rdelaage/postgresql
1
0
Fork 0
mirror of https://git.postgresql.org/git/postgresql.git synced 2024-07-16 06:31:10 +02:00
Code Issues Packages Projects Releases Wiki Activity

Maintain valid md.c state when FileClose() fails.

Browse Source 
FileClose() failure ordinarily causes a PANIC.  Suppose the user
disables that PANIC via data_sync_retry=on.  After mdclose() issued a
FileClose() that failed, calls into md.c raised SIGSEGV.  This fix adds
repalloc() calls during mdclose(); update a comment about ignoring
repalloc() cost.  The rate of relation segment count change is a minor
factor; more relevant to overall performance is the rate of mdclose()
and subsequent re-opening of segments.  Back-patch to v10, where commit
45e191e3aa introduced the bug.

Reviewed by Kyotaro Horiguchi.

Discussion: https://postgr.es/m/20191222091930.GA1280238@rfd.leadboat.com
This commit is contained in:
Noah Misch 2020-01-10 18:31:22 -08:00
parent 2696434b7f
commit 2e86e154da
1 changed files with 6 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
src/backend/storage/smgr/md.c
View File

@ -638,18 +638,10 @@ mdclose(SMgrRelation reln, ForkNumber forknum)
{ {
MdfdVec *v = &reln->md_seg_fds[forknum][nopensegs - 1]; MdfdVec *v = &reln->md_seg_fds[forknum][nopensegs - 1];
/* if not closed already */ FileClose(v->mdfd_vfd);
if (v->mdfd_vfd >= 0) _fdvec_resize(reln, forknum, nopensegs - 1);
{
FileClose(v->mdfd_vfd);
v->mdfd_vfd = -1;
}
nopensegs--; nopensegs--;
} }
/* resize just once, avoids pointless reallocations */
_fdvec_resize(reln, forknum, 0);
} }
/* /*
@ -1773,10 +1765,10 @@ _fdvec_resize(SMgrRelation reln,
else else
{ {
/* /*
* It doesn't seem worthwhile complicating the code by having a more * It doesn't seem worthwhile complicating the code to amortize
* aggressive growth strategy here; the number of segments doesn't * repalloc() calls. Those are far faster than PathNameOpenFile() or
* grow that fast, and the memory context internally will sometimes * FileClose(), and the memory context internally will sometimes avoid
* avoid doing an actual reallocation. * doing an actual reallocation.
*/ */
reln->md_seg_fds[forknum] = reln->md_seg_fds[forknum] =
repalloc(reln->md_seg_fds[forknum], repalloc(reln->md_seg_fds[forknum],