Hi Suman,

On 12/23/2013 08:48 PM, Anna, Suman wrote:

Hi Florian,

On 12/17/2013 06:53 AM, Florian Vaussard wrote:

quoted

As OMAP2+ is moving to a full DT boot for 3.14, commit 7ce93f3
"ARM: OMAP2+: Fix more missing data for omap3.dtsi file" adds
basic DT bits. But the driver is not yet converted, so this will
not work and driver will not be probed. Convert it!

Apart from standard bindings, this patch uses 'dma-window' (already
used by Tegra SMMU) and adds a custom 'ti,#tlb-entries' binding.

Signed-off-by: Florian Vaussard <redacted>
---
  .../devicetree/bindings/iommu/ti,omap-iommu.txt    | 19 ++++++++++++
  arch/arm/mach-omap2/omap-iommu.c                   |  5 +++
  drivers/iommu/omap-iommu.c                         | 36
+++++++++++++++++++---
  3 files changed, 55 insertions(+), 5 deletions(-)
  create mode 100644
Documentation/devicetree/bindings/iommu/ti,omap-iommu.txt

diff --git a/Documentation/devicetree/bindings/iommu/ti,omap-iommu.txt

b/Documentation/devicetree/bindings/iommu/ti,omap-iommu.txt
new file mode 100644
index 0000000..4e5027c

--- /dev/null
+++ b/Documentation/devicetree/bindings/iommu/ti,omap-iommu.txt

@@ -0,0 +1,19 @@
+OMAP3 IOMMU used by camera subsystem

As I mentioned in comments on your cover-letter, the current binding is
only limited to OMAP3 ISP, but the driver is common for all OMAP2+ SoCs.
This binding an update to handle all SoCs.

To summarize the differences, OMAP2/3 has the same MMU register set for
all remote processor MMUs (IVA/DSP/ISP) with the only difference being
the number of TLBs supported (8 for ISP on OMAP2/3 and 32 for all other
instances on all OMAP2+ SoCs). Depending on whether the compatibility
property is defined for an SoC or for the processor, "ti,#tlb-entries"
can be retrieved directly from the match data or made as an optional
property, with the default value being 32.

It makes sense.

OMAP4 and OMAP5 have some additional registers (MMU_FAULT_PC,
MMU_FAULT_STATUS, MMU_GP_REG/DSPSS_MMU_GPR) on top of the OMAP2/OMAP3
ones, and DRA7 a further superset of OMAP4/OMAP5. The only difference is
the definition of the MMU_GPR register, which is different between the
IPU and DSP processors, but present at the same offset, and dictates the
validity of the MMU_FAULT_PC and MMU_FAULT_STATUS registers. The MMU_GPR
register is also not defined on OMAP4430 ES1.0, but is present on all
other newer SoCs.

Thank you for this detailed comparison ! I do not know much about
OMAP4 and OMAP5. In the current implementation, I was unable to find
references to these specificities. Is this currently supported?

quoted

+
+Required properties:
+- compatible : "ti,omap3-mmu-isp"

My suggestion would be to use
    "ti,omap2-iommu" for OMAP2/OMAP3 MMUs with the optional
"ti,#tlb-entries" to distinguish ISP MMU.
    "ti,omap4-iommu" for OMAP4/OMAP5 MMUs with an additional boolean
property to distinguish the presence/functionality of the MMU_GPR register
    "ti,dra7-iommu" for DRA7 MMUs

Mark, Tony,
Do you have any other recommendations given the above summary?

quoted

+- ti,hwmods : "mmu_isp"

Replace with general description

You mean something like "Name of the hwmod associated to the IOMMU"?

quoted

+- reg : address space for the configuration registers
+- interrupts : interrupt line
+- dma-window : IOVA start address and length.
+- ti,#tlb-entries : number of entries in the translation look-aside
buffer
+
+Example:
+    mmu_isp: mmu_isp at 480bd400 {
+        compatible = "ti,omap3-mmu-isp";
+        ti,hwmods = "mmu_isp";
+        reg = <0x480bd400 0x80>;
+        interrupts = <8>;
+        dma-window = <0 0xfffff000>;
+        ti,#tlb-entries = <8>;
+    };

diff --git a/arch/arm/mach-omap2/omap-iommu.c

b/arch/arm/mach-omap2/omap-iommu.c
index f6daae8..f1fab56 100644

--- a/arch/arm/mach-omap2/omap-iommu.c
+++ b/arch/arm/mach-omap2/omap-iommu.c

@@ -10,6 +10,7 @@
   * published by the Free Software Foundation.
   */

+#include <linux/of.h>
  #include <linux/module.h>
  #include <linux/platform_device.h>
  #include <linux/err.h>

@@ -58,6 +59,10 @@ static int __init omap_iommu_dev_init(struct

omap_hwmod *oh, void *unused)

  static int __init omap_iommu_init(void)
  {
+    /* If dtb is there, the devices will be created dynamically */
+    if (of_have_populated_dt())
+        return -ENODEV;
+
      return omap_hwmod_for_each_by_class("mmu", omap_iommu_dev_init,
NULL);
  }
  /* must be ready before omap3isp is probed */

diff --git a/drivers/iommu/omap-iommu.c b/drivers/iommu/omap-iommu.c
index 385bf5e..51efcc4 100644
--- a/drivers/iommu/omap-iommu.c
+++ b/drivers/iommu/omap-iommu.c

@@ -23,6 +23,9 @@
  #include <linux/spinlock.h>
  #include <linux/io.h>
  #include <linux/pm_runtime.h>
+#include <linux/of.h>
+#include <linux/of_iommu.h>
+#include <linux/of_irq.h>

  #include <asm/cacheflush.h>

@@ -1171,20 +1174,30 @@ static int omap_iommu_probe(struct

platform_device *pdev)
  {
      int err = -ENODEV;
      int irq;
+    size_t len;
      struct omap_iommu *obj;
      struct resource *res;
      struct iommu_platform_data *pdata = pdev->dev.platform_data;
+    struct device_node *of = pdev->dev.of_node;

      obj = kzalloc(sizeof(*obj) + MMU_REG_SIZE, GFP_KERNEL);
      if (!obj)
          return -ENOMEM;

-    obj->nr_tlb_entries = pdata->nr_tlb_entries;
-    obj->name = pdata->name;
+    if (of) {
+        obj->name = of->name;
+        of_property_read_u32(of, "ti,#tlb-entries",
+                     &obj->nr_tlb_entries);
+        of_get_dma_window(of, NULL, 0, NULL, &obj->da_start, &len);

There is no error checking performed on both the of functions.

Indeed.

quoted

+        obj->da_end = obj->da_start + len;
+    } else {
+        obj->nr_tlb_entries = pdata->nr_tlb_entries;
+        obj->name = pdata->name;
+        obj->da_start = pdata->da_start;
+        obj->da_end = pdata->da_end;
+    }
      obj->dev = &pdev->dev;
      obj->ctx = (void *)obj + sizeof(*obj);
-    obj->da_start = pdata->da_start;
-    obj->da_end = pdata->da_end;

      spin_lock_init(&obj->iommu_lock);
      mutex_init(&obj->mmap_lock);

@@ -1210,7 +1223,11 @@ static int omap_iommu_probe(struct

platform_device *pdev)
          goto err_ioremap;
      }

-    irq = platform_get_irq(pdev, 0);
+    if (of)
+        irq = irq_of_parse_and_map(of, 0);
+    else
+        irq = platform_get_irq(pdev, 0);

The platform_get_irq should still work just fine, so this change can
probably be left out. We can switch once the driver supports DT-only
devices completely.

Ok I will test when back at my office.

Thank you!

Regards,
Florian