From e3890d05b34266f5981876c65a6a97fc0d0d0ccb Mon Sep 17 00:00:00 2001 From: Rodrigo Vivi Date: Wed, 7 Feb 2018 23:32:19 -0800 Subject: drm/i915/cnl: Sync PCI ID with Spec. Add one missing PCI ID and sort them in a way that gets easier to review and compare against spec's table. When trying to sync libdrm and mesa id list with kernel and spec I noticed something was wrong and we were missing a pci id. So to make our lives easier when checking against spec let's simplify and sort like spec does. BSpec: 13621 Cc: Lucas De Marchi Cc: James Ausmus Signed-off-by: Rodrigo Vivi Reviewed-by: James Ausmus Link: https://patchwork.freedesktop.org/patch/msgid/20180208073219.27860-1-rodrigo.vivi@intel.com --- include/drm/i915_pciids.h | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'include/drm') diff --git a/include/drm/i915_pciids.h b/include/drm/i915_pciids.h index 9e1fe6634424..0b2ba46fa00b 100644 --- a/include/drm/i915_pciids.h +++ b/include/drm/i915_pciids.h @@ -416,18 +416,19 @@ /* CNL */ #define INTEL_CNL_IDS(info) \ - INTEL_VGA_DEVICE(0x5A52, info), \ - INTEL_VGA_DEVICE(0x5A5A, info), \ - INTEL_VGA_DEVICE(0x5A42, info), \ - INTEL_VGA_DEVICE(0x5A4A, info), \ INTEL_VGA_DEVICE(0x5A51, info), \ INTEL_VGA_DEVICE(0x5A59, info), \ INTEL_VGA_DEVICE(0x5A41, info), \ INTEL_VGA_DEVICE(0x5A49, info), \ - INTEL_VGA_DEVICE(0x5A71, info), \ - INTEL_VGA_DEVICE(0x5A79, info), \ + INTEL_VGA_DEVICE(0x5A52, info), \ + INTEL_VGA_DEVICE(0x5A5A, info), \ + INTEL_VGA_DEVICE(0x5A42, info), \ + INTEL_VGA_DEVICE(0x5A4A, info), \ + INTEL_VGA_DEVICE(0x5A50, info), \ + INTEL_VGA_DEVICE(0x5A40, info), \ INTEL_VGA_DEVICE(0x5A54, info), \ INTEL_VGA_DEVICE(0x5A5C, info), \ - INTEL_VGA_DEVICE(0x5A44, info) + INTEL_VGA_DEVICE(0x5A44, info), \ + INTEL_VGA_DEVICE(0x5A4C, info) #endif /* _I915_PCIIDS_H */ -- cgit v1.2.3 From 3b765c0b765d2cc03ef02276f1af2658a03b3ced Mon Sep 17 00:00:00 2001 From: Dhinakaran Pandiyan Date: Fri, 2 Feb 2018 21:12:53 -0800 Subject: drm/vblank: Data type fixes for 64-bit vblank sequences. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit drm_vblank_count() has an u32 type returning what is a 64-bit vblank count. The effect of this is when drm_wait_vblank_ioctl() tries to widen the user space requested vblank sequence using this clipped 32-bit count(when the value is >= 2^32) as reference, the requested sequence remains a 32-bit value and gets queued like that. However, the code that checks if the requested sequence has passed compares this against the 64-bit vblank count. With drm_vblank_count() returning all bits of the vblank count, update drm_crtc_accurate_vblank_count() so that drm_crtc_arm_vblank_event() queues the correct sequence. Otherwise, this leads to prolonged waits for a vblank sequence when the current count is >=2^32. Finally, fix drm_wait_one_vblank() too. v2: Commit message fix (Keith) Squash commits (Rodrigo) Fixes: 570e86963a51 ("drm: Widen vblank count to 64-bits [v3]") Cc: Keith Packard Cc: Michel Dänzer Cc: Daniel Vetter Cc: Rodrigo Vivi Signed-off-by: Dhinakaran Pandiyan Acked-by: Daniel Vetter Reviewed-by: Keith Packard Signed-off-by: Rodrigo Vivi Link: https://patchwork.freedesktop.org/patch/msgid/20180203051302.9974-1-dhinakaran.pandiyan@intel.com --- drivers/gpu/drm/drm_vblank.c | 8 ++++---- include/drm/drm_vblank.h | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'include/drm') diff --git a/drivers/gpu/drm/drm_vblank.c b/drivers/gpu/drm/drm_vblank.c index 32d9bcf5be7f..f0d3ed5f2528 100644 --- a/drivers/gpu/drm/drm_vblank.c +++ b/drivers/gpu/drm/drm_vblank.c @@ -271,7 +271,7 @@ static void drm_update_vblank_count(struct drm_device *dev, unsigned int pipe, store_vblank(dev, pipe, diff, t_vblank, cur_vblank); } -static u32 drm_vblank_count(struct drm_device *dev, unsigned int pipe) +static u64 drm_vblank_count(struct drm_device *dev, unsigned int pipe) { struct drm_vblank_crtc *vblank = &dev->vblank[pipe]; @@ -292,11 +292,11 @@ static u32 drm_vblank_count(struct drm_device *dev, unsigned int pipe) * This is mostly useful for hardware that can obtain the scanout position, but * doesn't have a hardware frame counter. */ -u32 drm_crtc_accurate_vblank_count(struct drm_crtc *crtc) +u64 drm_crtc_accurate_vblank_count(struct drm_crtc *crtc) { struct drm_device *dev = crtc->dev; unsigned int pipe = drm_crtc_index(crtc); - u32 vblank; + u64 vblank; unsigned long flags; WARN_ONCE(drm_debug & DRM_UT_VBL && !dev->driver->get_vblank_timestamp, @@ -1055,7 +1055,7 @@ void drm_wait_one_vblank(struct drm_device *dev, unsigned int pipe) { struct drm_vblank_crtc *vblank = &dev->vblank[pipe]; int ret; - u32 last; + u64 last; if (WARN_ON(pipe >= dev->num_crtcs)) return; diff --git a/include/drm/drm_vblank.h b/include/drm/drm_vblank.h index 848b463a0af5..a4c3b0a0a197 100644 --- a/include/drm/drm_vblank.h +++ b/include/drm/drm_vblank.h @@ -179,7 +179,7 @@ void drm_crtc_wait_one_vblank(struct drm_crtc *crtc); void drm_crtc_vblank_off(struct drm_crtc *crtc); void drm_crtc_vblank_reset(struct drm_crtc *crtc); void drm_crtc_vblank_on(struct drm_crtc *crtc); -u32 drm_crtc_accurate_vblank_count(struct drm_crtc *crtc); +u64 drm_crtc_accurate_vblank_count(struct drm_crtc *crtc); bool drm_calc_vbltimestamp_from_scanoutpos(struct drm_device *dev, unsigned int pipe, int *max_error, -- cgit v1.2.3 From f4c0468e4bd64b7cd00e2db309925728fc72bf8b Mon Sep 17 00:00:00 2001 From: Dhinakaran Pandiyan Date: Fri, 2 Feb 2018 21:12:59 -0800 Subject: drm/atomic: Handle 64-bit return from drm_crtc_vblank_count() 570e86963a51 ("drm: Widen vblank count to 64-bits [v3]") changed the return type for drm_crtc_vblank_count() to u64. The flip ioctl receives a 32-bit target sequence from user space and is compared against the current sequence from drm_crtc_vblank_count(). So, typecast return from drm_crtc_vblank_count() explicitly to add clarity. __drm_crtcs_state.last_vblank_count however only ever stores the value from drm_crtc_vblank_count() and can be upgraded to u64. Cc: Keith Packard Cc: Rodrigo Vivi Signed-off-by: Dhinakaran Pandiyan Reviewed-by: Keith Packard Signed-off-by: Rodrigo Vivi Link: https://patchwork.freedesktop.org/patch/msgid/20180203051302.9974-7-dhinakaran.pandiyan@intel.com --- drivers/gpu/drm/drm_plane.c | 2 +- include/drm/drm_atomic.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'include/drm') diff --git a/drivers/gpu/drm/drm_plane.c b/drivers/gpu/drm/drm_plane.c index 2c90519576a3..c3419ee09ff0 100644 --- a/drivers/gpu/drm/drm_plane.c +++ b/drivers/gpu/drm/drm_plane.c @@ -944,7 +944,7 @@ int drm_mode_page_flip_ioctl(struct drm_device *dev, if (r) return r; - current_vblank = drm_crtc_vblank_count(crtc); + current_vblank = (u32)drm_crtc_vblank_count(crtc); switch (page_flip->flags & DRM_MODE_PAGE_FLIP_TARGET) { case DRM_MODE_PAGE_FLIP_TARGET_ABSOLUTE: diff --git a/include/drm/drm_atomic.h b/include/drm/drm_atomic.h index 1c27526c499e..6649baa19b65 100644 --- a/include/drm/drm_atomic.h +++ b/include/drm/drm_atomic.h @@ -145,7 +145,7 @@ struct __drm_crtcs_state { struct drm_crtc *ptr; struct drm_crtc_state *state, *old_state, *new_state; s32 __user *out_fence_ptr; - unsigned last_vblank_count; + u64 last_vblank_count; }; struct __drm_connnectors_state { -- cgit v1.2.3 From d0bb96b4be69feea97f16de5306c35e280658931 Mon Sep 17 00:00:00 2001 From: Dhinakaran Pandiyan Date: Fri, 2 Feb 2018 21:13:01 -0800 Subject: drm/vblank: Restoring vblank counts after device PM events. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The HW frame counter can get reset if device enters a low power state after vblank interrupts were disabled. This messes up any following vblank count update as a negative diff (huge unsigned diff) is calculated from the HW frame counter change. We cannot ignore negative diffs altogther as there could be legitimate wrap arounds. So, allow drivers to update vblank->count with missed vblanks for the time interrupts were disabled. This is similar to _crtc_vblank_on() except that vblanks interrupts are not enabled at the end as this function is expected to be called from the driver _enable_vblank() vfunc. v2: drm_crtc_vblank_restore should take crtc as arg. (Chris) Add docs and sprinkle some asserts. Cc: Daniel Vetter Cc: Chris Wilson Cc: Michel Dänzer Signed-off-by: Dhinakaran Pandiyan Reviewed-by: Rodrigo Vivi Acked-by: Daniel Vetter Signed-off-by: Rodrigo Vivi Link: https://patchwork.freedesktop.org/patch/msgid/20180203051302.9974-9-dhinakaran.pandiyan@intel.com --- drivers/gpu/drm/drm_vblank.c | 59 ++++++++++++++++++++++++++++++++++++++++++++ include/drm/drm_vblank.h | 2 ++ 2 files changed, 61 insertions(+) (limited to 'include/drm') diff --git a/drivers/gpu/drm/drm_vblank.c b/drivers/gpu/drm/drm_vblank.c index 913954765d9e..c781cb426bf1 100644 --- a/drivers/gpu/drm/drm_vblank.c +++ b/drivers/gpu/drm/drm_vblank.c @@ -1237,6 +1237,65 @@ void drm_crtc_vblank_on(struct drm_crtc *crtc) } EXPORT_SYMBOL(drm_crtc_vblank_on); +/** + * drm_vblank_restore - estimated vblanks using timestamps and update it. + * + * Power manamement features can cause frame counter resets between vblank + * disable and enable. Drivers can then use this function in their + * &drm_crtc_funcs.enable_vblank implementation to estimate the vblanks since + * the last &drm_crtc_funcs.disable_vblank. + * + * This function is the legacy version of drm_crtc_vblank_restore(). + */ +void drm_vblank_restore(struct drm_device *dev, unsigned int pipe) +{ + ktime_t t_vblank; + struct drm_vblank_crtc *vblank; + int framedur_ns; + u64 diff_ns; + u32 cur_vblank, diff = 1; + int count = DRM_TIMESTAMP_MAXRETRIES; + + if (WARN_ON(pipe >= dev->num_crtcs)) + return; + + assert_spin_locked(&dev->vbl_lock); + assert_spin_locked(&dev->vblank_time_lock); + + vblank = &dev->vblank[pipe]; + WARN_ONCE((drm_debug & DRM_UT_VBL) && !vblank->framedur_ns, + "Cannot compute missed vblanks without frame duration\n"); + framedur_ns = vblank->framedur_ns; + + do { + cur_vblank = __get_vblank_counter(dev, pipe); + drm_get_last_vbltimestamp(dev, pipe, &t_vblank, false); + } while (cur_vblank != __get_vblank_counter(dev, pipe) && --count > 0); + + diff_ns = ktime_to_ns(ktime_sub(t_vblank, vblank->time)); + if (framedur_ns) + diff = DIV_ROUND_CLOSEST_ULL(diff_ns, framedur_ns); + + + DRM_DEBUG_VBL("missed %d vblanks in %lld ns, frame duration=%d ns, hw_diff=%d\n", + diff, diff_ns, framedur_ns, cur_vblank - vblank->last); + store_vblank(dev, pipe, diff, t_vblank, cur_vblank); +} +EXPORT_SYMBOL(drm_vblank_restore); + +/** + * drm_crtc_vblank_restore - estimate vblanks using timestamps and update it. + * Power manamement features can cause frame counter resets between vblank + * disable and enable. Drivers can then use this function in their + * &drm_crtc_funcs.enable_vblank implementation to estimate the vblanks since + * the last &drm_crtc_funcs.disable_vblank. + */ +void drm_crtc_vblank_restore(struct drm_crtc *crtc) +{ + drm_vblank_restore(crtc->dev, drm_crtc_index(crtc)); +} +EXPORT_SYMBOL(drm_crtc_vblank_restore); + static void drm_legacy_vblank_pre_modeset(struct drm_device *dev, unsigned int pipe) { diff --git a/include/drm/drm_vblank.h b/include/drm/drm_vblank.h index a4c3b0a0a197..16d46e2a6854 100644 --- a/include/drm/drm_vblank.h +++ b/include/drm/drm_vblank.h @@ -180,6 +180,8 @@ void drm_crtc_vblank_off(struct drm_crtc *crtc); void drm_crtc_vblank_reset(struct drm_crtc *crtc); void drm_crtc_vblank_on(struct drm_crtc *crtc); u64 drm_crtc_accurate_vblank_count(struct drm_crtc *crtc); +void drm_vblank_restore(struct drm_device *dev, unsigned int pipe); +void drm_crtc_vblank_restore(struct drm_crtc *crtc); bool drm_calc_vbltimestamp_from_scanoutpos(struct drm_device *dev, unsigned int pipe, int *max_error, -- cgit v1.2.3 From c0a51fd07b1dd50f31a413f0e7bb5e4499de2042 Mon Sep 17 00:00:00 2001 From: Christian König Date: Fri, 16 Feb 2018 13:43:38 +0100 Subject: drm: move read_domains and write_domain into i915 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit i915 is the only driver using those fields in the drm_gem_object structure, so they only waste memory for all other drivers. Move the fields into drm_i915_gem_object instead and patch the i915 code with the following sed commands: sed -i "s/obj->base.read_domains/obj->read_domains/g" drivers/gpu/drm/i915/*.c drivers/gpu/drm/i915/*/*.c sed -i "s/obj->base.write_domain/obj->write_domain/g" drivers/gpu/drm/i915/*.c drivers/gpu/drm/i915/*/*.c Change is only compile tested. v2: move fields around as suggested by Chris. Signed-off-by: Christian König Reviewed-by: Chris Wilson Link: https://patchwork.freedesktop.org/patch/msgid/20180216124338.9087-1-christian.koenig@amd.com Signed-off-by: Chris Wilson --- drivers/gpu/drm/i915/gvt/dmabuf.c | 4 +- drivers/gpu/drm/i915/i915_debugfs.c | 4 +- drivers/gpu/drm/i915/i915_gem.c | 60 +++++++++++------------ drivers/gpu/drm/i915/i915_gem_clflush.c | 2 +- drivers/gpu/drm/i915/i915_gem_dmabuf.c | 4 +- drivers/gpu/drm/i915/i915_gem_execbuffer.c | 10 ++-- drivers/gpu/drm/i915/i915_gem_internal.c | 4 +- drivers/gpu/drm/i915/i915_gem_object.h | 15 ++++++ drivers/gpu/drm/i915/i915_gem_stolen.c | 2 +- drivers/gpu/drm/i915/i915_gem_userptr.c | 4 +- drivers/gpu/drm/i915/i915_gpu_error.c | 4 +- drivers/gpu/drm/i915/selftests/huge_gem_object.c | 4 +- drivers/gpu/drm/i915/selftests/huge_pages.c | 8 +-- drivers/gpu/drm/i915/selftests/i915_gem_context.c | 4 +- drivers/gpu/drm/i915/selftests/i915_gem_gtt.c | 4 +- include/drm/drm_gem.h | 15 ------ 16 files changed, 74 insertions(+), 74 deletions(-) (limited to 'include/drm') diff --git a/drivers/gpu/drm/i915/gvt/dmabuf.c b/drivers/gpu/drm/i915/gvt/dmabuf.c index 2ab584f97dfb..de3e076dcb31 100644 --- a/drivers/gpu/drm/i915/gvt/dmabuf.c +++ b/drivers/gpu/drm/i915/gvt/dmabuf.c @@ -162,8 +162,8 @@ static struct drm_i915_gem_object *vgpu_create_gem(struct drm_device *dev, info->size << PAGE_SHIFT); i915_gem_object_init(obj, &intel_vgpu_gem_ops); - obj->base.read_domains = I915_GEM_DOMAIN_GTT; - obj->base.write_domain = 0; + obj->read_domains = I915_GEM_DOMAIN_GTT; + obj->write_domain = 0; if (IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv)) { unsigned int tiling_mode = 0; unsigned int stride = 0; diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c index 3560eb567ca7..0cbe154e517d 100644 --- a/drivers/gpu/drm/i915/i915_debugfs.c +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -150,8 +150,8 @@ describe_obj(struct seq_file *m, struct drm_i915_gem_object *obj) get_global_flag(obj), get_pin_mapped_flag(obj), obj->base.size / 1024, - obj->base.read_domains, - obj->base.write_domain, + obj->read_domains, + obj->write_domain, i915_cache_level_str(dev_priv, obj->cache_level), obj->mm.dirty ? " dirty" : "", obj->mm.madv == I915_MADV_DONTNEED ? " purgeable" : ""); diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index fc68b35854df..f530cd247724 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -240,8 +240,8 @@ err_phys: static void __start_cpu_write(struct drm_i915_gem_object *obj) { - obj->base.read_domains = I915_GEM_DOMAIN_CPU; - obj->base.write_domain = I915_GEM_DOMAIN_CPU; + obj->read_domains = I915_GEM_DOMAIN_CPU; + obj->write_domain = I915_GEM_DOMAIN_CPU; if (cpu_write_needs_clflush(obj)) obj->cache_dirty = true; } @@ -257,7 +257,7 @@ __i915_gem_object_release_shmem(struct drm_i915_gem_object *obj, obj->mm.dirty = false; if (needs_clflush && - (obj->base.read_domains & I915_GEM_DOMAIN_CPU) == 0 && + (obj->read_domains & I915_GEM_DOMAIN_CPU) == 0 && !(obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_READ)) drm_clflush_sg(pages); @@ -703,10 +703,10 @@ flush_write_domain(struct drm_i915_gem_object *obj, unsigned int flush_domains) struct drm_i915_private *dev_priv = to_i915(obj->base.dev); struct i915_vma *vma; - if (!(obj->base.write_domain & flush_domains)) + if (!(obj->write_domain & flush_domains)) return; - switch (obj->base.write_domain) { + switch (obj->write_domain) { case I915_GEM_DOMAIN_GTT: i915_gem_flush_ggtt_writes(dev_priv); @@ -731,7 +731,7 @@ flush_write_domain(struct drm_i915_gem_object *obj, unsigned int flush_domains) break; } - obj->base.write_domain = 0; + obj->write_domain = 0; } static inline int @@ -831,7 +831,7 @@ int i915_gem_obj_prepare_shmem_read(struct drm_i915_gem_object *obj, * anyway again before the next pread happens. */ if (!obj->cache_dirty && - !(obj->base.read_domains & I915_GEM_DOMAIN_CPU)) + !(obj->read_domains & I915_GEM_DOMAIN_CPU)) *needs_clflush = CLFLUSH_BEFORE; out: @@ -890,7 +890,7 @@ int i915_gem_obj_prepare_shmem_write(struct drm_i915_gem_object *obj, * Same trick applies to invalidate partially written * cachelines read before writing. */ - if (!(obj->base.read_domains & I915_GEM_DOMAIN_CPU)) + if (!(obj->read_domains & I915_GEM_DOMAIN_CPU)) *needs_clflush |= CLFLUSH_BEFORE; } @@ -2391,8 +2391,8 @@ static int i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj) * wasn't in the GTT, there shouldn't be any way it could have been in * a GPU cache */ - GEM_BUG_ON(obj->base.read_domains & I915_GEM_GPU_DOMAINS); - GEM_BUG_ON(obj->base.write_domain & I915_GEM_GPU_DOMAINS); + GEM_BUG_ON(obj->read_domains & I915_GEM_GPU_DOMAINS); + GEM_BUG_ON(obj->write_domain & I915_GEM_GPU_DOMAINS); st = kmalloc(sizeof(*st), GFP_KERNEL); if (st == NULL) @@ -3703,7 +3703,7 @@ static void __i915_gem_object_flush_for_display(struct drm_i915_gem_object *obj) flush_write_domain(obj, ~I915_GEM_DOMAIN_CPU); if (obj->cache_dirty) i915_gem_clflush_object(obj, I915_CLFLUSH_FORCE); - obj->base.write_domain = 0; + obj->write_domain = 0; } void i915_gem_object_flush_if_display(struct drm_i915_gem_object *obj) @@ -3740,7 +3740,7 @@ i915_gem_object_set_to_wc_domain(struct drm_i915_gem_object *obj, bool write) if (ret) return ret; - if (obj->base.write_domain == I915_GEM_DOMAIN_WC) + if (obj->write_domain == I915_GEM_DOMAIN_WC) return 0; /* Flush and acquire obj->pages so that we are coherent through @@ -3761,17 +3761,17 @@ i915_gem_object_set_to_wc_domain(struct drm_i915_gem_object *obj, bool write) * coherent writes from the GPU, by effectively invalidating the * WC domain upon first access. */ - if ((obj->base.read_domains & I915_GEM_DOMAIN_WC) == 0) + if ((obj->read_domains & I915_GEM_DOMAIN_WC) == 0) mb(); /* It should now be out of any other write domains, and we can update * the domain values for our changes. */ - GEM_BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_WC) != 0); - obj->base.read_domains |= I915_GEM_DOMAIN_WC; + GEM_BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_WC) != 0); + obj->read_domains |= I915_GEM_DOMAIN_WC; if (write) { - obj->base.read_domains = I915_GEM_DOMAIN_WC; - obj->base.write_domain = I915_GEM_DOMAIN_WC; + obj->read_domains = I915_GEM_DOMAIN_WC; + obj->write_domain = I915_GEM_DOMAIN_WC; obj->mm.dirty = true; } @@ -3803,7 +3803,7 @@ i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, bool write) if (ret) return ret; - if (obj->base.write_domain == I915_GEM_DOMAIN_GTT) + if (obj->write_domain == I915_GEM_DOMAIN_GTT) return 0; /* Flush and acquire obj->pages so that we are coherent through @@ -3824,17 +3824,17 @@ i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, bool write) * coherent writes from the GPU, by effectively invalidating the * GTT domain upon first access. */ - if ((obj->base.read_domains & I915_GEM_DOMAIN_GTT) == 0) + if ((obj->read_domains & I915_GEM_DOMAIN_GTT) == 0) mb(); /* It should now be out of any other write domains, and we can update * the domain values for our changes. */ - GEM_BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_GTT) != 0); - obj->base.read_domains |= I915_GEM_DOMAIN_GTT; + GEM_BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_GTT) != 0); + obj->read_domains |= I915_GEM_DOMAIN_GTT; if (write) { - obj->base.read_domains = I915_GEM_DOMAIN_GTT; - obj->base.write_domain = I915_GEM_DOMAIN_GTT; + obj->read_domains = I915_GEM_DOMAIN_GTT; + obj->write_domain = I915_GEM_DOMAIN_GTT; obj->mm.dirty = true; } @@ -4146,7 +4146,7 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj, /* It should now be out of any other write domains, and we can update * the domain values for our changes. */ - obj->base.read_domains |= I915_GEM_DOMAIN_GTT; + obj->read_domains |= I915_GEM_DOMAIN_GTT; return vma; @@ -4199,15 +4199,15 @@ i915_gem_object_set_to_cpu_domain(struct drm_i915_gem_object *obj, bool write) flush_write_domain(obj, ~I915_GEM_DOMAIN_CPU); /* Flush the CPU cache if it's still invalid. */ - if ((obj->base.read_domains & I915_GEM_DOMAIN_CPU) == 0) { + if ((obj->read_domains & I915_GEM_DOMAIN_CPU) == 0) { i915_gem_clflush_object(obj, I915_CLFLUSH_SYNC); - obj->base.read_domains |= I915_GEM_DOMAIN_CPU; + obj->read_domains |= I915_GEM_DOMAIN_CPU; } /* It should now be out of any other write domains, and we can update * the domain values for our changes. */ - GEM_BUG_ON(obj->base.write_domain & ~I915_GEM_DOMAIN_CPU); + GEM_BUG_ON(obj->write_domain & ~I915_GEM_DOMAIN_CPU); /* If we're writing through the CPU, then the GPU read domains will * need to be invalidated at next use. @@ -4643,8 +4643,8 @@ i915_gem_object_create(struct drm_i915_private *dev_priv, u64 size) i915_gem_object_init(obj, &i915_gem_object_ops); - obj->base.write_domain = I915_GEM_DOMAIN_CPU; - obj->base.read_domains = I915_GEM_DOMAIN_CPU; + obj->write_domain = I915_GEM_DOMAIN_CPU; + obj->read_domains = I915_GEM_DOMAIN_CPU; if (HAS_LLC(dev_priv)) /* On some devices, we can have the GPU use the LLC (the CPU @@ -5702,7 +5702,7 @@ i915_gem_object_create_from_data(struct drm_i915_private *dev_priv, if (IS_ERR(obj)) return obj; - GEM_BUG_ON(obj->base.write_domain != I915_GEM_DOMAIN_CPU); + GEM_BUG_ON(obj->write_domain != I915_GEM_DOMAIN_CPU); file = obj->base.filp; offset = 0; diff --git a/drivers/gpu/drm/i915/i915_gem_clflush.c b/drivers/gpu/drm/i915/i915_gem_clflush.c index b9b53ac14176..f5c570d35b2a 100644 --- a/drivers/gpu/drm/i915/i915_gem_clflush.c +++ b/drivers/gpu/drm/i915/i915_gem_clflush.c @@ -177,7 +177,7 @@ bool i915_gem_clflush_object(struct drm_i915_gem_object *obj, } else if (obj->mm.pages) { __i915_do_clflush(obj); } else { - GEM_BUG_ON(obj->base.write_domain != I915_GEM_DOMAIN_CPU); + GEM_BUG_ON(obj->write_domain != I915_GEM_DOMAIN_CPU); } obj->cache_dirty = false; diff --git a/drivers/gpu/drm/i915/i915_gem_dmabuf.c b/drivers/gpu/drm/i915/i915_gem_dmabuf.c index 864439a214c8..69a7aec49e84 100644 --- a/drivers/gpu/drm/i915/i915_gem_dmabuf.c +++ b/drivers/gpu/drm/i915/i915_gem_dmabuf.c @@ -330,8 +330,8 @@ struct drm_gem_object *i915_gem_prime_import(struct drm_device *dev, * write-combined buffer or a delay through the chipset for GTT * writes that do require us to treat GTT as a separate cache domain.) */ - obj->base.read_domains = I915_GEM_DOMAIN_GTT; - obj->base.write_domain = 0; + obj->read_domains = I915_GEM_DOMAIN_GTT; + obj->write_domain = 0; return &obj->base; diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c index ed6e9db51e67..51f3c32c64bf 100644 --- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c @@ -1073,7 +1073,7 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, u32 *cmd; int err; - GEM_BUG_ON(vma->obj->base.write_domain & I915_GEM_DOMAIN_CPU); + GEM_BUG_ON(vma->obj->write_domain & I915_GEM_DOMAIN_CPU); obj = i915_gem_batch_pool_get(&eb->engine->batch_pool, PAGE_SIZE); if (IS_ERR(obj)) @@ -1861,16 +1861,16 @@ void i915_vma_move_to_active(struct i915_vma *vma, i915_gem_active_set(&vma->last_read[idx], req); list_move_tail(&vma->vm_link, &vma->vm->active_list); - obj->base.write_domain = 0; + obj->write_domain = 0; if (flags & EXEC_OBJECT_WRITE) { - obj->base.write_domain = I915_GEM_DOMAIN_RENDER; + obj->write_domain = I915_GEM_DOMAIN_RENDER; if (intel_fb_obj_invalidate(obj, ORIGIN_CS)) i915_gem_active_set(&obj->frontbuffer_write, req); - obj->base.read_domains = 0; + obj->read_domains = 0; } - obj->base.read_domains |= I915_GEM_GPU_DOMAINS; + obj->read_domains |= I915_GEM_GPU_DOMAINS; if (flags & EXEC_OBJECT_NEEDS_FENCE) i915_gem_active_set(&vma->last_fence, req); diff --git a/drivers/gpu/drm/i915/i915_gem_internal.c b/drivers/gpu/drm/i915/i915_gem_internal.c index 8301c06c952f..0d0144b2104c 100644 --- a/drivers/gpu/drm/i915/i915_gem_internal.c +++ b/drivers/gpu/drm/i915/i915_gem_internal.c @@ -201,8 +201,8 @@ i915_gem_object_create_internal(struct drm_i915_private *i915, drm_gem_private_object_init(&i915->drm, &obj->base, size); i915_gem_object_init(obj, &i915_gem_object_internal_ops); - obj->base.read_domains = I915_GEM_DOMAIN_CPU; - obj->base.write_domain = I915_GEM_DOMAIN_CPU; + obj->read_domains = I915_GEM_DOMAIN_CPU; + obj->write_domain = I915_GEM_DOMAIN_CPU; cache_level = HAS_LLC(i915) ? I915_CACHE_LLC : I915_CACHE_NONE; i915_gem_object_set_cache_coherency(obj, cache_level); diff --git a/drivers/gpu/drm/i915/i915_gem_object.h b/drivers/gpu/drm/i915/i915_gem_object.h index 05e89e1c0a08..ca2b3b62569d 100644 --- a/drivers/gpu/drm/i915/i915_gem_object.h +++ b/drivers/gpu/drm/i915/i915_gem_object.h @@ -148,6 +148,21 @@ struct drm_i915_gem_object { #define I915_BO_CACHE_COHERENT_FOR_WRITE BIT(1) unsigned int cache_dirty:1; + /** + * @read_domains: Read memory domains. + * + * These monitor which caches contain read/write data related to the + * object. When transitioning from one set of domains to another, + * the driver is called to ensure that caches are suitably flushed and + * invalidated. + */ + u16 read_domains; + + /** + * @write_domain: Corresponding unique write memory domain. + */ + u16 write_domain; + atomic_t frontbuffer_bits; unsigned int frontbuffer_ggtt_origin; /* write once */ struct i915_gem_active frontbuffer_write; diff --git a/drivers/gpu/drm/i915/i915_gem_stolen.c b/drivers/gpu/drm/i915/i915_gem_stolen.c index f18da9e2be8e..62aa67960bf4 100644 --- a/drivers/gpu/drm/i915/i915_gem_stolen.c +++ b/drivers/gpu/drm/i915/i915_gem_stolen.c @@ -516,7 +516,7 @@ _i915_gem_object_create_stolen(struct drm_i915_private *dev_priv, i915_gem_object_init(obj, &i915_gem_object_stolen_ops); obj->stolen = stolen; - obj->base.read_domains = I915_GEM_DOMAIN_CPU | I915_GEM_DOMAIN_GTT; + obj->read_domains = I915_GEM_DOMAIN_CPU | I915_GEM_DOMAIN_GTT; cache_level = HAS_LLC(dev_priv) ? I915_CACHE_LLC : I915_CACHE_NONE; i915_gem_object_set_cache_coherency(obj, cache_level); diff --git a/drivers/gpu/drm/i915/i915_gem_userptr.c b/drivers/gpu/drm/i915/i915_gem_userptr.c index 1f9d24021cbb..d596a8302ca3 100644 --- a/drivers/gpu/drm/i915/i915_gem_userptr.c +++ b/drivers/gpu/drm/i915/i915_gem_userptr.c @@ -798,8 +798,8 @@ i915_gem_userptr_ioctl(struct drm_device *dev, drm_gem_private_object_init(dev, &obj->base, args->user_size); i915_gem_object_init(obj, &i915_gem_userptr_ops); - obj->base.read_domains = I915_GEM_DOMAIN_CPU; - obj->base.write_domain = I915_GEM_DOMAIN_CPU; + obj->read_domains = I915_GEM_DOMAIN_CPU; + obj->write_domain = I915_GEM_DOMAIN_CPU; i915_gem_object_set_cache_coherency(obj, I915_CACHE_LLC); obj->userptr.ptr = args->user_ptr; diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c index 161d9103a65e..65c0bef73ee5 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.c +++ b/drivers/gpu/drm/i915/i915_gpu_error.c @@ -1021,8 +1021,8 @@ static void capture_bo(struct drm_i915_error_buffer *err, err->engine = __active_get_engine_id(&obj->frontbuffer_write); err->gtt_offset = vma->node.start; - err->read_domains = obj->base.read_domains; - err->write_domain = obj->base.write_domain; + err->read_domains = obj->read_domains; + err->write_domain = obj->write_domain; err->fence_reg = vma->fence ? vma->fence->id : -1; err->tiling = i915_gem_object_get_tiling(obj); err->dirty = obj->mm.dirty; diff --git a/drivers/gpu/drm/i915/selftests/huge_gem_object.c b/drivers/gpu/drm/i915/selftests/huge_gem_object.c index a2632df39173..391f3d9ffdf1 100644 --- a/drivers/gpu/drm/i915/selftests/huge_gem_object.c +++ b/drivers/gpu/drm/i915/selftests/huge_gem_object.c @@ -129,8 +129,8 @@ huge_gem_object(struct drm_i915_private *i915, drm_gem_private_object_init(&i915->drm, &obj->base, dma_size); i915_gem_object_init(obj, &huge_ops); - obj->base.read_domains = I915_GEM_DOMAIN_CPU; - obj->base.write_domain = I915_GEM_DOMAIN_CPU; + obj->read_domains = I915_GEM_DOMAIN_CPU; + obj->write_domain = I915_GEM_DOMAIN_CPU; cache_level = HAS_LLC(i915) ? I915_CACHE_LLC : I915_CACHE_NONE; i915_gem_object_set_cache_coherency(obj, cache_level); obj->scratch = phys_size; diff --git a/drivers/gpu/drm/i915/selftests/huge_pages.c b/drivers/gpu/drm/i915/selftests/huge_pages.c index 2ea69394f428..52b1bd17bf46 100644 --- a/drivers/gpu/drm/i915/selftests/huge_pages.c +++ b/drivers/gpu/drm/i915/selftests/huge_pages.c @@ -178,8 +178,8 @@ huge_pages_object(struct drm_i915_private *i915, drm_gem_private_object_init(&i915->drm, &obj->base, size); i915_gem_object_init(obj, &huge_page_ops); - obj->base.write_domain = I915_GEM_DOMAIN_CPU; - obj->base.read_domains = I915_GEM_DOMAIN_CPU; + obj->write_domain = I915_GEM_DOMAIN_CPU; + obj->read_domains = I915_GEM_DOMAIN_CPU; obj->cache_level = I915_CACHE_NONE; obj->mm.page_mask = page_mask; @@ -329,8 +329,8 @@ fake_huge_pages_object(struct drm_i915_private *i915, u64 size, bool single) else i915_gem_object_init(obj, &fake_ops); - obj->base.write_domain = I915_GEM_DOMAIN_CPU; - obj->base.read_domains = I915_GEM_DOMAIN_CPU; + obj->write_domain = I915_GEM_DOMAIN_CPU; + obj->read_domains = I915_GEM_DOMAIN_CPU; obj->cache_level = I915_CACHE_NONE; return obj; diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_context.c b/drivers/gpu/drm/i915/selftests/i915_gem_context.c index 56a803d11916..6da2a2f29c54 100644 --- a/drivers/gpu/drm/i915/selftests/i915_gem_context.c +++ b/drivers/gpu/drm/i915/selftests/i915_gem_context.c @@ -215,8 +215,8 @@ static int cpu_fill(struct drm_i915_gem_object *obj, u32 value) } i915_gem_obj_finish_shmem_access(obj); - obj->base.read_domains = I915_GEM_DOMAIN_GTT | I915_GEM_DOMAIN_CPU; - obj->base.write_domain = 0; + obj->read_domains = I915_GEM_DOMAIN_GTT | I915_GEM_DOMAIN_CPU; + obj->write_domain = 0; return 0; } diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c index 89b6ca9b14a7..f7dc926f4ef1 100644 --- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c @@ -113,8 +113,8 @@ fake_dma_object(struct drm_i915_private *i915, u64 size) drm_gem_private_object_init(&i915->drm, &obj->base, size); i915_gem_object_init(obj, &fake_ops); - obj->base.write_domain = I915_GEM_DOMAIN_CPU; - obj->base.read_domains = I915_GEM_DOMAIN_CPU; + obj->write_domain = I915_GEM_DOMAIN_CPU; + obj->read_domains = I915_GEM_DOMAIN_CPU; obj->cache_level = I915_CACHE_NONE; /* Preallocate the "backing storage" */ diff --git a/include/drm/drm_gem.h b/include/drm/drm_gem.h index 9c55c2acaa2b..3583b98a1718 100644 --- a/include/drm/drm_gem.h +++ b/include/drm/drm_gem.h @@ -115,21 +115,6 @@ struct drm_gem_object { */ int name; - /** - * @read_domains: - * - * Read memory domains. These monitor which caches contain read/write data - * related to the object. When transitioning from one set of domains - * to another, the driver is called to ensure that caches are suitably - * flushed and invalidated. - */ - uint32_t read_domains; - - /** - * @write_domain: Corresponding unique write memory domain. - */ - uint32_t write_domain; - /** * @dma_buf: * -- cgit v1.2.3