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

mpr/mps: Fix a race in diagnostic reset

Browse Source 
There's a small race in freezing the simq when performing a diagnostic
reset. During this time, a transaction can slip through and encounter
the target id of 0. If we're still in diagnostic reset when we detect
this, return a CAM_DEVICE_NOT_THERE status. Instead, freeze the queue
and return a requeue status, similar to what we do when we're resetting
a target and a transaction get here. The race is unavoidable due to
separate locks for queue and SIM, but easy enough to detect and make
harmless.

Sponsored by:		Netflix
Reviewed by:		scottl, mav
Differential Revision:	https://reviews.freebsd.org/D34017
This commit is contained in:
Warner Losh 2022-01-25 16:23:03 -07:00
parent 5fcb5ae8dc
commit e35816c1c9
2 changed files with 18 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
sys/dev/mpr/mpr_sas.c
View File

@ -1864,6 +1864,15 @@ mprsas_action_scsiio(struct mprsas_softc *sassc, union ccb *ccb)
targ = &sassc->targets[csio->ccb_h.target_id];
mpr_dprint(sc, MPR_TRACE, "ccb %p target flag %x\n", ccb, targ->flags);
if (targ->handle == 0x0) {
if (targ->flags & MPRSAS_TARGET_INDIAGRESET) {
mpr_dprint(sc, MPR_ERROR,
"%s NULL handle for target %u in diag reset freezing queue\n",
__func__, csio->ccb_h.target_id);
ccb->ccb_h.status = CAM_REQUEUE_REQ | CAM_DEV_QFRZN;
xpt_freeze_devq(ccb->ccb_h.path, 1);
xpt_done(ccb);
return;
}
mpr_dprint(sc, MPR_ERROR, "%s NULL handle for target %u\n",
__func__, csio->ccb_h.target_id);
mprsas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE);

9
sys/dev/mps/mps_sas.c
View File

@ -1637,6 +1637,15 @@ mpssas_action_scsiio(struct mpssas_softc *sassc, union ccb *ccb)
targ = &sassc->targets[csio->ccb_h.target_id];
mps_dprint(sc, MPS_TRACE, "ccb %p target flag %x\n", ccb, targ->flags);
if (targ->handle == 0x0) {
if (targ->flags & MPSSAS_TARGET_INDIAGRESET) {
mps_dprint(sc, MPS_ERROR,
"%s NULL handle for target %u in diag reset freezing queue\n",
__func__, csio->ccb_h.target_id);
ccb->ccb_h.status = CAM_REQUEUE_REQ | CAM_DEV_QFRZN;
xpt_freeze_devq(ccb->ccb_h.path, 1);
xpt_done(ccb);
return;
}
mps_dprint(sc, MPS_ERROR, "%s NULL handle for target %u\n",
__func__, csio->ccb_h.target_id);
mpssas_set_ccbstatus(ccb, CAM_DEV_NOT_THERE);