nvmet-fc: Fix a missed _irqsave version of spin_lock in 'nvmet_fc_fod_op_done()'
diff mbox series

Message ID 20200821075819.152474-1-christophe.jaillet@wanadoo.fr
State Accepted
Commit 70e37988db94aba607d5491a94f80ba08e399b6b
Headers show
Series
  • nvmet-fc: Fix a missed _irqsave version of spin_lock in 'nvmet_fc_fod_op_done()'
Related show

Commit Message

Christophe JAILLET Aug. 21, 2020, 7:58 a.m. UTC
The way 'spin_lock()' and 'spin_lock_irqsave()' are used is not consistent
in this function.

Use 'spin_lock_irqsave()' also here, as there is no guarantee that
interruptions are disabled at that point, according to surrounding code.

Fixes: a97ec51b37ef ("nvmet_fc: Rework target side abort handling")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
Not tested, only based on what looks logical to me according to
surrounding code
---
 drivers/nvme/target/fc.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Comments

Sagi Grimberg Aug. 24, 2020, 8 p.m. UTC | #1
> The way 'spin_lock()' and 'spin_lock_irqsave()' are used is not consistent
> in this function.
> 
> Use 'spin_lock_irqsave()' also here, as there is no guarantee that
> interruptions are disabled at that point, according to surrounding code.
> 
> Fixes: a97ec51b37ef ("nvmet_fc: Rework target side abort handling")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
> Not tested, only based on what looks logical to me according to
> surrounding code
> ---
>   drivers/nvme/target/fc.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/nvme/target/fc.c b/drivers/nvme/target/fc.c
> index 55bafd56166a..e6861cc10e7d 100644
> --- a/drivers/nvme/target/fc.c
> +++ b/drivers/nvme/target/fc.c
> @@ -2342,9 +2342,9 @@ nvmet_fc_fod_op_done(struct nvmet_fc_fcp_iod *fod)
>   			return;
>   		if (fcpreq->fcp_error ||
>   		    fcpreq->transferred_length != fcpreq->transfer_length) {
> -			spin_lock(&fod->flock);
> +			spin_lock_irqsave(&fod->flock, flags);
>   			fod->abort = true;
> -			spin_unlock(&fod->flock);
> +			spin_unlock_irqrestore(&fod->flock, flags);
>   
>   			nvmet_req_complete(&fod->req, NVME_SC_INTERNAL);
>   			return;
> 

James, can I get a reviewed-by from you on this?
Keith Busch Aug. 24, 2020, 9:16 p.m. UTC | #2
On Mon, Aug 24, 2020 at 01:00:11PM -0700, Sagi Grimberg wrote:
> > The way 'spin_lock()' and 'spin_lock_irqsave()' are used is not consistent
> > in this function.
> > 
> > Use 'spin_lock_irqsave()' also here, as there is no guarantee that
> > interruptions are disabled at that point, according to surrounding code.
> > 
> > Fixes: a97ec51b37ef ("nvmet_fc: Rework target side abort handling")
> > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> > ---
> > Not tested, only based on what looks logical to me according to
> > surrounding code
> > ---
> >   drivers/nvme/target/fc.c | 4 ++--
> >   1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/nvme/target/fc.c b/drivers/nvme/target/fc.c
> > index 55bafd56166a..e6861cc10e7d 100644
> > --- a/drivers/nvme/target/fc.c
> > +++ b/drivers/nvme/target/fc.c
> > @@ -2342,9 +2342,9 @@ nvmet_fc_fod_op_done(struct nvmet_fc_fcp_iod *fod)
> >   			return;
> >   		if (fcpreq->fcp_error ||
> >   		    fcpreq->transferred_length != fcpreq->transfer_length) {
> > -			spin_lock(&fod->flock);
> > +			spin_lock_irqsave(&fod->flock, flags);
> >   			fod->abort = true;
> > -			spin_unlock(&fod->flock);
> > +			spin_unlock_irqrestore(&fod->flock, flags);
> >   			nvmet_req_complete(&fod->req, NVME_SC_INTERNAL);
> >   			return;
> > 
> 
> James, can I get a reviewed-by from you on this?

afaics, the lock just serializes single writes, in which
WRITE/READ_ONCE() can handle that without a lock, right?
Christoph Hellwig Aug. 25, 2020, 7:09 a.m. UTC | #3
On Mon, Aug 24, 2020 at 02:16:30PM -0700, Keith Busch wrote:
> On Mon, Aug 24, 2020 at 01:00:11PM -0700, Sagi Grimberg wrote:
> > > The way 'spin_lock()' and 'spin_lock_irqsave()' are used is not consistent
> > > in this function.
> > > 
> > > Use 'spin_lock_irqsave()' also here, as there is no guarantee that
> > > interruptions are disabled at that point, according to surrounding code.
> > > 
> > > Fixes: a97ec51b37ef ("nvmet_fc: Rework target side abort handling")
> > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> > > ---
> > > Not tested, only based on what looks logical to me according to
> > > surrounding code
> > > ---
> > >   drivers/nvme/target/fc.c | 4 ++--
> > >   1 file changed, 2 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/nvme/target/fc.c b/drivers/nvme/target/fc.c
> > > index 55bafd56166a..e6861cc10e7d 100644
> > > --- a/drivers/nvme/target/fc.c
> > > +++ b/drivers/nvme/target/fc.c
> > > @@ -2342,9 +2342,9 @@ nvmet_fc_fod_op_done(struct nvmet_fc_fcp_iod *fod)
> > >   			return;
> > >   		if (fcpreq->fcp_error ||
> > >   		    fcpreq->transferred_length != fcpreq->transfer_length) {
> > > -			spin_lock(&fod->flock);
> > > +			spin_lock_irqsave(&fod->flock, flags);
> > >   			fod->abort = true;
> > > -			spin_unlock(&fod->flock);
> > > +			spin_unlock_irqrestore(&fod->flock, flags);
> > >   			nvmet_req_complete(&fod->req, NVME_SC_INTERNAL);
> > >   			return;
> > > 
> > 
> > James, can I get a reviewed-by from you on this?
> 
> afaics, the lock just serializes single writes, in which
> WRITE/READ_ONCE() can handle that without a lock, right?

It synchronizes a few fields, sometimes taken over multiple updates.
Christoph Hellwig Aug. 25, 2020, 7:10 a.m. UTC | #4
On Fri, Aug 21, 2020 at 09:58:19AM +0200, Christophe JAILLET wrote:
> The way 'spin_lock()' and 'spin_lock_irqsave()' are used is not consistent
> in this function.
> 
> Use 'spin_lock_irqsave()' also here, as there is no guarantee that
> interruptions are disabled at that point, according to surrounding code.
> 
> Fixes: a97ec51b37ef ("nvmet_fc: Rework target side abort handling")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
> Not tested, only based on what looks logical to me according to
> surrounding code

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

Patch
diff mbox series

diff --git a/drivers/nvme/target/fc.c b/drivers/nvme/target/fc.c
index 55bafd56166a..e6861cc10e7d 100644
--- a/drivers/nvme/target/fc.c
+++ b/drivers/nvme/target/fc.c
@@ -2342,9 +2342,9 @@  nvmet_fc_fod_op_done(struct nvmet_fc_fcp_iod *fod)
 			return;
 		if (fcpreq->fcp_error ||
 		    fcpreq->transferred_length != fcpreq->transfer_length) {
-			spin_lock(&fod->flock);
+			spin_lock_irqsave(&fod->flock, flags);
 			fod->abort = true;
-			spin_unlock(&fod->flock);
+			spin_unlock_irqrestore(&fod->flock, flags);
 
 			nvmet_req_complete(&fod->req, NVME_SC_INTERNAL);
 			return;