| Message ID | 20180222193020.30276-1-bart.vanassche@wdc.com (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
On Thu, 22 Feb 2018 11:30:20 -0800 Bart Van Assche <bart.vanassche@wdc.com> wrote: > Avoid that the recently introduced call_rcu() call in the SCSI core > triggers a double call_rcu() call. This patch also prevents my machine from hanging. However, the ATA error messages that I previously have had are gone, so I don't know if it is because the underlying problem is gone, or if the error messages are simply not reported. -nc > > Reported-by: Natanael Copa <ncopa@alpinelinux.org> > Reported-by: Damien Le Moal <damien.lemoal@wdc.com> > References: https://bugzilla.kernel.org/show_bug.cgi?id=198861 > Fixes: 3bd6f43f5cb3 ("scsi: core: Ensure that the SCSI error handler gets woken up") > Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> > Cc: Natanael Copa <ncopa@alpinelinux.org> > Cc: Damien Le Moal <damien.lemoal@wdc.com> > Cc: Alexandre Oliva <oliva@gnu.org> > Cc: Pavel Tikhomirov <ptikhomirov@virtuozzo.com> > Cc: Hannes Reinecke <hare@suse.com> > Cc: Johannes Thumshirn <jthumshirn@suse.de> > Cc: <stable@vger.kernel.org> > --- > drivers/scsi/hosts.c | 3 --- > drivers/scsi/scsi_error.c | 5 +++-- > drivers/scsi/scsi_lib.c | 2 ++ > include/scsi/scsi_cmnd.h | 3 +++ > include/scsi/scsi_host.h | 2 -- > 5 files changed, 8 insertions(+), 7 deletions(-) > > diff --git a/drivers/scsi/hosts.c b/drivers/scsi/hosts.c > index a0a7e4ff255c..7279d3d2e941 100644 > --- a/drivers/scsi/hosts.c > +++ b/drivers/scsi/hosts.c > @@ -334,8 +334,6 @@ static void scsi_host_dev_release(struct device *dev) > if (shost->work_q) > destroy_workqueue(shost->work_q); > > - destroy_rcu_head(&shost->rcu); > - > if (shost->shost_state == SHOST_CREATED) { > /* > * Free the shost_dev device name here if scsi_host_alloc() > @@ -404,7 +402,6 @@ struct Scsi_Host *scsi_host_alloc(struct scsi_host_template *sht, int privsize) > INIT_LIST_HEAD(&shost->starved_list); > init_waitqueue_head(&shost->host_wait); > mutex_init(&shost->scan_mutex); > - init_rcu_head(&shost->rcu); > > index = ida_simple_get(&host_index_ida, 0, 0, GFP_KERNEL); > if (index < 0) > diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c > index 96f988a7efda..9b0242f84407 100644 > --- a/drivers/scsi/scsi_error.c > +++ b/drivers/scsi/scsi_error.c > @@ -229,7 +229,8 @@ static void scsi_eh_reset(struct scsi_cmnd *scmd) > > static void scsi_eh_inc_host_failed(struct rcu_head *head) > { > - struct Scsi_Host *shost = container_of(head, typeof(*shost), rcu); > + struct scsi_cmnd *scmd = container_of(head, typeof(*scmd), rcu); > + struct Scsi_Host *shost = scmd->device->host; > unsigned long flags; > > spin_lock_irqsave(shost->host_lock, flags); > @@ -265,7 +266,7 @@ void scsi_eh_scmd_add(struct scsi_cmnd *scmd) > * Ensure that all tasks observe the host state change before the > * host_failed change. > */ > - call_rcu(&shost->rcu, scsi_eh_inc_host_failed); > + call_rcu(&scmd->rcu, scsi_eh_inc_host_failed); > } > > /** > diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c > index aea5a1ae318b..e1ca2160aa40 100644 > --- a/drivers/scsi/scsi_lib.c > +++ b/drivers/scsi/scsi_lib.c > @@ -671,6 +671,7 @@ static bool scsi_end_request(struct request *req, blk_status_t error, > if (!blk_rq_is_scsi(req)) { > WARN_ON_ONCE(!(cmd->flags & SCMD_INITIALIZED)); > cmd->flags &= ~SCMD_INITIALIZED; > + destroy_rcu_head(&cmd->rcu); > } > > if (req->mq_ctx) { > @@ -1151,6 +1152,7 @@ static void scsi_initialize_rq(struct request *rq) > struct scsi_cmnd *cmd = blk_mq_rq_to_pdu(rq); > > scsi_req_init(&cmd->req); > + init_rcu_head(&cmd->rcu); > cmd->jiffies_at_alloc = jiffies; > cmd->retries = 0; > } > diff --git a/include/scsi/scsi_cmnd.h b/include/scsi/scsi_cmnd.h > index d8d4a902a88d..2280b2351739 100644 > --- a/include/scsi/scsi_cmnd.h > +++ b/include/scsi/scsi_cmnd.h > @@ -68,6 +68,9 @@ struct scsi_cmnd { > struct list_head list; /* scsi_cmnd participates in queue lists */ > struct list_head eh_entry; /* entry for the host eh_cmd_q */ > struct delayed_work abort_work; > + > + struct rcu_head rcu; > + > int eh_eflags; /* Used by error handlr */ > > /* > diff --git a/include/scsi/scsi_host.h b/include/scsi/scsi_host.h > index 19317585ae48..7aedb6776b4f 100644 > --- a/include/scsi/scsi_host.h > +++ b/include/scsi/scsi_host.h > @@ -577,8 +577,6 @@ struct Scsi_Host { > struct blk_mq_tag_set tag_set; > }; > > - struct rcu_head rcu; > - > atomic_t host_busy; /* commands actually active on low-level */ > atomic_t host_blocked; >
Hi, On Fri, Feb 23, 2018 at 04:46:30PM +0100, Natanael Copa wrote: > On Thu, 22 Feb 2018 11:30:20 -0800 > Bart Van Assche <bart.vanassche@wdc.com> wrote: > > > Avoid that the recently introduced call_rcu() call in the SCSI core > > triggers a double call_rcu() call. > > This patch also prevents my machine from hanging. However, the ATA > error messages that I previously have had are gone, so I don't know if > it is because the underlying problem is gone, or if the error messages > are simply not reported. For reference, the patch has as well been confirmed to solve the problem for Jean-Francois Pirus in https://bugs.debian.org/891467 (on top of 4.15.4) Regards, Salvatore
On Thu, 2018-02-22 at 11:30 -0800, Bart Van Assche wrote: > Avoid that the recently introduced call_rcu() call in the SCSI core > triggers a double call_rcu() call. > [ ... ] Can anyone review this patch? Multiple users have confirmed independently that this patch fixes the double call_rcu() issue for them. Thanks, Bart.
On 2018/02/27 10:53, Bart Van Assche wrote: > On Thu, 2018-02-22 at 11:30 -0800, Bart Van Assche wrote: >> Avoid that the recently introduced call_rcu() call in the SCSI core >> triggers a double call_rcu() call. >> [ ... ] > > Can anyone review this patch? Multiple users have confirmed independently > that this patch fixes the double call_rcu() issue for them. > > Thanks, > > Bart. Please feel free to add: Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Tested-by: Damien Le Moal <damien.lemoal@wdc.com> -- Damien Le Moal Western Digital Research
Bart, > Avoid that the recently introduced call_rcu() call in the SCSI core > triggers a double call_rcu() call. Applied to 4.16/scsi-fixes. Thank you!
diff --git a/drivers/scsi/hosts.c b/drivers/scsi/hosts.c index a0a7e4ff255c..7279d3d2e941 100644 --- a/drivers/scsi/hosts.c +++ b/drivers/scsi/hosts.c @@ -334,8 +334,6 @@ static void scsi_host_dev_release(struct device *dev) if (shost->work_q) destroy_workqueue(shost->work_q); - destroy_rcu_head(&shost->rcu); - if (shost->shost_state == SHOST_CREATED) { /* * Free the shost_dev device name here if scsi_host_alloc() @@ -404,7 +402,6 @@ struct Scsi_Host *scsi_host_alloc(struct scsi_host_template *sht, int privsize) INIT_LIST_HEAD(&shost->starved_list); init_waitqueue_head(&shost->host_wait); mutex_init(&shost->scan_mutex); - init_rcu_head(&shost->rcu); index = ida_simple_get(&host_index_ida, 0, 0, GFP_KERNEL); if (index < 0) diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c index 96f988a7efda..9b0242f84407 100644 --- a/drivers/scsi/scsi_error.c +++ b/drivers/scsi/scsi_error.c @@ -229,7 +229,8 @@ static void scsi_eh_reset(struct scsi_cmnd *scmd) static void scsi_eh_inc_host_failed(struct rcu_head *head) { - struct Scsi_Host *shost = container_of(head, typeof(*shost), rcu); + struct scsi_cmnd *scmd = container_of(head, typeof(*scmd), rcu); + struct Scsi_Host *shost = scmd->device->host; unsigned long flags; spin_lock_irqsave(shost->host_lock, flags); @@ -265,7 +266,7 @@ void scsi_eh_scmd_add(struct scsi_cmnd *scmd) * Ensure that all tasks observe the host state change before the * host_failed change. */ - call_rcu(&shost->rcu, scsi_eh_inc_host_failed); + call_rcu(&scmd->rcu, scsi_eh_inc_host_failed); } /** diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c index aea5a1ae318b..e1ca2160aa40 100644 --- a/drivers/scsi/scsi_lib.c +++ b/drivers/scsi/scsi_lib.c @@ -671,6 +671,7 @@ static bool scsi_end_request(struct request *req, blk_status_t error, if (!blk_rq_is_scsi(req)) { WARN_ON_ONCE(!(cmd->flags & SCMD_INITIALIZED)); cmd->flags &= ~SCMD_INITIALIZED; + destroy_rcu_head(&cmd->rcu); } if (req->mq_ctx) { @@ -1151,6 +1152,7 @@ static void scsi_initialize_rq(struct request *rq) struct scsi_cmnd *cmd = blk_mq_rq_to_pdu(rq); scsi_req_init(&cmd->req); + init_rcu_head(&cmd->rcu); cmd->jiffies_at_alloc = jiffies; cmd->retries = 0; } diff --git a/include/scsi/scsi_cmnd.h b/include/scsi/scsi_cmnd.h index d8d4a902a88d..2280b2351739 100644 --- a/include/scsi/scsi_cmnd.h +++ b/include/scsi/scsi_cmnd.h @@ -68,6 +68,9 @@ struct scsi_cmnd { struct list_head list; /* scsi_cmnd participates in queue lists */ struct list_head eh_entry; /* entry for the host eh_cmd_q */ struct delayed_work abort_work; + + struct rcu_head rcu; + int eh_eflags; /* Used by error handlr */ /* diff --git a/include/scsi/scsi_host.h b/include/scsi/scsi_host.h index 19317585ae48..7aedb6776b4f 100644 --- a/include/scsi/scsi_host.h +++ b/include/scsi/scsi_host.h @@ -577,8 +577,6 @@ struct Scsi_Host { struct blk_mq_tag_set tag_set; }; - struct rcu_head rcu; - atomic_t host_busy; /* commands actually active on low-level */ atomic_t host_blocked;
Avoid that the recently introduced call_rcu() call in the SCSI core triggers a double call_rcu() call. Reported-by: Natanael Copa <ncopa@alpinelinux.org> Reported-by: Damien Le Moal <damien.lemoal@wdc.com> References: https://bugzilla.kernel.org/show_bug.cgi?id=198861 Fixes: 3bd6f43f5cb3 ("scsi: core: Ensure that the SCSI error handler gets woken up") Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> Cc: Natanael Copa <ncopa@alpinelinux.org> Cc: Damien Le Moal <damien.lemoal@wdc.com> Cc: Alexandre Oliva <oliva@gnu.org> Cc: Pavel Tikhomirov <ptikhomirov@virtuozzo.com> Cc: Hannes Reinecke <hare@suse.com> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: <stable@vger.kernel.org> --- drivers/scsi/hosts.c | 3 --- drivers/scsi/scsi_error.c | 5 +++-- drivers/scsi/scsi_lib.c | 2 ++ include/scsi/scsi_cmnd.h | 3 +++ include/scsi/scsi_host.h | 2 -- 5 files changed, 8 insertions(+), 7 deletions(-)