project-base/server-skynet-source-3rd-jemalloc
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix opt_zero-triggered in-place huge reallocation zeroing.

Browse Source 
Fix huge_ralloc_no_move_expand() to update the extent's zeroed attribute
based on the intersection of the previous value and that of the newly
merged trailing extent.
This commit is contained in:
Jason Evans 2016-05-17 17:12:13 -07:00
parent 9aea58d9a2
commit f442254bdf
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
src/huge.c
View File

@ -223,11 +223,10 @@ huge_ralloc_no_move_expand(tsdn_t *tsdn, extent_t *extent, void *ptr,
malloc_mutex_unlock(tsdn, &arena->huge_mtx);
/*
* Copy zero into is_zeroed_chunk and pass the copy to chunk_alloc(), so
* that it is possible to make correct junk/zero fill decisions below.
* Use is_zeroed_chunk to detect whether the trailing memory is zeroed,
* update extent's zeroed field, and zero as necessary.
*/
is_zeroed_chunk = zero;
is_zeroed_chunk = false;
if (arena_chunk_ralloc_huge_expand(tsdn, arena, ptr, oldsize, usize,
&is_zeroed_chunk))
return (true);
@ -236,6 +235,7 @@ huge_ralloc_no_move_expand(tsdn_t *tsdn, extent_t *extent, void *ptr,
chunk_deregister(tsdn, ptr, extent);
malloc_mutex_lock(tsdn, &arena->huge_mtx);
extent_size_set(extent, usize);
extent_zeroed_set(extent, extent_zeroed_get(extent) && is_zeroed_chunk);
malloc_mutex_unlock(tsdn, &arena->huge_mtx);
chunk_reregister(tsdn, ptr, extent);