[lkml]   [2009]   [Mar]   [31]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
Patch in this message
Subject[PATCH] dmaengine: fix regression introduced by d6103085dfd83c13db65c3bd7e182f021d77c541
chan is an index variable, used to loop over a list of channels, and here 
it is used _after_ the loop, in which case it doesn't point to a DMA
channel struct anymore. Dereferencing it leads to a corruption of a random
memory location, which in my case was a pointer inside a clock struct. Fix
it by using a local variable pointing to the DMA device.

Signed-off-by: Guennadi Liakhovetski <>

Guys, this was a _real_ bad joke, cost me almost a day, and that patch has
been reviewed by two persons...

So far 2.6.29(-next) has been very bad for me, regressions all over the
place, lots of wasted time hunting them down:-(

diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c
index 59e0fb2..92438e9 100644
--- a/drivers/dma/dmaengine.c
+++ b/drivers/dma/dmaengine.c
@@ -726,7 +726,7 @@ int dma_async_device_register(struct dma_device *device)
list_add_tail_rcu(&device->global_node, &dma_device_list);
if (dma_has_cap(DMA_PRIVATE, device->cap_mask))
- chan->device->privatecnt++; /* Always private */
+ device->privatecnt++; /* Always private */

 \ /
  Last update: 2009-03-31 18:09    [W:0.032 / U:0.460 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site