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

Fix a long-standing deadlock issue with vnode backed md(4) devices:

Browse Source 
On vnode backed md(4) devices over a certain, currently undetermined
size relative to the buffer cache our "lemming-syncer" can provoke
a buffer starvation which puts the md thread to sleep on wdrain.

This generally tends to grind the entire system to a stop because the
event that is supposed to wake up the thread will not happen until a fair
bit of the piled up I/O requests in the system finish, and since a lot
of those are on a md(4) vnode backed device which is currently waiting
on wdrain until a fair amount of the piled up ... you get the picture.

The cure is to issue all VOP_WRITES on the vnode backing the device
with IO_SYNC.

In addition to more closely emulating a real disk device with a
non-lying write-cache, this makes the writes exempt from rate-limited
(there to avoid starving the buffer cache) and consequently prevents
the deadlock.

Unfortunately performance takes a hit.

Add "async" option to give people who know what they are doing the
old behaviour.
This commit is contained in:
Poul-Henning Kamp 2004-03-10 20:41:09 +00:00
parent 724e52cd0d
commit 7a6b2b6429
4 changed files with 14 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
sbin/mdconfig/mdconfig.8
View File

@ -141,6 +141,9 @@ other devices.
.It Fl o Oo Cm no Oc Ns Ar option
Set or reset options.
.Bl -tag -width indent
.It Oo Cm no Oc Ns Cm async
For vnode backed devices: avoid IO_SYNC for increased performance but
at the risk of deadlocking the entire kernel.
.It Oo Cm no Oc Ns Cm reserve
Allocate and reserve all needed storage from the start, rather than as needed.
.It Oo Cm no Oc Ns Cm cluster

6
sbin/mdconfig/mdconfig.c
View File

@ -117,7 +117,11 @@ main(int argc, char **argv)
case 'o':
if (cmdline != 2)
usage();
if (!strcmp(optarg, "cluster"))
if (!strcmp(optarg, "async"))
mdio.md_options |= MD_ASYNC;
else if (!strcmp(optarg, "noasync"))
mdio.md_options &= ~MD_ASYNC;
else if (!strcmp(optarg, "cluster"))
mdio.md_options |= MD_CLUSTER;
else if (!strcmp(optarg, "nocluster"))
mdio.md_options &= ~MD_CLUSTER;

8
sys/dev/md/md.c
View File

@ -503,13 +503,15 @@ mdstart_vnode(struct md_s *sc, struct bio *bp)
if (bp->bio_cmd == BIO_READ) {
vn_lock(sc->vnode, LK_EXCLUSIVE | LK_RETRY, curthread);
error = VOP_READ(sc->vnode, &auio, IO_DIRECT, sc->cred);
VOP_UNLOCK(sc->vnode, 0, curthread);
} else {
(void) vn_start_write(sc->vnode, &mp, V_WAIT);
vn_lock(sc->vnode, LK_EXCLUSIVE | LK_RETRY, curthread);
error = VOP_WRITE(sc->vnode, &auio, 0, sc->cred);
error = VOP_WRITE(sc->vnode, &auio,
sc->flags & MD_ASYNC ? 0 : IO_SYNC, sc->cred);
VOP_UNLOCK(sc->vnode, 0, curthread);
vn_finished_write(mp);
}
VOP_UNLOCK(sc->vnode, 0, curthread);
bp->bio_resid = auio.uio_resid;
return (error);
}
@ -938,7 +940,7 @@ mdcreate_vnode(struct md_ioctl *mdio, struct thread *td)
if (mdio->md_fwheads != 0)
sc->fwheads = mdio->md_fwheads;
sc->type = MD_VNODE;
sc->flags = mdio->md_options & MD_FORCE;
sc->flags = mdio->md_options & (MD_FORCE | MD_ASYNC);
if (!(flags & FWRITE))
sc->flags |= MD_READONLY;
sc->secsize = DEV_BSIZE;

1
sys/sys/mdioctl.h
View File

@ -90,5 +90,6 @@ struct md_ioctl {
#define MD_READONLY 0x08 /* Readonly mode */
#define MD_COMPRESS 0x10 /* Compression mode */
#define MD_FORCE 0x20 /* Don't try to prevent foot-shooting */
#define MD_ASYNC 0x40 /* Don't try to prevent foot-shooting */
#endif /* _SYS_MDIOCTL_H_*/