4 files changed, 32 insertions, 36 deletions

diff --git a/arch/x86/include/asm/amd_iommu_types.h b/arch/x86/include/asm/amd_iommu_types.h
index 1a30c0440c6b..ac302a2fa339 100644
--- a/arch/x86/include/asm/amd_iommu_types.h
+++ b/arch/x86/include/asm/amd_iommu_types.h
@@ -251,13 +251,6 @@ struct amd_iommu {
         /* Pointer to PCI device of this IOMMU */
         struct pci_dev *dev;
 
-        /*
-         * Capability pointer. There could be more than one IOMMU per PCI
-         * device function if there are more than one AMD IOMMU capability
-         * pointers.
-         */
-        u16 cap_ptr;
-
         /* physical address of MMIO space */
         u64 mmio_phys;
         /* virtual address of MMIO space */
@@ -266,6 +259,13 @@ struct amd_iommu {
         /* capabilities of that IOMMU read from ACPI */
         u32 cap;
 
+        /*
+         * Capability pointer. There could be more than one IOMMU per PCI
+         * device function if there are more than one AMD IOMMU capability
+         * pointers.
+         */
+        u16 cap_ptr;
+
         /* pci domain of this IOMMU */
         u16 pci_seg;
 
@@ -284,19 +284,19 @@ struct amd_iommu {
         /* size of command buffer */
         u32 cmd_buf_size;
 
-        /* event buffer virtual address */
-        u8 *evt_buf;
         /* size of event buffer */
         u32 evt_buf_size;
+        /* event buffer virtual address */
+        u8 *evt_buf;
         /* MSI number for event interrupt */
         u16 evt_msi_num;
 
-        /* if one, we need to send a completion wait command */
-        int need_sync;
-
         /* true if interrupts for this IOMMU are already enabled */
         bool int_enabled;
 
+        /* if one, we need to send a completion wait command */
+        int need_sync;
+
         /* default dma_ops domain for that IOMMU */
         struct dma_ops_domain *default_dom;
 };
diff --git a/arch/x86/include/asm/dma-mapping.h b/arch/x86/include/asm/dma-mapping.h
index 7f225a4b2a26..097794ff6b79 100644
--- a/arch/x86/include/asm/dma-mapping.h
+++ b/arch/x86/include/asm/dma-mapping.h
@@ -71,15 +71,13 @@ static inline struct dma_mapping_ops *get_dma_ops(struct device *dev)
 /* Make sure we keep the same behaviour */
 static inline int dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
 {
-#ifdef CONFIG_X86_32
-        return 0;
-#else
+#ifdef CONFIG_X86_64
         struct dma_mapping_ops *ops = get_dma_ops(dev);
         if (ops->mapping_error)
                 return ops->mapping_error(dev, dma_addr);
 
-        return (dma_addr == bad_dma_address);
 #endif
+        return (dma_addr == bad_dma_address);
 }
 
 #define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f)
diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c
index e4899e0e8787..5662e226b0c9 100644
--- a/arch/x86/kernel/amd_iommu.c
+++ b/arch/x86/kernel/amd_iommu.c
@@ -187,6 +187,8 @@ static int iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd)
 
         spin_lock_irqsave(&iommu->lock, flags);
         ret = __iommu_queue_command(iommu, cmd);
+        if (!ret)
+                iommu->need_sync = 1;
         spin_unlock_irqrestore(&iommu->lock, flags);
 
         return ret;
@@ -210,10 +212,13 @@ static int iommu_completion_wait(struct amd_iommu *iommu)
         cmd.data[0] = CMD_COMPL_WAIT_INT_MASK;
         CMD_SET_TYPE(&cmd, CMD_COMPL_WAIT);
 
-        iommu->need_sync = 0;
-
         spin_lock_irqsave(&iommu->lock, flags);
 
+        if (!iommu->need_sync)
+                goto out;
+
+        iommu->need_sync = 0;
+
         ret = __iommu_queue_command(iommu, &cmd);
 
         if (ret)
@@ -254,8 +259,6 @@ static int iommu_queue_inv_dev_entry(struct amd_iommu *iommu, u16 devid)
 
         ret = iommu_queue_command(iommu, &cmd);
 
-        iommu->need_sync = 1;
-
         return ret;
 }
 
@@ -281,8 +284,6 @@ static int iommu_queue_inv_iommu_pages(struct amd_iommu *iommu,
 
         ret = iommu_queue_command(iommu, &cmd);
 
-        iommu->need_sync = 1;
-
         return ret;
 }
 
@@ -762,8 +763,6 @@ static void set_device_domain(struct amd_iommu *iommu,
         write_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
 
         iommu_queue_inv_dev_entry(iommu, devid);
-
-        iommu->need_sync = 1;
 }
 
 /*****************************************************************************
@@ -858,6 +857,9 @@ static int get_device_resources(struct device *dev,
                 print_devid(_bdf, 1);
         }
 
+        if (domain_for_device(_bdf) == NULL)
+                set_device_domain(*iommu, *domain, _bdf);
+
         return 1;
 }
 
@@ -1031,8 +1033,7 @@ static dma_addr_t map_single(struct device *dev, phys_addr_t paddr,
         if (addr == bad_dma_address)
                 goto out;
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
 out:
         spin_unlock_irqrestore(&domain->lock, flags);
@@ -1060,8 +1061,7 @@ static void unmap_single(struct device *dev, dma_addr_t dma_addr,
 
         __unmap_single(iommu, domain->priv, dma_addr, size, dir);
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
         spin_unlock_irqrestore(&domain->lock, flags);
 }
@@ -1127,8 +1127,7 @@ static int map_sg(struct device *dev, struct scatterlist *sglist,
                         goto unmap;
         }
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
 out:
         spin_unlock_irqrestore(&domain->lock, flags);
@@ -1173,8 +1172,7 @@ static void unmap_sg(struct device *dev, struct scatterlist *sglist,
                 s->dma_address = s->dma_length = 0;
         }
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
         spin_unlock_irqrestore(&domain->lock, flags);
 }
@@ -1225,8 +1223,7 @@ static void *alloc_coherent(struct device *dev, size_t size,
                 goto out;
         }
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
 out:
         spin_unlock_irqrestore(&domain->lock, flags);
@@ -1257,8 +1254,7 @@ static void free_coherent(struct device *dev, size_t size,
 
         __unmap_single(iommu, domain->priv, dma_addr, size, DMA_BIDIRECTIONAL);
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
         spin_unlock_irqrestore(&domain->lock, flags);
 
diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c
index a42b02b4df68..ba7ad83e20a8 100644
--- a/arch/x86/kernel/pci-gart_64.c
+++ b/arch/x86/kernel/pci-gart_64.c
@@ -123,6 +123,8 @@ static void free_iommu(unsigned long offset, int size)
 
         spin_lock_irqsave(&iommu_bitmap_lock, flags);
         iommu_area_free(iommu_gart_bitmap, offset, size);
+        if (offset >= next_bit)
+                next_bit = offset + size;
         spin_unlock_irqrestore(&iommu_bitmap_lock, flags);
 }