WIP: Clean memory sanitizer warnings around GEOS code #265
Conversation
| @@ -378,6 +378,8 @@ static char * lwdouble_to_dms(double val, const char *pos_dir_symbol, const char | |||
|
|
|||
| /* Allocate space for the result. Leave plenty of room for excess digits, negative sign, etc.*/ | |||
| result = (char*)lwalloc(format_length + WORK_SIZE); | |||
| memset(result, 0, format_length + WORK_SIZE); | |||
There was a problem hiding this comment.
Do you think we should add a lwcalloc function that delegates to calloc on the lwgeom side and palloc0 on the postgis side?
There was a problem hiding this comment.
I've thought about that and it might help, but at the end of of the day if you don't use memset you probably won't use lwcalloc either.
Also, this case looks like a false positive as I don't see any way for the memory outside of the set string to be read, but since it's an output function and memset shouldn't any noticeable impact in performance I've decided to set it and make the sanitizer output clearer.
There was a problem hiding this comment.
grep -R -B1 "memset(" --include "*.c" | grep alloc | wc shows 20 or so usages. Seems worth it to me.
There was a problem hiding this comment.
Most of those seem uses to set simple structs, but I wouldn't oppose for it to be added in a followup issue.
|
Trac issue: https://trac.osgeo.org/postgis/ticket/4118 |
Algunenano
force-pushed
the
sanitizer_memory_geos
branch
from
e2aa181
to
8fa848f
Compare
Shut up some of the warnings I get when running unit tests with -fsanitizer=memory