644 --- a/drivers/gpu/drm/nouveau/nouveau_object.c +++ b/drivers/gpu/drm/nouveau/nouveau_object.c @@ -885,11 +885,12 @@ int nouveau_gpuobj_sw_new(struct nouveau_channel *chan, int class, struct nouveau_gpuobj **gpuobj_ret) { - struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct drm_nouveau_private *dev_priv; struct nouveau_gpuobj *gpuobj; if (!chan || !gpuobj_ret || *gpuobj_ret != NULL) return -EINVAL; + dev_priv = chan->dev->dev_private; gpuobj = kzalloc(sizeof(*gpuobj), GFP_KERNEL); if (!gpuobj) diff --git a/drivers/gpu/drm/nouveau/nouveau_reg.h b/drivers/gpu/drm/nouveau/nouveau_reg.h index 251f1b3b38b9..aa9b310e41be 100644 --- a/drivers/gpu/drm/nouveau/nouveau_reg.h +++ b/drivers/gpu/drm/nouveau/nouveau_reg.h @@ -99,6 +99,7 @@ * the card will hang early on in the X init process. */ # define NV_PMC_ENABLE_UNK13 (1<<13) +#define NV40_PMC_GRAPH_UNITS 0x00001540 #define NV40_PMC_BACKLIGHT 0x000015f0 # define NV40_PMC_BACKLIGHT_MASK 0x001f0000 #define NV40_PMC_1700 0x00001700 diff --git a/drivers/gpu/drm/nouveau/nouveau_sgdma.c b/drivers/gpu/drm/nouveau/nouveau_sgdma.c index 4c7f1e403e80..ed1590577b6c 100644 --- a/drivers/gpu/drm/nouveau/nouveau_sgdma.c +++ b/drivers/gpu/drm/nouveau/nouveau_sgdma.c @@ -54,11 +54,12 @@ static void nouveau_sgdma_clear(struct ttm_backend *be) { struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; - struct drm_device *dev = nvbe->dev; - - NV_DEBUG(nvbe->dev, "\n"); + struct drm_device *dev; if (nvbe && nvbe->pages) { + dev = nvbe->dev; + NV_DEBUG(dev, "\n"); + if (nvbe->bound) be->func->unbind(be); diff --git a/drivers/gpu/drm/nouveau/nouveau_state.c b/drivers/gpu/drm/nouveau/nouveau_state.c index f2d0187ba152..516a8d36cb10 100644 --- a/drivers/gpu/drm/nouveau/nouveau_state.c +++ b/drivers/gpu/drm/nouveau/nouveau_state.c @@ -310,6 +310,14 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) static unsigned int nouveau_vga_set_decode(void *priv, bool state) { + struct drm_device *dev = priv; + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset >= 0x40) + nv_wr32(dev, 0x88054, state); + else + nv_wr32(dev, 0x1854, state); + if (state) return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM | VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM; @@ -383,6 +391,7 @@ nouveau_card_init(struct drm_device *dev) goto out; engine = &dev_priv->engine; dev_priv->init_state = NOUVEAU_CARD_INIT_FAILED; + spin_lock_init(&dev_priv->context_switch_lock); /* Parse BIOS tables / Run init tables if card not POSTed */ if (drm_core_check_feature(dev, DRIVER_MODESET)) { @@ -427,15 +436,19 @@ nouveau_card_init(struct drm_device *dev) if (ret) goto out_timer; - /* PGRAPH */ - ret = engine->graph.init(dev); - if (ret) - goto out_fb; + if (nouveau_noaccel) + engine->graph.accel_blocked = true; + else { + /* PGRAPH */ + ret = engine->graph.init(dev); + if (ret) + goto out_fb; - /* PFIFO */ - ret = engine->fifo.init(dev); - if (ret) - goto out_graph; + /* PFIFO */ + ret = engine->fifo.init(dev); + if (ret) + goto out_graph; + } /* this call irq_preinstall, register irq handler and * call irq_postinstall @@ -479,9 +492,11 @@ nouveau_card_init(struct drm_device *dev) out_irq: drm_irq_uninstall(dev); out_fifo: - engine->fifo.takedown(dev); + if (!nouveau_noaccel) + engine->fifo.takedown(dev); out_graph: - engine->graph.takedown(dev); + if (!nouveau_noaccel) + engine->graph.takedown(dev); out_fb: engine->fb.takedown(dev); out_timer: @@ -518,8 +533,10 @@ static void nouveau_card_takedown(struct drm_device *dev) dev_priv->channel = NULL; } - engine->fifo.takedown(dev); - engine->graph.takedown(dev); + if (!nouveau_noaccel) { + engine->fifo.takedown(dev); + engine->graph.takedown(dev); + } engine->fb.takedown(dev); engine->timer.takedown(dev); engine->mc.takedown(dev); @@ -760,13 +777,6 @@ int nouveau_unload(struct drm_device *dev) return 0; } -int -nouveau_ioctl_card_init(struct drm_device *dev, void *data, - struct drm_file *file_priv) -{ - return nouveau_card_init(dev); -} - int nouveau_ioctl_getparam(struct drm_device *dev, void *data, struct drm_file *file_priv) { @@ -817,6 +827,15 @@ int nouveau_ioctl_getparam(struct drm_device *dev, void *data, case NOUVEAU_GETPARAM_VM_VRAM_BASE: getparam->value = dev_priv->vm_vram_base; break; + case NOUVEAU_GETPARAM_GRAPH_UNITS: + /* NV40 and NV50 versions are quite different, but register + * address is the same. User is supposed to know the card + * family anyway... */ + if (dev_priv->chipset >= 0x40) { + getparam->value = nv_rd32(dev, NV40_PMC_GRAPH_UNITS); + break; + } + /* FALLTHRU */ default: NV_ERROR(dev, "unknown parameter %lld\n", getparam->param); return -EINVAL; diff --git a/drivers/gpu/drm/nouveau/nv04_crtc.c b/drivers/gpu/drm/nouveau/nv04_crtc.c index d2f143ed97c1..a1d1ebb073d9 100644 --- a/drivers/gpu/drm/nouveau/nv04_crtc.c +++ b/drivers/gpu/drm/nouveau/nv04_crtc.c @@ -926,9 +926,7 @@ nv04_crtc_cursor_set(struct drm_crtc *crtc, struct drm_file *file_priv, nv_crtc->cursor.set_offset(nv_crtc, nv_crtc->cursor.offset); nv_crtc->cursor.show(nv_crtc, true); out: - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gem); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gem); return ret; } diff --git a/drivers/gpu/drm/nouveau/nv04_dac.c b/drivers/gpu/drm/nouveau/nv04_dac.c index d0e038d28948..1cb19e3acb55 100644 --- a/drivers/gpu/drm/nouveau/nv04_dac.c +++ b/drivers/gpu/drm/nouveau/nv04_dac.c @@ -119,7 +119,7 @@ static enum drm_connector_status nv04_dac_detect(struct drm_encoder *encoder, struct drm_connector *connector) { struct drm_device *dev = encoder->dev; - uint8_t saved_seq1, saved_pi, saved_rpc1; + uint8_t saved_seq1, saved_pi, saved_rpc1, saved_cr_mode; uint8_t saved_palette0[3], saved_palette_mask; uint32_t saved_rtest_ctrl, saved_rgen_ctrl; int i; @@ -135,6 +135,9 @@ static enum drm_connector_status nv04_dac_detect(struct drm_encoder *encoder, /* only implemented for head A for now */ NVSetOwner(dev, 0); + saved_cr_mode = NVReadVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX); + NVWriteVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX, saved_cr_mode | 0x80); + saved_seq1 = NVReadVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX); NVWriteVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX, saved_seq1 & ~0x20); @@ -203,6 +206,7 @@ out: NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_PIXEL_INDEX, saved_pi); NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_RPC1_INDEX, saved_rpc1); NVWriteVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX, saved_seq1); + NVWriteVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX, saved_cr_mode); if (blue == 0x18) { NV_INFO(dev, "Load detected on head A\n"); @@ -226,13 +230,13 @@ uint32_t nv17_dac_sample_load(struct drm_encoder *encoder) if (dcb->type == OUTPUT_TV) { testval = RGB_TEST_DATA(0xa0, 0xa0, 0xa0); - if (dev_priv->vbios->tvdactestval) - testval = dev_priv->vbios->tvdactestval; + if (dev_priv->vbios.tvdactestval) + testval = dev_priv->vbios.tvdactestval; } else { testval = RGB_TEST_DATA(0x140, 0x140, 0x140); /* 0x94050140 */ - if (dev_priv->vbios->dactestval) - testval = dev_priv->vbios->dactestval; + if (dev_priv->vbios.dactestval) + testval = dev_priv->vbios.dactestval; } saved_rtest_ctrl = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset); diff --git a/drivers/gpu/drm/nouveau/nv04_dfp.c b/drivers/gpu/drm/nouveau/nv04_dfp.c index 483f875bdb6a..41634d4752fe 100644 --- a/drivers/gpu/drm/nouveau/nv04_dfp.c +++ b/drivers/gpu/drm/nouveau/nv04_dfp.c @@ -269,10 +269,10 @@ static void nv04_dfp_mode_set(struct drm_encoder *encoder, regp->fp_horiz_regs[FP_TOTAL] = output_mode->htotal - 1; if (!nv_gf4_disp_arch(dev) || (output_mode->hsync_start - output_mode->hdisplay) >= - dev_priv->vbios->digital_min_front_porch) + dev_priv->vbios.digital_min_front_porch) regp->fp_horiz_regs[FP_CRTC] = output_mode->hdisplay; else - regp->fp_horiz_regs[FP_CRTC] = output_mode->hsync_start - dev_priv->vbios->digital_min_front_porch - 1; + regp->fp_horiz_regs[FP_CRTC] = output_mode->hsync_start - dev_priv->vbios.digital_min_front_porch - 1; regp->fp_horiz_regs[FP_SYNC_START] = output_mode->hsync_start - 1; regp->fp_horiz_regs[FP_SYNC_END] = output_mode->hsync_end - 1; regp->fp_horiz_regs[FP_VALID_START] = output_mode->hskew; diff --git a/drivers/gpu/drm/nouveau/nv04_display.c b/drivers/gpu/drm/nouveau/nv04_display.c index ef77215fa5b9..c7898b4f6dfb 100644 --- a/drivers/gpu/drm/nouveau/nv04_display.c +++ b/drivers/gpu/drm/nouveau/nv04_display.c @@ -93,10 +93,9 @@ int nv04_display_create(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct parsed_dcb *dcb = dev_priv->vbios->dcb; + struct dcb_table *dcb = &dev_priv->vbios.dcb; struct drm_encoder *encoder; struct drm_crtc *crtc; - uint16_t connector[16] = { 0 }; int i, ret; NV_DEBUG_KMS(dev, "\n"); @@ -154,52 +153,10 @@ nv04_display_create(struct drm_device *dev) if (ret) continue; - - connector[dcbent->connector] |= (1 << dcbent->type); } - for (i = 0; i < dcb->entries; i++) { - struct dcb_entry *dcbent = &dcb->entry[i]; - uint16_t encoders; - int type; - - encoders = connector[dcbent->connector]; - if (!(encoders & (1 << dcbent->type))) - continue; - connector[dcbent->connector] = 0; - - switch (dcbent->type) { - case OUTPUT_ANALOG: - if (!MULTIPLE_ENCODERS(encoders)) - type = DRM_MODE_CONNECTOR_VGA; - else - type = DRM_MODE_CONNECTOR_DVII; - break; - case OUTPUT_TMDS: - if (!MULTIPLE_ENCODERS(encoders)) - type = DRM_MODE_CONNECTOR_DVID; - else - type = DRM_MODE_CONNECTOR_DVII; - break; - case OUTPUT_LVDS: - type = DRM_MODE_CONNECTOR_LVDS; -#if 0 - /* don't create i2c adapter when lvds ddc not allowed */ - if (dcbent->lvdsconf.use_straps_for_mode || - dev_priv->vbios->fp_no_ddc) - i2c_index = 0xf; -#endif - break; - case OUTPUT_TV: - type = DRM_MODE_CONNECTOR_TV; - break; - default: - type = DRM_MODE_CONNECTOR_Unknown; - continue; - } - - nouveau_connector_create(dev, dcbent->connector, type); - } + for (i = 0; i < dcb->connector.entries; i++) + nouveau_connector_create(dev, &dcb->connector.entry[i]); /* Save previous state */ NVLockVgaCrtcs(dev, false); diff --git a/drivers/gpu/drm/nouveau/nv04_fbcon.c b/drivers/gpu/drm/nouveau/nv04_fbcon.c index d910873c1368..3da90c2c4e63 100644 --- a/drivers/gpu/drm/nouveau/nv04_fbcon.c +++ b/drivers/gpu/drm/nouveau/nv04_fbcon.c @@ -27,7 +27,7 @@ #include "nouveau_dma.h" #include "nouveau_fbcon.h" -static void +void nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) { struct nouveau_fbcon_par *par = info->par; @@ -54,7 +54,7 @@ nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) FIRE_RING(chan); } -static void +void nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) { struct nouveau_fbcon_par *par = info->par; @@ -88,7 +88,7 @@ nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) FIRE_RING(chan); } -static void +void nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) { struct nouveau_fbcon_par *par = info->par; @@ -118,7 +118,7 @@ nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) return; } - width = (image->width + 31) & ~31; + width = ALIGN(image->width, 32); dsize = (width * image->height) >> 5; if (info->fix.visual == FB_VISUAL_TRUECOLOR || @@ -307,9 +307,6 @@ nv04_fbcon_accel_init(struct fb_info *info) FIRE_RING(chan); - info->fbops->fb_fillrect = nv04_fbcon_fillrect; - info->fbops->fb_copyarea = nv04_fbcon_copyarea; - info->fbops->fb_imageblit = nv04_fbcon_imageblit; return 0; } diff --git a/drivers/gpu/drm/nouveau/nv04_fifo.c b/drivers/gpu/drm/nouveau/nv04_fifo.c index f31347b8c9b0..66fe55983b6e 100644 --- a/drivers/gpu/drm/nouveau/nv04_fifo.c +++ b/drivers/gpu/drm/nouveau/nv04_fifo.c @@ -117,6 +117,7 @@ nv04_fifo_create_context(struct nouveau_channel *chan) { struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; int ret; ret = nouveau_gpuobj_new_fake(dev, NV04_RAMFC(chan->id), ~0, @@ -127,6 +128,8 @@ nv04_fifo_create_context(struct nouveau_channel *chan) if (ret) return ret; + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + /* Setup initial state */ dev_priv->engine.instmem.prepare_access(dev, true); RAMFC_WR(DMA_PUT, chan->pushbuf_base); @@ -144,6 +147,8 @@ nv04_fifo_create_context(struct nouveau_channel *chan) /* enable the fifo dma operation */ nv_wr32(dev, NV04_PFIFO_MODE, nv_rd32(dev, NV04_PFIFO_MODE) | (1 << chan->id)); + + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); return 0; } diff --git a/drivers/gpu/drm/nouveau/nv04_tv.c b/drivers/gpu/drm/nouveau/nv04_tv.c index 9c63099e9c42..c4e3404337d4 100644 --- a/drivers/gpu/drm/nouveau/nv04_tv.c +++ b/drivers/gpu/drm/nouveau/nv04_tv.c @@ -262,7 +262,7 @@ int nv04_tv_create(struct drm_device *dev, struct dcb_entry *entry) nv_encoder->or = ffs(entry->or) - 1; /* Run the slave-specific initialization */ - adap = &dev_priv->vbios->dcb->i2c[i2c_index].chan->adapter; + adap = &dev_priv->vbios.dcb.i2c[i2c_index].chan->adapter; was_locked = NVLockVgaCrtcs(dev, false); diff --git a/drivers/gpu/drm/nouveau/nv17_tv.c b/drivers/gpu/drm/nouveau/nv17_tv.c index 58b917c3341b..74c880374fb9 100644 --- a/drivers/gpu/drm/nouveau/nv17_tv.c +++ b/drivers/gpu/drm/nouveau/nv17_tv.c @@ -45,8 +45,8 @@ static uint32_t nv42_tv_sample_load(struct drm_encoder *encoder) #define RGB_TEST_DATA(r, g, b) (r << 0 | g << 10 | b << 20) testval = RGB_TEST_DATA(0x82, 0xeb, 0x82); - if (dev_priv->vbios->tvdactestval) - testval = dev_priv->vbios->tvdactestval; + if (dev_priv->vbios.tvdactestval) + testval = dev_priv->vbios.tvdactestval; dacclk = NVReadRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset); head = (dacclk & 0x100) >> 8; @@ -367,7 +367,7 @@ static void nv17_tv_prepare(struct drm_encoder *encoder) !enc->crtc && nv04_dfp_get_bound_head(dev, dcb) == head) { nv04_dfp_bind_head(dev, dcb, head ^ 1, - dev_priv->VBIOS.fp.dual_link); + dev_priv->vbios.fp.dual_link); } } @@ -579,6 +579,8 @@ static void nv17_tv_restore(struct drm_encoder *encoder) nouveau_encoder(encoder)->restore.output); nv17_tv_state_load(dev, &to_tv_enc(encoder)->saved_state); + + nouveau_encoder(encoder)->last_dpms = NV_DPMS_CLEARED; } static int nv17_tv_create_resources(struct drm_encoder *encoder, diff --git a/drivers/gpu/drm/nouveau/nv40_fifo.c b/drivers/gpu/drm/nouveau/nv40_fifo.c index b4f19ccb8b41..6b2ef4a9fce1 100644 --- a/drivers/gpu/drm/nouveau/nv40_fifo.c +++ b/drivers/gpu/drm/nouveau/nv40_fifo.c @@ -37,6 +37,7 @@ nv40_fifo_create_context(struct nouveau_channel *chan) struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; uint32_t fc = NV40_RAMFC(chan->id); + unsigned long flags; int ret; ret = nouveau_gpuobj_new_fake(dev, NV40_RAMFC(chan->id), ~0, @@ -45,6 +46,8 @@ nv40_fifo_create_context(struct nouveau_channel *chan) if (ret) return ret; + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + dev_priv->engine.instmem.prepare_access(dev, true); nv_wi32(dev, fc + 0, chan->pushbuf_base); nv_wi32(dev, fc + 4, chan->pushbuf_base); @@ -63,6 +66,8 @@ nv40_fifo_create_context(struct nouveau_channel *chan) /* enable the fifo dma operation */ nv_wr32(dev, NV04_PFIFO_MODE, nv_rd32(dev, NV04_PFIFO_MODE) | (1 << chan->id)); + + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); return 0; } diff --git a/drivers/gpu/drm/nouveau/nv50_crtc.c b/drivers/gpu/drm/nouveau/nv50_crtc.c index 40b7360841f8..cfabeb974a56 100644 --- a/drivers/gpu/drm/nouveau/nv50_crtc.c +++ b/drivers/gpu/drm/nouveau/nv50_crtc.c @@ -298,14 +298,17 @@ nv50_crtc_set_clock(struct drm_device *dev, int head, int pclk) static void nv50_crtc_destroy(struct drm_crtc *crtc) { - struct drm_device *dev = crtc->dev; - struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); - - NV_DEBUG_KMS(dev, "\n"); + struct drm_device *dev; + struct nouveau_crtc *nv_crtc; if (!crtc) return; + dev = crtc->dev; + nv_crtc = nouveau_crtc(crtc); + + NV_DEBUG_KMS(dev, "\n"); + drm_crtc_cleanup(&nv_crtc->base); nv50_cursor_fini(nv_crtc); @@ -355,9 +358,7 @@ nv50_crtc_cursor_set(struct drm_crtc *crtc, struct drm_file *file_priv, nv_crtc->cursor.show(nv_crtc, true); out: - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gem); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gem); return ret; } diff --git a/drivers/gpu/drm/nouveau/nv50_dac.c b/drivers/gpu/drm/nouveau/nv50_dac.c index f08f042a8e10..1fd9537beff6 100644 --- a/drivers/gpu/drm/nouveau/nv50_dac.c +++ b/drivers/gpu/drm/nouveau/nv50_dac.c @@ -79,8 +79,8 @@ nv50_dac_detect(struct drm_encoder *encoder, struct drm_connector *connector) } /* Use bios provided value if possible. */ - if (dev_priv->vbios->dactestval) { - load_pattern = dev_priv->vbios->dactestval; + if (dev_priv->vbios.dactestval) { + load_pattern = dev_priv->vbios.dactestval; NV_DEBUG_KMS(dev, "Using bios provided load_pattern of %d\n", load_pattern); } else { diff --git a/drivers/gpu/drm/nouveau/nv50_display.c b/drivers/gpu/drm/nouveau/nv50_display.c index 90f0bf59fbcd..61a89f2dc553 100644 --- a/drivers/gpu/drm/nouveau/nv50_display.c +++ b/drivers/gpu/drm/nouveau/nv50_display.c @@ -370,9 +370,7 @@ nv50_display_init(struct drm_device *dev) struct nouveau_connector *conn = nouveau_connector(connector); struct dcb_gpio_entry *gpio; - if (connector->connector_type != DRM_MODE_CONNECTOR_DVII && - connector->connector_type != DRM_MODE_CONNECTOR_DVID && - connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort) + if (conn->dcb->gpio_tag == 0xff) continue; gpio = nouveau_bios_gpio_entry(dev, conn->dcb->gpio_tag); @@ -465,8 +463,7 @@ static int nv50_display_disable(struct drm_device *dev) int nv50_display_create(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct parsed_dcb *dcb = dev_priv->vbios->dcb; - uint32_t connector[16] = {}; + struct dcb_table *dcb = &dev_priv->vbios.dcb; int ret, i; NV_DEBUG_KMS(dev, "\n"); @@ -522,44 +519,13 @@ int nv50_display_create(struct drm_device *dev) NV_WARN(dev, "DCB encoder %d unknown\n", entry->type); continue; } - - connector[entry->connector] |= (1 << entry->type); } - /* It appears that DCB 3.0+ VBIOS has a connector table, however, - * I'm not 100% certain how to decode it correctly yet so just - * look at what encoders are present on each connector index and - * attempt to derive the connector type from that. - */ - for (i = 0 ; i < dcb->entries; i++) { - struct dcb_entry *entry = &dcb->entry[i]; - uint16_t encoders; - int type; - - encoders = connector[entry->connector]; - if (!(encoders & (1 << entry->type))) + for (i = 0 ; i < dcb->connector.entries; i++) { + if (i != 0 && dcb->connector.entry[i].index == + dcb->connector.entry[i - 1].index) continue; - connector[entry->connector] = 0; - - if (encoders & (1 << OUTPUT_DP)) { - type = DRM_MODE_CONNECTOR_DisplayPort; - } else if (encoders & (1 << OUTPUT_TMDS)) { - if (encoders & (1 << OUTPUT_ANALOG)) - type = DRM_MODE_CONNECTOR_DVII; - else - type = DRM_MODE_CONNECTOR_DVID; - } else if (encoders & (1 << OUTPUT_ANALOG)) { - type = DRM_MODE_CONNECTOR_VGA; - } else if (encoders & (1 << OUTPUT_LVDS)) { - type = DRM_MODE_CONNECTOR_LVDS; - } else { - type = DRM_MODE_CONNECTOR_Unknown; - } - - if (type == DRM_MODE_CONNECTOR_Unknown) - continue; - - nouveau_connector_create(dev, entry->connector, type); + nouveau_connector_create(dev, &dcb->connector.entry[i]); } ret = nv50_display_init(dev); @@ -667,8 +633,8 @@ nv50_display_irq_head(struct drm_device *dev, int *phead, return -1; } - for (i = 0; i < dev_priv->vbios->dcb->entries; i++) { - struct dcb_entry *dcbent = &dev_priv->vbios->dcb->entry[i]; + for (i = 0; i < dev_priv->vbios.dcb.entries; i++) { + struct dcb_entry *dcbent = &dev_priv->vbios.dcb.entry[i]; if (dcbent->type != type) continue; @@ -692,7 +658,7 @@ nv50_display_script_select(struct drm_device *dev, struct dcb_entry *dcbent, struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_connector *nv_connector = NULL; struct drm_encoder *encoder; - struct nvbios *bios = &dev_priv->VBIOS; + struct nvbios *bios = &dev_priv->vbios; uint32_t mc, script = 0, or; list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { @@ -710,7 +676,7 @@ nv50_display_script_select(struct drm_device *dev, struct dcb_entry *dcbent, switch (dcbent->type) { case OUTPUT_LVDS: script = (mc >> 8) & 0xf; - if (bios->pub.fp_no_ddc) { + if (bios->fp_no_ddc) { if (bios->fp.dual_link) script |= 0x0100; if (bios->fp.if_is_24bit) diff --git a/drivers/gpu/drm/nouveau/nv50_fbcon.c b/drivers/gpu/drm/nouveau/nv50_fbcon.c index e4f279ee61cf..993c7126fbde 100644 --- a/drivers/gpu/drm/nouveau/nv50_fbcon.c +++ b/drivers/gpu/drm/nouveau/nv50_fbcon.c @@ -3,7 +3,7 @@ #include "nouveau_dma.h" #include "nouveau_fbcon.h" -static void +void nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) { struct nouveau_fbcon_par *par = info->par; @@ -46,7 +46,7 @@ nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) FIRE_RING(chan); } -static void +void nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) { struct nouveau_fbcon_par *par = info->par; @@ -81,7 +81,7 @@ nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) FIRE_RING(chan); } -static void +void nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) { struct nouveau_fbcon_par *par = info->par; @@ -109,7 +109,7 @@ nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) return; } - width = (image->width + 31) & ~31; + width = ALIGN(image->width, 32); dwords = (width * image->height) >> 5; BEGIN_RING(chan, NvSub2D, 0x0814, 2); @@ -262,9 +262,6 @@ nv50_fbcon_accel_init(struct fb_info *info) OUT_RING(chan, info->fix.smem_start - dev_priv->fb_phys + dev_priv->vm_vram_base); - info->fbops->fb_fillrect = nv50_fbcon_fillrect; - info->fbops->fb_copyarea = nv50_fbcon_copyarea; - info->fbops->fb_imageblit = nv50_fbcon_imageblit; return 0; } diff --git a/drivers/gpu/drm/nouveau/nv50_fifo.c b/drivers/gpu/drm/nouveau/nv50_fifo.c index 32b244bcb482..e20c0e2474f3 100644 --- a/drivers/gpu/drm/nouveau/nv50_fifo.c +++ b/drivers/gpu/drm/nouveau/nv50_fifo.c @@ -243,6 +243,7 @@ nv50_fifo_create_context(struct nouveau_channel *chan) struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_gpuobj *ramfc = NULL; + unsigned long flags; int ret; NV_DEBUG(dev, "ch%d\n", chan->id); @@ -278,19 +279,21 @@ nv50_fifo_create_context(struct nouveau_channel *chan) return ret; } + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + dev_priv->engine.instmem.prepare_access(dev, true); - nv_wo32(dev, ramfc, 0x08/4, chan->pushbuf_base); - nv_wo32(dev, ramfc, 0x10/4, chan->pushbuf_base); nv_wo32(dev, ramfc, 0x48/4, chan->pushbuf->instance >> 4); nv_wo32(dev, ramfc, 0x80/4, (0xc << 24) | (chan->ramht->instance >> 4)); - nv_wo32(dev, ramfc, 0x3c/4, 0x00086078); nv_wo32(dev, ramfc, 0x44/4, 0x2101ffff); nv_wo32(dev, ramfc, 0x60/4, 0x7fffffff); nv_wo32(dev, ramfc, 0x40/4, 0x00000000); nv_wo32(dev, ramfc, 0x7c/4, 0x30000001); nv_wo32(dev, ramfc, 0x78/4, 0x00000000); - nv_wo32(dev, ramfc, 0x4c/4, 0xffffffff); + nv_wo32(dev, ramfc, 0x3c/4, 0x403f6078); + nv_wo32(dev, ramfc, 0x50/4, chan->pushbuf_base + + chan->dma.ib_base * 4); + nv_wo32(dev, ramfc, 0x54/4, drm_order(chan->dma.ib_max + 1) << 16); if (!IS_G80) { nv_wo32(dev, chan->ramin->gpuobj, 0, chan->id); @@ -306,10 +309,12 @@ nv50_fifo_create_context(struct nouveau_channel *chan) ret = nv50_fifo_channel_enable(dev, chan->id, false); if (ret) { NV_ERROR(dev, "error enabling ch%d: %d\n", chan->id, ret); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); nouveau_gpuobj_ref_del(dev, &chan->ramfc); return ret; } + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); return 0; } @@ -317,17 +322,20 @@ void nv50_fifo_destroy_context(struct nouveau_channel *chan) { struct drm_device *dev = chan->dev; + struct nouveau_gpuobj_ref *ramfc = chan->ramfc; NV_DEBUG(dev, "ch%d\n", chan->id); - nouveau_gpuobj_ref_del(dev, &chan->ramfc); - nouveau_gpuobj_ref_del(dev, &chan->cache); - + /* This will ensure the channel is seen as disabled. */ + chan->ramfc = NULL; nv50_fifo_channel_disable(dev, chan->id, false); /* Dummy channel, also used on ch 127 */ if (chan->id == 0) nv50_fifo_channel_disable(dev, 127, false); + + nouveau_gpuobj_ref_del(dev, &ramfc); + nouveau_gpuobj_ref_del(dev, &chan->cache); } int diff --git a/drivers/gpu/drm/nouveau/nv50_graph.c b/drivers/gpu/drm/nouveau/nv50_graph.c index 20319e59d368..857a09671a39 100644 --- a/drivers/gpu/drm/nouveau/nv50_graph.c +++ b/drivers/gpu/drm/nouveau/nv50_graph.c @@ -28,30 +28,7 @@ #include "drm.h" #include "nouveau_drv.h" -MODULE_FIRMWARE("nouveau/nv50.ctxprog"); -MODULE_FIRMWARE("nouveau/nv50.ctxvals"); -MODULE_FIRMWARE("nouveau/nv84.ctxprog"); -MODULE_FIRMWARE("nouveau/nv84.ctxvals"); -MODULE_FIRMWARE("nouveau/nv86.ctxprog"); -MODULE_FIRMWARE("nouveau/nv86.ctxvals"); -MODULE_FIRMWARE("nouveau/nv92.ctxprog"); -MODULE_FIRMWARE("nouveau/nv92.ctxvals"); -MODULE_FIRMWARE("nouveau/nv94.ctxprog"); -MODULE_FIRMWARE("nouveau/nv94.ctxvals"); -MODULE_FIRMWARE("nouveau/nv96.ctxprog"); -MODULE_FIRMWARE("nouveau/nv96.ctxvals"); -MODULE_FIRMWARE("nouveau/nv98.ctxprog"); -MODULE_FIRMWARE("nouveau/nv98.ctxvals"); -MODULE_FIRMWARE("nouveau/nva0.ctxprog"); -MODULE_FIRMWARE("nouveau/nva0.ctxvals"); -MODULE_FIRMWARE("nouveau/nva5.ctxprog"); -MODULE_FIRMWARE("nouveau/nva5.ctxvals"); -MODULE_FIRMWARE("nouveau/nva8.ctxprog"); -MODULE_FIRMWARE("nouveau/nva8.ctxvals"); -MODULE_FIRMWARE("nouveau/nvaa.ctxprog"); -MODULE_FIRMWARE("nouveau/nvaa.ctxvals"); -MODULE_FIRMWARE("nouveau/nvac.ctxprog"); -MODULE_FIRMWARE("nouveau/nvac.ctxvals"); +#include "nouveau_grctx.h" #define IS_G80 ((dev_priv->chipset & 0xf0) == 0x50) @@ -111,9 +88,34 @@ nv50_graph_init_ctxctl(struct drm_device *dev) NV_DEBUG(dev, "\n"); - nouveau_grctx_prog_load(dev); - if (!dev_priv->engine.graph.ctxprog) - dev_priv->engine.graph.accel_blocked = true; + if (nouveau_ctxfw) { + nouveau_grctx_prog_load(dev); + dev_priv->engine.graph.grctx_size = 0x70000; + } + if (!dev_priv->engine.graph.ctxprog) { + struct nouveau_grctx ctx = {}; + uint32_t *cp = kmalloc(512 * 4, GFP_KERNEL); + int i; + if (!cp) { + NV_ERROR(dev, "Couldn't alloc ctxprog! Disabling acceleration.\n"); + dev_priv->engine.graph.accel_blocked = true; + return 0; + } + ctx.dev = dev; + ctx.mode = NOUVEAU_GRCTX_PROG; + ctx.data = cp; + ctx.ctxprog_max = 512; + if (!nv50_grctx_init(&ctx)) { + dev_priv->engine.graph.grctx_size = ctx.ctxvals_pos * 4; + + nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0); + for (i = 0; i < ctx.ctxprog_len; i++) + nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_DATA, cp[i]); + } else { + dev_priv->engine.graph.accel_blocked = true; + } + kfree(cp); + } nv_wr32(dev, 0x400320, 4); nv_wr32(dev, NV40_PGRAPH_CTXCTL_CUR, 0); @@ -165,6 +167,12 @@ nv50_graph_channel(struct drm_device *dev) uint32_t inst; int i; + /* Be sure we're not in the middle of a context switch or bad things + * will happen, such as unloading the wrong pgraph context. + */ + if (!nv_wait(0x400300, 0x00000001, 0x00000000)) + NV_ERROR(dev, "Ctxprog is still running\n"); + inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR); if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED)) return NULL; @@ -187,13 +195,13 @@ nv50_graph_create_context(struct nouveau_channel *chan) struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_gpuobj *ramin = chan->ramin->gpuobj; struct nouveau_gpuobj *ctx; - uint32_t grctx_size = 0x70000; + struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; int hdr, ret; NV_DEBUG(dev, "ch%d\n", chan->id); - ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, grctx_size, 0x1000, - NVOBJ_FLAG_ZERO_ALLOC | + ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, pgraph->grctx_size, + 0x1000, NVOBJ_FLAG_ZERO_ALLOC | NVOBJ_FLAG_ZERO_FREE, &chan->ramin_grctx); if (ret) return ret; @@ -203,7 +211,7 @@ nv50_graph_create_context(struct nouveau_channel *chan) dev_priv->engine.instmem.prepare_access(dev, true); nv_wo32(dev, ramin, (hdr + 0x00)/4, 0x00190002); nv_wo32(dev, ramin, (hdr + 0x04)/4, chan->ramin_grctx->instance + - grctx_size - 1); + pgraph->grctx_size - 1); nv_wo32(dev, ramin, (hdr + 0x08)/4, chan->ramin_grctx->instance); nv_wo32(dev, ramin, (hdr + 0x0c)/4, 0); nv_wo32(dev, ramin, (hdr + 0x10)/4, 0); @@ -211,7 +219,15 @@ nv50_graph_create_context(struct nouveau_channel *chan) dev_priv->engine.instmem.finish_access(dev); dev_priv->engine.instmem.prepare_access(dev, true); - nouveau_grctx_vals_load(dev, ctx); + if (!pgraph->ctxprog) { + struct nouveau_grctx ctx = {}; + ctx.dev = chan->dev; + ctx.mode = NOUVEAU_GRCTX_VALS; + ctx.data = chan->ramin_grctx->gpuobj; + nv50_grctx_init(&ctx); + } else { + nouveau_grctx_vals_load(dev, ctx); + } nv_wo32(dev, ctx, 0x00000/4, chan->ramin->instance >> 12); if ((dev_priv->chipset & 0xf0) == 0xa0) nv_wo32(dev, ctx, 0x00004/4, 0x00000000); @@ -275,7 +291,7 @@ nv50_graph_load_context(struct nouveau_channel *chan) int nv50_graph_unload_context(struct drm_device *dev) { - uint32_t inst, fifo = nv_rd32(dev, 0x400500); + uint32_t inst; inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR); if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED)) @@ -283,12 +299,10 @@ nv50_graph_unload_context(struct drm_device *dev) inst &= NV50_PGRAPH_CTXCTL_CUR_INSTANCE; nouveau_wait_for_idle(dev); - nv_wr32(dev, 0x400500, fifo & ~1); nv_wr32(dev, 0x400784, inst); nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x20); nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 0x01); nouveau_wait_for_idle(dev); - nv_wr32(dev, 0x400500, fifo); nv_wr32(dev, NV50_PGRAPH_CTXCTL_CUR, inst); return 0; diff --git a/drivers/gpu/drm/nouveau/nv50_grctx.c b/drivers/gpu/drm/nouveau/nv50_grctx.c new file mode 100644 index 000000000000..d105fcd42ca0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_grctx.c @@ -0,0 +1,2367 @@ +/* + * Copyright 2009 Marcin KoÅ›cielnicki + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#define CP_FLAG_CLEAR 0 +#define CP_FLAG_SET 1 +#define CP_FLAG_SWAP_DIRECTION ((0 * 32) + 0) +#define CP_FLAG_SWAP_DIRECTION_LOAD 0 +#define CP_FLAG_SWAP_DIRECTION_SAVE 1 +#define CP_FLAG_UNK01 ((0 * 32) + 1) +#define CP_FLAG_UNK01_CLEAR 0 +#define CP_FLAG_UNK01_SET 1 +#define CP_FLAG_UNK03 ((0 * 32) + 3) +#define CP_FLAG_UNK03_CLEAR 0 +#define CP_FLAG_UNK03_SET 1 +#define CP_FLAG_USER_SAVE ((0 * 32) + 5) +#define CP_FLAG_USER_SAVE_NOT_PENDING 0 +#define CP_FLAG_USER_SAVE_PENDING 1 +#define CP_FLAG_USER_LOAD ((0 * 32) + 6) +#define CP_FLAG_USER_LOAD_NOT_PENDING 0 +#define CP_FLAG_USER_LOAD_PENDING 1 +#define CP_FLAG_UNK0B ((0 * 32) + 0xb) +#define CP_FLAG_UNK0B_CLEAR 0 +#define CP_FLAG_UNK0B_SET 1 +#define CP_FLAG_UNK1D ((0 * 32) + 0x1d) +#define CP_FLAG_UNK1D_CLEAR 0 +#define CP_FLAG_UNK1D_SET 1 +#define CP_FLAG_UNK20 ((1 * 32) + 0) +#define CP_FLAG_UNK20_CLEAR 0 +#define CP_FLAG_UNK20_SET 1 +#define CP_FLAG_STATUS ((2 * 32) + 0) +#define CP_FLAG_STATUS_BUSY 0 +#define CP_FLAG_STATUS_IDLE 1 +#define CP_FLAG_AUTO_SAVE ((2 * 32) + 4) +#define CP_FLAG_AUTO_SAVE_NOT_PENDING 0 +#define CP_FLAG_AUTO_SAVE_PENDING 1 +#define CP_FLAG_AUTO_LOAD ((2 * 32) + 5) +#define CP_FLAG_AUTO_LOAD_NOT_PENDING 0 +#define CP_FLAG_AUTO_LOAD_PENDING 1 +#define CP_FLAG_XFER ((2 * 32) + 11) +#define CP_FLAG_XFER_IDLE 0 +#define CP_FLAG_XFER_BUSY 1 +#define CP_FLAG_NEWCTX ((2 * 32) + 12) +#define CP_FLAG_NEWCTX_BUSY 0 +#define CP_FLAG_NEWCTX_DONE 1 +#define CP_FLAG_ALWAYS ((2 * 32) + 13) +#define CP_FLAG_ALWAYS_FALSE 0 +#define CP_FLAG_ALWAYS_TRUE 1 + +#define CP_CTX 0x00100000 +#define CP_CTX_COUNT 0x000f0000 +#define CP_CTX_COUNT_SHIFT 16 +#define CP_CTX_REG 0x00003fff +#define CP_LOAD_SR 0x00200000 +#define CP_LOAD_SR_VALUE 0x000fffff +#define CP_BRA 0x00400000 +#define CP_BRA_IP 0x0001ff00 +#define CP_BRA_IP_SHIFT 8 +#define CP_BRA_IF_CLEAR 0x00000080 +#define CP_BRA_FLAG 0x0000007f +#define CP_WAIT 0x00500000 +#define CP_WAIT_SET 0x00000080 +#define CP_WAIT_FLAG 0x0000007f +#define CP_SET 0x00700000 +#define CP_SET_1 0x00000080 +#define CP_SET_FLAG 0x0000007f +#define CP_NEWCTX 0x00600004 +#define CP_NEXT_TO_SWAP 0x00600005 +#define CP_SET_CONTEXT_POINTER 0x00600006 +#define CP_SET_XFER_POINTER 0x00600007 +#define CP_ENABLE 0x00600009 +#define CP_END 0x0060000c +#define CP_NEXT_TO_CURRENT 0x0060000d +#define CP_DISABLE1 0x0090ffff +#define CP_DISABLE2 0x0091ffff +#define CP_XFER_1 0x008000ff +#define CP_XFER_2 0x008800ff +#define CP_SEEK_1 0x00c000ff +#define CP_SEEK_2 0x00c800ff + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_grctx.h" + +/* + * This code deals with PGRAPH contexts on NV50 family cards. Like NV40, it's + * the GPU itself that does context-switching, but it needs a special + * microcode to do it. And it's the driver's task to supply this microcode, + * further known as ctxprog, as well as the initial context values, known + * as ctxvals. + * + * Without ctxprog, you cannot switch contexts. Not even in software, since + * the majority of context [xfer strands] isn't accessible directly. You're + * stuck with a single channel, and you also suffer all the problems resulting + * from missing ctxvals, since you cannot load them. + * + * Without ctxvals, you're stuck with PGRAPH's default context. It's enough to + * run 2d operations, but trying to utilise 3d or CUDA will just lock you up, + * since you don't have... some sort of needed setup. + * + * Nouveau will just disable acceleration if not given ctxprog + ctxvals, since + * it's too much hassle to handle no-ctxprog as a special case. + */ + +/* + * How ctxprogs work. + * + * The ctxprog is written in its own kind of microcode, with very small and + * crappy set of available commands. You upload it to a small [512 insns] + * area of memory on PGRAPH, and it'll be run when PFIFO wants PGRAPH to + * switch channel. or when the driver explicitely requests it. Stuff visible + * to ctxprog consists of: PGRAPH MMIO registers, PGRAPH context strands, + * the per-channel context save area in VRAM [known as ctxvals or grctx], + * 4 flags registers, a scratch register, two grctx pointers, plus many + * random poorly-understood details. + * + * When ctxprog runs, it's supposed to check what operations are asked of it, + * save old context if requested, optionally reset PGRAPH and switch to the + * new channel, and load the new context. Context consists of three major + * parts: subset of MMIO registers and two "xfer areas". + */ + +/* TODO: + * - document unimplemented bits compared to nvidia + * - NVAx: make a TP subroutine, use it. + * - use 0x4008fc instead of 0x1540? + */ + +enum cp_label { + cp_check_load = 1, + cp_setup_auto_load, + cp_setup_load, + cp_setup_save, + cp_swap_state, + cp_prepare_exit, + cp_exit, +}; + +static void nv50_graph_construct_mmio(struct nouveau_grctx *ctx); +static void nv50_graph_construct_xfer1(struct nouveau_grctx *ctx); +static void nv50_graph_construct_xfer2(struct nouveau_grctx *ctx); + +/* Main function: construct the ctxprog skeleton, call the other functions. */ + +int +nv50_grctx_init(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + + switch (dev_priv->chipset) { + case 0x50: + case 0x84: + case 0x86: + case 0x92: + case 0x94: + case 0x96: + case 0x98: + case 0xa0: + case 0xa5: + case 0xa8: + case 0xaa: + case 0xac: + break; + default: + NV_ERROR(ctx->dev, "I don't know how to make a ctxprog for " + "your NV%x card.\n", dev_priv->chipset); + NV_ERROR(ctx->dev, "Disabling acceleration. Please contact " + "the devs.\n"); + return -ENOSYS; + } + /* decide whether we're loading/unloading the context */ + cp_bra (ctx, AUTO_SAVE, PENDING, cp_setup_save); + cp_bra (ctx, USER_SAVE, PENDING, cp_setup_save); + + cp_name(ctx, cp_check_load); + cp_bra (ctx, AUTO_LOAD, PENDING, cp_setup_auto_load); + cp_bra (ctx, USER_LOAD, PENDING, cp_setup_load); + cp_bra (ctx, ALWAYS, TRUE, cp_exit); + + /* setup for context load */ + cp_name(ctx, cp_setup_auto_load); + cp_out (ctx, CP_DISABLE1); + cp_out (ctx, CP_DISABLE2); + cp_out (ctx, CP_ENABLE); + cp_out (ctx, CP_NEXT_TO_SWAP); + cp_set (ctx, UNK01, SET); + cp_name(ctx, cp_setup_load); + cp_out (ctx, CP_NEWCTX); + cp_wait(ctx, NEWCTX, BUSY); + cp_set (ctx, UNK1D, CLEAR); + cp_set (ctx, SWAP_DIRECTION, LOAD); + cp_bra (ctx, UNK0B, SET, cp_prepare_exit); + cp_bra (ctx, ALWAYS, TRUE, cp_swap_state); + + /* setup for context save */ + cp_name(ctx, cp_setup_save); + cp_set (ctx, UNK1D, SET); + cp_wait(ctx, STATUS, BUSY); + cp_set (ctx, UNK01, SET); + cp_set (ctx, SWAP_DIRECTION, SAVE); + + /* general PGRAPH state */ + cp_name(ctx, cp_swap_state); + cp_set (ctx, UNK03, SET); + cp_pos (ctx, 0x00004/4); + cp_ctx (ctx, 0x400828, 1); /* needed. otherwise, flickering happens. */ + cp_pos (ctx, 0x00100/4); + nv50_graph_construct_mmio(ctx); + nv50_graph_construct_xfer1(ctx); + nv50_graph_construct_xfer2(ctx); + + cp_bra (ctx, SWAP_DIRECTION, SAVE, cp_check_load); + + cp_set (ctx, UNK20, SET); + cp_set (ctx, SWAP_DIRECTION, SAVE); /* no idea why this is needed, but fixes at least one lockup. */ + cp_lsr (ctx, ctx->ctxvals_base); + cp_out (ctx, CP_SET_XFER_POINTER); + cp_lsr (ctx, 4); + cp_out (ctx, CP_SEEK_1); + cp_out (ctx, CP_XFER_1); + cp_wait(ctx, XFER, BUSY); + + /* pre-exit state updates */ + cp_name(ctx, cp_prepare_exit); + cp_set (ctx, UNK01, CLEAR); + cp_set (ctx, UNK03, CLEAR); + cp_set (ctx, UNK1D, CLEAR); + + cp_bra (ctx, USER_SAVE, PENDING, cp_exit); + cp_out (ctx, CP_NEXT_TO_CURRENT); + + cp_name(ctx, cp_exit); + cp_set (ctx, USER_SAVE, NOT_PENDING); + cp_set (ctx, USER_LOAD, NOT_PENDING); + cp_out (ctx, CP_END); + ctx->ctxvals_pos += 0x400; /* padding... no idea why you need it */ + + return 0; +} + +/* + * Constructs MMIO part of ctxprog and ctxvals. Just a matter of knowing which + * registers to save/restore and the default values for them. + */ + +static void +nv50_graph_construct_mmio(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i, j; + int offset, base; + uint32_t units = nv_rd32 (ctx->dev, 0x1540); + + /* 0800 */ + cp_ctx(ctx, 0x400808, 7); + gr_def(ctx, 0x400814, 0x00000030); + cp_ctx(ctx, 0x400834, 0x32); + if (dev_priv->chipset == 0x50) { + gr_def(ctx, 0x400834, 0xff400040); + gr_def(ctx, 0x400838, 0xfff00080); + gr_def(ctx, 0x40083c, 0xfff70090); + gr_def(ctx, 0x400840, 0xffe806a8); + } + gr_def(ctx, 0x400844, 0x00000002); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + gr_def(ctx, 0x400894, 0x00001000); + gr_def(ctx, 0x4008e8, 0x00000003); + gr_def(ctx, 0x4008ec, 0x00001000); + if (dev_priv->chipset == 0x50) + cp_ctx(ctx, 0x400908, 0xb); + else if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, 0x400908, 0xc); + else + cp_ctx(ctx, 0x400908, 0xe); + + if (dev_priv->chipset >= 0xa0) + cp_ctx(ctx, 0x400b00, 0x1); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + cp_ctx(ctx, 0x400b10, 0x1); + gr_def(ctx, 0x400b10, 0x0001629d); + cp_ctx(ctx, 0x400b20, 0x1); + gr_def(ctx, 0x400b20, 0x0001629d); + } + + /* 0C00 */ + cp_ctx(ctx, 0x400c08, 0x2); + gr_def(ctx, 0x400c08, 0x0000fe0c); + + /* 1000 */ + if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, 0x401008, 0x4); + gr_def(ctx, 0x401014, 0x00001000); + } else if (dev_priv->chipset == 0xa0 || dev_priv->chipset >= 0xaa) { + cp_ctx(ctx, 0x401008, 0x5); + gr_def(ctx, 0x401018, 0x00001000); + } else { + cp_ctx(ctx, 0x401008, 0x5); + gr_def(ctx, 0x401018, 0x00004000); + } + + /* 1400 */ + cp_ctx(ctx, 0x401400, 0x8); + cp_ctx(ctx, 0x401424, 0x3); + if (dev_priv->chipset == 0x50) + gr_def(ctx, 0x40142c, 0x0001fd87); + else + gr_def(ctx, 0x40142c, 0x00000187); + cp_ctx(ctx, 0x401540, 0x5); + gr_def(ctx, 0x401550, 0x00001018); + + /* 1800 */ + cp_ctx(ctx, 0x401814, 0x1); + gr_def(ctx, 0x401814, 0x000000ff); + if (dev_priv->chipset == 0x50) { + cp_ctx(ctx, 0x40181c, 0xe); + gr_def(ctx, 0x401850, 0x00000004); + } else if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, 0x40181c, 0xf); + gr_def(ctx, 0x401854, 0x00000004); + } else { + cp_ctx(ctx, 0x40181c, 0x13); + gr_def(ctx, 0x401864, 0x00000004); + } + + /* 1C00 */ + cp_ctx(ctx, 0x401c00, 0x1); + switch (dev_priv->chipset) { + case 0x50: + gr_def(ctx, 0x401c00, 0x0001005f); + break; + case 0x84: + case 0x86: + case 0x94: + gr_def(ctx, 0x401c00, 0x044d00df); + break; + case 0x92: + case 0x96: + case 0x98: + case 0xa0: + case 0xaa: + case 0xac: + gr_def(ctx, 0x401c00, 0x042500df); + break; + case 0xa5: + case 0xa8: + gr_def(ctx, 0x401c00, 0x142500df); + break; + } + + /* 2400 */ + cp_ctx(ctx, 0x402400, 0x1); + if (dev_priv->chipset == 0x50) + cp_ctx(ctx, 0x402408, 0x1); + else + cp_ctx(ctx, 0x402408, 0x2); + gr_def(ctx, 0x402408, 0x00000600); + + /* 2800 */ + cp_ctx(ctx, 0x402800, 0x1); + if (dev_priv->chipset == 0x50) + gr_def(ctx, 0x402800, 0x00000006); + + /* 2C00 */ + cp_ctx(ctx, 0x402c08, 0x6); + if (dev_priv->chipset != 0x50) + gr_def(ctx, 0x402c14, 0x01000000); + gr_def(ctx, 0x402c18, 0x000000ff); + if (dev_priv->chipset == 0x50) + cp_ctx(ctx, 0x402ca0, 0x1); + else + cp_ctx(ctx, 0x402ca0, 0x2); + if (dev_priv->chipset < 0xa0) + gr_def(ctx, 0x402ca0, 0x00000400); + else if (dev_priv->chipset == 0xa0 || dev_priv->chipset >= 0xaa) + gr_def(ctx, 0x402ca0, 0x00000800); + else + gr_def(ctx, 0x402ca0, 0x00000400); + cp_ctx(ctx, 0x402cac, 0x4); + + /* 3000 */ + cp_ctx(ctx, 0x403004, 0x1); + gr_def(ctx, 0x403004, 0x00000001); + + /* 3404 */ + if (dev_priv->chipset >= 0xa0) { + cp_ctx(ctx, 0x403404, 0x1); + gr_def(ctx, 0x403404, 0x00000001); + } + + /* 5000 */ + cp_ctx(ctx, 0x405000, 0x1); + switch (dev_priv->chipset) { + case 0x50: + gr_def(ctx, 0x405000, 0x00300080); + break; + case 0x84: + case 0xa0: + case 0xa5: + case 0xa8: + case 0xaa: + case 0xac: + gr_def(ctx, 0x405000, 0x000e0080); + break; + case 0x86: + case 0x92: + case 0x94: + case 0x96: + case 0x98: + gr_def(ctx, 0x405000, 0x00000080); + break; + } + cp_ctx(ctx, 0x405014, 0x1); + gr_def(ctx, 0x405014, 0x00000004); + cp_ctx(ctx, 0x40501c, 0x1); + cp_ctx(ctx, 0x405024, 0x1); + cp_ctx(ctx, 0x40502c, 0x1); + + /* 5400 or maybe 4800 */ + if (dev_priv->chipset == 0x50) { + offset = 0x405400; + cp_ctx(ctx, 0x405400, 0xea); + } else if (dev_priv->chipset < 0x94) { + offset = 0x405400; + cp_ctx(ctx, 0x405400, 0xcb); + } else if (dev_priv->chipset < 0xa0) { + offset = 0x405400; + cp_ctx(ctx, 0x405400, 0xcc); + } else if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + offset = 0x404800; + cp_ctx(ctx, 0x404800, 0xda); + } else { + offset = 0x405400; + cp_ctx(ctx, 0x405400, 0xd4); + } + gr_def(ctx, offset + 0x0c, 0x00000002); + gr_def(ctx, offset + 0x10, 0x00000001); + if (dev_priv->chipset >= 0x94) + offset += 4; + gr_def(ctx, offset + 0x1c, 0x00000001); + gr_def(ctx, offset + 0x20, 0x00000100); + gr_def(ctx, offset + 0x38, 0x00000002); + gr_def(ctx, offset + 0x3c, 0x00000001); + gr_def(ctx, offset + 0x40, 0x00000001); + gr_def(ctx, offset + 0x50, 0x00000001); + gr_def(ctx, offset + 0x54, 0x003fffff); + gr_def(ctx, offset + 0x58, 0x00001fff); + gr_def(ctx, offset + 0x60, 0x00000001); + gr_def(ctx, offset + 0x64, 0x00000001); + gr_def(ctx, offset + 0x6c, 0x00000001); + gr_def(ctx, offset + 0x70, 0x00000001); + gr_def(ctx, offset + 0x74, 0x00000001); + gr_def(ctx, offset + 0x78, 0x00000004); + gr_def(ctx, offset + 0x7c, 0x00000001); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + offset += 4; + gr_def(ctx, offset + 0x80, 0x00000001); + gr_def(ctx, offset + 0x84, 0x00000001); + gr_def(ctx, offset + 0x88, 0x00000007); + gr_def(ctx, offset + 0x8c, 0x00000001); + gr_def(ctx, offset + 0x90, 0x00000007); + gr_def(ctx, offset + 0x94, 0x00000001); + gr_def(ctx, offset + 0x98, 0x00000001); + gr_def(ctx, offset + 0x9c, 0x00000001); + if (dev_priv->chipset == 0x50) { + gr_def(ctx, offset + 0xb0, 0x00000001); + gr_def(ctx, offset + 0xb4, 0x00000001); + gr_def(ctx, offset + 0xbc, 0x00000001); + gr_def(ctx, offset + 0xc0, 0x0000000a); + gr_def(ctx, offset + 0xd0, 0x00000040); + gr_def(ctx, offset + 0xd8, 0x00000002); + gr_def(ctx, offset + 0xdc, 0x00000100); + gr_def(ctx, offset + 0xe0, 0x00000001); + gr_def(ctx, offset + 0xe4, 0x00000100); + gr_def(ctx, offset + 0x100, 0x00000001); + gr_def(ctx, offset + 0x124, 0x00000004); + gr_def(ctx, offset + 0x13c, 0x00000001); + gr_def(ctx, offset + 0x140, 0x00000100); + gr_def(ctx, offset + 0x148, 0x00000001); + gr_def(ctx, offset + 0x154, 0x00000100); + gr_def(ctx, offset + 0x158, 0x00000001); + gr_def(ctx, offset + 0x15c, 0x00000100); + gr_def(ctx, offset + 0x164, 0x00000001); + gr_def(ctx, offset + 0x170, 0x00000100); + gr_def(ctx, offset + 0x174, 0x00000001); + gr_def(ctx, offset + 0x17c, 0x00000001); + gr_def(ctx, offset + 0x188, 0x00000002); + gr_def(ctx, offset + 0x190, 0x00000001); + gr_def(ctx, offset + 0x198, 0x00000001); + gr_def(ctx, offset + 0x1ac, 0x00000003); + offset += 0xd0; + } else { + gr_def(ctx, offset + 0xb0, 0x00000001); + gr_def(ctx, offset + 0xb4, 0x00000100); + gr_def(ctx, offset + 0xbc, 0x00000001); + gr_def(ctx, offset + 0xc8, 0x00000100); + gr_def(ctx, offset + 0xcc, 0x00000001); + gr_def(ctx, offset + 0xd0, 0x00000100); + gr_def(ctx, offset + 0xd8, 0x00000001); + gr_def(ctx, offset + 0xe4, 0x00000100); + } + gr_def(ctx, offset + 0xf8, 0x00000004); + gr_def(ctx, offset + 0xfc, 0x00000070); + gr_def(ctx, offset + 0x100, 0x00000080); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + offset += 4; + gr_def(ctx, offset + 0x114, 0x0000000c); + if (dev_priv->chipset == 0x50) + offset -= 4; + gr_def(ctx, offset + 0x11c, 0x00000008); + gr_def(ctx, offset + 0x120, 0x00000014); + if (dev_priv->chipset == 0x50) { + gr_def(ctx, offset + 0x124, 0x00000026); + offset -= 0x18; + } else { + gr_def(ctx, offset + 0x128, 0x00000029); + gr_def(ctx, offset + 0x12c, 0x00000027); + gr_def(ctx, offset + 0x130, 0x00000026); + gr_def(ctx, offset + 0x134, 0x00000008); + gr_def(ctx, offset + 0x138, 0x00000004); + gr_def(ctx, offset + 0x13c, 0x00000027); + } + gr_def(ctx, offset + 0x148, 0x00000001); + gr_def(ctx, offset + 0x14c, 0x00000002); + gr_def(ctx, offset + 0x150, 0x00000003); + gr_def(ctx, offset + 0x154, 0x00000004); + gr_def(ctx, offset + 0x158, 0x00000005); + gr_def(ctx, offset + 0x15c, 0x00000006); + gr_def(ctx, offset + 0x160, 0x00000007); + gr_def(ctx, offset + 0x164, 0x00000001); + gr_def(ctx, offset + 0x1a8, 0x000000cf); + if (dev_priv->chipset == 0x50) + offset -= 4; + gr_def(ctx, offset + 0x1d8, 0x00000080); + gr_def(ctx, offset + 0x1dc, 0x00000004); + gr_def(ctx, offset + 0x1e0, 0x00000004); + if (dev_priv->chipset == 0x50) + offset -= 4; + else + gr_def(ctx, offset + 0x1e4, 0x00000003); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + gr_def(ctx, offset + 0x1ec, 0x00000003); + offset += 8; + } + gr_def(ctx, offset + 0x1e8, 0x00000001); + if (dev_priv->chipset == 0x50) + offset -= 4; + gr_def(ctx, offset + 0x1f4, 0x00000012); + gr_def(ctx, offset + 0x1f8, 0x00000010); + gr_def(ctx, offset + 0x1fc, 0x0000000c); + gr_def(ctx, offset + 0x200, 0x00000001); + gr_def(ctx, offset + 0x210, 0x00000004); + gr_def(ctx, offset + 0x214, 0x00000002); + gr_def(ctx, offset + 0x218, 0x00000004); + if (dev_priv->chipset >= 0xa0) + offset += 4; + gr_def(ctx, offset + 0x224, 0x003fffff); + gr_def(ctx, offset + 0x228, 0x00001fff); + if (dev_priv->chipset == 0x50) + offset -= 0x20; + else if (dev_priv->chipset >= 0xa0) { + gr_def(ctx, offset + 0x250, 0x00000001); + gr_def(ctx, offset + 0x254, 0x00000001); + gr_def(ctx, offset + 0x258, 0x00000002); + offset += 0x10; + } + gr_def(ctx, offset + 0x250, 0x00000004); + gr_def(ctx, offset + 0x254, 0x00000014); + gr_def(ctx, offset + 0x258, 0x00000001); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + offset += 4; + gr_def(ctx, offset + 0x264, 0x00000002); + if (dev_priv->chipset >= 0xa0) + offset += 8; + gr_def(ctx, offset + 0x270, 0x00000001); + gr_def(ctx, offset + 0x278, 0x00000002); + gr_def(ctx, offset + 0x27c, 0x00001000); + if (dev_priv->chipset == 0x50) + offset -= 0xc; + else { + gr_def(ctx, offset + 0x280, 0x00000e00); + gr_def(ctx, offset + 0x284, 0x00001000); + gr_def(ctx, offset + 0x288, 0x00001e00); + } + gr_def(ctx, offset + 0x290, 0x00000001); + gr_def(ctx, offset + 0x294, 0x00000001); + gr_def(ctx, offset + 0x298, 0x00000001); + gr_def(ctx, offset + 0x29c, 0x00000001); + gr_def(ctx, offset + 0x2a0, 0x00000001); + gr_def(ctx, offset + 0x2b0, 0x00000200); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + gr_def(ctx, offset + 0x2b4, 0x00000200); + offset += 4; + } + if (dev_priv->chipset < 0xa0) { + gr_def(ctx, offset + 0x2b8, 0x00000001); + gr_def(ctx, offset + 0x2bc, 0x00000070); + gr_def(ctx, offset + 0x2c0, 0x00000080); + gr_def(ctx, offset + 0x2cc, 0x00000001); + gr_def(ctx, offset + 0x2d0, 0x00000070); + gr_def(ctx, offset + 0x2d4, 0x00000080); + } else { + gr_def(ctx, offset + 0x2b8, 0x00000001); + gr_def(ctx, offset + 0x2bc, 0x000000f0); + gr_def(ctx, offset + 0x2c0, 0x000000ff); + gr_def(ctx, offset + 0x2cc, 0x00000001); + gr_def(ctx, offset + 0x2d0, 0x000000f0); + gr_def(ctx, offset + 0x2d4, 0x000000ff); + gr_def(ctx, offset + 0x2dc, 0x00000009); + offset += 4; + } + gr_def(ctx, offset + 0x2e4, 0x00000001); + gr_def(ctx, offset + 0x2e8, 0x000000cf); + gr_def(ctx, offset + 0x2f0, 0x00000001); + gr_def(ctx, offset + 0x300, 0x000000cf); + gr_def(ctx, offset + 0x308, 0x00000002); + gr_def(ctx, offset + 0x310, 0x00000001); + gr_def(ctx, offset + 0x318, 0x00000001); + gr_def(ctx, offset + 0x320, 0x000000cf); + gr_def(ctx, offset + 0x324, 0x000000cf); + gr_def(ctx, offset + 0x328, 0x00000001); + + /* 6000? */ + if (dev_priv->chipset == 0x50) + cp_ctx(ctx, 0x4063e0, 0x1); + + /* 6800 */ + if (dev_priv->chipset < 0x90) { + cp_ctx(ctx, 0x406814, 0x2b); + gr_def(ctx, 0x406818, 0x00000f80); + gr_def(ctx, 0x406860, 0x007f0080); + gr_def(ctx, 0x40689c, 0x007f0080); + } else { + cp_ctx(ctx, 0x406814, 0x4); + if (dev_priv->chipset == 0x98) + gr_def(ctx, 0x406818, 0x00000f80); + else + gr_def(ctx, 0x406818, 0x00001f80); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + gr_def(ctx, 0x40681c, 0x00000030); + cp_ctx(ctx, 0x406830, 0x3); + } + + /* 7000: per-ROP group state */ + for (i = 0; i < 8; i++) { + if (units & (1<<(i+16))) { + cp_ctx(ctx, 0x407000 + (i<<8), 3); + if (dev_priv->chipset == 0x50) + gr_def(ctx, 0x407000 + (i<<8), 0x1b74f820); + else if (dev_priv->chipset != 0xa5) + gr_def(ctx, 0x407000 + (i<<8), 0x3b74f821); + else + gr_def(ctx, 0x407000 + (i<<8), 0x7b74f821); + gr_def(ctx, 0x407004 + (i<<8), 0x89058001); + + if (dev_priv->chipset == 0x50) { + cp_ctx(ctx, 0x407010 + (i<<8), 1); + } else if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, 0x407010 + (i<<8), 2); + gr_def(ctx, 0x407010 + (i<<8), 0x00001000); + gr_def(ctx, 0x407014 + (i<<8), 0x0000001f); + } else { + cp_ctx(ctx, 0x407010 + (i<<8), 3); + gr_def(ctx, 0x407010 + (i<<8), 0x00001000); + if (dev_priv->chipset != 0xa5) + gr_def(ctx, 0x407014 + (i<<8), 0x000000ff); + else + gr_def(ctx, 0x407014 + (i<<8), 0x000001ff); + } + + cp_ctx(ctx, 0x407080 + (i<<8), 4); + if (dev_priv->chipset != 0xa5) + gr_def(ctx, 0x407080 + (i<<8), 0x027c10fa); + else + gr_def(ctx, 0x407080 + (i<<8), 0x827c10fa); + if (dev_priv->chipset == 0x50) + gr_def(ctx, 0x407084 + (i<<8), 0x000000c0); + else + gr_def(ctx, 0x407084 + (i<<8), 0x400000c0); + gr_def(ctx, 0x407088 + (i<<8), 0xb7892080); + + if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, 0x407094 + (i<<8), 1); + else if (dev_priv->chipset <= 0xa0 || dev_priv->chipset >= 0xaa) + cp_ctx(ctx, 0x407094 + (i<<8), 3); + else { + cp_ctx(ctx, 0x407094 + (i<<8), 4); + gr_def(ctx, 0x4070a0 + (i<<8), 1); + } + } + } + + cp_ctx(ctx, 0x407c00, 0x3); + if (dev_priv->chipset < 0x90) + gr_def(ctx, 0x407c00, 0x00010040); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, 0x407c00, 0x00390040); + else + gr_def(ctx, 0x407c00, 0x003d0040); + gr_def(ctx, 0x407c08, 0x00000022); + if (dev_priv->chipset >= 0xa0) { + cp_ctx(ctx, 0x407c10, 0x3); + cp_ctx(ctx, 0x407c20, 0x1); + cp_ctx(ctx, 0x407c2c, 0x1); + } + + if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, 0x407d00, 0x9); + } else { + cp_ctx(ctx, 0x407d00, 0x15); + } + if (dev_priv->chipset == 0x98) + gr_def(ctx, 0x407d08, 0x00380040); + else { + if (dev_priv->chipset < 0x90) + gr_def(ctx, 0x407d08, 0x00010040); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, 0x407d08, 0x00390040); + else + gr_def(ctx, 0x407d08, 0x003d0040); + gr_def(ctx, 0x407d0c, 0x00000022); + } + + /* 8000+: per-TP state */ + for (i = 0; i < 10; i++) { + if (units & (1<<i)) { + if (dev_priv->chipset < 0xa0) + base = 0x408000 + (i<<12); + else + base = 0x408000 + (i<<11); + if (dev_priv->chipset < 0xa0) + offset = base + 0xc00; + else + offset = base + 0x80; + cp_ctx(ctx, offset + 0x00, 1); + gr_def(ctx, offset + 0x00, 0x0000ff0a); + cp_ctx(ctx, offset + 0x08, 1); + + /* per-MP state */ + for (j = 0; j < (dev_priv->chipset < 0xa0 ? 2 : 4); j++) { + if (!(units & (1 << (j+24)))) continue; + if (dev_priv->chipset < 0xa0) + offset = base + 0x200 + (j<<7); + else + offset = base + 0x100 + (j<<7); + cp_ctx(ctx, offset, 0x20); + gr_def(ctx, offset + 0x00, 0x01800000); + gr_def(ctx, offset + 0x04, 0x00160000); + gr_def(ctx, offset + 0x08, 0x01800000); + gr_def(ctx, offset + 0x18, 0x0003ffff); + switch (dev_priv->chipset) { + case 0x50: + gr_def(ctx, offset + 0x1c, 0x00080000); + break; + case 0x84: + gr_def(ctx, offset + 0x1c, 0x00880000); + break; + case 0x86: + gr_def(ctx, offset + 0x1c, 0x008c0000); + break; + case 0x92: + case 0x96: + case 0x98: + gr_def(ctx, offset + 0x1c, 0x118c0000); + break; + case 0x94: + gr_def(ctx, offset + 0x1c, 0x10880000); + break; + case 0xa0: + case 0xa5: + gr_def(ctx, offset + 0x1c, 0x310c0000); + break; + case 0xa8: + case 0xaa: + case 0xac: + gr_def(ctx, offset + 0x1c, 0x300c0000); + break; + } + gr_def(ctx, offset + 0x40, 0x00010401); + if (dev_priv->chipset == 0x50) + gr_def(ctx, offset + 0x48, 0x00000040); + else + gr_def(ctx, offset + 0x48, 0x00000078); + gr_def(ctx, offset + 0x50, 0x000000bf); + gr_def(ctx, offset + 0x58, 0x00001210); + if (dev_priv->chipset == 0x50) + gr_def(ctx, offset + 0x5c, 0x00000080); + else + gr_def(ctx, offset + 0x5c, 0x08000080); + if (dev_priv->chipset >= 0xa0) + gr_def(ctx, offset + 0x68, 0x0000003e); + } + + if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, base + 0x300, 0x4); + else + cp_ctx(ctx, base + 0x300, 0x5); + if (dev_priv->chipset == 0x50) + gr_def(ctx, base + 0x304, 0x00007070); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, base + 0x304, 0x00027070); + else if (dev_priv->chipset <= 0xa0 || dev_priv->chipset >= 0xaa) + gr_def(ctx, base + 0x304, 0x01127070); + else + gr_def(ctx, base + 0x304, 0x05127070); + + if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, base + 0x318, 1); + else + cp_ctx(ctx, base + 0x320, 1); + if (dev_priv->chipset == 0x50) + gr_def(ctx, base + 0x318, 0x0003ffff); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, base + 0x318, 0x03ffffff); + else + gr_def(ctx, base + 0x320, 0x07ffffff); + + if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, base + 0x324, 5); + else + cp_ctx(ctx, base + 0x328, 4); + + if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, base + 0x340, 9); + offset = base + 0x340; + } else if (dev_priv->chipset <= 0xa0 || dev_priv->chipset >= 0xaa) { + cp_ctx(ctx, base + 0x33c, 0xb); + offset = base + 0x344; + } else { + cp_ctx(ctx, base + 0x33c, 0xd); + offset = base + 0x344; + } + gr_def(ctx, offset + 0x0, 0x00120407); + gr_def(ctx, offset + 0x4, 0x05091507); + if (dev_priv->chipset == 0x84) + gr_def(ctx, offset + 0x8, 0x05100202); + else + gr_def(ctx, offset + 0x8, 0x05010202); + gr_def(ctx, offset + 0xc, 0x00030201); + + cp_ctx(ctx, base + 0x400, 2); + gr_def(ctx, base + 0x404, 0x00000040); + cp_ctx(ctx, base + 0x40c, 2); + gr_def(ctx, base + 0x40c, 0x0d0c0b0a); + gr_def(ctx, base + 0x410, 0x00141210); + + if (dev_priv->chipset < 0xa0) + offset = base + 0x800; + else + offset = base + 0x500; + cp_ctx(ctx, offset, 6); + gr_def(ctx, offset + 0x0, 0x000001f0); + gr_def(ctx, offset + 0x4, 0x00000001); + gr_def(ctx, offset + 0x8, 0x00000003); + if (dev_priv->chipset == 0x50 || dev_priv->chipset >= 0xaa) + gr_def(ctx, offset + 0xc, 0x00008000); + gr_def(ctx, offset + 0x14, 0x00039e00); + cp_ctx(ctx, offset + 0x1c, 2); + if (dev_priv->chipset == 0x50) + gr_def(ctx, offset + 0x1c, 0x00000040); + else + gr_def(ctx, offset + 0x1c, 0x00000100); + gr_def(ctx, offset + 0x20, 0x00003800); + + if (dev_priv->chipset >= 0xa0) { + cp_ctx(ctx, base + 0x54c, 2); + if (dev_priv->chipset <= 0xa0 || dev_priv->chipset >= 0xaa) + gr_def(ctx, base + 0x54c, 0x003fe006); + else + gr_def(ctx, base + 0x54c, 0x003fe007); + gr_def(ctx, base + 0x550, 0x003fe000); + } + + if (dev_priv->chipset < 0xa0) + offset = base + 0xa00; + else + offset = base + 0x680; + cp_ctx(ctx, offset, 1); + gr_def(ctx, offset, 0x00404040); + + if (dev_priv->chipset < 0xa0) + offset = base + 0xe00; + else + offset = base + 0x700; + cp_ctx(ctx, offset, 2); + if (dev_priv->chipset < 0xa0) + gr_def(ctx, offset, 0x0077f005); + else if (dev_priv->chipset == 0xa5) + gr_def(ctx, offset, 0x6cf7f007); + else if (dev_priv->chipset == 0xa8) + gr_def(ctx, offset, 0x6cfff007); + else if (dev_priv->chipset == 0xac) + gr_def(ctx, offset, 0x0cfff007); + else + gr_def(ctx, offset, 0x0cf7f007); + if (dev_priv->chipset == 0x50) + gr_def(ctx, offset + 0x4, 0x00007fff); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, offset + 0x4, 0x003f7fff); + else + gr_def(ctx, offset + 0x4, 0x02bf7fff); + cp_ctx(ctx, offset + 0x2c, 1); + if (dev_priv->chipset == 0x50) { + cp_ctx(ctx, offset + 0x50, 9); + gr_def(ctx, offset + 0x54, 0x000003ff); + gr_def(ctx, offset + 0x58, 0x00000003); + gr_def(ctx, offset + 0x5c, 0x00000003); + gr_def(ctx, offset + 0x60, 0x000001ff); + gr_def(ctx, offset + 0x64, 0x0000001f); + gr_def(ctx, offset + 0x68, 0x0000000f); + gr_def(ctx, offset + 0x6c, 0x0000000f); + } else if(dev_priv->chipset < 0xa0) { + cp_ctx(ctx, offset + 0x50, 1); + cp_ctx(ctx, offset + 0x70, 1); + } else { + cp_ctx(ctx, offset + 0x50, 1); + cp_ctx(ctx, offset + 0x60, 5); + } + } + } +} + +/* + * xfer areas. These are a pain. + * + * There are 2 xfer areas: the first one is big and contains all sorts of + * stuff, the second is small and contains some per-TP context. + * + * Each area is split into 8 "strands". The areas, when saved to grctx, + * are made of 8-word blocks. Each block contains a single word from + * each strand. The strands are independent of each other, their + * addresses are unrelated to each other, and data in them is closely + * packed together. The strand layout varies a bit between cards: here + * and there, a single word is thrown out in the middle and the whole + * strand is offset by a bit from corresponding one on another chipset. + * For this reason, addresses of stuff in strands are almost useless. + * Knowing sequence of stuff and size of gaps between them is much more + * useful, and that's how we build the strands in our generator. + * + * NVA0 takes this mess to a whole new level by cutting the old strands + * into a few dozen pieces [known as genes], rearranging them randomly, + * and putting them back together to make new strands. Hopefully these + * genes correspond more or less directly to the same PGRAPH subunits + * as in 400040 register. + * + * The most common value in default context is 0, and when the genes + * are separated by 0's, gene bounduaries are quite speculative... + * some of them can be clearly deduced, others can be guessed, and yet + * others won't be resolved without figuring out the real meaning of + * given ctxval. For the same reason, ending point of each strand + * is unknown. Except for strand 0, which is the longest strand and + * its end corresponds to end of the whole xfer. + * + * An unsolved mystery is the seek instruction: it takes an argument + * in bits 8-18, and that argument is clearly the place in strands to + * seek to... but the offsets don't seem to correspond to offsets as + * seen in grctx. Perhaps there's another, real, not randomly-changing + * addressing in strands, and the xfer insn just happens to skip over + * the unused bits? NV10-NV30 PIPE comes to mind... + * + * As far as I know, there's no way to access the xfer areas directly + * without the help of ctxprog. + */ + +static inline void +xf_emit(struct nouveau_grctx *ctx, int num, uint32_t val) { + int i; + if (val && ctx->mode == NOUVEAU_GRCTX_VALS) + for (i = 0; i < num; i++) + nv_wo32(ctx->dev, ctx->data, ctx->ctxvals_pos + (i << 3), val); + ctx->ctxvals_pos += num << 3; +} + +/* Gene declarations... */ + +static void nv50_graph_construct_gene_m2mf(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk1(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk2(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk3(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk4(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk5(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk6(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk7(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk8(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk9(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk10(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_ropc(struct nouveau_grctx *ctx); +static void nv50_graph_construct_xfer_tp(struct nouveau_grctx *ctx); + +static void +nv50_graph_construct_xfer1(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + int offset; + int size = 0; + uint32_t units = nv_rd32 (ctx->dev, 0x1540); + + offset = (ctx->ctxvals_pos+0x3f)&~0x3f; + ctx->ctxvals_base = offset; + + if (dev_priv->chipset < 0xa0) { + /* Strand 0 */ + ctx->ctxvals_pos = offset; + switch (dev_priv->chipset) { + case 0x50: + xf_emit(ctx, 0x99, 0); + break; + case 0x84: + case 0x86: + xf_emit(ctx, 0x384, 0); + break; + case 0x92: + case 0x94: + case 0x96: + case 0x98: + xf_emit(ctx, 0x380, 0); + break; + } + nv50_graph_construct_gene_m2mf (ctx); + switch (dev_priv->chipset) { + case 0x50: + case 0x84: + case 0x86: + case 0x98: + xf_emit(ctx, 0x4c4, 0); + break; + case 0x92: + case 0x94: + case 0x96: + xf_emit(ctx, 0x984, 0); + break; + } + nv50_graph_construct_gene_unk5(ctx); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 0xa, 0); + else + xf_emit(ctx, 0xb, 0); + nv50_graph_construct_gene_unk4(ctx); + nv50_graph_construct_gene_unk3(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 1 */ + ctx->ctxvals_pos = offset + 0x1; + nv50_graph_construct_gene_unk6(ctx); + nv50_graph_construct_gene_unk7(ctx); + nv50_graph_construct_gene_unk8(ctx); + switch (dev_priv->chipset) { + case 0x50: + case 0x92: + xf_emit(ctx, 0xfb, 0); + break; + case 0x84: + xf_emit(ctx, 0xd3, 0); + break; + case 0x94: + case 0x96: + xf_emit(ctx, 0xab, 0); + break; + case 0x86: + case 0x98: + xf_emit(ctx, 0x6b, 0); + break; + } + xf_emit(ctx, 2, 0x4e3bfdf); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 0xb, 0); + xf_emit(ctx, 2, 0x4e3bfdf); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 2 */ + ctx->ctxvals_pos = offset + 0x2; + switch (dev_priv->chipset) { + case 0x50: + case 0x92: + xf_emit(ctx, 0xa80, 0); + break; + case 0x84: + xf_emit(ctx, 0xa7e, 0); + break; + case 0x94: + case 0x96: + xf_emit(ctx, 0xa7c, 0); + break; + case 0x86: + case 0x98: + xf_emit(ctx, 0xa7a, 0); + break; + } + xf_emit(ctx, 1, 0x3fffff); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x1fff); + xf_emit(ctx, 0xe, 0); + nv50_graph_construct_gene_unk9(ctx); + nv50_graph_construct_gene_unk2(ctx); + nv50_graph_construct_gene_unk1(ctx); + nv50_graph_construct_gene_unk10(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 3: per-ROP group state */ + ctx->ctxvals_pos = offset + 3; + for (i = 0; i < 6; i++) + if (units & (1 << (i + 16))) + nv50_graph_construct_gene_ropc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strands 4-7: per-TP state */ + for (i = 0; i < 4; i++) { + ctx->ctxvals_pos = offset + 4 + i; + if (units & (1 << (2 * i))) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << (2 * i + 1))) + nv50_graph_construct_xfer_tp(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + } + } else { + /* Strand 0 */ + ctx->ctxvals_pos = offset; + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0x385, 0); + else + xf_emit(ctx, 0x384, 0); + nv50_graph_construct_gene_m2mf(ctx); + xf_emit(ctx, 0x950, 0); + nv50_graph_construct_gene_unk10(ctx); + xf_emit(ctx, 1, 0x0fac6881); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + xf_emit(ctx, 1, 1); + xf_emit(ctx, 3, 0); + } + nv50_graph_construct_gene_unk8(ctx); + if (dev_priv->chipset == 0xa0) + xf_emit(ctx, 0x189, 0); + else if (dev_priv->chipset < 0xa8) + xf_emit(ctx, 0x99, 0); + else if (dev_priv->chipset == 0xaa) + xf_emit(ctx, 0x65, 0); + else + xf_emit(ctx, 0x6d, 0); + nv50_graph_construct_gene_unk9(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 1 */ + ctx->ctxvals_pos = offset + 1; + nv50_graph_construct_gene_unk1(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 2 */ + ctx->ctxvals_pos = offset + 2; + if (dev_priv->chipset == 0xa0) { + nv50_graph_construct_gene_unk2(ctx); + } + xf_emit(ctx, 0x36, 0); + nv50_graph_construct_gene_unk5(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 3 */ + ctx->ctxvals_pos = offset + 3; + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + nv50_graph_construct_gene_unk6(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 4 */ + ctx->ctxvals_pos = offset + 4; + if (dev_priv->chipset == 0xa0) + xf_emit(ctx, 0xa80, 0); + else + xf_emit(ctx, 0xa7a, 0); + xf_emit(ctx, 1, 0x3fffff); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x1fff); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 5 */ + ctx->ctxvals_pos = offset + 5; + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 0xb, 0); + xf_emit(ctx, 2, 0x4e3bfdf); + xf_emit(ctx, 3, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 2, 0x4e3bfdf); + xf_emit(ctx, 2, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 0); + for (i = 0; i < 8; i++) + if (units & (1<<(i+16))) + nv50_graph_construct_gene_ropc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 6 */ + ctx->ctxvals_pos = offset + 6; + nv50_graph_construct_gene_unk3(ctx); + xf_emit(ctx, 0xb, 0); + nv50_graph_construct_gene_unk4(ctx); + nv50_graph_construct_gene_unk7(ctx); + if (units & (1 << 0)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 1)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 2)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 3)) + nv50_graph_construct_xfer_tp(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 7 */ + ctx->ctxvals_pos = offset + 7; + if (dev_priv->chipset == 0xa0) { + if (units & (1 << 4)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 5)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 6)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 7)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 8)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 9)) + nv50_graph_construct_xfer_tp(ctx); + } else { + nv50_graph_construct_gene_unk2(ctx); + } + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + } + + ctx->ctxvals_pos = offset + size * 8; + ctx->ctxvals_pos = (ctx->ctxvals_pos+0x3f)&~0x3f; + cp_lsr (ctx, offset); + cp_out (ctx, CP_SET_XFER_POINTER); + cp_lsr (ctx, size); + cp_out (ctx, CP_SEEK_1); + cp_out (ctx, CP_XFER_1); + cp_wait(ctx, XFER, BUSY); +} + +/* + * non-trivial demagiced parts of ctx init go here + */ + +static void +nv50_graph_construct_gene_m2mf(struct nouveau_grctx *ctx) +{ + /* m2mf state */ + xf_emit (ctx, 1, 0); /* DMA_NOTIFY instance >> 4 */ + xf_emit (ctx, 1, 0); /* DMA_BUFFER_IN instance >> 4 */ + xf_emit (ctx, 1, 0); /* DMA_BUFFER_OUT instance >> 4 */ + xf_emit (ctx, 1, 0); /* OFFSET_IN */ + xf_emit (ctx, 1, 0); /* OFFSET_OUT */ + xf_emit (ctx, 1, 0); /* PITCH_IN */ + xf_emit (ctx, 1, 0); /* PITCH_OUT */ + xf_emit (ctx, 1, 0); /* LINE_LENGTH */ + xf_emit (ctx, 1, 0); /* LINE_COUNT */ + xf_emit (ctx, 1, 0x21); /* FORMAT: bits 0-4 INPUT_INC, bits 5-9 OUTPUT_INC */ + xf_emit (ctx, 1, 1); /* LINEAR_IN */ + xf_emit (ctx, 1, 0x2); /* TILING_MODE_IN: bits 0-2 y tiling, bits 3-5 z tiling */ + xf_emit (ctx, 1, 0x100); /* TILING_PITCH_IN */ + xf_emit (ctx, 1, 0x100); /* TILING_HEIGHT_IN */ + xf_emit (ctx, 1, 1); /* TILING_DEPTH_IN */ + xf_emit (ctx, 1, 0); /* TILING_POSITION_IN_Z */ + xf_emit (ctx, 1, 0); /* TILING_POSITION_IN */ + xf_emit (ctx, 1, 1); /* LINEAR_OUT */ + xf_emit (ctx, 1, 0x2); /* TILING_MODE_OUT: bits 0-2 y tiling, bits 3-5 z tiling */ + xf_emit (ctx, 1, 0x100); /* TILING_PITCH_OUT */ + xf_emit (ctx, 1, 0x100); /* TILING_HEIGHT_OUT */ + xf_emit (ctx, 1, 1); /* TILING_DEPTH_OUT */ + xf_emit (ctx, 1, 0); /* TILING_POSITION_OUT_Z */ + xf_emit (ctx, 1, 0); /* TILING_POSITION_OUT */ + xf_emit (ctx, 1, 0); /* OFFSET_IN_HIGH */ + xf_emit (ctx, 1, 0); /* OFFSET_OUT_HIGH */ +} + +static void +nv50_graph_construct_gene_unk1(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* end of area 2 on pre-NVA0, area 1 on NVAx */ + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x80); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0x80c14); + xf_emit(ctx, 1, 0); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0x3ff); + else + xf_emit(ctx, 1, 0x7ff); + switch (dev_priv->chipset) { + case 0x50: + case 0x86: + case 0x98: + case 0xaa: + case 0xac: + xf_emit(ctx, 0x542, 0); + break; + case 0x84: + case 0x92: + case 0x94: + case 0x96: + xf_emit(ctx, 0x942, 0); + break; + case 0xa0: + xf_emit(ctx, 0x2042, 0); + break; + case 0xa5: + case 0xa8: + xf_emit(ctx, 0x842, 0); + break; + } + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x80); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x27); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x26); + xf_emit(ctx, 3, 0); +} + +static void +nv50_graph_construct_gene_unk10(struct nouveau_grctx *ctx) +{ + /* end of area 2 on pre-NVA0, area 1 on NVAx */ + xf_emit(ctx, 0x10, 0x04000000); + xf_emit(ctx, 0x24, 0); + xf_emit(ctx, 2, 0x04e3bfdf); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x1fe21); +} + +static void +nv50_graph_construct_gene_unk2(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* middle of area 2 on pre-NVA0, beginning of area 2 on NVA0, area 7 on >NVA0 */ + if (dev_priv->chipset != 0x50) { + xf_emit(ctx, 5, 0); + xf_emit(ctx, 1, 0x80c14); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x804); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0x8100c12); + } + xf_emit(ctx, 1, 0); + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x10); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 3, 0); + else + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 0x804); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x1a); + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 0x7f); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x80c14); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x8100c12); + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x10); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x8100c12); + xf_emit(ctx, 6, 0); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0x3ff); + else + xf_emit(ctx, 1, 0x7ff); + xf_emit(ctx, 1, 0x80c14); + xf_emit(ctx, 0x38, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x10); + xf_emit(ctx, 0x38, 0); + xf_emit(ctx, 2, 0x88); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 0x16, 0); + xf_emit(ctx, 1, 0x26); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x3f800000); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 4, 0); + else + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x1a); + xf_emit(ctx, 1, 0x10); + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 0x28, 0); + else + xf_emit(ctx, 0x25, 0); + xf_emit(ctx, 1, 0x52); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x26); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x1a); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x00ffff00); + xf_emit(ctx, 1, 0); +} + +static void +nv50_graph_construct_gene_unk3(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* end of area 0 on pre-NVA0, beginning of area 6 on NVAx */ + xf_emit(ctx, 1, 0x3f); + xf_emit(ctx, 0xa, 0); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 2, 0x04000000); + xf_emit(ctx, 8, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 4); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 0x10, 0); + else + xf_emit(ctx, 0x11, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x1001); + xf_emit(ctx, 4, 0xffff); + xf_emit(ctx, 0x20, 0); + xf_emit(ctx, 0x10, 0x3f800000); + xf_emit(ctx, 1, 0x10); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0); + else + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 3); + xf_emit(ctx, 2, 0); +} + +static void +nv50_graph_construct_gene_unk4(struct nouveau_grctx *ctx) +{ + /* middle of area 0 on pre-NVA0, middle of area 6 on NVAx */ + xf_emit(ctx, 2, 0x04000000); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x80); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x80); + xf_emit(ctx, 1, 0); +} + +static void +nv50_graph_construct_gene_unk5(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* middle of area 0 on pre-NVA0 [after m2mf], end of area 2 on NVAx */ + xf_emit(ctx, 2, 4); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0x1c4d, 0); + else + xf_emit(ctx, 0x1c4b, 0); + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0x8100c12); + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 3); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x8100c12); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x80c14); + xf_emit(ctx, 1, 1); + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0x80c14); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x8100c12); + xf_emit(ctx, 1, 0x27); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0x3c1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0x16, 0); + xf_emit(ctx, 1, 0x8100c12); + xf_emit(ctx, 1, 0); +} + +static void +nv50_graph_construct_gene_unk6(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* beginning of area 1 on pre-NVA0 [after m2mf], area 3 on NVAx */ + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 0xf); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 8, 0); + else + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 0x20); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0x11, 0); + else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 0xf, 0); + else + xf_emit(ctx, 0xe, 0); + xf_emit(ctx, 1, 0x1a); + xf_emit(ctx, 0xd, 0); + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 8); + xf_emit(ctx, 1, 0); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0x3ff); + else + xf_emit(ctx, 1, 0x7ff); + if (dev_priv->chipset == 0xa8) + xf_emit(ctx, 1, 0x1e00); + xf_emit(ctx, 0xc, 0); + xf_emit(ctx, 1, 0xf); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 0x125, 0); + else if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0x126, 0); + else if (dev_priv->chipset == 0xa0 || dev_priv->chipset >= 0xaa) + xf_emit(ctx, 0x124, 0); + else + xf_emit(ctx, 0x1f7, 0); + xf_emit(ctx, 1, 0xf); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 3, 0); + else + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0xa1, 0); + else + xf_emit(ctx, 0x5a, 0); + xf_emit(ctx, 1, 0xf); + if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0x834, 0); + else if (dev_priv->chipset == 0xa0) + xf_emit(ctx, 0x1873, 0); + else if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0x8ba, 0); + else + xf_emit(ctx, 0x833, 0); + xf_emit(ctx, 1, 0xf); + xf_emit(ctx, 0xf, 0); +} + +static void +nv50_graph_construct_gene_unk7(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* middle of area 1 on pre-NVA0 [after m2mf], middle of area 6 on NVAx */ + xf_emit(ctx, 2, 0); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 2, 1); + else + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0x100); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 8); + xf_emit(ctx, 5, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 3, 1); + xf_emit(ctx, 1, 0xcf); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 6, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 3, 1); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x15); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x4444480); + xf_emit(ctx, 0x37, 0); +} + +static void +nv50_graph_construct_gene_unk8(struct nouveau_grctx *ctx) +{ + /* middle of area 1 on pre-NVA0 [after m2mf], middle of area 0 on NVAx */ + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 0x8100c12); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 0x100); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x10001); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x10001); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x10001); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 2); +} + +static void +nv50_graph_construct_gene_unk9(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* middle of area 2 on pre-NVA0 [after m2mf], end of area 0 on NVAx */ + xf_emit(ctx, 1, 0x3f800000); + xf_emit(ctx, 6, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0x1a); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0x12, 0); + xf_emit(ctx, 1, 0x00ffff00); + xf_emit(ctx, 6, 0); + xf_emit(ctx, 1, 0xf); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 0xf, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 2, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 3); + else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 2, 0x04000000); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 5); + xf_emit(ctx, 1, 0x52); + if (dev_priv->chipset == 0x50) { + xf_emit(ctx, 0x13, 0); + } else { + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 1); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0x11, 0); + else + xf_emit(ctx, 0x10, 0); + } + xf_emit(ctx, 0x10, 0x3f800000); + xf_emit(ctx, 1, 0x10); + xf_emit(ctx, 0x26, 0); + xf_emit(ctx, 1, 0x8100c12); + xf_emit(ctx, 1, 5); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 4, 0xffff); + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 3); + if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0x1f, 0); + else if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0xc, 0); + else + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x00ffff00); + xf_emit(ctx, 1, 0x1a); + if (dev_priv->chipset != 0x50) { + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 3); + } + if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0x26, 0); + else + xf_emit(ctx, 0x3c, 0); + xf_emit(ctx, 1, 0x102); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 4, 4); + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 8, 0); + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0x3ff); + else + xf_emit(ctx, 1, 0x7ff); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x102); + xf_emit(ctx, 9, 0); + xf_emit(ctx, 4, 4); + xf_emit(ctx, 0x2c, 0); +} + +static void +nv50_graph_construct_gene_ropc(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int magic2; + if (dev_priv->chipset == 0x50) { + magic2 = 0x00003e60; + } else if (dev_priv->chipset <= 0xa0 || dev_priv->chipset >= 0xaa) { + magic2 = 0x001ffe67; + } else { + magic2 = 0x00087e67; + } + xf_emit(ctx, 8, 0); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, magic2); + xf_emit(ctx, 4, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 1); + xf_emit(ctx, 7, 0); + if (dev_priv->chipset >= 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 0x15); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x10); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 4, 0); + if (dev_priv->chipset == 0x86 || dev_priv->chipset == 0x92 || dev_priv->chipset == 0x98 || dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0x400); + xf_emit(ctx, 1, 0x300); + xf_emit(ctx, 1, 0x1001); + if (dev_priv->chipset != 0xa0) { + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 0); + else + xf_emit(ctx, 1, 0x15); + } + xf_emit(ctx, 3, 0); + } + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 8, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x10); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0x13, 0); + xf_emit(ctx, 1, 0x10); + xf_emit(ctx, 0x10, 0); + xf_emit(ctx, 0x10, 0x3f800000); + xf_emit(ctx, 0x19, 0); + xf_emit(ctx, 1, 0x10); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x3f); + xf_emit(ctx, 6, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x1001); + xf_emit(ctx, 0xb, 0); + } else { + xf_emit(ctx, 0xc, 0); + } + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0xf); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x11); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 4, 0); + else + xf_emit(ctx, 6, 0); + xf_emit(ctx, 3, 1); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, magic2); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x0fac6881); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + xf_emit(ctx, 1, 0); + xf_emit(ctx, 0x18, 1); + xf_emit(ctx, 8, 2); + xf_emit(ctx, 8, 1); + xf_emit(ctx, 8, 2); + xf_emit(ctx, 8, 1); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 5, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0x16, 0); + } else { + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 0x1b, 0); + else + xf_emit(ctx, 0x15, 0); + } + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 2, 1); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 2, 1); + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 4, 0); + else + xf_emit(ctx, 3, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + xf_emit(ctx, 0x10, 1); + xf_emit(ctx, 8, 2); + xf_emit(ctx, 0x10, 1); + xf_emit(ctx, 8, 2); + xf_emit(ctx, 8, 1); + xf_emit(ctx, 3, 0); + } + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0x5b, 0); +} + +static void +nv50_graph_construct_xfer_tp_x1(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int magic3; + if (dev_priv->chipset == 0x50) + magic3 = 0x1000; + else if (dev_priv->chipset == 0x86 || dev_priv->chipset == 0x98 || dev_priv->chipset >= 0xa8) + magic3 = 0x1e00; + else + magic3 = 0; + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 4); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0x24, 0); + else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 0x14, 0); + else + xf_emit(ctx, 0x15, 0); + xf_emit(ctx, 2, 4); + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0x03020100); + else + xf_emit(ctx, 1, 0x00608080); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 2, 4); + xf_emit(ctx, 1, 0x80); + if (magic3) + xf_emit(ctx, 1, magic3); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 0x24, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0x80); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0x03020100); + xf_emit(ctx, 1, 3); + if (magic3) + xf_emit(ctx, 1, magic3); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 3); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 4); + if (dev_priv->chipset == 0x94 || dev_priv->chipset == 0x96) + xf_emit(ctx, 0x1024, 0); + else if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0xa24, 0); + else if (dev_priv->chipset == 0xa0 || dev_priv->chipset >= 0xaa) + xf_emit(ctx, 0x214, 0); + else + xf_emit(ctx, 0x414, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 3); + xf_emit(ctx, 2, 0); +} + +static void +nv50_graph_construct_xfer_tp_x2(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int magic1, magic2; + if (dev_priv->chipset == 0x50) { + magic1 = 0x3ff; + magic2 = 0x00003e60; + } else if (dev_priv->chipset <= 0xa0 || dev_priv->chipset >= 0xaa) { + magic1 = 0x7ff; + magic2 = 0x001ffe67; + } else { + magic1 = 0x7ff; + magic2 = 0x00087e67; + } + xf_emit(ctx, 3, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0xc, 0); + xf_emit(ctx, 1, 0xf); + xf_emit(ctx, 0xb, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 4, 0xffff); + xf_emit(ctx, 8, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 5, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + xf_emit(ctx, 1, 3); + xf_emit(ctx, 1, 0); + } else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0xa, 0); + xf_emit(ctx, 2, 1); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 2, 1); + xf_emit(ctx, 1, 2); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + xf_emit(ctx, 1, 0); + xf_emit(ctx, 0x18, 1); + xf_emit(ctx, 8, 2); + xf_emit(ctx, 8, 1); + xf_emit(ctx, 8, 2); + xf_emit(ctx, 8, 1); + xf_emit(ctx, 1, 0); + } + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 3, 0xcf); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0xa, 0); + xf_emit(ctx, 2, 1); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 2, 1); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 8, 1); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 1, 0xf); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, magic2); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x11); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 2, 1); + else + xf_emit(ctx, 1, 1); + if(dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0); + else + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 5, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, magic1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0x28, 0); + xf_emit(ctx, 8, 8); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 8, 0x400); + xf_emit(ctx, 8, 0x300); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0xf); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x20); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 0x100); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x40); + xf_emit(ctx, 1, 0x100); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 3); + xf_emit(ctx, 4, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, magic2); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 9, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x400); + xf_emit(ctx, 1, 0x300); + xf_emit(ctx, 1, 0x1001); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 4, 0); + else + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 1, 0xf); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + xf_emit(ctx, 0x15, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 3, 0); + } else + xf_emit(ctx, 0x17, 0); + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0x0fac6881); + xf_emit(ctx, 1, magic2); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 2, 1); + xf_emit(ctx, 3, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 2, 1); + else + xf_emit(ctx, 1, 1); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 2, 0); + else if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 0); +} + +static void +nv50_graph_construct_xfer_tp_x3(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 2, 0); + else + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0x2a712488); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x4085c000); + xf_emit(ctx, 1, 0x40); + xf_emit(ctx, 1, 0x100); + xf_emit(ctx, 1, 0x10100); + xf_emit(ctx, 1, 0x02800000); +} + +static void +nv50_graph_construct_xfer_tp_x4(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 2, 0x04e3bfdf); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x00ffff00); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 2, 1); + else + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x00ffff00); + xf_emit(ctx, 8, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0x30201000); + xf_emit(ctx, 1, 0x70605040); + xf_emit(ctx, 1, 0xb8a89888); + xf_emit(ctx, 1, 0xf8e8d8c8); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x1a); +} + +static void +nv50_graph_construct_xfer_tp_x5(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 0xfac6881); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 2, 1); + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 1); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 0xb, 0); + else + xf_emit(ctx, 0xa, 0); + xf_emit(ctx, 8, 1); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0xfac6881); + xf_emit(ctx, 1, 0xf); + xf_emit(ctx, 7, 0); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 1); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + xf_emit(ctx, 6, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 6, 0); + } else { + xf_emit(ctx, 0xb, 0); + } +} + +static void +nv50_graph_construct_xfer_tp(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + if (dev_priv->chipset < 0xa0) { + nv50_graph_construct_xfer_tp_x1(ctx); + nv50_graph_construct_xfer_tp_x2(ctx); + nv50_graph_construct_xfer_tp_x3(ctx); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 0xf, 0); + else + xf_emit(ctx, 0x12, 0); + nv50_graph_construct_xfer_tp_x4(ctx); + } else { + nv50_graph_construct_xfer_tp_x3(ctx); + if (dev_priv->chipset < 0xaa) + xf_emit(ctx, 0xc, 0); + else + xf_emit(ctx, 0xa, 0); + nv50_graph_construct_xfer_tp_x2(ctx); + nv50_graph_construct_xfer_tp_x5(ctx); + nv50_graph_construct_xfer_tp_x4(ctx); + nv50_graph_construct_xfer_tp_x1(ctx); + } +} + +static void +nv50_graph_construct_xfer_tp2(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i, mpcnt; + if (dev_priv->chipset == 0x98 || dev_priv->chipset == 0xaa) + mpcnt = 1; + else if (dev_priv->chipset < 0xa0 || dev_priv->chipset >= 0xa8) + mpcnt = 2; + else + mpcnt = 3; + for (i = 0; i < mpcnt; i++) { + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x80); + xf_emit(ctx, 1, 0x80007004); + xf_emit(ctx, 1, 0x04000400); + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0xc0); + xf_emit(ctx, 1, 0x1000); + xf_emit(ctx, 2, 0); + if (dev_priv->chipset == 0x86 || dev_priv->chipset == 0x98 || dev_priv->chipset >= 0xa8) { + xf_emit(ctx, 1, 0xe00); + xf_emit(ctx, 1, 0x1e00); + } + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 2, 0x1000); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 2); + if (dev_priv->chipset >= 0xaa) + xf_emit(ctx, 0xb, 0); + else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 0xc, 0); + else + xf_emit(ctx, 0xa, 0); + } + xf_emit(ctx, 1, 0x08100c12); + xf_emit(ctx, 1, 0); + if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 1, 0x1fe21); + } + xf_emit(ctx, 5, 0); + xf_emit(ctx, 4, 0xffff); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 2, 0x10001); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 0x1fe21); + xf_emit(ctx, 1, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 1); + xf_emit(ctx, 4, 0); + xf_emit(ctx, 1, 0x08100c12); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 8, 0); + xf_emit(ctx, 1, 0xfac6881); + xf_emit(ctx, 1, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) + xf_emit(ctx, 1, 3); + xf_emit(ctx, 3, 0); + xf_emit(ctx, 1, 4); + xf_emit(ctx, 9, 0); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 2, 1); + xf_emit(ctx, 1, 2); + xf_emit(ctx, 3, 1); + xf_emit(ctx, 1, 0); + if (dev_priv->chipset > 0xa0 && dev_priv->chipset < 0xaa) { + xf_emit(ctx, 8, 2); + xf_emit(ctx, 0x10, 1); + xf_emit(ctx, 8, 2); + xf_emit(ctx, 0x18, 1); + xf_emit(ctx, 3, 0); + } + xf_emit(ctx, 1, 4); + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 0x3a0, 0); + else if (dev_priv->chipset < 0x94) + xf_emit(ctx, 0x3a2, 0); + else if (dev_priv->chipset == 0x98 || dev_priv->chipset == 0xaa) + xf_emit(ctx, 0x39f, 0); + else + xf_emit(ctx, 0x3a3, 0); + xf_emit(ctx, 1, 0x11); + xf_emit(ctx, 1, 0); + xf_emit(ctx, 1, 1); + xf_emit(ctx, 0x2d, 0); +} + +static void +nv50_graph_construct_xfer2(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + uint32_t offset; + uint32_t units = nv_rd32 (ctx->dev, 0x1540); + int size = 0; + + offset = (ctx->ctxvals_pos+0x3f)&~0x3f; + + if (dev_priv->chipset < 0xa0) { + for (i = 0; i < 8; i++) { + ctx->ctxvals_pos = offset + i; + if (i == 0) + xf_emit(ctx, 1, 0x08100c12); + if (units & (1 << i)) + nv50_graph_construct_xfer_tp2(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + } + } else { + /* Strand 0: TPs 0, 1 */ + ctx->ctxvals_pos = offset; + xf_emit(ctx, 1, 0x08100c12); + if (units & (1 << 0)) + nv50_graph_construct_xfer_tp2(ctx); + if (units & (1 << 1)) + nv50_graph_construct_xfer_tp2(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 0: TPs 2, 3 */ + ctx->ctxvals_pos = offset + 1; + if (units & (1 << 2)) + nv50_graph_construct_xfer_tp2(ctx); + if (units & (1 << 3)) + nv50_graph_construct_xfer_tp2(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 0: TPs 4, 5, 6 */ + ctx->ctxvals_pos = offset + 2; + if (units & (1 << 4)) + nv50_graph_construct_xfer_tp2(ctx); + if (units & (1 << 5)) + nv50_graph_construct_xfer_tp2(ctx); + if (units & (1 << 6)) + nv50_graph_construct_xfer_tp2(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 0: TPs 7, 8, 9 */ + ctx->ctxvals_pos = offset + 3; + if (units & (1 << 7)) + nv50_graph_construct_xfer_tp2(ctx); + if (units & (1 << 8)) + nv50_graph_construct_xfer_tp2(ctx); + if (units & (1 << 9)) + nv50_graph_construct_xfer_tp2(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + } + ctx->ctxvals_pos = offset + size * 8; + ctx->ctxvals_pos = (ctx->ctxvals_pos+0x3f)&~0x3f; + cp_lsr (ctx, offset); + cp_out (ctx, CP_SET_XFER_POINTER); + cp_lsr (ctx, size); + cp_out (ctx, CP_SEEK_2); + cp_out (ctx, CP_XFER_2); + cp_wait(ctx, XFER, BUSY); +} diff --git a/drivers/gpu/drm/nouveau/nv50_instmem.c b/drivers/gpu/drm/nouveau/nv50_instmem.c index 94400f777e7f..de1f5b0062c5 100644 --- a/drivers/gpu/drm/nouveau/nv50_instmem.c +++ b/drivers/gpu/drm/nouveau/nv50_instmem.c @@ -76,6 +76,11 @@ nv50_instmem_init(struct drm_device *dev) for (i = 0x1700; i <= 0x1710; i += 4) priv->save1700[(i-0x1700)/4] = nv_rd32(dev, i); + if (dev_priv->chipset == 0xaa || dev_priv->chipset == 0xac) + dev_priv->vram_sys_base = nv_rd32(dev, 0x100e10) << 12; + else + dev_priv->vram_sys_base = 0; + /* Reserve the last MiB of VRAM, we should probably try to avoid * setting up the below tables over the top of the VBIOS image at * some point. @@ -172,16 +177,28 @@ nv50_instmem_init(struct drm_device *dev) * We map the entire fake channel into the start of the PRAMIN BAR */ ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, pt_size, 0x1000, - 0, &priv->pramin_pt); + 0, &priv->pramin_pt); if (ret) return ret; - for (i = 0, v = c_offset; i < pt_size; i += 8, v += 0x1000) { - if (v < (c_offset + c_size)) - BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, v | 1); - else - BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, 0x00000009); + v = c_offset | 1; + if (dev_priv->vram_sys_base) { + v += dev_priv->vram_sys_base; + v |= 0x30; + } + + i = 0; + while (v < dev_priv->vram_sys_base + c_offset + c_size) { + BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, v); + BAR0_WI32(priv->pramin_pt->gpuobj, i + 4, 0x00000000); + v += 0x1000; + i += 8; + } + + while (i < pt_size) { + BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, 0x00000000); BAR0_WI32(priv->pramin_pt->gpuobj, i + 4, 0x00000000); + i += 8; } BAR0_WI32(chan->vm_pd, 0x00, priv->pramin_pt->instance | 0x63); @@ -373,7 +390,7 @@ nv50_instmem_populate(struct drm_device *dev, struct nouveau_gpuobj *gpuobj, if (gpuobj->im_backing) return -EINVAL; - *sz = (*sz + (NV50_INSTMEM_PAGE_SIZE-1)) & ~(NV50_INSTMEM_PAGE_SIZE-1); + *sz = ALIGN(*sz, NV50_INSTMEM_PAGE_SIZE); if (*sz == 0) return -EINVAL; @@ -416,7 +433,9 @@ nv50_instmem_bind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj) { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nv50_instmem_priv *priv = dev_priv->engine.instmem.priv; - uint32_t pte, pte_end, vram; + struct nouveau_gpuobj *pramin_pt = priv->pramin_pt->gpuobj; + uint32_t pte, pte_end; + uint64_t vram; if (!gpuobj->im_backing || !gpuobj->im_pramin || gpuobj->im_bound) return -EINVAL; @@ -424,20 +443,24 @@ nv50_instmem_bind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj) NV_DEBUG(dev, "st=0x%0llx sz=0x%0llx\n", gpuobj->im_pramin->start, gpuobj->im_pramin->size); - pte = (gpuobj->im_pramin->start >> 12) << 3; - pte_end = ((gpuobj->im_pramin->size >> 12) << 3) + pte; + pte = (gpuobj->im_pramin->start >> 12) << 1; + pte_end = ((gpuobj->im_pramin->size >> 12) << 1) + pte; vram = gpuobj->im_backing_start; NV_DEBUG(dev, "pramin=0x%llx, pte=%d, pte_end=%d\n", gpuobj->im_pramin->start, pte, pte_end); NV_DEBUG(dev, "first vram page: 0x%08x\n", gpuobj->im_backing_start); + vram |= 1; + if (dev_priv->vram_sys_base) { + vram += dev_priv->vram_sys_base; + vram |= 0x30; + } + dev_priv->engine.instmem.prepare_access(dev, true); while (pte < pte_end) { - nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 0)/4, vram | 1); - nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 4)/4, 0x00000000); - - pte += 8; + nv_wo32(dev, pramin_pt, pte++, lower_32_bits(vram)); + nv_wo32(dev, pramin_pt, pte++, upper_32_bits(vram)); vram += NV50_INSTMEM_PAGE_SIZE; } dev_priv->engine.instmem.finish_access(dev); @@ -470,14 +493,13 @@ nv50_instmem_unbind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj) if (gpuobj->im_bound == 0) return -EINVAL; - pte = (gpuobj->im_pramin->start >> 12) << 3; - pte_end = ((gpuobj->im_pramin->size >> 12) << 3) + pte; + pte = (gpuobj->im_pramin->start >> 12) << 1; + pte_end = ((gpuobj->im_pramin->size >> 12) << 1) + pte; dev_priv->engine.instmem.prepare_access(dev, true); while (pte < pte_end) { - nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 0)/4, 0x00000009); - nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 4)/4, 0x00000000); - pte += 8; + nv_wo32(dev, priv->pramin_pt->gpuobj, pte++, 0x00000000); + nv_wo32(dev, priv->pramin_pt->gpuobj, pte++, 0x00000000); } dev_priv->engine.instmem.finish_access(dev); diff --git a/drivers/gpu/drm/nouveau/nv50_sor.c b/drivers/gpu/drm/nouveau/nv50_sor.c index ecf1936b8224..c2fff543b06f 100644 --- a/drivers/gpu/drm/nouveau/nv50_sor.c +++ b/drivers/gpu/drm/nouveau/nv50_sor.c @@ -101,6 +101,7 @@ nv50_sor_dpms(struct drm_encoder *encoder, int mode) struct nouveau_encoder *nvenc = nouveau_encoder(enc); if (nvenc == nv_encoder || + nvenc->disconnect != nv50_sor_disconnect || nvenc->dcb->or != nv_encoder->dcb->or) continue; diff --git a/drivers/gpu/drm/radeon/Kconfig b/drivers/gpu/drm/radeon/Kconfig index 5982321be4d5..1c02d23f6fcc 100644 --- a/drivers/gpu/drm/radeon/Kconfig +++ b/drivers/gpu/drm/radeon/Kconfig @@ -1,10 +1,14 @@ config DRM_RADEON_KMS - bool "Enable modesetting on radeon by default" + bool "Enable modesetting on radeon by default - NEW DRIVER" depends on DRM_RADEON help - Choose this option if you want kernel modesetting enabled by default, - and you have a new enough userspace to support this. Running old - userspaces with this enabled will cause pain. + Choose this option if you want kernel modesetting enabled by default. + + This is a completely new driver. It's only part of the existing drm + for compatibility reasons. It requires an entirely different graphics + stack above it and works very differently from the old drm stack. + i.e. don't enable this unless you know what you are doing it may + cause issues or bugs compared to the previous userspace driver stack. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed diff --git a/drivers/gpu/drm/radeon/atom.c b/drivers/gpu/drm/radeon/atom.c index bd0c843872b2..d75788feac6c 100644 --- a/drivers/gpu/drm/radeon/atom.c +++ b/drivers/gpu/drm/radeon/atom.c @@ -24,6 +24,7 @@ #include <linux/module.h> #include <linux/sched.h> +#include <asm/unaligned.h> #define ATOM_DEBUG @@ -212,7 +213,9 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr, case ATOM_ARG_PS: idx = U8(*ptr); (*ptr)++; - val = le32_to_cpu(ctx->ps[idx]); + /* get_unaligned_le32 avoids unaligned accesses from atombios + * tables, noticed on a DEC Alpha. */ + val = get_unaligned_le32((u32 *)&ctx->ps[idx]); if (print) DEBUG("PS[0x%02X,0x%04X]", idx, val); break; @@ -641,7 +644,6 @@ static void atom_op_delay(atom_exec_context *ctx, int *ptr, int arg) SDEBUG(" count: %d\n", count); if (arg == ATOM_UNIT_MICROSEC) udelay(count); - // schedule_timeout_uninterruptible(usecs_to_jiffies(count)); else schedule_timeout_uninterruptible(msecs_to_jiffies(count)); } @@ -879,8 +881,6 @@ static void atom_op_shl(atom_exec_context *ctx, int *ptr, int arg) uint8_t attr = U8((*ptr)++), shift; uint32_t saved, dst; int dptr = *ptr; - attr &= 0x38; - attr |= atom_def_dst[attr >> 3] << 6; SDEBUG(" dst: "); dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); shift = atom_get_src(ctx, attr, ptr); @@ -895,8 +895,6 @@ static void atom_op_shr(atom_exec_context *ctx, int *ptr, int arg) uint8_t attr = U8((*ptr)++), shift; uint32_t saved, dst; int dptr = *ptr; - attr &= 0x38; - attr |= atom_def_dst[attr >> 3] << 6; SDEBUG(" dst: "); dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); shift = atom_get_src(ctx, attr, ptr); diff --git a/drivers/gpu/drm/radeon/atombios_dp.c b/drivers/gpu/drm/radeon/atombios_dp.c index 0b6f2cef1c52..8a133bda00a2 100644 --- a/drivers/gpu/drm/radeon/atombios_dp.c +++ b/drivers/gpu/drm/radeon/atombios_dp.c @@ -336,11 +336,13 @@ bool radeon_process_aux_ch(struct radeon_i2c_chan *chan, u8 *req_bytes, union aux_channel_transaction args; int index = GetIndexIntoMasterTable(COMMAND, ProcessAuxChannelTransaction); unsigned char *base; + int retry_count = 0; memset(&args, 0, sizeof(args)); base = (unsigned char *)rdev->mode_info.atom_context->scratch; +retry: memcpy(base, req_bytes, num_bytes); args.v1.lpAuxRequest = 0; @@ -353,10 +355,12 @@ bool radeon_process_aux_ch(struct radeon_i2c_chan *chan, u8 *req_bytes, atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); - if (args.v1.ucReplyStatus) { - DRM_DEBUG("failed to get auxch %02x%02x %02x %02x 0x%02x %02x\n", + if (args.v1.ucReplyStatus && !args.v1.ucDataOutLen) { + if (args.v1.ucReplyStatus == 0x20 && retry_count++ < 10) + goto retry; + DRM_DEBUG("failed to get auxch %02x%02x %02x %02x 0x%02x %02x after %d retries\n", req_bytes[1], req_bytes[0], req_bytes[2], req_bytes[3], - chan->rec.i2c_id, args.v1.ucReplyStatus); + chan->rec.i2c_id, args.v1.ucReplyStatus, retry_count); return false; } diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index f9a83358aa5a..c52290197292 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -1970,6 +1970,13 @@ int r600_resume(struct radeon_device *rdev) DRM_ERROR("radeon: failled testing IB (%d).\n", r); return r; } + + r = r600_audio_init(rdev); + if (r) { + DRM_ERROR("radeon: audio resume failed\n"); + return r; + } + return r; } @@ -1977,6 +1984,7 @@ int r600_suspend(struct radeon_device *rdev) { int r; + r600_audio_fini(rdev); /* FIXME: we should wait for ring to be empty */ r600_cp_stop(rdev); rdev->cp.ready = false; diff --git a/drivers/gpu/drm/radeon/r600_audio.c b/drivers/gpu/drm/radeon/r600_audio.c index d7f6909afc01..387abaa275a4 100644 --- a/drivers/gpu/drm/radeon/r600_audio.c +++ b/drivers/gpu/drm/radeon/r600_audio.c @@ -270,6 +270,7 @@ void r600_audio_fini(struct radeon_device *rdev) return; del_timer(&rdev->audio_timer); + WREG32_P(R600_AUDIO_ENABLE, 0x0, ~0x81000000); r600_audio_engine_enable(rdev, false); } diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c index 6f8619cd1a0d..93783b15c81d 100644 --- a/drivers/gpu/drm/radeon/radeon_atombios.c +++ b/drivers/gpu/drm/radeon/radeon_atombios.c @@ -213,6 +213,15 @@ static bool radeon_atom_apply_quirks(struct drm_device *dev, *connector_type = DRM_MODE_CONNECTOR_DVID; } + /* Asrock RS600 board lists the DVI port as HDMI */ + if ((dev->pdev->device == 0x7941) && + (dev->pdev->subsystem_vendor == 0x1849) && + (dev->pdev->subsystem_device == 0x7941)) { + if ((*connector_type == DRM_MODE_CONNECTOR_HDMIA) && + (supported_device == ATOM_DEVICE_DFP3_SUPPORT)) + *connector_type = DRM_MODE_CONNECTOR_DVID; + } + /* a-bit f-i90hd - ciaranm on #radeonhd - this board has no DVI */ if ((dev->pdev->device == 0x7941) && (dev->pdev->subsystem_vendor == 0x147b) && @@ -294,6 +303,15 @@ static bool radeon_atom_apply_quirks(struct drm_device *dev, *connector_type = DRM_MODE_CONNECTOR_DVID; } + /* XFX Pine Group device rv730 reports no VGA DDC lines + * even though they are wired up to record 0x93 + */ + if ((dev->pdev->device == 0x9498) && + (dev->pdev->subsystem_vendor == 0x1682) && + (dev->pdev->subsystem_device == 0x2452)) { + struct radeon_device *rdev = dev->dev_private; + *i2c_bus = radeon_lookup_i2c_gpio(rdev, 0x93); + } return true; } diff --git a/drivers/gpu/drm/radeon/radeon_benchmark.c b/drivers/gpu/drm/radeon/radeon_benchmark.c index 4ddfd4b5bc51..7932dc4d6b90 100644 --- a/drivers/gpu/drm/radeon/radeon_benchmark.c +++ b/drivers/gpu/drm/radeon/radeon_benchmark.c @@ -65,31 +65,42 @@ void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize, if (r) { goto out_cleanup; } - start_jiffies = jiffies; - for (i = 0; i < n; i++) { - r = radeon_fence_create(rdev, &fence); - if (r) { - goto out_cleanup; + + /* r100 doesn't have dma engine so skip the test */ + if (rdev->asic->copy_dma) { + + start_jiffies = jiffies; + for (i = 0; i < n; i++) { + r = radeon_fence_create(rdev, &fence); + if (r) { + goto out_cleanup; + } + + r = radeon_copy_dma(rdev, saddr, daddr, + size / RADEON_GPU_PAGE_SIZE, fence); + + if (r) { + goto out_cleanup; + } + r = radeon_fence_wait(fence, false); + if (r) { + goto out_cleanup; + } + radeon_fence_unref(&fence); } - r = radeon_copy_dma(rdev, saddr, daddr, size / RADEON_GPU_PAGE_SIZE, fence); - if (r) { - goto out_cleanup; + end_jiffies = jiffies; + time = end_jiffies - start_jiffies; + time = jiffies_to_msecs(time); + if (time > 0) { + i = ((n * size) >> 10) / time; + printk(KERN_INFO "radeon: dma %u bo moves of %ukb from" + " %d to %d in %lums (%ukb/ms %ukb/s %uM/s)\n", + n, size >> 10, + sdomain, ddomain, time, + i, i * 1000, (i * 1000) / 1024); } - r = radeon_fence_wait(fence, false); - if (r) { - goto out_cleanup; - } - radeon_fence_unref(&fence); - } - end_jiffies = jiffies; - time = end_jiffies - start_jiffies; - time = jiffies_to_msecs(time); - if (time > 0) { - i = ((n * size) >> 10) / time; - printk(KERN_INFO "radeon: dma %u bo moves of %ukb from %d to %d" - " in %lums (%ukb/ms %ukb/s %uM/s)\n", n, size >> 10, - sdomain, ddomain, time, i, i * 1000, (i * 1000) / 1024); } + start_jiffies = jiffies; for (i = 0; i < n; i++) { r = radeon_fence_create(rdev, &fence); diff --git a/drivers/gpu/drm/radeon/radeon_connectors.c b/drivers/gpu/drm/radeon/radeon_connectors.c index 6e9e7b59d67e..ee0083f982d8 100644 --- a/drivers/gpu/drm/radeon/radeon_connectors.c +++ b/drivers/gpu/drm/radeon/radeon_connectors.c @@ -578,14 +578,15 @@ static enum drm_connector_status radeon_vga_detect(struct drm_connector *connect struct radeon_connector *radeon_connector = to_radeon_connector(connector); struct drm_encoder *encoder; struct drm_encoder_helper_funcs *encoder_funcs; - bool dret; + bool dret = false; enum drm_connector_status ret = connector_status_disconnected; encoder = radeon_best_single_encoder(connector); if (!encoder) ret = connector_status_disconnected; - dret = radeon_ddc_probe(radeon_connector); + if (radeon_connector->ddc_bus) + dret = radeon_ddc_probe(radeon_connector); if (dret) { if (radeon_connector->edid) { kfree(radeon_connector->edid); @@ -734,9 +735,10 @@ static enum drm_connector_status radeon_dvi_detect(struct drm_connector *connect struct drm_mode_object *obj; int i; enum drm_connector_status ret = connector_status_disconnected; - bool dret; + bool dret = false; - dret = radeon_ddc_probe(radeon_connector); + if (radeon_connector->ddc_bus) + dret = radeon_ddc_probe(radeon_connector); if (dret) { if (radeon_connector->edid) { kfree(radeon_connector->edid); @@ -766,7 +768,7 @@ static enum drm_connector_status radeon_dvi_detect(struct drm_connector *connect * connected and the DVI port disconnected. If the edid doesn't * say HDMI, vice versa. */ - if (radeon_connector->shared_ddc && connector_status_connected) { + if (radeon_connector->shared_ddc && (ret == connector_status_connected)) { struct drm_device *dev = connector->dev; struct drm_connector *list_connector; struct radeon_connector *list_radeon_connector; @@ -1044,8 +1046,7 @@ radeon_add_atom_connector(struct drm_device *dev, return; } if (radeon_connector->ddc_bus && i2c_bus->valid) { - if (memcmp(&radeon_connector->ddc_bus->rec, i2c_bus, - sizeof(struct radeon_i2c_bus_rec)) == 0) { + if (radeon_connector->ddc_bus->rec.i2c_id == i2c_bus->i2c_id) { radeon_connector->shared_ddc = true; shared_ddc = true; } diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c index e9d085021c1f..70ba02ed7723 100644 --- a/drivers/gpu/drm/radeon/radeon_cs.c +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -194,11 +194,8 @@ static void radeon_cs_parser_fini(struct radeon_cs_parser *parser, int error) } radeon_bo_list_unreserve(&parser->validated); for (i = 0; i < parser->nrelocs; i++) { - if (parser->relocs[i].gobj) { - mutex_lock(&parser->rdev->ddev->struct_mutex); - drm_gem_object_unreference(parser->relocs[i].gobj); - mutex_unlock(&parser->rdev->ddev->struct_mutex); - } + if (parser->relocs[i].gobj) + drm_gem_object_unreference_unlocked(parser->relocs[i].gobj); } kfree(parser->track); kfree(parser->relocs); diff --git a/drivers/gpu/drm/radeon/radeon_cursor.c b/drivers/gpu/drm/radeon/radeon_cursor.c index 7ecf5e1b39c1..b7023fff89eb 100644 --- a/drivers/gpu/drm/radeon/radeon_cursor.c +++ b/drivers/gpu/drm/radeon/radeon_cursor.c @@ -186,17 +186,13 @@ int radeon_crtc_cursor_set(struct drm_crtc *crtc, unpin: if (radeon_crtc->cursor_bo) { radeon_gem_object_unpin(radeon_crtc->cursor_bo); - mutex_lock(&crtc->dev->struct_mutex); - drm_gem_object_unreference(radeon_crtc->cursor_bo); - mutex_unlock(&crtc->dev->struct_mutex); + drm_gem_object_unreference_unlocked(radeon_crtc->cursor_bo); } radeon_crtc->cursor_bo = obj; return 0; fail: - mutex_lock(&crtc->dev->struct_mutex); - drm_gem_object_unreference(obj); - mutex_unlock(&crtc->dev->struct_mutex); + drm_gem_object_unreference_unlocked(obj); return 0; } diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c index e35cc3da8f22..ba8d806dcf39 100644 --- a/drivers/gpu/drm/radeon/radeon_display.c +++ b/drivers/gpu/drm/radeon/radeon_display.c @@ -310,7 +310,7 @@ static void radeon_print_display_setup(struct drm_device *dev) DRM_INFO(" %s\n", connector_names[connector->connector_type]); if (radeon_connector->hpd.hpd != RADEON_HPD_NONE) DRM_INFO(" %s\n", hpd_names[radeon_connector->hpd.hpd]); - if (radeon_connector->ddc_bus) + if (radeon_connector->ddc_bus) { DRM_INFO(" DDC: 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x\n", radeon_connector->ddc_bus->rec.mask_clk_reg, radeon_connector->ddc_bus->rec.mask_data_reg, @@ -320,6 +320,15 @@ static void radeon_print_display_setup(struct drm_device *dev) radeon_connector->ddc_bus->rec.en_data_reg, radeon_connector->ddc_bus->rec.y_clk_reg, radeon_connector->ddc_bus->rec.y_data_reg); + } else { + if (connector->connector_type == DRM_MODE_CONNECTOR_VGA || + connector->connector_type == DRM_MODE_CONNECTOR_DVII || + connector->connector_type == DRM_MODE_CONNECTOR_DVID || + connector->connector_type == DRM_MODE_CONNECTOR_DVIA || + connector->connector_type == DRM_MODE_CONNECTOR_HDMIA || + connector->connector_type == DRM_MODE_CONNECTOR_HDMIB) + DRM_INFO(" DDC: no ddc bus - possible BIOS bug - please report to xorg-driver-ati@lists.x.org\n"); + } DRM_INFO(" Encoders:\n"); list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { radeon_encoder = to_radeon_encoder(encoder); @@ -801,11 +810,8 @@ static void radeon_user_framebuffer_destroy(struct drm_framebuffer *fb) if (fb->fbdev) radeonfb_remove(dev, fb); - if (radeon_fb->obj) { - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(radeon_fb->obj); - mutex_unlock(&dev->struct_mutex); - } + if (radeon_fb->obj) + drm_gem_object_unreference_unlocked(radeon_fb->obj); drm_framebuffer_cleanup(fb); kfree(radeon_fb); } diff --git a/drivers/gpu/drm/radeon/radeon_drv.h b/drivers/gpu/drm/radeon/radeon_drv.h index ee484b61bc5c..cf911859eac2 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.h +++ b/drivers/gpu/drm/radeon/radeon_drv.h @@ -106,9 +106,10 @@ * 1.29- R500 3D cmd buffer support * 1.30- Add support for occlusion queries * 1.31- Add support for num Z pipes from GET_PARAM + * 1.32- fixes for rv740 setup */ #define DRIVER_MAJOR 1 -#define DRIVER_MINOR 31 +#define DRIVER_MINOR 32 #define DRIVER_PATCHLEVEL 0 enum radeon_cp_microcode_version { diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c index c39ddda13840..cda112cc7a6c 100644 --- a/drivers/gpu/drm/radeon/radeon_fb.c +++ b/drivers/gpu/drm/radeon/radeon_fb.c @@ -59,7 +59,7 @@ static struct fb_ops radeonfb_ops = { }; /** - * Curretly it is assumed that the old framebuffer is reused. + * Currently it is assumed that the old framebuffer is reused. * * LOCKING * caller should hold the mode config lock. @@ -243,7 +243,7 @@ int radeonfb_create(struct drm_device *dev, if (ret) goto out_unref; - memset_io(fbptr, 0xff, aligned_size); + memset_io(fbptr, 0x0, aligned_size); strcpy(info->fix.id, "radeondrmfb"); diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index db8e9a355a01..ef92d147d8f0 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -69,9 +69,7 @@ int radeon_gem_object_create(struct radeon_device *rdev, int size, if (r != -ERESTARTSYS) DRM_ERROR("Failed to allocate GEM object (%d, %d, %u, %d)\n", size, initial_domain, alignment, r); - mutex_lock(&rdev->ddev->struct_mutex); - drm_gem_object_unreference(gobj); - mutex_unlock(&rdev->ddev->struct_mutex); + drm_gem_object_unreference_unlocked(gobj); return r; } gobj->driver_private = robj; @@ -202,14 +200,10 @@ int radeon_gem_create_ioctl(struct drm_device *dev, void *data, } r = drm_gem_handle_create(filp, gobj, &handle); if (r) { - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gobj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gobj); return r; } - mutex_lock(&dev->struct_mutex); - drm_gem_object_handle_unreference(gobj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_handle_unreference_unlocked(gobj); args->handle = handle; return 0; } @@ -236,9 +230,7 @@ int radeon_gem_set_domain_ioctl(struct drm_device *dev, void *data, r = radeon_gem_set_domain(gobj, args->read_domains, args->write_domain); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gobj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gobj); return r; } @@ -255,9 +247,7 @@ int radeon_gem_mmap_ioctl(struct drm_device *dev, void *data, } robj = gobj->driver_private; args->addr_ptr = radeon_bo_mmap_offset(robj); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gobj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gobj); return 0; } @@ -288,9 +278,7 @@ int radeon_gem_busy_ioctl(struct drm_device *dev, void *data, default: break; } - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gobj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gobj); return r; } @@ -311,9 +299,7 @@ int radeon_gem_wait_idle_ioctl(struct drm_device *dev, void *data, /* callback hw specific functions if any */ if (robj->rdev->asic->ioctl_wait_idle) robj->rdev->asic->ioctl_wait_idle(robj->rdev, robj); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gobj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gobj); return r; } @@ -331,9 +317,7 @@ int radeon_gem_set_tiling_ioctl(struct drm_device *dev, void *data, return -EINVAL; robj = gobj->driver_private; r = radeon_bo_set_tiling_flags(robj, args->tiling_flags, args->pitch); - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gobj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gobj); return r; } @@ -356,8 +340,6 @@ int radeon_gem_get_tiling_ioctl(struct drm_device *dev, void *data, radeon_bo_get_tiling_flags(rbo, &args->tiling_flags, &args->pitch); radeon_bo_unreserve(rbo); out: - mutex_lock(&dev->struct_mutex); - drm_gem_object_unreference(gobj); - mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(gobj); return r; } diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c index 38fa14429320..e50513a62735 100644 --- a/drivers/gpu/drm/radeon/radeon_ring.c +++ b/drivers/gpu/drm/radeon/radeon_ring.c @@ -130,6 +130,8 @@ void radeon_ib_free(struct radeon_device *rdev, struct radeon_ib **ib) if (tmp == NULL) { return; } + if (!tmp->fence->emited) + radeon_fence_unref(&tmp->fence); mutex_lock(&rdev->ib_pool.mutex); tmp->free = true; mutex_unlock(&rdev->ib_pool.mutex); @@ -222,6 +224,7 @@ void radeon_ib_pool_fini(struct radeon_device *rdev) } mutex_lock(&rdev->ib_pool.mutex); radeon_ib_bogus_cleanup(rdev); + if (rdev->ib_pool.robj) { r = radeon_bo_reserve(rdev->ib_pool.robj, false); if (likely(r == 0)) { diff --git a/drivers/gpu/drm/radeon/radeon_state.c b/drivers/gpu/drm/radeon/radeon_state.c index 8e69dc485dcb..3c32f840dcd2 100644 --- a/drivers/gpu/drm/radeon/radeon_state.c +++ b/drivers/gpu/drm/radeon/radeon_state.c @@ -1978,7 +1978,7 @@ static void radeon_apply_surface_regs(int surf_index, * Note that refcount can be at most 2, since during a free refcount=3 * might mean we have to allocate a new surface which might not always * be available. - * For example : we allocate three contigous surfaces ABC. If B is + * For example : we allocate three contiguous surfaces ABC. If B is * freed, we suddenly need two surfaces to store A and C, which might * not always be available. */ diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index 1157e0f758fa..43c5ab34b634 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -409,7 +409,7 @@ static int radeon_bo_move(struct ttm_buffer_object *bo, new_mem->mem_type == TTM_PL_SYSTEM) || (old_mem->mem_type == TTM_PL_SYSTEM && new_mem->mem_type == TTM_PL_TT)) { - /* bind is enought */ + /* bind is enough */ radeon_move_null(bo, new_mem); return 0; } diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 1a3e909b7bba..c7320ce4567d 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -1020,6 +1020,12 @@ static int ttm_bo_mem_compat(struct ttm_placement *placement, struct ttm_mem_reg *mem) { int i; + struct drm_mm_node *node = mem->mm_node; + + if (node && placement->lpfn != 0 && + (node->start < placement->fpfn || + node->start + node->size > placement->lpfn)) + return -1; for (i = 0; i < placement->num_placement; i++) { if ((placement->placement[i] & mem->placement & diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c index 3f72fe1ddf8b..5ca37a58a98c 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_util.c +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c @@ -424,7 +424,7 @@ static int ttm_bo_kmap_ttm(struct ttm_buffer_object *bo, /* * We need to use vmap to get the desired page protection - * or to make the buffer object look contigous. + * or to make the buffer object look contiguous. */ prot = (mem->placement & TTM_PL_FLAG_CACHED) ? PAGE_KERNEL : diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index e2123af7775a..a759170763bb 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -196,14 +196,15 @@ EXPORT_SYMBOL(ttm_tt_populate); #ifdef CONFIG_X86 static inline int ttm_tt_set_page_caching(struct page *p, - enum ttm_caching_state c_state) + enum ttm_caching_state c_old, + enum ttm_caching_state c_new) { int ret = 0; if (PageHighMem(p)) return 0; - if (get_page_memtype(p) != -1) { + if (c_old != tt_cached) { /* p isn't in the default caching state, set it to * writeback first to free its current memtype. */ @@ -212,16 +213,17 @@ static inline int ttm_tt_set_page_caching(struct page *p, return ret; } - if (c_state == tt_wc) + if (c_new == tt_wc) ret = set_memory_wc((unsigned long) page_address(p), 1); - else if (c_state == tt_uncached) + else if (c_new == tt_uncached) ret = set_pages_uc(p, 1); return ret; } #else /* CONFIG_X86 */ static inline int ttm_tt_set_page_caching(struct page *p, - enum ttm_caching_state c_state) + enum ttm_caching_state c_old, + enum ttm_caching_state c_new) { return 0; } @@ -254,7 +256,9 @@ static int ttm_tt_set_caching(struct ttm_tt *ttm, for (i = 0; i < ttm->num_pages; ++i) { cur_page = ttm->pages[i]; if (likely(cur_page != NULL)) { - ret = ttm_tt_set_page_caching(cur_page, c_state); + ret = ttm_tt_set_page_caching(cur_page, + ttm->caching_state, + c_state); if (unlikely(ret != 0)) goto out_err; } @@ -268,7 +272,7 @@ out_err: for (j = 0; j < i; ++j) { cur_page = ttm->pages[j]; if (likely(cur_page != NULL)) { - (void)ttm_tt_set_page_caching(cur_page, + (void)ttm_tt_set_page_caching(cur_page, c_state, ttm->caching_state); } } @@ -476,7 +480,7 @@ static int ttm_tt_swapin(struct ttm_tt *ttm) void *from_virtual; void *to_virtual; int i; - int ret; + int ret = -ENOMEM; if (ttm->page_flags & TTM_PAGE_FLAG_USER) { ret = ttm_tt_set_user(ttm, ttm->tsk, ttm->start, @@ -495,8 +499,10 @@ static int ttm_tt_swapin(struct ttm_tt *ttm) for (i = 0; i < ttm->num_pages; ++i) { from_page = read_mapping_page(swap_space, i, NULL); - if (IS_ERR(from_page)) + if (IS_ERR(from_page)) { + ret = PTR_ERR(from_page); goto out_err; + } to_page = __ttm_tt_get_page(ttm, i); if (unlikely(to_page == NULL)) goto out_err; @@ -519,7 +525,7 @@ static int ttm_tt_swapin(struct ttm_tt *ttm) return 0; out_err: ttm_tt_free_alloced_pages(ttm); - return -ENOMEM; + return ret; } int ttm_tt_swapout(struct ttm_tt *ttm, struct file *persistant_swap_storage) @@ -531,6 +537,7 @@ int ttm_tt_swapout(struct ttm_tt *ttm, struct file *persistant_swap_storage) void *from_virtual; void *to_virtual; int i; + int ret = -ENOMEM; BUG_ON(ttm->state != tt_unbound && ttm->state != tt_unpopulated); BUG_ON(ttm->caching_state != tt_cached); @@ -553,7 +560,7 @@ int ttm_tt_swapout(struct ttm_tt *ttm, struct file *persistant_swap_storage) 0); if (unlikely(IS_ERR(swap_storage))) { printk(KERN_ERR "Failed allocating swap storage.\n"); - return -ENOMEM; + return PTR_ERR(swap_storage); } } else swap_storage = persistant_swap_storage; @@ -565,9 +572,10 @@ int ttm_tt_swapout(struct ttm_tt *ttm, struct file *persistant_swap_storage) if (unlikely(from_page == NULL)) continue; to_page = read_mapping_page(swap_space, i, NULL); - if (unlikely(to_page == NULL)) + if (unlikely(IS_ERR(to_page))) { + ret = PTR_ERR(to_page); goto out_err; - + } preempt_disable(); from_virtual = kmap_atomic(from_page, KM_USER0); to_virtual = kmap_atomic(to_page, KM_USER1); @@ -591,5 +599,5 @@ out_err: if (!persistant_swap_storage) fput(swap_storage); - return -ENOMEM; + return ret; } diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index a6e8f687fa64..0c9c0811f42d 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -348,22 +348,19 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset) */ DRM_INFO("It appears like vesafb is loaded. " - "Ignore above error if any. Entering stealth mode.\n"); + "Ignore above error if any.\n"); ret = pci_request_region(dev->pdev, 2, "vmwgfx stealth probe"); if (unlikely(ret != 0)) { DRM_ERROR("Failed reserving the SVGA MMIO resource.\n"); goto out_no_device; } - vmw_kms_init(dev_priv); - vmw_overlay_init(dev_priv); - } else { - ret = vmw_request_device(dev_priv); - if (unlikely(ret != 0)) - goto out_no_device; - vmw_kms_init(dev_priv); - vmw_overlay_init(dev_priv); - vmw_fb_init(dev_priv); } + ret = vmw_request_device(dev_priv); + if (unlikely(ret != 0)) + goto out_no_device; + vmw_kms_init(dev_priv); + vmw_overlay_init(dev_priv); + vmw_fb_init(dev_priv); dev_priv->pm_nb.notifier_call = vmwgfx_pm_notifier; register_pm_notifier(&dev_priv->pm_nb); @@ -406,17 +403,15 @@ static int vmw_driver_unload(struct drm_device *dev) unregister_pm_notifier(&dev_priv->pm_nb); - if (!dev_priv->stealth) { - vmw_fb_close(dev_priv); - vmw_kms_close(dev_priv); - vmw_overlay_close(dev_priv); - vmw_release_device(dev_priv); - pci_release_regions(dev->pdev); - } else { - vmw_kms_close(dev_priv); - vmw_overlay_close(dev_priv); + vmw_fb_close(dev_priv); + vmw_kms_close(dev_priv); + vmw_overlay_close(dev_priv); + vmw_release_device(dev_priv); + if (dev_priv->stealth) pci_release_region(dev->pdev, 2); - } + else + pci_release_regions(dev->pdev); + if (dev_priv->capabilities & SVGA_CAP_IRQMASK) drm_irq_uninstall(dev_priv->dev); if (dev->devname == vmw_devname) @@ -585,11 +580,6 @@ static int vmw_master_set(struct drm_device *dev, int ret = 0; DRM_INFO("Master set.\n"); - if (dev_priv->stealth) { - ret = vmw_request_device(dev_priv); - if (unlikely(ret != 0)) - return ret; - } if (active) { BUG_ON(active != &dev_priv->fbdev_master); @@ -649,18 +639,11 @@ static void vmw_master_drop(struct drm_device *dev, ttm_lock_set_kill(&vmaster->lock, true, SIGTERM); - if (dev_priv->stealth) { - ret = ttm_bo_evict_mm(&dev_priv->bdev, TTM_PL_VRAM); - if (unlikely(ret != 0)) - DRM_ERROR("Unable to clean VRAM on master drop.\n"); - vmw_release_device(dev_priv); - } dev_priv->active_master = &dev_priv->fbdev_master; ttm_lock_set_kill(&dev_priv->fbdev_master.lock, false, SIGTERM); ttm_vt_unlock(&dev_priv->fbdev_master.lock); - if (!dev_priv->stealth) - vmw_fb_on(dev_priv); + vmw_fb_on(dev_priv); } diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h index 135be9688c90..356dc935ec13 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h @@ -39,10 +39,10 @@ #include "ttm/ttm_execbuf_util.h" #include "ttm/ttm_module.h" -#define VMWGFX_DRIVER_DATE "20090724" -#define VMWGFX_DRIVER_MAJOR 0 -#define VMWGFX_DRIVER_MINOR 1 -#define VMWGFX_DRIVER_PATCHLEVEL 2 +#define VMWGFX_DRIVER_DATE "20100209" +#define VMWGFX_DRIVER_MAJOR 1 +#define VMWGFX_DRIVER_MINOR 0 +#define VMWGFX_DRIVER_PATCHLEVEL 0 #define VMWGFX_FILE_PAGE_OFFSET 0x00100000 #define VMWGFX_FIFO_STATIC_SIZE (1024*1024) #define VMWGFX_MAX_RELOCATIONS 2048 @@ -113,6 +113,7 @@ struct vmw_fifo_state { unsigned long static_buffer_size; bool using_bounce_buffer; uint32_t capabilities; + struct mutex fifo_mutex; struct rw_semaphore rwsem; }; @@ -213,7 +214,7 @@ struct vmw_private { * Fencing and IRQs. */ - uint32_t fence_seq; + atomic_t fence_seq; wait_queue_head_t fence_queue; wait_queue_head_t fifo_queue; atomic_t fence_queue_waiters; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c index d69caf92ffe7..0897359b3e4e 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c @@ -182,25 +182,19 @@ static int vmw_cmd_present_check(struct vmw_private *dev_priv, return vmw_cmd_sid_check(dev_priv, sw_context, &cmd->body.sid); } -static int vmw_cmd_dma(struct vmw_private *dev_priv, - struct vmw_sw_context *sw_context, - SVGA3dCmdHeader *header) +static int vmw_translate_guest_ptr(struct vmw_private *dev_priv, + struct vmw_sw_context *sw_context, + SVGAGuestPtr *ptr, + struct vmw_dma_buffer **vmw_bo_p) { - uint32_t handle; struct vmw_dma_buffer *vmw_bo = NULL; struct ttm_buffer_object *bo; - struct vmw_surface *srf = NULL; - struct vmw_dma_cmd { - SVGA3dCmdHeader header; - SVGA3dCmdSurfaceDMA dma; - } *cmd; + uint32_t handle = ptr->gmrId; struct vmw_relocation *reloc; - int ret; uint32_t cur_validate_node; struct ttm_validate_buffer *val_buf; + int ret; - cmd = container_of(header, struct vmw_dma_cmd, header); - handle = cmd->dma.guest.ptr.gmrId; ret = vmw_user_dmabuf_lookup(sw_context->tfile, handle, &vmw_bo); if (unlikely(ret != 0)) { DRM_ERROR("Could not find or use GMR region.\n"); @@ -209,14 +203,14 @@ static int vmw_cmd_dma(struct vmw_private *dev_priv, bo = &vmw_bo->base; if (unlikely(sw_context->cur_reloc >= VMWGFX_MAX_RELOCATIONS)) { - DRM_ERROR("Max number of DMA commands per submission" + DRM_ERROR("Max number relocations per submission" " exceeded\n"); ret = -EINVAL; goto out_no_reloc; } reloc = &sw_context->relocs[sw_context->cur_reloc++]; - reloc->location = &cmd->dma.guest.ptr; + reloc->location = ptr; cur_validate_node = vmw_dmabuf_validate_node(bo, sw_context->cur_val_buf); if (unlikely(cur_validate_node >= VMWGFX_MAX_GMRS)) { @@ -234,7 +228,89 @@ static int vmw_cmd_dma(struct vmw_private *dev_priv, list_add_tail(&val_buf->head, &sw_context->validate_nodes); ++sw_context->cur_val_buf; } + *vmw_bo_p = vmw_bo; + return 0; + +out_no_reloc: + vmw_dmabuf_unreference(&vmw_bo); + vmw_bo_p = NULL; + return ret; +} + +static int vmw_cmd_end_query(struct vmw_private *dev_priv, + struct vmw_sw_context *sw_context, + SVGA3dCmdHeader *header) +{ + struct vmw_dma_buffer *vmw_bo; + struct vmw_query_cmd { + SVGA3dCmdHeader header; + SVGA3dCmdEndQuery q; + } *cmd; + int ret; + + cmd = container_of(header, struct vmw_query_cmd, header); + ret = vmw_cmd_cid_check(dev_priv, sw_context, header); + if (unlikely(ret != 0)) + return ret; + + ret = vmw_translate_guest_ptr(dev_priv, sw_context, + &cmd->q.guestResult, + &vmw_bo); + if (unlikely(ret != 0)) + return ret; + + vmw_dmabuf_unreference(&vmw_bo); + return 0; +} +static int vmw_cmd_wait_query(struct vmw_private *dev_priv, + struct vmw_sw_context *sw_context, + SVGA3dCmdHeader *header) +{ + struct vmw_dma_buffer *vmw_bo; + struct vmw_query_cmd { + SVGA3dCmdHeader header; + SVGA3dCmdWaitForQuery q; + } *cmd; + int ret; + + cmd = container_of(header, struct vmw_query_cmd, header); + ret = vmw_cmd_cid_check(dev_priv, sw_context, header); + if (unlikely(ret != 0)) + return ret; + + ret = vmw_translate_guest_ptr(dev_priv, sw_context, + &cmd->q.guestResult, + &vmw_bo); + if (unlikely(ret != 0)) + return ret; + + vmw_dmabuf_unreference(&vmw_bo); + return 0; +} + + +static int vmw_cmd_dma(struct vmw_private *dev_priv, + struct vmw_sw_context *sw_context, + SVGA3dCmdHeader *header) +{ + struct vmw_dma_buffer *vmw_bo = NULL; + struct ttm_buffer_object *bo; + struct vmw_surface *srf = NULL; + struct vmw_dma_cmd { + SVGA3dCmdHeader header; + SVGA3dCmdSurfaceDMA dma; + } *cmd; + int ret; + + cmd = container_of(header, struct vmw_dma_cmd, header); + ret = vmw_translate_guest_ptr(dev_priv, sw_context, + &cmd->dma.guest.ptr, + &vmw_bo); + if (unlikely(ret != 0)) + return ret; + + bo = &vmw_bo->base; ret = vmw_user_surface_lookup_handle(dev_priv, sw_context->tfile, cmd->dma.host.sid, &srf); if (ret) { @@ -379,8 +455,8 @@ static vmw_cmd_func vmw_cmd_funcs[SVGA_3D_CMD_MAX] = { VMW_CMD_DEF(SVGA_3D_CMD_DRAW_PRIMITIVES, &vmw_cmd_draw), VMW_CMD_DEF(SVGA_3D_CMD_SETSCISSORRECT, &vmw_cmd_cid_check), VMW_CMD_DEF(SVGA_3D_CMD_BEGIN_QUERY, &vmw_cmd_cid_check), - VMW_CMD_DEF(SVGA_3D_CMD_END_QUERY, &vmw_cmd_cid_check), - VMW_CMD_DEF(SVGA_3D_CMD_WAIT_FOR_QUERY, &vmw_cmd_cid_check), + VMW_CMD_DEF(SVGA_3D_CMD_END_QUERY, &vmw_cmd_end_query), + VMW_CMD_DEF(SVGA_3D_CMD_WAIT_FOR_QUERY, &vmw_cmd_wait_query), VMW_CMD_DEF(SVGA_3D_CMD_PRESENT_READBACK, &vmw_cmd_ok), VMW_CMD_DEF(SVGA_3D_CMD_BLIT_SURFACE_TO_SCREEN, &vmw_cmd_blt_surf_screen_check) diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c index 4f4f6432be8b..a93367041cdc 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c @@ -559,6 +559,9 @@ int vmw_fb_init(struct vmw_private *vmw_priv) info->pixmap.scan_align = 1; #endif + info->aperture_base = vmw_priv->vram_start; + info->aperture_size = vmw_priv->vram_size; + /* * Dirty & Deferred IO */ diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c index 4157547cc6e4..39d43a01d846 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c @@ -74,6 +74,7 @@ int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo) fifo->reserved_size = 0; fifo->using_bounce_buffer = false; + mutex_init(&fifo->fifo_mutex); init_rwsem(&fifo->rwsem); /* @@ -117,7 +118,7 @@ int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo) (unsigned int) min, (unsigned int) fifo->capabilities); - dev_priv->fence_seq = dev_priv->last_read_sequence; + atomic_set(&dev_priv->fence_seq, dev_priv->last_read_sequence); iowrite32(dev_priv->last_read_sequence, fifo_mem + SVGA_FIFO_FENCE); return vmw_fifo_send_fence(dev_priv, &dummy); @@ -283,7 +284,7 @@ void *vmw_fifo_reserve(struct vmw_private *dev_priv, uint32_t bytes) uint32_t reserveable = fifo_state->capabilities & SVGA_FIFO_CAP_RESERVE; int ret; - down_write(&fifo_state->rwsem); + mutex_lock(&fifo_state->fifo_mutex); max = ioread32(fifo_mem + SVGA_FIFO_MAX); min = ioread32(fifo_mem + SVGA_FIFO_MIN); next_cmd = ioread32(fifo_mem + SVGA_FIFO_NEXT_CMD); @@ -351,7 +352,7 @@ void *vmw_fifo_reserve(struct vmw_private *dev_priv, uint32_t bytes) } out_err: fifo_state->reserved_size = 0; - up_write(&fifo_state->rwsem); + mutex_unlock(&fifo_state->fifo_mutex); return NULL; } @@ -426,6 +427,7 @@ void vmw_fifo_commit(struct vmw_private *dev_priv, uint32_t bytes) } + down_write(&fifo_state->rwsem); if (fifo_state->using_bounce_buffer || reserveable) { next_cmd += bytes; if (next_cmd >= max) @@ -437,8 +439,9 @@ void vmw_fifo_commit(struct vmw_private *dev_priv, uint32_t bytes) if (reserveable) iowrite32(0, fifo_mem + SVGA_FIFO_RESERVED); mb(); - vmw_fifo_ping_host(dev_priv, SVGA_SYNC_GENERIC); up_write(&fifo_state->rwsem); + vmw_fifo_ping_host(dev_priv, SVGA_SYNC_GENERIC); + mutex_unlock(&fifo_state->fifo_mutex); } int vmw_fifo_send_fence(struct vmw_private *dev_priv, uint32_t *sequence) @@ -451,9 +454,7 @@ int vmw_fifo_send_fence(struct vmw_private *dev_priv, uint32_t *sequence) fm = vmw_fifo_reserve(dev_priv, bytes); if (unlikely(fm == NULL)) { - down_write(&fifo_state->rwsem); - *sequence = dev_priv->fence_seq; - up_write(&fifo_state->rwsem); + *sequence = atomic_read(&dev_priv->fence_seq); ret = -ENOMEM; (void)vmw_fallback_wait(dev_priv, false, true, *sequence, false, 3*HZ); @@ -461,7 +462,7 @@ int vmw_fifo_send_fence(struct vmw_private *dev_priv, uint32_t *sequence) } do { - *sequence = dev_priv->fence_seq++; + *sequence = atomic_add_return(1, &dev_priv->fence_seq); } while (*sequence == 0); if (!(fifo_state->capabilities & SVGA_FIFO_CAP_FENCE)) { diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c index 778851f9f1d6..1c7a316454d8 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c @@ -48,6 +48,12 @@ int vmw_getparam_ioctl(struct drm_device *dev, void *data, case DRM_VMW_PARAM_FIFO_OFFSET: param->value = dev_priv->mmio_start; break; + case DRM_VMW_PARAM_HW_CAPS: + param->value = dev_priv->capabilities; + break; + case DRM_VMW_PARAM_FIFO_CAPS: + param->value = dev_priv->fifo.capabilities; + break; default: DRM_ERROR("Illegal vmwgfx get param request: %d\n", param->param); diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c b/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c index d40086fc8647..4d7cb5393860 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c @@ -85,19 +85,12 @@ bool vmw_fence_signaled(struct vmw_private *dev_priv, return true; /** - * Below is to signal stale fences that have wrapped. - * First, block fence submission. - */ - - down_read(&fifo_state->rwsem); - - /** * Then check if the sequence is higher than what we've actually * emitted. Then the fence is stale and signaled. */ - ret = ((dev_priv->fence_seq - sequence) > VMW_FENCE_WRAP); - up_read(&fifo_state->rwsem); + ret = ((atomic_read(&dev_priv->fence_seq) - sequence) + > VMW_FENCE_WRAP); return ret; } @@ -127,7 +120,7 @@ int vmw_fallback_wait(struct vmw_private *dev_priv, if (fifo_idle) down_read(&fifo_state->rwsem); - signal_seq = dev_priv->fence_seq; + signal_seq = atomic_read(&dev_priv->fence_seq); ret = 0; for (;;) { diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c index eeba6d1d06e4..31f9afed0a63 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c @@ -769,10 +769,10 @@ int vmw_kms_init(struct vmw_private *dev_priv) drm_mode_config_init(dev); dev->mode_config.funcs = &vmw_kms_funcs; - dev->mode_config.min_width = 640; - dev->mode_config.min_height = 480; - dev->mode_config.max_width = 2048; - dev->mode_config.max_height = 2048; + dev->mode_config.min_width = 1; + dev->mode_config.min_height = 1; + dev->mode_config.max_width = dev_priv->fb_max_width; + dev->mode_config.max_height = dev_priv->fb_max_height; ret = vmw_kms_init_legacy_display_system(dev_priv); diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c index c7efbd47ab84..f8fbbc67a406 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c @@ -35,11 +35,6 @@ #define VMW_RES_SURFACE ttm_driver_type1 #define VMW_RES_STREAM ttm_driver_type2 -/* XXX: This isn't a real hardware flag, but just a hack for kernel to - * know about primary surfaces. Find a better way to accomplish this. - */ -#define SVGA3D_SURFACE_HINT_SCANOUT (1 << 9) - struct vmw_user_context { struct ttm_base_object base; struct vmw_resource res; @@ -579,6 +574,7 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data, srf->flags = req->flags; srf->format = req->format; + srf->scanout = req->scanout; memcpy(srf->mip_levels, req->mip_levels, sizeof(srf->mip_levels)); srf->num_sizes = 0; for (i = 0; i < DRM_VMW_MAX_SURFACE_FACES; ++i) @@ -604,16 +600,6 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data, if (unlikely(ret != 0)) goto out_err1; - if (srf->flags & SVGA3D_SURFACE_HINT_SCANOUT) { - /* we should not send this flag down to hardware since - * its not a official one - */ - srf->flags &= ~SVGA3D_SURFACE_HINT_SCANOUT; - srf->scanout = true; - } else { - srf->scanout = false; - } - if (srf->scanout && srf->num_sizes == 1 && srf->sizes[0].width == 64 && diff --git a/drivers/gpu/vga/vgaarb.c b/drivers/gpu/vga/vgaarb.c index 1ac0c93603c9..2f6cf69ecb39 100644 --- a/drivers/gpu/vga/vgaarb.c +++ b/drivers/gpu/vga/vgaarb.c @@ -961,7 +961,7 @@ static ssize_t vga_arb_write(struct file *file, const char __user * buf, remaining -= 7; pr_devel("client 0x%p called 'target'\n", priv); /* if target is default */ - if (!strncmp(buf, "default", 7)) + if (!strncmp(curr_pos, "default", 7)) pdev = pci_dev_get(vga_default_device()); else { if (!vga_pci_str_to_vars(curr_pos, remaining, |