| commit | d09cb7c2fd2425eabe4bca9b9b85838b8beec7cf | |
| author | Dan Carpenter <error27@gmail.com> | |
| Thu, 23 Apr 2009 15:59:52 +0000 (23 18:59 +0300) | ||
| committer | Dan Carpenter <error27@gmail.com> | |
| Thu, 23 Apr 2009 15:59:52 +0000 (23 18:59 +0300) | ||
| tree | a39ba81a4c2ea339a4f66b1a5eb8d7a16ca7107c | treesnapshot (tar.gz zip) |
| parent | e1729dd11e7d8b9862575698b63881a8e63aa7f4 | commitdiff |
Set things in smatch extra instead of smatch implied.
When I reverted 4469500cba9 it broke things.
The bug is like this:
family is either 2 or 10 in cur_slist.
if (family == PF_INET) {
smatch extra sets the true path to 2 and the false path to 10. But _also_
that sets the cur_slist to 2. Smatch extra sees that cur_slist can only
be 2 so it sets family to the empty set in the false path.
When I reverted 4469500cba9 I mentioned that it would be nice to
preserve the ->my_pools. Actually, that is wrong. We want the most recent
branches in ->my_pools.
With this patch I still get what I need for handling switch statements.
Signed-off-by: Dan Carpenter <error27@gmail.com>
When I reverted 4469500cba9 it broke things.
The bug is like this:
family is either 2 or 10 in cur_slist.
if (family == PF_INET) {
smatch extra sets the true path to 2 and the false path to 10. But _also_
that sets the cur_slist to 2. Smatch extra sees that cur_slist can only
be 2 so it sets family to the empty set in the false path.
When I reverted 4469500cba9 I mentioned that it would be nice to
preserve the ->my_pools. Actually, that is wrong. We want the most recent
branches in ->my_pools.
With this patch I still get what I need for handling switch statements.
Signed-off-by: Dan Carpenter <error27@gmail.com>
| smatch_implied.c | diffblobblamehistory | |
| smatch_slist.c | diffblobblamehistory |