romenskiy2012/jemalloc
1
0
Fork 0
mirror of https://github.com/jemalloc/jemalloc.git synced 2026-05-15 09:26:21 +03:00
Code Issues Projects Releases Packages Wiki Activity

Avoid a resource leak down extent split failure paths.

Browse source 
Previously, we would leak the extent and memory associated with a salvageable
portion of an extent that we were trying to split in three, in the case where
the first split attempt succeeded and the second failed.
This commit is contained in:
David Goldblatt 2018-04-17 13:16:42 -07:00 • committed by David Goldblatt
parent a62e42baeb
commit c95284df1a
1 changed files with 8 additions and 10 deletions
Download patch file Download diff file Expand all files Collapse all files

18
src/extent.c
View file

@ -1319,21 +1319,19 @@ extent_grow_retained(tsdn_t *tsdn, arena_t *arena,
* cant_alloc case should not occur.
*/
assert(result == extent_split_interior_error);
if (to_salvage != NULL) {
if (config_prof) {
extent_gdump_add(tsdn, to_salvage);
}
extent_record(tsdn, arena, r_extent_hooks,
&arena->extents_retained, to_salvage, true);
}
if (to_leak != NULL) {
extent_deregister_no_gdump_sub(tsdn, to_leak);
extents_leak(tsdn, arena, r_extent_hooks,
&arena->extents_retained, to_leak, true);
goto label_err;
}
/*
* Note: we don't handle the non-NULL to_salvage case at all.
* This maintains the behavior that was present when the
* refactor pulling extent_split_interior into a helper function
* was added. I think this is actually a bug (we leak both the
* memory and the extent_t in that case), but since this code is
* getting deleted very shortly (in a subsequent commit),
* ensuring correctness down this path isn't worth the effort.
*/
goto label_err;
}
if (*commit && !extent_committed_get(extent)) {