d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix possible live-lock under heavy load where we can't allocate more

Browse Source 
memory for request.
I was sure graid3 should handle such situations well, but green@ reported
it is not and we want to fix it before 6.0.

Submitted by:	green
This commit is contained in:
Pawel Jakub Dawidek 2005-10-28 20:25:02 +00:00
parent ee042e1ba6
commit a65a0da2f9
Notes: svn2git 2020-12-20 02:59:44 +00:00 
svn path=/head/; revision=151822
1 changed files with 12 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
sys/geom/raid3/g_raid3.c
View File

@ -966,7 +966,7 @@ g_raid3_clone_bio(struct g_raid3_softc *sc, struct bio *pbp)
if (cbp->bio_data == NULL) {
if (size > 16384)
g_raid3_64k_failed++;
if (size > 4096)
else if (size > 4096)
g_raid3_16k_failed++;
else
g_raid3_4k_failed++;
@ -1560,6 +1560,17 @@ g_raid3_register_request(struct bio *pbp)
if (cbp == NULL) {
while ((cbp = G_RAID3_HEAD_BIO(pbp)) != NULL)
g_raid3_destroy_bio(sc, cbp);
/*
* To prevent deadlock, we must run back up
* with the ENOMEM for failed requests of any
* of our consumers. Our own sync requests
* can stick around, as they are finite.
*/
if ((pbp->bio_cflags &
G_RAID3_BIO_CFLAG_REGULAR) != 0) {
g_io_deliver(pbp, ENOMEM);
return (0);
}
return (ENOMEM);
}
cbp->bio_offset = offset;