Merge tag 'iommu-updates-v3.20' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu

Pull IOMMU updates from Joerg Roedel:
 "This time with:

   - Generic page-table framework for ARM IOMMUs using the LPAE
     page-table format, ARM-SMMU and Renesas IPMMU make use of it
     already.

   - Break out the IO virtual address allocator from the Intel IOMMU so
     that it can be used by other DMA-API implementations too.  The
     first user will be the ARM64 common DMA-API implementation for
     IOMMUs

   - Device tree support for Renesas IPMMU

   - Various fixes and cleanups all over the place"

* tag 'iommu-updates-v3.20' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (36 commits)
  iommu/amd: Convert non-returned local variable to boolean when relevant
  iommu: Update my email address
  iommu/amd: Use wait_event in put_pasid_state_wait
  iommu/amd: Fix amd_iommu_free_device()
  iommu/arm-smmu: Avoid build warning
  iommu/fsl: Various cleanups
  iommu/fsl: Use %pa to print phys_addr_t
  iommu/omap: Print phys_addr_t using %pa
  iommu: Make more drivers depend on COMPILE_TEST
  iommu/ipmmu-vmsa: Fix IOMMU lookup when multiple IOMMUs are registered
  iommu: Disable on !MMU builds
  iommu/fsl: Remove unused fsl_of_pamu_ids[]
  iommu/fsl: Fix section mismatch
  iommu/ipmmu-vmsa: Use the ARM LPAE page table allocator
  iommu: Fix trace_map() to report original iova and original size
  iommu/arm-smmu: add support for iova_to_phys through ATS1PR
  iopoll: Introduce memory-mapped IO polling macros
  iommu/arm-smmu: don't touch the secure STLBIALL register
  iommu/arm-smmu: make use of generic LPAE allocator
  iommu: io-pgtable-arm: add non-secure quirk
  ...

1 files changed, 49 insertions, 4 deletions

diff --git a/drivers/iommu/iova.c b/drivers/iommu/iova.c
index f6b17e6af2fb..9dd8208312c2 100644
--- a/drivers/iommu/iova.c
+++ b/drivers/iommu/iova.c
@@ -18,13 +18,58 @@
  */
 
 #include <linux/iova.h>
+#include <linux/slab.h>
+
+static struct kmem_cache *iommu_iova_cache;
+
+int iommu_iova_cache_init(void)
+{
+	int ret = 0;
+
+	iommu_iova_cache = kmem_cache_create("iommu_iova",
+					 sizeof(struct iova),
+					 0,
+					 SLAB_HWCACHE_ALIGN,
+					 NULL);
+	if (!iommu_iova_cache) {
+		pr_err("Couldn't create iova cache\n");
+		ret = -ENOMEM;
+	}
+
+	return ret;
+}
+
+void iommu_iova_cache_destroy(void)
+{
+	kmem_cache_destroy(iommu_iova_cache);
+}
+
+struct iova *alloc_iova_mem(void)
+{
+	return kmem_cache_alloc(iommu_iova_cache, GFP_ATOMIC);
+}
+
+void free_iova_mem(struct iova *iova)
+{
+	kmem_cache_free(iommu_iova_cache, iova);
+}
 
 void
-init_iova_domain(struct iova_domain *iovad, unsigned long pfn_32bit)
+init_iova_domain(struct iova_domain *iovad, unsigned long granule,
+	unsigned long start_pfn, unsigned long pfn_32bit)
 {
+	/*
+	 * IOVA granularity will normally be equal to the smallest
+	 * supported IOMMU page size; both *must* be capable of
+	 * representing individual CPU pages exactly.
+	 */
+	BUG_ON((granule > PAGE_SIZE) || !is_power_of_2(granule));
+
 	spin_lock_init(&iovad->iova_rbtree_lock);
 	iovad->rbroot = RB_ROOT;
 	iovad->cached32_node = NULL;
+	iovad->granule = granule;
+	iovad->start_pfn = start_pfn;
 	iovad->dma_32bit_pfn = pfn_32bit;
 }
 
@@ -127,7 +172,7 @@ move_left:
 	if (!curr) {
 		if (size_aligned)
 			pad_size = iova_get_pad_size(size, limit_pfn);
-		if ((IOVA_START_PFN + size + pad_size) > limit_pfn) {
+		if ((iovad->start_pfn + size + pad_size) > limit_pfn) {
 			spin_unlock_irqrestore(&iovad->iova_rbtree_lock, flags);
 			return -ENOMEM;
 		}
@@ -202,8 +247,8 @@ iova_insert_rbtree(struct rb_root *root, struct iova *iova)
  * @size: - size of page frames to allocate
  * @limit_pfn: - max limit address
  * @size_aligned: - set if size_aligned address range is required
- * This function allocates an iova in the range limit_pfn to IOVA_START_PFN
- * looking from limit_pfn instead from IOVA_START_PFN. If the size_aligned
+ * This function allocates an iova in the range iovad->start_pfn to limit_pfn,
+ * searching top-down from limit_pfn to iovad->start_pfn. If the size_aligned
  * flag is set then the allocated address iova->pfn_lo will be naturally
  * aligned on roundup_power_of_two(size).
  */