sparc: Revert generic IOMMU allocator.

I applied the wrong version of this patch series, V4 instead
of V10, due to a patchwork bundling snafu.

Signed-off-by: David S. Miller <davem@davemloft.net>

5 files changed, 324 insertions, 257 deletions

diff --git a/arch/sparc/include/asm/iommu_64.h b/arch/sparc/include/asm/iommu_64.h
index e3cd4493d81d..2b9321ab064d 100644
--- a/arch/sparc/include/asm/iommu_64.h
+++ b/arch/sparc/include/asm/iommu_64.h
@@ -16,7 +16,6 @@
 #define IOPTE_WRITE   0x0000000000000002UL
 
 #define IOMMU_NUM_CTXS	4096
-#include <linux/iommu-common.h>
 
 struct iommu_arena {
 	unsigned long	*map;
@@ -25,10 +24,11 @@ struct iommu_arena {
 };
 
 struct iommu {
-	struct iommu_table	tbl;
 	spinlock_t		lock;
-	u32			dma_addr_mask;
+	struct iommu_arena	arena;
+	void			(*flush_all)(struct iommu *);
 	iopte_t			*page_table;
+	u32			page_table_map_base;
 	unsigned long		iommu_control;
 	unsigned long		iommu_tsbbase;
 	unsigned long		iommu_flush;
@@ -40,6 +40,7 @@ struct iommu {
 	unsigned long		dummy_page_pa;
 	unsigned long		ctx_lowest_free;
 	DECLARE_BITMAP(ctx_bitmap, IOMMU_NUM_CTXS);
+	u32			dma_addr_mask;
 };
 
 struct strbuf {
diff --git a/arch/sparc/kernel/iommu.c b/arch/sparc/kernel/iommu.c
index 9b16b341b6ae..bfa4d0c2df42 100644
--- a/arch/sparc/kernel/iommu.c
+++ b/arch/sparc/kernel/iommu.c
@@ -13,15 +13,11 @@
 #include <linux/errno.h>
 #include <linux/iommu-helper.h>
 #include <linux/bitmap.h>
-#include <linux/hash.h>
-#include <linux/iommu-common.h>
 
 #ifdef CONFIG_PCI
 #include <linux/pci.h>
 #endif
 
-static	DEFINE_PER_CPU(unsigned int, iommu_pool_hash);
-
 #include <asm/iommu.h>
 
 #include "iommu_common.h"
@@ -49,9 +45,8 @@ static	DEFINE_PER_CPU(unsigned int, iommu_pool_hash);
 			       "i" (ASI_PHYS_BYPASS_EC_E))
 
 /* Must be invoked under the IOMMU lock. */
-static void iommu_flushall(struct iommu_table *iommu_table)
+static void iommu_flushall(struct iommu *iommu)
 {
-	struct iommu *iommu = container_of(iommu_table, struct iommu, tbl);
 	if (iommu->iommu_flushinv) {
 		iommu_write(iommu->iommu_flushinv, ~(u64)0);
 	} else {
@@ -92,22 +87,93 @@ static inline void iopte_make_dummy(struct iommu *iommu, iopte_t *iopte)
 	iopte_val(*iopte) = val;
 }
 
-static struct iommu_tbl_ops iommu_sparc_ops = {
-	.reset	= iommu_flushall
-};
-
-static void setup_iommu_pool_hash(void)
+/* Based almost entirely upon the ppc64 iommu allocator.  If you use the 'handle'
+ * facility it must all be done in one pass while under the iommu lock.
+ *
+ * On sun4u platforms, we only flush the IOMMU once every time we've passed
+ * over the entire page table doing allocations.  Therefore we only ever advance
+ * the hint and cannot backtrack it.
+ */
+unsigned long iommu_range_alloc(struct device *dev,
+				struct iommu *iommu,
+				unsigned long npages,
+				unsigned long *handle)
 {
-	unsigned int i;
-	static bool do_once;
+	unsigned long n, end, start, limit, boundary_size;
+	struct iommu_arena *arena = &iommu->arena;
+	int pass = 0;
 
-	if (do_once)
-		return;
-	do_once = true;
-	for_each_possible_cpu(i)
-		per_cpu(iommu_pool_hash, i) = hash_32(i, IOMMU_POOL_HASHBITS);
+	/* This allocator was derived from x86_64's bit string search */
+
+	/* Sanity check */
+	if (unlikely(npages == 0)) {
+		if (printk_ratelimit())
+			WARN_ON(1);
+		return DMA_ERROR_CODE;
+	}
+
+	if (handle && *handle)
+		start = *handle;
+	else
+		start = arena->hint;
+
+	limit = arena->limit;
+
+	/* The case below can happen if we have a small segment appended
+	 * to a large, or when the previous alloc was at the very end of
+	 * the available space. If so, go back to the beginning and flush.
+	 */
+	if (start >= limit) {
+		start = 0;
+		if (iommu->flush_all)
+			iommu->flush_all(iommu);
+	}
+
+ again:
+
+	if (dev)
+		boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
+				      1 << IO_PAGE_SHIFT);
+	else
+		boundary_size = ALIGN(1UL << 32, 1 << IO_PAGE_SHIFT);
+
+	n = iommu_area_alloc(arena->map, limit, start, npages,
+			     iommu->page_table_map_base >> IO_PAGE_SHIFT,
+			     boundary_size >> IO_PAGE_SHIFT, 0);
+	if (n == -1) {
+		if (likely(pass < 1)) {
+			/* First failure, rescan from the beginning.  */
+			start = 0;
+			if (iommu->flush_all)
+				iommu->flush_all(iommu);
+			pass++;
+			goto again;
+		} else {
+			/* Second failure, give up */
+			return DMA_ERROR_CODE;
+		}
+	}
+
+	end = n + npages;
+
+	arena->hint = end;
+
+	/* Update handle for SG allocations */
+	if (handle)
+		*handle = end;
+
+	return n;
 }
 
+void iommu_range_free(struct iommu *iommu, dma_addr_t dma_addr, unsigned long npages)
+{
+	struct iommu_arena *arena = &iommu->arena;
+	unsigned long entry;
+
+	entry = (dma_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT;
+
+	bitmap_clear(arena->map, entry, npages);
+}
 
 int iommu_table_init(struct iommu *iommu, int tsbsize,
 		     u32 dma_offset, u32 dma_addr_mask,
@@ -121,22 +187,22 @@ int iommu_table_init(struct iommu *iommu, int tsbsize,
 	/* Setup initial software IOMMU state. */
 	spin_lock_init(&iommu->lock);
 	iommu->ctx_lowest_free = 1;
-	iommu->tbl.page_table_map_base = dma_offset;
+	iommu->page_table_map_base = dma_offset;
 	iommu->dma_addr_mask = dma_addr_mask;
 
 	/* Allocate and initialize the free area map.  */
 	sz = num_tsb_entries / 8;
 	sz = (sz + 7UL) & ~7UL;
-	iommu->tbl.map = kmalloc_node(sz, GFP_KERNEL, numa_node);
-	if (!iommu->tbl.map)
+	iommu->arena.map = kmalloc_node(sz, GFP_KERNEL, numa_node);
+	if (!iommu->arena.map) {
+		printk(KERN_ERR "IOMMU: Error, kmalloc(arena.map) failed.\n");
 		return -ENOMEM;
-	memset(iommu->tbl.map, 0, sz);
-	if (tlb_type != hypervisor)
-		iommu_sparc_ops.reset = NULL; /* not needed on on sun4v */
+	}
+	memset(iommu->arena.map, 0, sz);
+	iommu->arena.limit = num_tsb_entries;
 
-	setup_iommu_pool_hash();
-	iommu_tbl_pool_init(&iommu->tbl, num_tsb_entries, IO_PAGE_SHIFT,
-			    &iommu_sparc_ops, false, 1);
+	if (tlb_type != hypervisor)
+		iommu->flush_all = iommu_flushall;
 
 	/* Allocate and initialize the dummy page which we
 	 * set inactive IO PTEs to point to.
@@ -169,20 +235,18 @@ out_free_dummy_page:
 	iommu->dummy_page = 0UL;
 
 out_free_map:
-	kfree(iommu->tbl.map);
-	iommu->tbl.map = NULL;
+	kfree(iommu->arena.map);
+	iommu->arena.map = NULL;
 
 	return -ENOMEM;
 }
 
-static inline iopte_t *alloc_npages(struct device *dev,
-				    struct iommu *iommu,
+static inline iopte_t *alloc_npages(struct device *dev, struct iommu *iommu,
 				    unsigned long npages)
 {
 	unsigned long entry;
 
-	entry = iommu_tbl_range_alloc(dev, &iommu->tbl, npages, NULL,
-				      __this_cpu_read(iommu_pool_hash));
+	entry = iommu_range_alloc(dev, iommu, npages, NULL);
 	if (unlikely(entry == DMA_ERROR_CODE))
 		return NULL;
 
@@ -220,7 +284,7 @@ static void *dma_4u_alloc_coherent(struct device *dev, size_t size,
 				   dma_addr_t *dma_addrp, gfp_t gfp,
 				   struct dma_attrs *attrs)
 {
-	unsigned long order, first_page;
+	unsigned long flags, order, first_page;
 	struct iommu *iommu;
 	struct page *page;
 	int npages, nid;
@@ -242,14 +306,16 @@ static void *dma_4u_alloc_coherent(struct device *dev, size_t size,
 
 	iommu = dev->archdata.iommu;
 
+	spin_lock_irqsave(&iommu->lock, flags);
 	iopte = alloc_npages(dev, iommu, size >> IO_PAGE_SHIFT);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	if (unlikely(iopte == NULL)) {
 		free_pages(first_page, order);
 		return NULL;
 	}
 
-	*dma_addrp = (iommu->tbl.page_table_map_base +
+	*dma_addrp = (iommu->page_table_map_base +
 		      ((iopte - iommu->page_table) << IO_PAGE_SHIFT));
 	ret = (void *) first_page;
 	npages = size >> IO_PAGE_SHIFT;
@@ -270,12 +336,16 @@ static void dma_4u_free_coherent(struct device *dev, size_t size,
 				 struct dma_attrs *attrs)
 {
 	struct iommu *iommu;
-	unsigned long order, npages;
+	unsigned long flags, order, npages;
 
 	npages = IO_PAGE_ALIGN(size) >> IO_PAGE_SHIFT;
 	iommu = dev->archdata.iommu;
 
-	iommu_tbl_range_free(&iommu->tbl, dvma, npages, false, NULL);
+	spin_lock_irqsave(&iommu->lock, flags);
+
+	iommu_range_free(iommu, dvma, npages);
+
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	order = get_order(size);
 	if (order < 10)
@@ -305,8 +375,8 @@ static dma_addr_t dma_4u_map_page(struct device *dev, struct page *page,
 	npages = IO_PAGE_ALIGN(oaddr + sz) - (oaddr & IO_PAGE_MASK);
 	npages >>= IO_PAGE_SHIFT;
 
-	base = alloc_npages(dev, iommu, npages);
 	spin_lock_irqsave(&iommu->lock, flags);
+	base = alloc_npages(dev, iommu, npages);
 	ctx = 0;
 	if (iommu->iommu_ctxflush)
 		ctx = iommu_alloc_ctx(iommu);
@@ -315,7 +385,7 @@ static dma_addr_t dma_4u_map_page(struct device *dev, struct page *page,
 	if (unlikely(!base))
 		goto bad;
 
-	bus_addr = (iommu->tbl.page_table_map_base +
+	bus_addr = (iommu->page_table_map_base +
 		    ((base - iommu->page_table) << IO_PAGE_SHIFT));
 	ret = bus_addr | (oaddr & ~IO_PAGE_MASK);
 	base_paddr = __pa(oaddr & IO_PAGE_MASK);
@@ -426,7 +496,7 @@ static void dma_4u_unmap_page(struct device *dev, dma_addr_t bus_addr,
 	npages = IO_PAGE_ALIGN(bus_addr + sz) - (bus_addr & IO_PAGE_MASK);
 	npages >>= IO_PAGE_SHIFT;
 	base = iommu->page_table +
-		((bus_addr - iommu->tbl.page_table_map_base) >> IO_PAGE_SHIFT);
+		((bus_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
 	bus_addr &= IO_PAGE_MASK;
 
 	spin_lock_irqsave(&iommu->lock, flags);
@@ -445,11 +515,11 @@ static void dma_4u_unmap_page(struct device *dev, dma_addr_t bus_addr,
 	for (i = 0; i < npages; i++)
 		iopte_make_dummy(iommu, base + i);
 
+	iommu_range_free(iommu, bus_addr, npages);
+
 	iommu_free_ctx(iommu, ctx);
-	spin_unlock_irqrestore(&iommu->lock, flags);
 
-	iommu_tbl_range_free(&iommu->tbl, bus_addr, npages,
-			     false, NULL);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 }
 
 static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
@@ -497,7 +567,7 @@ static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
 	max_seg_size = dma_get_max_seg_size(dev);
 	seg_boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
 				  IO_PAGE_SIZE) >> IO_PAGE_SHIFT;
-	base_shift = iommu->tbl.page_table_map_base >> IO_PAGE_SHIFT;
+	base_shift = iommu->page_table_map_base >> IO_PAGE_SHIFT;
 	for_each_sg(sglist, s, nelems, i) {
 		unsigned long paddr, npages, entry, out_entry = 0, slen;
 		iopte_t *base;
@@ -511,8 +581,7 @@ static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
 		/* Allocate iommu entries for that segment */
 		paddr = (unsigned long) SG_ENT_PHYS_ADDRESS(s);
 		npages = iommu_num_pages(paddr, slen, IO_PAGE_SIZE);
-		entry = iommu_tbl_range_alloc(dev, &iommu->tbl, npages, &handle,
-					      __this_cpu_read(iommu_pool_hash));
+		entry = iommu_range_alloc(dev, iommu, npages, &handle);
 
 		/* Handle failure */
 		if (unlikely(entry == DMA_ERROR_CODE)) {
@@ -525,7 +594,7 @@ static int dma_4u_map_sg(struct device *dev, struct scatterlist *sglist,
 		base = iommu->page_table + entry;
 
 		/* Convert entry to a dma_addr_t */
-		dma_addr = iommu->tbl.page_table_map_base +
+		dma_addr = iommu->page_table_map_base +
 			(entry << IO_PAGE_SHIFT);
 		dma_addr |= (s->offset & ~IO_PAGE_MASK);
 
@@ -585,17 +654,15 @@ iommu_map_failed:
 			vaddr = s->dma_address & IO_PAGE_MASK;
 			npages = iommu_num_pages(s->dma_address, s->dma_length,
 						 IO_PAGE_SIZE);
+			iommu_range_free(iommu, vaddr, npages);
 
-			entry = (vaddr - iommu->tbl.page_table_map_base)
+			entry = (vaddr - iommu->page_table_map_base)
 				>> IO_PAGE_SHIFT;
 			base = iommu->page_table + entry;
 
 			for (j = 0; j < npages; j++)
 				iopte_make_dummy(iommu, base + j);
 
-			iommu_tbl_range_free(&iommu->tbl, vaddr, npages,
-					     false, NULL);
-
 			s->dma_address = DMA_ERROR_CODE;
 			s->dma_length = 0;
 		}
@@ -610,19 +677,17 @@ iommu_map_failed:
 /* If contexts are being used, they are the same in all of the mappings
  * we make for a particular SG.
  */
-static unsigned long fetch_sg_ctx(struct iommu *iommu,
-				  struct scatterlist *sg)
+static unsigned long fetch_sg_ctx(struct iommu *iommu, struct scatterlist *sg)
 {
 	unsigned long ctx = 0;
 
 	if (iommu->iommu_ctxflush) {
 		iopte_t *base;
 		u32 bus_addr;
-		struct iommu_table *tbl = &iommu->tbl;
 
 		bus_addr = sg->dma_address & IO_PAGE_MASK;
 		base = iommu->page_table +
-		       ((bus_addr - tbl->page_table_map_base) >> IO_PAGE_SHIFT);
+			((bus_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
 
 		ctx = (iopte_val(*base) & IOPTE_CONTEXT) >> 47UL;
 	}
@@ -658,8 +723,9 @@ static void dma_4u_unmap_sg(struct device *dev, struct scatterlist *sglist,
 		if (!len)
 			break;
 		npages = iommu_num_pages(dma_handle, len, IO_PAGE_SIZE);
+		iommu_range_free(iommu, dma_handle, npages);
 
-		entry = ((dma_handle - iommu->tbl.page_table_map_base)
+		entry = ((dma_handle - iommu->page_table_map_base)
 			 >> IO_PAGE_SHIFT);
 		base = iommu->page_table + entry;
 
@@ -671,8 +737,6 @@ static void dma_4u_unmap_sg(struct device *dev, struct scatterlist *sglist,
 		for (i = 0; i < npages; i++)
 			iopte_make_dummy(iommu, base + i);
 
-		iommu_tbl_range_free(&iommu->tbl, dma_handle, npages, false,
-				     NULL);
 		sg = sg_next(sg);
 	}
 
@@ -706,10 +770,9 @@ static void dma_4u_sync_single_for_cpu(struct device *dev,
 	if (iommu->iommu_ctxflush &&
 	    strbuf->strbuf_ctxflush) {
 		iopte_t *iopte;
-		struct iommu_table *tbl = &iommu->tbl;
 
 		iopte = iommu->page_table +
-			((bus_addr - tbl->page_table_map_base)>>IO_PAGE_SHIFT);
+			((bus_addr - iommu->page_table_map_base)>>IO_PAGE_SHIFT);
 		ctx = (iopte_val(*iopte) & IOPTE_CONTEXT) >> 47UL;
 	}
 
@@ -742,10 +805,9 @@ static void dma_4u_sync_sg_for_cpu(struct device *dev,
 	if (iommu->iommu_ctxflush &&
 	    strbuf->strbuf_ctxflush) {
 		iopte_t *iopte;
-		struct iommu_table *tbl = &iommu->tbl;
 
-		iopte = iommu->page_table + ((sglist[0].dma_address -
-			tbl->page_table_map_base) >> IO_PAGE_SHIFT);
+		iopte = iommu->page_table +
+			((sglist[0].dma_address - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
 		ctx = (iopte_val(*iopte) & IOPTE_CONTEXT) >> 47UL;
 	}
 
diff --git a/arch/sparc/kernel/iommu_common.h b/arch/sparc/kernel/iommu_common.h
index f4be0d724fc6..1ec0de4156e7 100644
--- a/arch/sparc/kernel/iommu_common.h
+++ b/arch/sparc/kernel/iommu_common.h
@@ -48,4 +48,12 @@ static inline int is_span_boundary(unsigned long entry,
 	return iommu_is_span_boundary(entry, nr, shift, boundary_size);
 }
 
+unsigned long iommu_range_alloc(struct device *dev,
+				struct iommu *iommu,
+				unsigned long npages,
+				unsigned long *handle);
+void iommu_range_free(struct iommu *iommu,
+		      dma_addr_t dma_addr,
+		      unsigned long npages);
+
 #endif /* _IOMMU_COMMON_H */
diff --git a/arch/sparc/kernel/ldc.c b/arch/sparc/kernel/ldc.c
index d485697c37c0..274a9f59d95c 100644
--- a/arch/sparc/kernel/ldc.c
+++ b/arch/sparc/kernel/ldc.c
@@ -15,8 +15,6 @@
 #include <linux/list.h>
 #include <linux/init.h>
 #include <linux/bitmap.h>
-#include <linux/hash.h>
-#include <linux/iommu-common.h>
 
 #include <asm/hypervisor.h>
 #include <asm/iommu.h>
@@ -29,11 +27,6 @@
 #define DRV_MODULE_VERSION	"1.1"
 #define DRV_MODULE_RELDATE	"July 22, 2008"
 
-#define COOKIE_PGSZ_CODE	0xf000000000000000ULL
-#define COOKIE_PGSZ_CODE_SHIFT	60ULL
-
-static DEFINE_PER_CPU(unsigned int, ldc_pool_hash);
-
 static char version[] =
 	DRV_MODULE_NAME ".c:v" DRV_MODULE_VERSION " (" DRV_MODULE_RELDATE ")\n";
 #define LDC_PACKET_SIZE		64
@@ -105,10 +98,10 @@ static const struct ldc_mode_ops stream_ops;
 int ldom_domaining_enabled;
 
 struct ldc_iommu {
-	/* Protects ldc_unmap.  */
+	/* Protects arena alloc/free.  */
 	spinlock_t			lock;
+	struct iommu_arena		arena;
 	struct ldc_mtable_entry		*page_table;
-	struct iommu_table		iommu_table;
 };
 
 struct ldc_channel {
@@ -1005,85 +998,31 @@ static void free_queue(unsigned long num_entries, struct ldc_packet *q)
 	free_pages((unsigned long)q, order);
 }
 
-static unsigned long ldc_cookie_to_index(u64 cookie, void *arg)
-{
-	u64 szcode = cookie >> COOKIE_PGSZ_CODE_SHIFT;
-	/* struct ldc_iommu *ldc_iommu = (struct ldc_iommu *)arg; */
-
-	cookie &= ~COOKIE_PGSZ_CODE;
-
-	return (cookie >> (13ULL + (szcode * 3ULL)));
-}
-
-struct ldc_demap_arg {
-	struct ldc_iommu *ldc_iommu;
-	u64 cookie;
-	unsigned long id;
-};
-
-static void ldc_demap(void *arg, unsigned long entry, unsigned long npages)
-{
-	struct ldc_demap_arg *ldc_demap_arg = arg;
-	struct ldc_iommu *iommu = ldc_demap_arg->ldc_iommu;
-	unsigned long id = ldc_demap_arg->id;
-	u64 cookie = ldc_demap_arg->cookie;
-	struct ldc_mtable_entry *base;
-	unsigned long i, shift;
-
-	shift = (cookie >> COOKIE_PGSZ_CODE_SHIFT) * 3;
-	base = iommu->page_table + entry;
-	for (i = 0; i < npages; i++) {
-		if (base->cookie)
-			sun4v_ldc_revoke(id, cookie + (i << shift),
-					 base->cookie);
-		base->mte = 0;
-	}
-}
-
 /* XXX Make this configurable... XXX */
 #define LDC_IOTABLE_SIZE	(8 * 1024)
 
-struct iommu_tbl_ops ldc_iommu_ops = {
-	.cookie_to_index = ldc_cookie_to_index,
-	.demap = ldc_demap,
-};
-
-static void setup_ldc_pool_hash(void)
-{
-	unsigned int i;
-	static bool do_once;
-
-	if (do_once)
-		return;
-	do_once = true;
-	for_each_possible_cpu(i)
-		per_cpu(ldc_pool_hash, i) = hash_32(i, IOMMU_POOL_HASHBITS);
-}
-
-
-static int ldc_iommu_init(const char *name, struct ldc_channel *lp)
+static int ldc_iommu_init(struct ldc_channel *lp)
 {
 	unsigned long sz, num_tsb_entries, tsbsize, order;
-	struct ldc_iommu *ldc_iommu = &lp->iommu;
-	struct iommu_table *iommu = &ldc_iommu->iommu_table;
+	struct ldc_iommu *iommu = &lp->iommu;
 	struct ldc_mtable_entry *table;
 	unsigned long hv_err;
 	int err;
 
 	num_tsb_entries = LDC_IOTABLE_SIZE;
 	tsbsize = num_tsb_entries * sizeof(struct ldc_mtable_entry);
-	setup_ldc_pool_hash();
-	spin_lock_init(&ldc_iommu->lock);
+
+	spin_lock_init(&iommu->lock);
 
 	sz = num_tsb_entries / 8;
 	sz = (sz + 7UL) & ~7UL;
-	iommu->map = kzalloc(sz, GFP_KERNEL);
-	if (!iommu->map) {
+	iommu->arena.map = kzalloc(sz, GFP_KERNEL);
+	if (!iommu->arena.map) {
 		printk(KERN_ERR PFX "Alloc of arena map failed, sz=%lu\n", sz);
 		return -ENOMEM;
 	}
-	iommu_tbl_pool_init(iommu, num_tsb_entries, PAGE_SHIFT,
-			    &ldc_iommu_ops, false, 1);
+
+	iommu->arena.limit = num_tsb_entries;
 
 	order = get_order(tsbsize);
 
@@ -1098,7 +1037,7 @@ static int ldc_iommu_init(const char *name, struct ldc_channel *lp)
 
 	memset(table, 0, PAGE_SIZE << order);
 
-	ldc_iommu->page_table = table;
+	iommu->page_table = table;
 
 	hv_err = sun4v_ldc_set_map_table(lp->id, __pa(table),
 					 num_tsb_entries);
@@ -1110,32 +1049,31 @@ static int ldc_iommu_init(const char *name, struct ldc_channel *lp)
 
 out_free_table:
 	free_pages((unsigned long) table, order);
-	ldc_iommu->page_table = NULL;
+	iommu->page_table = NULL;
 
 out_free_map:
-	kfree(iommu->map);
-	iommu->map = NULL;
+	kfree(iommu->arena.map);
+	iommu->arena.map = NULL;
 
 	return err;
 }
 
 static void ldc_iommu_release(struct ldc_channel *lp)
 {
-	struct ldc_iommu *ldc_iommu = &lp->iommu;
-	struct iommu_table *iommu = &ldc_iommu->iommu_table;
+	struct ldc_iommu *iommu = &lp->iommu;
 	unsigned long num_tsb_entries, tsbsize, order;
 
 	(void) sun4v_ldc_set_map_table(lp->id, 0, 0);
 
-	num_tsb_entries = iommu->poolsize * iommu->nr_pools;
+	num_tsb_entries = iommu->arena.limit;
 	tsbsize = num_tsb_entries * sizeof(struct ldc_mtable_entry);
 	order = get_order(tsbsize);
 
-	free_pages((unsigned long) ldc_iommu->page_table, order);
-	ldc_iommu->page_table = NULL;
+	free_pages((unsigned long) iommu->page_table, order);
+	iommu->page_table = NULL;
 
-	kfree(iommu->map);
-	iommu->map = NULL;
+	kfree(iommu->arena.map);
+	iommu->arena.map = NULL;
 }
 
 struct ldc_channel *ldc_alloc(unsigned long id,
@@ -1202,7 +1140,7 @@ struct ldc_channel *ldc_alloc(unsigned long id,
 
 	lp->id = id;
 
-	err = ldc_iommu_init(name, lp);
+	err = ldc_iommu_init(lp);
 	if (err)
 		goto out_free_ldc;
 
@@ -1947,6 +1885,40 @@ int ldc_read(struct ldc_channel *lp, void *buf, unsigned int size)
 }
 EXPORT_SYMBOL(ldc_read);
 
+static long arena_alloc(struct ldc_iommu *iommu, unsigned long npages)
+{
+	struct iommu_arena *arena = &iommu->arena;
+	unsigned long n, start, end, limit;
+	int pass;
+
+	limit = arena->limit;
+	start = arena->hint;
+	pass = 0;
+
+again:
+	n = bitmap_find_next_zero_area(arena->map, limit, start, npages, 0);
+	end = n + npages;
+	if (unlikely(end >= limit)) {
+		if (likely(pass < 1)) {
+			limit = start;
+			start = 0;
+			pass++;
+			goto again;
+		} else {
+			/* Scanned the whole thing, give up. */
+			return -1;
+		}
+	}
+	bitmap_set(arena->map, n, npages);
+
+	arena->hint = end;
+
+	return n;
+}
+
+#define COOKIE_PGSZ_CODE	0xf000000000000000ULL
+#define COOKIE_PGSZ_CODE_SHIFT	60ULL
+
 static u64 pagesize_code(void)
 {
 	switch (PAGE_SIZE) {
@@ -1973,14 +1945,23 @@ static u64 make_cookie(u64 index, u64 pgsz_code, u64 page_offset)
 		page_offset);
 }
 
+static u64 cookie_to_index(u64 cookie, unsigned long *shift)
+{
+	u64 szcode = cookie >> COOKIE_PGSZ_CODE_SHIFT;
+
+	cookie &= ~COOKIE_PGSZ_CODE;
+
+	*shift = szcode * 3;
+
+	return (cookie >> (13ULL + (szcode * 3ULL)));
+}
 
 static struct ldc_mtable_entry *alloc_npages(struct ldc_iommu *iommu,
 					     unsigned long npages)
 {
 	long entry;
 
-	entry = iommu_tbl_range_alloc(NULL, &iommu->iommu_table, npages,
-				     NULL, __this_cpu_read(ldc_pool_hash));
+	entry = arena_alloc(iommu, npages);
 	if (unlikely(entry < 0))
 		return NULL;
 
@@ -2109,7 +2090,7 @@ int ldc_map_sg(struct ldc_channel *lp,
 	       struct ldc_trans_cookie *cookies, int ncookies,
 	       unsigned int map_perm)
 {
-	unsigned long i, npages;
+	unsigned long i, npages, flags;
 	struct ldc_mtable_entry *base;
 	struct cookie_state state;
 	struct ldc_iommu *iommu;
@@ -2128,7 +2109,9 @@ int ldc_map_sg(struct ldc_channel *lp,
 
 	iommu = &lp->iommu;
 
+	spin_lock_irqsave(&iommu->lock, flags);
 	base = alloc_npages(iommu, npages);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	if (!base)
 		return -ENOMEM;
@@ -2153,7 +2136,7 @@ int ldc_map_single(struct ldc_channel *lp,
 		   struct ldc_trans_cookie *cookies, int ncookies,
 		   unsigned int map_perm)
 {
-	unsigned long npages, pa;
+	unsigned long npages, pa, flags;
 	struct ldc_mtable_entry *base;
 	struct cookie_state state;
 	struct ldc_iommu *iommu;
@@ -2169,7 +2152,9 @@ int ldc_map_single(struct ldc_channel *lp,
 
 	iommu = &lp->iommu;
 
+	spin_lock_irqsave(&iommu->lock, flags);
 	base = alloc_npages(iommu, npages);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	if (!base)
 		return -ENOMEM;
@@ -2187,29 +2172,35 @@ int ldc_map_single(struct ldc_channel *lp,
 }
 EXPORT_SYMBOL(ldc_map_single);
 
-
 static void free_npages(unsigned long id, struct ldc_iommu *iommu,
 			u64 cookie, u64 size)
 {
-	unsigned long npages;
-	struct ldc_demap_arg demap_arg;
-
-	demap_arg.ldc_iommu = iommu;
-	demap_arg.cookie = cookie;
-	demap_arg.id = id;
+	struct iommu_arena *arena = &iommu->arena;
+	unsigned long i, shift, index, npages;
+	struct ldc_mtable_entry *base;
 
 	npages = PAGE_ALIGN(((cookie & ~PAGE_MASK) + size)) >> PAGE_SHIFT;
-	iommu_tbl_range_free(&iommu->iommu_table, cookie, npages, true,
-			     &demap_arg);
+	index = cookie_to_index(cookie, &shift);
+	base = iommu->page_table + index;
+
+	BUG_ON(index > arena->limit ||
+	       (index + npages) > arena->limit);
 
+	for (i = 0; i < npages; i++) {
+		if (base->cookie)
+			sun4v_ldc_revoke(id, cookie + (i << shift),
+					 base->cookie);
+		base->mte = 0;
+		__clear_bit(index + i, arena->map);
+	}
 }
 
 void ldc_unmap(struct ldc_channel *lp, struct ldc_trans_cookie *cookies,
 	       int ncookies)
 {
 	struct ldc_iommu *iommu = &lp->iommu;
-	int i;
 	unsigned long flags;
+	int i;
 
 	spin_lock_irqsave(&iommu->lock, flags);
 	for (i = 0; i < ncookies; i++) {
diff --git a/arch/sparc/kernel/pci_sun4v.c b/arch/sparc/kernel/pci_sun4v.c
index 9b76b9d639e1..47ddbd496a1e 100644
--- a/arch/sparc/kernel/pci_sun4v.c
+++ b/arch/sparc/kernel/pci_sun4v.c
@@ -15,8 +15,6 @@
 #include <linux/export.h>
 #include <linux/log2.h>
 #include <linux/of_device.h>
-#include <linux/hash.h>
-#include <linux/iommu-common.h>
 
 #include <asm/iommu.h>
 #include <asm/irq.h>
@@ -30,7 +28,6 @@
 
 #define DRIVER_NAME	"pci_sun4v"
 #define PFX		DRIVER_NAME ": "
-static DEFINE_PER_CPU(unsigned int, iommu_pool_hash);
 
 static unsigned long vpci_major = 1;
 static unsigned long vpci_minor = 1;
@@ -158,13 +155,14 @@ static void *dma_4v_alloc_coherent(struct device *dev, size_t size,
 
 	iommu = dev->archdata.iommu;
 
-	entry = iommu_tbl_range_alloc(dev, &iommu->tbl, npages, NULL,
-				      __this_cpu_read(iommu_pool_hash));
+	spin_lock_irqsave(&iommu->lock, flags);
+	entry = iommu_range_alloc(dev, iommu, npages, NULL);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	if (unlikely(entry == DMA_ERROR_CODE))
 		goto range_alloc_fail;
 
-	*dma_addrp = (iommu->tbl.page_table_map_base +
+	*dma_addrp = (iommu->page_table_map_base +
 		      (entry << IO_PAGE_SHIFT));
 	ret = (void *) first_page;
 	first_page = __pa(first_page);
@@ -190,46 +188,45 @@ static void *dma_4v_alloc_coherent(struct device *dev, size_t size,
 	return ret;
 
 iommu_map_fail:
-	iommu_tbl_range_free(&iommu->tbl, *dma_addrp, npages, false, NULL);
+	/* Interrupts are disabled.  */
+	spin_lock(&iommu->lock);
+	iommu_range_free(iommu, *dma_addrp, npages);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 range_alloc_fail:
 	free_pages(first_page, order);
 	return NULL;
 }
 
-static void dma_4v_iommu_demap(void *demap_arg, unsigned long entry,
-			       unsigned long npages)
-{
-	u32 devhandle = *(u32 *)demap_arg;
-	unsigned long num, flags;
-
-	local_irq_save(flags);
-	do {
-		num = pci_sun4v_iommu_demap(devhandle,
-					    HV_PCI_TSBID(0, entry),
-					    npages);
-
-		entry += num;
-		npages -= num;
-	} while (npages != 0);
-	local_irq_restore(flags);
-}
-
 static void dma_4v_free_coherent(struct device *dev, size_t size, void *cpu,
 				 dma_addr_t dvma, struct dma_attrs *attrs)
 {
 	struct pci_pbm_info *pbm;
 	struct iommu *iommu;
-	unsigned long order, npages, entry;
+	unsigned long flags, order, npages, entry;
 	u32 devhandle;
 
 	npages = IO_PAGE_ALIGN(size) >> IO_PAGE_SHIFT;
 	iommu = dev->archdata.iommu;
 	pbm = dev->archdata.host_controller;
 	devhandle = pbm->devhandle;
-	entry = ((dvma - iommu->tbl.page_table_map_base) >> IO_PAGE_SHIFT);
-	dma_4v_iommu_demap(&devhandle, entry, npages);
-	iommu_tbl_range_free(&iommu->tbl, dvma, npages, false, NULL);
+	entry = ((dvma - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
+
+	spin_lock_irqsave(&iommu->lock, flags);
+
+	iommu_range_free(iommu, dvma, npages);
+
+	do {
+		unsigned long num;
+
+		num = pci_sun4v_iommu_demap(devhandle, HV_PCI_TSBID(0, entry),
+					    npages);
+		entry += num;
+		npages -= num;
+	} while (npages != 0);
+
+	spin_unlock_irqrestore(&iommu->lock, flags);
+
 	order = get_order(size);
 	if (order < 10)
 		free_pages((unsigned long)cpu, order);
@@ -256,13 +253,14 @@ static dma_addr_t dma_4v_map_page(struct device *dev, struct page *page,
 	npages = IO_PAGE_ALIGN(oaddr + sz) - (oaddr & IO_PAGE_MASK);
 	npages >>= IO_PAGE_SHIFT;
 
-	entry = iommu_tbl_range_alloc(dev, &iommu->tbl, npages, NULL,
-				      __this_cpu_read(iommu_pool_hash));
+	spin_lock_irqsave(&iommu->lock, flags);
+	entry = iommu_range_alloc(dev, iommu, npages, NULL);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	if (unlikely(entry == DMA_ERROR_CODE))
 		goto bad;
 
-	bus_addr = (iommu->tbl.page_table_map_base +
+	bus_addr = (iommu->page_table_map_base +
 		    (entry << IO_PAGE_SHIFT));
 	ret = bus_addr | (oaddr & ~IO_PAGE_MASK);
 	base_paddr = __pa(oaddr & IO_PAGE_MASK);
@@ -292,7 +290,11 @@ bad:
 	return DMA_ERROR_CODE;
 
 iommu_map_fail:
-	iommu_tbl_range_free(&iommu->tbl, bus_addr, npages, false, NULL);
+	/* Interrupts are disabled.  */
+	spin_lock(&iommu->lock);
+	iommu_range_free(iommu, bus_addr, npages);
+	spin_unlock_irqrestore(&iommu->lock, flags);
+
 	return DMA_ERROR_CODE;
 }
 
@@ -302,7 +304,7 @@ static void dma_4v_unmap_page(struct device *dev, dma_addr_t bus_addr,
 {
 	struct pci_pbm_info *pbm;
 	struct iommu *iommu;
-	unsigned long npages;
+	unsigned long flags, npages;
 	long entry;
 	u32 devhandle;
 
@@ -319,9 +321,22 @@ static void dma_4v_unmap_page(struct device *dev, dma_addr_t bus_addr,
 	npages = IO_PAGE_ALIGN(bus_addr + sz) - (bus_addr & IO_PAGE_MASK);
 	npages >>= IO_PAGE_SHIFT;
 	bus_addr &= IO_PAGE_MASK;
-	entry = (bus_addr - iommu->tbl.page_table_map_base) >> IO_PAGE_SHIFT;
-	dma_4v_iommu_demap(&devhandle, entry, npages);
-	iommu_tbl_range_free(&iommu->tbl, bus_addr, npages, false, NULL);
+
+	spin_lock_irqsave(&iommu->lock, flags);
+
+	iommu_range_free(iommu, bus_addr, npages);
+
+	entry = (bus_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT;
+	do {
+		unsigned long num;
+
+		num = pci_sun4v_iommu_demap(devhandle, HV_PCI_TSBID(0, entry),
+					    npages);
+		entry += num;
+		npages -= num;
+	} while (npages != 0);
+
+	spin_unlock_irqrestore(&iommu->lock, flags);
 }
 
 static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
@@ -356,14 +371,14 @@ static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
 	/* Init first segment length for backout at failure */
 	outs->dma_length = 0;
 
-	local_irq_save(flags);
+	spin_lock_irqsave(&iommu->lock, flags);
 
 	iommu_batch_start(dev, prot, ~0UL);
 
 	max_seg_size = dma_get_max_seg_size(dev);
 	seg_boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
 				  IO_PAGE_SIZE) >> IO_PAGE_SHIFT;
-	base_shift = iommu->tbl.page_table_map_base >> IO_PAGE_SHIFT;
+	base_shift = iommu->page_table_map_base >> IO_PAGE_SHIFT;
 	for_each_sg(sglist, s, nelems, i) {
 		unsigned long paddr, npages, entry, out_entry = 0, slen;
 
@@ -376,8 +391,7 @@ static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
 		/* Allocate iommu entries for that segment */
 		paddr = (unsigned long) SG_ENT_PHYS_ADDRESS(s);
 		npages = iommu_num_pages(paddr, slen, IO_PAGE_SIZE);
-		entry = iommu_tbl_range_alloc(dev, &iommu->tbl, npages, &handle,
-				      __this_cpu_read(iommu_pool_hash));
+		entry = iommu_range_alloc(dev, iommu, npages, &handle);
 
 		/* Handle failure */
 		if (unlikely(entry == DMA_ERROR_CODE)) {
@@ -390,7 +404,7 @@ static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
 		iommu_batch_new_entry(entry);
 
 		/* Convert entry to a dma_addr_t */
-		dma_addr = iommu->tbl.page_table_map_base +
+		dma_addr = iommu->page_table_map_base +
 			(entry << IO_PAGE_SHIFT);
 		dma_addr |= (s->offset & ~IO_PAGE_MASK);
 
@@ -437,7 +451,7 @@ static int dma_4v_map_sg(struct device *dev, struct scatterlist *sglist,
 	if (unlikely(err < 0L))
 		goto iommu_map_failed;
 
-	local_irq_restore(flags);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	if (outcount < incount) {
 		outs = sg_next(outs);
@@ -455,8 +469,7 @@ iommu_map_failed:
 			vaddr = s->dma_address & IO_PAGE_MASK;
 			npages = iommu_num_pages(s->dma_address, s->dma_length,
 						 IO_PAGE_SIZE);
-			iommu_tbl_range_free(&iommu->tbl, vaddr, npages,
-					     false, NULL);
+			iommu_range_free(iommu, vaddr, npages);
 			/* XXX demap? XXX */
 			s->dma_address = DMA_ERROR_CODE;
 			s->dma_length = 0;
@@ -464,7 +477,7 @@ iommu_map_failed:
 		if (s == outs)
 			break;
 	}
-	local_irq_restore(flags);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	return 0;
 }
@@ -476,7 +489,7 @@ static void dma_4v_unmap_sg(struct device *dev, struct scatterlist *sglist,
 	struct pci_pbm_info *pbm;
 	struct scatterlist *sg;
 	struct iommu *iommu;
-	unsigned long flags, entry;
+	unsigned long flags;
 	u32 devhandle;
 
 	BUG_ON(direction == DMA_NONE);
@@ -485,27 +498,33 @@ static void dma_4v_unmap_sg(struct device *dev, struct scatterlist *sglist,
 	pbm = dev->archdata.host_controller;
 	devhandle = pbm->devhandle;
 	
-	local_irq_save(flags);
+	spin_lock_irqsave(&iommu->lock, flags);
 
 	sg = sglist;
 	while (nelems--) {
 		dma_addr_t dma_handle = sg->dma_address;
 		unsigned int len = sg->dma_length;
-		unsigned long npages;
-		struct iommu_table *tbl = &iommu->tbl;
-		unsigned long shift = IO_PAGE_SHIFT;
+		unsigned long npages, entry;
 
 		if (!len)
 			break;
 		npages = iommu_num_pages(dma_handle, len, IO_PAGE_SIZE);
-		entry = ((dma_handle - tbl->page_table_map_base) >> shift);
-		dma_4v_iommu_demap(&devhandle, entry, npages);
-		iommu_tbl_range_free(&iommu->tbl, dma_handle, npages,
-				     false, NULL);
+		iommu_range_free(iommu, dma_handle, npages);
+
+		entry = ((dma_handle - iommu->page_table_map_base) >> IO_PAGE_SHIFT);
+		while (npages) {
+			unsigned long num;
+
+			num = pci_sun4v_iommu_demap(devhandle, HV_PCI_TSBID(0, entry),
+						    npages);
+			entry += num;
+			npages -= num;
+		}
+
 		sg = sg_next(sg);
 	}
 
-	local_irq_restore(flags);
+	spin_unlock_irqrestore(&iommu->lock, flags);
 }
 
 static struct dma_map_ops sun4v_dma_ops = {
@@ -517,8 +536,6 @@ static struct dma_map_ops sun4v_dma_ops = {
 	.unmap_sg			= dma_4v_unmap_sg,
 };
 
-static struct iommu_tbl_ops dma_4v_iommu_ops;
-
 static void pci_sun4v_scan_bus(struct pci_pbm_info *pbm, struct device *parent)
 {
 	struct property *prop;
@@ -533,33 +550,30 @@ static void pci_sun4v_scan_bus(struct pci_pbm_info *pbm, struct device *parent)
 }
 
 static unsigned long probe_existing_entries(struct pci_pbm_info *pbm,
-					    struct iommu_table *iommu)
+					    struct iommu *iommu)
 {
-	struct iommu_pool *pool;
-	unsigned long i, pool_nr, cnt = 0;
+	struct iommu_arena *arena = &iommu->arena;
+	unsigned long i, cnt = 0;
 	u32 devhandle;
 
 	devhandle = pbm->devhandle;
-	for (pool_nr = 0; pool_nr < iommu->nr_pools; pool_nr++) {
-		pool = &(iommu->arena_pool[pool_nr]);
-		for (i = pool->start; i <= pool->end; i++) {
-			unsigned long ret, io_attrs, ra;
-
-			ret = pci_sun4v_iommu_getmap(devhandle,
-						     HV_PCI_TSBID(0, i),
-						     &io_attrs, &ra);
-			if (ret == HV_EOK) {
-				if (page_in_phys_avail(ra)) {
-					pci_sun4v_iommu_demap(devhandle,
-							      HV_PCI_TSBID(0,
-							      i), 1);
-				} else {
-					cnt++;
-					__set_bit(i, iommu->map);
-				}
+	for (i = 0; i < arena->limit; i++) {
+		unsigned long ret, io_attrs, ra;
+
+		ret = pci_sun4v_iommu_getmap(devhandle,
+					     HV_PCI_TSBID(0, i),
+					     &io_attrs, &ra);
+		if (ret == HV_EOK) {
+			if (page_in_phys_avail(ra)) {
+				pci_sun4v_iommu_demap(devhandle,
+						      HV_PCI_TSBID(0, i), 1);
+			} else {
+				cnt++;
+				__set_bit(i, arena->map);
 			}
 		}
 	}
+
 	return cnt;
 }
 
@@ -587,22 +601,22 @@ static int pci_sun4v_iommu_init(struct pci_pbm_info *pbm)
 	dma_offset = vdma[0];
 
 	/* Setup initial software IOMMU state. */
+	spin_lock_init(&iommu->lock);
 	iommu->ctx_lowest_free = 1;
-	iommu->tbl.page_table_map_base = dma_offset;
+	iommu->page_table_map_base = dma_offset;
 	iommu->dma_addr_mask = dma_mask;
 
 	/* Allocate and initialize the free area map.  */
 	sz = (num_tsb_entries + 7) / 8;
 	sz = (sz + 7UL) & ~7UL;
-	iommu->tbl.map = kzalloc(sz, GFP_KERNEL);
-	if (!iommu->tbl.map) {
+	iommu->arena.map = kzalloc(sz, GFP_KERNEL);
+	if (!iommu->arena.map) {
 		printk(KERN_ERR PFX "Error, kmalloc(arena.map) failed.\n");
 		return -ENOMEM;
 	}
-	iommu_tbl_pool_init(&iommu->tbl, num_tsb_entries, IO_PAGE_SHIFT,
-			    &dma_4v_iommu_ops, false /* no large_pool */,
-			    0 /* default npools */);
-	sz = probe_existing_entries(pbm, &iommu->tbl);
+	iommu->arena.limit = num_tsb_entries;
+
+	sz = probe_existing_entries(pbm, iommu);
 	if (sz)
 		printk("%s: Imported %lu TSB entries from OBP\n",
 		       pbm->name, sz);
@@ -1001,17 +1015,8 @@ static struct platform_driver pci_sun4v_driver = {
 	.probe		= pci_sun4v_probe,
 };
 
-static void setup_iommu_pool_hash(void)
-{
-	unsigned int i;
-
-	for_each_possible_cpu(i)
-		per_cpu(iommu_pool_hash, i) = hash_32(i, IOMMU_POOL_HASHBITS);
-}
-
 static int __init pci_sun4v_init(void)
 {
-	setup_iommu_pool_hash();
 	return platform_driver_register(&pci_sun4v_driver);
 }