| commit | 11413c9990dfb63bb3a2e1bf3db4431ad364447a | |
| author | Dan Carpenter <error27@gmail.com> | |
| Sun, 2 May 2010 21:16:57 +0000 (2 23:16 +0200) | ||
| committer | Dan Carpenter <error27@gmail.com> | |
| Sun, 2 May 2010 21:16:57 +0000 (2 23:16 +0200) | ||
| tree | 5f1141e2c4e5982c551bc9bed34577a51814d3de | treesnapshot (tar.gz zip) |
| parent | 4f16abe8ca57628bfff8a831ce353d9186cb826e | commitdiff |
*new* smatch_scripts/unlocked_paths.pl
I was looking for race conditions caused by not taking a lock so I wrote
this script.
Say if you are always supposed to call frob() with with @foo_lock held then
first generate a warns.txt using the --call-tree option. Then the command
is: "unlocked_paths.pl warns.txt foo_lock frob"
It doesn't handle calling function pointers. The base of the path will
probably be something that gets assigned to a function pointer actually.
Signed-off-by: Dan Carpenter <error27@gmail.com>
I was looking for race conditions caused by not taking a lock so I wrote
this script.
Say if you are always supposed to call frob() with with @foo_lock held then
first generate a warns.txt using the --call-tree option. Then the command
is: "unlocked_paths.pl warns.txt foo_lock frob"
It doesn't handle calling function pointers. The base of the path will
probably be something that gets assigned to a function pointer actually.
Signed-off-by: Dan Carpenter <error27@gmail.com>
| check_call_tree.c | diffblobblamehistory | |
| check_locking.c | diffblobblamehistory | |
| smatch.h | diffblobblamehistory | |
| smatch_scripts/unlocked_paths.pl | [new file with mode: 0755] | blob |