[PATCH] video: fbdev: Fix an errro handling path in 'au1200fb_drv_probe()'

If 'dmam_alloc_attrs()' fails, we must go through the error handling code,
as done elsewhere in this function. Otherwise, there is a resource leak.

Signed-off-by: Christophe JAILLET <redacted>
---
I'm also puzzled by the 'framebuffer_alloc()' call a few lines above.
'ret' is known to be 0 at this point. I guess that -ENOMEM should also be
returned.
---
 drivers/video/fbdev/au1200fb.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/video/fbdev/au1200fb.c b/drivers/video/fbdev/au1200fb.c
index 5f04b4096c42..99d6cfb168b5 100644
--- a/drivers/video/fbdev/au1200fb.c
+++ b/drivers/video/fbdev/au1200fb.c

@@ -1701,7 +1701,8 @@ static int au1200fb_drv_probe(struct platform_device *dev)
 		if (!fbdev->fb_mem) {
 			print_err("fail to allocate frambuffer (size: %dK))",
 				  fbdev->fb_len / 1024);
-			return -ENOMEM;
+			ret = -ENOMEM;
+			goto failed;
 		}
 
 		/*

-- 
2.11.0

[ added dri-devel ML to cc: ]

On Tuesday, September 12, 2017 07:39:30 AM Christophe JAILLET wrote:

If 'dmam_alloc_attrs()' fails, we must go through the error handling code,
as done elsewhere in this function. Otherwise, there is a resource leak.

Signed-off-by: Christophe JAILLET <redacted>
---
I'm also puzzled by the 'framebuffer_alloc()' call a few lines above.
'ret' is known to be 0 at this point. I guess that -ENOMEM should also be
returned.

Yes, moreover the "failed:" error path is incomplete (please take
a look at au1200fb_drv_remove() for comparison) and needs to be fixed.

Could you please take care of it?

Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics

Le 12/10/2017 à 18:25, Bartlomiej Zolnierkiewicz a écrit :

[ added dri-devel ML to cc: ]

On Tuesday, September 12, 2017 07:39:30 AM Christophe JAILLET wrote:

quoted

If 'dmam_alloc_attrs()' fails, we must go through the error handling code,
as done elsewhere in this function. Otherwise, there is a resource leak.

Signed-off-by: Christophe JAILLET <redacted>
---
I'm also puzzled by the 'framebuffer_alloc()' call a few lines above.
'ret' is known to be 0 at this point. I guess that -ENOMEM should also be
returned.

Yes, moreover the "failed:" error path is incomplete (please take
a look at au1200fb_drv_remove() for comparison) and needs to be fixed.

Could you please take care of it?

I will try, but be indulgent :)

My understanding of the code is that they are several issues all related 
to this error handling path (double free, missing memory release, 
invalid irq release...)
I will try to sort it out, but my first tries will likely be incomplete 
and/or broken.

CJ