| commit | 5b3c30b59544df43d9ade5cca84b9c2aec1c53e6 | |
| author | Dan Carpenter <dan.carpenter@oracle.com> | |
| Fri, 14 Sep 2018 14:46:24 +0000 (14 17:46 +0300) | ||
| committer | Dan Carpenter <dan.carpenter@oracle.com> | |
| Fri, 14 Sep 2018 14:46:24 +0000 (14 17:46 +0300) | ||
| tree | 4d51d9733462e45c680151f0c3e3f711905ddbc0 | treesnapshot (tar.gz zip) |
| parent | f02a2f5053f0f64f425eb621822c90960e356950 | commitdiff |
param_limit: make mtag limits more generic
Say a function is only called with a known mtag parameter. Then it has
a check for NULL "if (!p) return -EINVAL;" that is an impossible path and
we CULL it. Fine. But then the success path is limited to a specific
mtag.
So then we add a new caller, which passes a different mtag. Then smatch is
like this mtag is impossible on every return path... And it creates a
bunch of problems. So what this does is we replace mtag limits with
"valid pointer" instead.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Say a function is only called with a known mtag parameter. Then it has
a check for NULL "if (!p) return -EINVAL;" that is an impossible path and
we CULL it. Fine. But then the success path is limited to a specific
mtag.
So then we add a new caller, which passes a different mtag. Then smatch is
like this mtag is impossible on every return path... And it creates a
bunch of problems. So what this does is we replace mtag limits with
"valid pointer" instead.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
| smatch_param_limit.c | diffblobblamehistory |