]> bbs.cooldavid.org Git - net-next-2.6.git/commit
dma: dmatest: fix potential sign bug
authorKulikov Vasiliy <segooon@gmail.com>
Sat, 17 Jul 2010 15:19:48 +0000 (19:19 +0400)
committerDan Williams <dan.j.williams@intel.com>
Wed, 4 Aug 2010 21:27:47 +0000 (14:27 -0700)
commitb9033e682e86f3c6a66763f9b6a3935c5c64e145
tree8aeeb08aeae97b12f0c6cb7338d8cbcf15e6f8ac
parent556ab45f9a775bfa4762bacc0a4afb5b44b067bc
dma: dmatest: fix potential sign bug

'cnt' is unsigned, so this code may become wrong in future as
dmatest_add_threads() can return error code:

cnt = dmatest_add_threads(dtc, DMA_MEMCPY);
thread_count += cnt > 0 ? cnt : 0;
        ^^^^^^^

Now it can return only -EINVAL if and only if second argument of
dmatest_add_threads() is not one of DMA_MEMCPY, DMA_XOR, DMA_PQ.
So, now it is not wrong but may become wrong in future.

The semantic patch that finds this problem (many false-positive results):
(http://coccinelle.lip6.fr/)

// <smpl>
@ r1 @
identifier f;
@@
int f(...) { ... }

@@
identifier r1.f;
type T;
unsigned T x;
@@

*x = f(...)
 ...
*x > 0

Signed-off-by: Kulikov Vasiliy <segooon@gmail.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
drivers/dma/dmatest.c