Merge tag 'iommu-updates-v3.8' of git://git.kernel.org/pub/scm/linux/…

…kernel/git/joro/iommu

Pull IOMMU updates from Joerg Roedel:
 "A few new features this merge-window.  The most important one is
  probably, that dma-debug now warns if a dma-handle is not checked with
  dma_mapping_error by the device driver.  This requires minor changes
  to some architectures which make use of dma-debug.  Most of these
  changes have the respective Acks by the Arch-Maintainers.

  Besides that there are updates to the AMD IOMMU driver for refactor
  the IOMMU-Groups support and to make sure it does not trigger a
  hardware erratum.

  The OMAP changes (for which I pulled in a branch from Tony Lindgren's
  tree) have a conflict in linux-next with the arm-soc tree.  The
  conflict is in the file arch/arm/mach-omap2/clock44xx_data.c which is
  deleted in the arm-soc tree.  It is safe to delete the file too so
  solve the conflict.  Similar changes are done in the arm-soc tree in
  the common clock framework migration.  A missing hunk from the patch
  in the IOMMU tree will be submitted as a seperate patch when the
  merge-window is closed."

* tag 'iommu-updates-v3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (29 commits)
  ARM: dma-mapping: support debug_dma_mapping_error
  ARM: OMAP4: hwmod data: ipu and dsp to use parent clocks instead of leaf clocks
  iommu/omap: Adapt to runtime pm
  iommu/omap: Migrate to hwmod framework
  iommu/omap: Keep mmu enabled when requested
  iommu/omap: Remove redundant clock handling on ISR
  iommu/amd: Remove obsolete comment
  iommu/amd: Don't use 512GB pages
  iommu/tegra: smmu: Move bus_set_iommu after probe for multi arch
  iommu/tegra: gart: Move bus_set_iommu after probe for multi arch
  iommu/tegra: smmu: Remove unnecessary PTC/TLB flush all
  tile: dma_debug: add debug_dma_mapping_error support
  sh: dma_debug: add debug_dma_mapping_error support
  powerpc: dma_debug: add debug_dma_mapping_error support
  mips: dma_debug: add debug_dma_mapping_error support
  microblaze: dma-mapping: support debug_dma_mapping_error
  ia64: dma_debug: add debug_dma_mapping_error support
  c6x: dma_debug: add debug_dma_mapping_error support
  ARM64: dma_debug: add debug_dma_mapping_error support
  intel-iommu: Prevent devices with RMRRs from being placed into SI Domain
  ...

Documentation/DMA-API-HOWTO.txt

@@ -468,11 +468,46 @@ To map a single region, you do:
 	size_t size = buffer->len;
 
 	dma_handle = dma_map_single(dev, addr, size, direction);
+	if (dma_mapping_error(dma_handle)) {
+		/*
+		 * reduce current DMA mapping usage,
+		 * delay and try again later or
+		 * reset driver.
+		 */
+		goto map_error_handling;
+	}
 
 and to unmap it:
 
 	dma_unmap_single(dev, dma_handle, size, direction);
 
+You should call dma_mapping_error() as dma_map_single() could fail and return
+error. Not all dma implementations support dma_mapping_error() interface.
+However, it is a good practice to call dma_mapping_error() interface, which
+will invoke the generic mapping error check interface. Doing so will ensure
+that the mapping code will work correctly on all dma implementations without
+any dependency on the specifics of the underlying implementation. Using the
+returned address without checking for errors could result in failures ranging
+from panics to silent data corruption. Couple of example of incorrect ways to
+check for errors that make assumptions about the underlying dma implementation
+are as follows and these are applicable to dma_map_page() as well.
+
+Incorrect example 1:
+	dma_addr_t dma_handle;
+
+	dma_handle = dma_map_single(dev, addr, size, direction);
+	if ((dma_handle & 0xffff != 0) || (dma_handle >= 0x1000000)) {
+		goto map_error;
+	}
+
+Incorrect example 2:
+	dma_addr_t dma_handle;
+
+	dma_handle = dma_map_single(dev, addr, size, direction);
+	if (dma_handle == DMA_ERROR_CODE) {
+		goto map_error;
+	}
+
 You should call dma_unmap_single when the DMA activity is finished, e.g.
 from the interrupt which told you that the DMA transfer is done.
 
@@ -489,13 +524,27 @@ Specifically:
 	size_t size = buffer->len;
 
 	dma_handle = dma_map_page(dev, page, offset, size, direction);
+	if (dma_mapping_error(dma_handle)) {
+		/*
+		 * reduce current DMA mapping usage,
+		 * delay and try again later or
+		 * reset driver.
+		 */
+		goto map_error_handling;
+	}
 
 	...
 
 	dma_unmap_page(dev, dma_handle, size, direction);
 
 Here, "offset" means byte offset within the given page.
 
+You should call dma_mapping_error() as dma_map_page() could fail and return
+error as outlined under the dma_map_single() discussion.
+
+You should call dma_unmap_page when the DMA activity is finished, e.g.
+from the interrupt which told you that the DMA transfer is done.
+
 With scatterlists, you map a region gathered from several regions by:
 
 	int i, count = dma_map_sg(dev, sglist, nents, direction);
@@ -578,6 +627,14 @@ to use the dma_sync_*() interfaces.
 		dma_addr_t mapping;
 
 		mapping = dma_map_single(cp->dev, buffer, len, DMA_FROM_DEVICE);
+		if (dma_mapping_error(dma_handle)) {
+			/*
+			 * reduce current DMA mapping usage,
+			 * delay and try again later or
+			 * reset driver.
+			 */
+			goto map_error_handling;
+		}
 
 		cp->rx_buf = buffer;
 		cp->rx_len = len;
@@ -658,6 +715,75 @@ failure can be determined by:
 		 * delay and try again later or
 		 * reset driver.
 		 */
+		goto map_error_handling;
+	}
+
+- unmap pages that are already mapped, when mapping error occurs in the middle
+  of a multiple page mapping attempt. These example are applicable to
+  dma_map_page() as well.
+
+Example 1:
+	dma_addr_t dma_handle1;
+	dma_addr_t dma_handle2;
+
+	dma_handle1 = dma_map_single(dev, addr, size, direction);
+	if (dma_mapping_error(dev, dma_handle1)) {
+		/*
+		 * reduce current DMA mapping usage,
+		 * delay and try again later or
+		 * reset driver.
+		 */
+		goto map_error_handling1;
+	}
+	dma_handle2 = dma_map_single(dev, addr, size, direction);
+	if (dma_mapping_error(dev, dma_handle2)) {
+		/*
+		 * reduce current DMA mapping usage,
+		 * delay and try again later or
+		 * reset driver.
+		 */
+		goto map_error_handling2;
+	}
+
+	...
+
+	map_error_handling2:
+		dma_unmap_single(dma_handle1);
+	map_error_handling1:
+
+Example 2: (if buffers are allocated a loop, unmap all mapped buffers when
+	    mapping error is detected in the middle)
+
+	dma_addr_t dma_addr;
+	dma_addr_t array[DMA_BUFFERS];
+	int save_index = 0;
+
+	for (i = 0; i < DMA_BUFFERS; i++) {
+
+		...
+
+		dma_addr = dma_map_single(dev, addr, size, direction);
+		if (dma_mapping_error(dev, dma_addr)) {
+			/*
+			 * reduce current DMA mapping usage,
+			 * delay and try again later or
+			 * reset driver.
+			 */
+			goto map_error_handling;
+		}
+		array[i].dma_addr = dma_addr;
+		save_index++;
+	}
+
+	...
+
+	map_error_handling:
+
+	for (i = 0; i < save_index; i++) {
+
+		...
+
+		dma_unmap_single(array[i].dma_addr);
 	}
 
 Networking drivers must call dev_kfree_skb to free the socket buffer

Documentation/DMA-API.txt

@@ -678,3 +678,15 @@ out of dma_debug_entries. These entries are preallocated at boot. The number
 of preallocated entries is defined per architecture. If it is too low for you
 boot with 'dma_debug_entries=<your_desired_number>' to overwrite the
 architectural default.
+
+void debug_dmap_mapping_error(struct device *dev, dma_addr_t dma_addr);
+
+dma-debug interface debug_dma_mapping_error() to debug drivers that fail
+to check dma mapping errors on addresses returned by dma_map_single() and
+dma_map_page() interfaces. This interface clears a flag set by
+debug_dma_map_page() to indicate that dma_mapping_error() has been called by
+the driver. When driver does unmap, debug_dma_unmap() checks the flag and if
+this flag is still set, prints warning message that includes call trace that
+leads up to the unmap. This interface can be called from dma_mapping_error()
+routines to enable dma mapping error check debugging.
+

arch/arm/include/asm/dma-mapping.h

@@ -91,6 +91,7 @@ static inline dma_addr_t virt_to_dma(struct device *dev, void *addr)
  */
 static inline int dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
 {
+	debug_dma_mapping_error(dev, dma_addr);
 	return dma_addr == DMA_ERROR_CODE;
 }
 

arch/arm/mach-omap2/devices.c

@@ -226,7 +226,7 @@ static struct platform_device omap3isp_device = {
 };
 
 static struct omap_iommu_arch_data omap3_isp_iommu = {
-	.name = "isp",
+	.name = "mmu_isp",
 };
 
 int omap3_init_camera(struct isp_platform_data *pdata)

arch/arm/mach-omap2/omap-iommu.c

@@ -12,153 +12,60 @@
 
 #include <linux/module.h>
 #include <linux/platform_device.h>
+#include <linux/err.h>
+#include <linux/slab.h>
 
 #include <linux/platform_data/iommu-omap.h>
+#include <plat/omap_hwmod.h>
+#include <plat/omap_device.h>
 
-#include "soc.h"
-#include "common.h"
-
-struct iommu_device {
-	resource_size_t base;
-	int irq;
-	struct iommu_platform_data pdata;
-	struct resource res[2];
-};
-static struct iommu_device *devices;
-static int num_iommu_devices;
-
-#ifdef CONFIG_ARCH_OMAP3
-static struct iommu_device omap3_devices[] = {
-	{
-		.base = 0x480bd400,
-		.irq = 24 + OMAP_INTC_START,
-		.pdata = {
-			.name = "isp",
-			.nr_tlb_entries = 8,
-			.clk_name = "cam_ick",
-			.da_start = 0x0,
-			.da_end = 0xFFFFF000,
-		},
-	},
-#if defined(CONFIG_OMAP_IOMMU_IVA2)
-	{
-		.base = 0x5d000000,
-		.irq = 28 + OMAP_INTC_START,
-		.pdata = {
-			.name = "iva2",
-			.nr_tlb_entries = 32,
-			.clk_name = "iva2_ck",
-			.da_start = 0x11000000,
-			.da_end = 0xFFFFF000,
-		},
-	},
-#endif
-};
-#define NR_OMAP3_IOMMU_DEVICES ARRAY_SIZE(omap3_devices)
-static struct platform_device *omap3_iommu_pdev[NR_OMAP3_IOMMU_DEVICES];
-#else
-#define omap3_devices		NULL
-#define NR_OMAP3_IOMMU_DEVICES	0
-#define omap3_iommu_pdev	NULL
-#endif
-
-#ifdef CONFIG_ARCH_OMAP4
-static struct iommu_device omap4_devices[] = {
-	{
-		.base = OMAP4_MMU1_BASE,
-		.irq = 100 + OMAP44XX_IRQ_GIC_START,
-		.pdata = {
-			.name = "ducati",
-			.nr_tlb_entries = 32,
-			.clk_name = "ipu_fck",
-			.da_start = 0x0,
-			.da_end = 0xFFFFF000,
-		},
-	},
-	{
-		.base = OMAP4_MMU2_BASE,
-		.irq = 28 + OMAP44XX_IRQ_GIC_START,
-		.pdata = {
-			.name = "tesla",
-			.nr_tlb_entries = 32,
-			.clk_name = "dsp_fck",
-			.da_start = 0x0,
-			.da_end = 0xFFFFF000,
-		},
-	},
-};
-#define NR_OMAP4_IOMMU_DEVICES ARRAY_SIZE(omap4_devices)
-static struct platform_device *omap4_iommu_pdev[NR_OMAP4_IOMMU_DEVICES];
-#else
-#define omap4_devices		NULL
-#define NR_OMAP4_IOMMU_DEVICES	0
-#define omap4_iommu_pdev	NULL
-#endif
-
-static struct platform_device **omap_iommu_pdev;
-
-static int __init omap_iommu_init(void)
+static int __init omap_iommu_dev_init(struct omap_hwmod *oh, void *unused)
 {
-	int i, err;
-	struct resource res[] = {
-		{ .flags = IORESOURCE_MEM },
-		{ .flags = IORESOURCE_IRQ },
-	};
+	struct platform_device *pdev;
+	struct iommu_platform_data *pdata;
+	struct omap_mmu_dev_attr *a = (struct omap_mmu_dev_attr *)oh->dev_attr;
+	static int i;
+
+	pdata = kzalloc(sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	pdata->name = oh->name;
+	pdata->nr_tlb_entries = a->nr_tlb_entries;
+	pdata->da_start = a->da_start;
+	pdata->da_end = a->da_end;
+
+	if (oh->rst_lines_cnt == 1) {
+		pdata->reset_name = oh->rst_lines->name;
+		pdata->assert_reset = omap_device_assert_hardreset;
+		pdata->deassert_reset = omap_device_deassert_hardreset;
+	}
 
-	if (cpu_is_omap34xx()) {
-		devices = omap3_devices;
-		omap_iommu_pdev = omap3_iommu_pdev;
-		num_iommu_devices = NR_OMAP3_IOMMU_DEVICES;
-	} else if (cpu_is_omap44xx()) {
-		devices = omap4_devices;
-		omap_iommu_pdev = omap4_iommu_pdev;
-		num_iommu_devices = NR_OMAP4_IOMMU_DEVICES;
-	} else
-		return -ENODEV;
+	pdev = omap_device_build("omap-iommu", i, oh, pdata, sizeof(*pdata),
+				NULL, 0, 0);
 
-	for (i = 0; i < num_iommu_devices; i++) {
-		struct platform_device *pdev;
-		const struct iommu_device *d = &devices[i];
+	kfree(pdata);
 
-		pdev = platform_device_alloc("omap-iommu", i);
-		if (!pdev) {
-			err = -ENOMEM;
-			goto err_out;
-		}
+	if (IS_ERR(pdev)) {
+		pr_err("%s: device build err: %ld\n", __func__, PTR_ERR(pdev));
+		return PTR_ERR(pdev);
+	}
 
-		res[0].start = d->base;
-		res[0].end = d->base + MMU_REG_SIZE - 1;
-		res[1].start = res[1].end = d->irq;
+	i++;
 
-		err = platform_device_add_resources(pdev, res,
-						    ARRAY_SIZE(res));
-		if (err)
-			goto err_out;
-		err = platform_device_add_data(pdev, &d->pdata,
-					       sizeof(d->pdata));
-		if (err)
-			goto err_out;
-		err = platform_device_add(pdev);
-		if (err)
-			goto err_out;
-		omap_iommu_pdev[i] = pdev;
-	}
 	return 0;
+}
 
-err_out:
-	while (i--)
-		platform_device_put(omap_iommu_pdev[i]);
-	return err;
+static int __init omap_iommu_init(void)
+{
+	return omap_hwmod_for_each_by_class("mmu", omap_iommu_dev_init, NULL);
 }
 /* must be ready before omap3isp is probed */
 subsys_initcall(omap_iommu_init);
 
 static void __exit omap_iommu_exit(void)
 {
-	int i;
-
-	for (i = 0; i < num_iommu_devices; i++)
-		platform_device_unregister(omap_iommu_pdev[i]);
+	/* Do nothing */
 }
 module_exit(omap_iommu_exit);