[PATCH v2] gpu: drm: Use dev_err_probe instead of dev_err

[PATCH v2] gpu: drm: Use dev_err_probe instead of dev_err

[Wang Ming]

It is possible that dma_request_chan will return EPROBE_DEFER,
which means that disp->dev is not ready yet. In this case,
dev_err(disp->dev), there will be no output. This patch fixes the bug.

Signed-off-by: Wang Ming <machel@vivo.com>
---
 drivers/gpu/drm/xlnx/zynqmp_disp.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/xlnx/zynqmp_disp.c b/drivers/gpu/drm/xlnx/zynqmp_disp.c
index 3b87eebddc97..676f09fe8594 100644
--- a/drivers/gpu/drm/xlnx/zynqmp_disp.c
+++ b/drivers/gpu/drm/xlnx/zynqmp_disp.c
@@ -1094,8 +1094,8 @@ static int zynqmp_disp_layer_request_dma(struct zynqmp_disp *disp,
 			 "%s%u", dma_names[layer->id], i);
 		dma->chan = dma_request_chan(disp->dev, dma_channel_name);
 		if (IS_ERR(dma->chan)) {
-			dev_err(disp->dev, "failed to request dma channel\n");
-			ret = PTR_ERR(dma->chan);
+			ret = dev_err_probe(disp->dev, PTR_ERR(dma->chan),
+				"failed to request dma channel\n");
 			dma->chan = NULL;
 			return ret;
 		}
-- 
2.25.1

Re: [PATCH v2] gpu: drm: Use dev_err_probe instead of dev_err

[Laurent Pinchart]

Hi Wang,

Thank you for the patch.

On Wed, Jul 26, 2023 at 07:57:56PM +0800, Wang Ming wrote:
> It is possible that dma_request_chan will return EPROBE_DEFER,
> which means that disp->dev is not ready yet. In this case,
> dev_err(disp->dev), there will be no output. This patch fixes the bug.

It's not a bug. The existing code works as expected. dev_err_probe() is
nicer though, as it records the reason for the probe deferral. Here's a
proposal for a better commit message:

  It is possible that dma_request_chan() returns EPROBE_DEFER, in which
  case the driver defers probing without printing any message. Use
  dev_err_probe() to record the probe deferral cause and ease debugging.

> Signed-off-by: Wang Ming <machel@vivo.com>
> ---
>  drivers/gpu/drm/xlnx/zynqmp_disp.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xlnx/zynqmp_disp.c b/drivers/gpu/drm/xlnx/zynqmp_disp.c
> index 3b87eebddc97..676f09fe8594 100644
> --- a/drivers/gpu/drm/xlnx/zynqmp_disp.c
> +++ b/drivers/gpu/drm/xlnx/zynqmp_disp.c
> @@ -1094,8 +1094,8 @@ static int zynqmp_disp_layer_request_dma(struct zynqmp_disp *disp,
>  			 "%s%u", dma_names[layer->id], i);
>  		dma->chan = dma_request_chan(disp->dev, dma_channel_name);
>  		if (IS_ERR(dma->chan)) {
> -			dev_err(disp->dev, "failed to request dma channel\n");
> -			ret = PTR_ERR(dma->chan);
> +			ret = dev_err_probe(disp->dev, PTR_ERR(dma->chan),
> +				"failed to request dma channel\n");

The alignment is wrong, this should be

			ret = dev_err_probe(disp->dev, PTR_ERR(dma->chan),
					    "failed to request dma channel\n");

If you're fine with those changes, I can fix this locally, there's no
need to resubmit.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

>  			dma->chan = NULL;
>  			return ret;
>  		}

-- 
Regards,

Laurent Pinchart