Merge branch 'iommu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

* 'iommu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (61 commits)
  amd-iommu: remove unnecessary "AMD IOMMU: " prefix
  amd-iommu: detach device explicitly before attaching it to a new domain
  amd-iommu: remove BUS_NOTIFY_BOUND_DRIVER handling
  dma-debug: simplify logic in driver_filter()
  dma-debug: disable/enable irqs only once in device_dma_allocations
  dma-debug: use pr_* instead of printk(KERN_* ...)
  dma-debug: code style fixes
  dma-debug: comment style fixes
  dma-debug: change hash_bucket_find from first-fit to best-fit
  x86: enable GART-IOMMU only after setting up protection methods
  amd_iommu: fix lock imbalance
  dma-debug: add documentation for the driver filter
  dma-debug: add dma_debug_driver kernel command line
  dma-debug: add debugfs file for driver filter
  dma-debug: add variables and checks for driver filter
  dma-debug: fix debug_dma_sync_sg_for_cpu and debug_dma_sync_sg_for_device
  dma-debug: use sg_dma_len accessor
  dma-debug: use sg_dma_address accessor instead of using dma_address directly
  amd-iommu: don't free dma adresses below 512MB with CONFIG_IOMMU_STRESS
  amd-iommu: don't preallocate page tables with CONFIG_IOMMU_STRESS
  ...

1 files changed, 16 insertions, 39 deletions

diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c
index b284b58c035c..cfd9f9063896 100644
--- a/arch/x86/kernel/pci-gart_64.c
+++ b/arch/x86/kernel/pci-gart_64.c
@@ -144,48 +144,21 @@ static void flush_gart(void)
 }
 
 #ifdef CONFIG_IOMMU_LEAK
-
-#define SET_LEAK(x)                                                     \
-        do {                                                            \
-                if (iommu_leak_tab)                                     \
-                        iommu_leak_tab[x] = __builtin_return_address(0);\
-        } while (0)
-
-#define CLEAR_LEAK(x)                                                   \
-        do {                                                            \
-                if (iommu_leak_tab)                                     \
-                        iommu_leak_tab[x] = NULL;                       \
-        } while (0)
-
 /* Debugging aid for drivers that don't free their IOMMU tables */
-static void **iommu_leak_tab;
 static int leak_trace;
 static int iommu_leak_pages = 20;
 
 static void dump_leak(void)
 {
-        int i;
         static int dump;
 
-        if (dump || !iommu_leak_tab)
+        if (dump)
                 return;
         dump = 1;
-        show_stack(NULL, NULL);
 
-        /* Very crude. dump some from the end of the table too */
-        printk(KERN_DEBUG "Dumping %d pages from end of IOMMU:\n",
-               iommu_leak_pages);
-        for (i = 0; i < iommu_leak_pages; i += 2) {
-                printk(KERN_DEBUG "%lu: ", iommu_pages-i);
-                printk_address((unsigned long) iommu_leak_tab[iommu_pages-i],
-                                0);
-                printk(KERN_CONT "%c", (i+1)%2 == 0 ? '\n' : ' ');
-        }
-        printk(KERN_DEBUG "\n");
+        show_stack(NULL, NULL);
+        debug_dma_dump_mappings(NULL);
 }
-#else
-# define SET_LEAK(x)
-# define CLEAR_LEAK(x)
 #endif
 
 static void iommu_full(struct device *dev, size_t size, int dir)
@@ -248,7 +221,6 @@ static dma_addr_t dma_map_area(struct device *dev, dma_addr_t phys_mem,
 
         for (i = 0; i < npages; i++) {
                 iommu_gatt_base[iommu_page + i] = GPTE_ENCODE(phys_mem);
-                SET_LEAK(iommu_page + i);
                 phys_mem += PAGE_SIZE;
         }
         return iommu_bus_base + iommu_page*PAGE_SIZE + (phys_mem & ~PAGE_MASK);
@@ -294,7 +266,6 @@ static void gart_unmap_page(struct device *dev, dma_addr_t dma_addr,
         npages = iommu_num_pages(dma_addr, size, PAGE_SIZE);
         for (i = 0; i < npages; i++) {
                 iommu_gatt_base[iommu_page + i] = gart_unmapped_entry;
-                CLEAR_LEAK(iommu_page + i);
         }
         free_iommu(iommu_page, npages);
 }
@@ -377,7 +348,6 @@ static int __dma_map_cont(struct device *dev, struct scatterlist *start,
                 pages = iommu_num_pages(s->offset, s->length, PAGE_SIZE);
                 while (pages--) {
                         iommu_gatt_base[iommu_page] = GPTE_ENCODE(addr);
-                        SET_LEAK(iommu_page);
                         addr += PAGE_SIZE;
                         iommu_page++;
                 }
@@ -688,8 +658,6 @@ static __init int init_k8_gatt(struct agp_kern_info *info)
 
         agp_gatt_table = gatt;
 
-        enable_gart_translations();
-
         error = sysdev_class_register(&gart_sysdev_class);
         if (!error)
                 error = sysdev_register(&device_gart);
@@ -801,11 +769,12 @@ void __init gart_iommu_init(void)
 
 #ifdef CONFIG_IOMMU_LEAK
         if (leak_trace) {
-                iommu_leak_tab = (void *)__get_free_pages(GFP_KERNEL|__GFP_ZERO,
-                                  get_order(iommu_pages*sizeof(void *)));
-                if (!iommu_leak_tab)
+                int ret;
+
+                ret = dma_debug_resize_entries(iommu_pages);
+                if (ret)
                         printk(KERN_DEBUG
-                               "PCI-DMA: Cannot allocate leak trace area\n");
+                               "PCI-DMA: Cannot trace all the entries\n");
         }
 #endif
 
@@ -845,6 +814,14 @@ void __init gart_iommu_init(void)
          * the pages as Not-Present:
          */
         wbinvd();
+        
+        /*
+         * Now all caches are flushed and we can safely enable
+         * GART hardware.  Doing it early leaves the possibility
+         * of stale cache entries that can lead to GART PTE
+         * errors.
+         */
+        enable_gart_translations();
 
         /*
          * Try to workaround a bug (thanks to BenH):