Hi Steve,

On 2/26/2011 3:39 AM, Stepan Moskovchenko wrote:

On 2/24/2011 11:29 PM, Trilok Soni wrote:

quoted

Hi Steve,

Hello

quoted

quoted

@@ -130,117 +131,134 @@ static int msm_iommu_probe(struct platform_device *pdev)
  {
      struct resource *r, *r2;
      struct clk *iommu_clk;
+    struct clk *iommu_pclk;
      struct msm_iommu_drvdata *drvdata;
      struct msm_iommu_dev *iommu_dev = pdev->dev.platform_data;

const please.

I am not clear on what you mean. Please be more specific here.

I mean

const struct msm_iommu_dev *iommu_dev = pdev->dev.platform_data; 

Now I see that this change is not part of this patch so you can ignore my comment.

quoted

quoted

+    iommu_pclk = clk_get(NULL, "smmu_pclk");
+    if (IS_ERR(iommu_pclk)) {
+        ret = -ENODEV;
+        goto fail;
+    }

I am not a big fan of this when you have the "device" around. You should just do

iommu_pclk = clk_get(&pdev->dev, NULL);

...error logic...

iommu_clk = clk_get(&pdev->dev, "iommu_clk");

...error logic...

The pclk is a "special" bus clock and does not have a specific device instance associated with it, so passing a device would not be appropriate in this case. I pass the device for other clocks that are indeed associated with devices, but this is not one of them. I suppose we could create 11 or 12 aliases for the pclk and associate it with all the IOMMU devices, but I would prefer to avoid doing that as I believe the current approach is cleaner.

ok.

quoted

quoted

-            ret = -EBUSY;
-            goto fail;
-        }
+    len = r->end - r->start + 1;

resource_size please.

Ok

Thanks.

---Trilok Soni

-- 
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.