| Message ID | 1515020042-43590-1-git-send-email-ukrishn@linux.vnet.ibm.com (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
On 04/01/18 09:54, Uma Krishnan wrote: > In the event of a command failure, cxlflash returns the failure to the > upper layers to process. After processing the error, when the command is > queued again, the private command structure will not be zeroed and the > ioasc could be stale. Per the SISLite specification, the AFU only sets the > ioasc in the presence of a failure. Thus, even though the original command > succeeds the second time, the command is considered a failure due to stale > ioasc. This cycle repeats indefinitely and can cause a hang or IO failure. > > To fix the issue, clear the ioasc before queuing any command. > > Fixes: 479ad8e9d48c ("scsi: cxlflash: Remove zeroing of private command > data") > Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com> Should this go to stable? > --- > drivers/scsi/cxlflash/main.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c > index 38b3a9c..48d3663 100644 > --- a/drivers/scsi/cxlflash/main.c > +++ b/drivers/scsi/cxlflash/main.c > @@ -620,6 +620,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp) > cmd->parent = afu; > cmd->hwq_index = hwq_index; > > + cmd->sa.ioasc = 0; > cmd->rcb.ctx_id = hwq->ctx_hndl; > cmd->rcb.msi = SISL_MSI_RRQ_UPDATED; > cmd->rcb.port_sel = CHAN2PORTMASK(scp->device->channel); >
On Wed, Jan 03, 2018 at 04:54:02PM -0600, Uma Krishnan wrote: > In the event of a command failure, cxlflash returns the failure to the > upper layers to process. After processing the error, when the command is > queued again, the private command structure will not be zeroed and the > ioasc could be stale. Per the SISLite specification, the AFU only sets the > ioasc in the presence of a failure. Thus, even though the original command > succeeds the second time, the command is considered a failure due to stale > ioasc. This cycle repeats indefinitely and can cause a hang or IO failure. > > To fix the issue, clear the ioasc before queuing any command. > > Fixes: 479ad8e9d48c ("scsi: cxlflash: Remove zeroing of private command > data") > Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com> Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
On Thu, Jan 04, 2018 at 05:33:48PM +1100, Andrew Donnellan wrote: > On 04/01/18 09:54, Uma Krishnan wrote: > >In the event of a command failure, cxlflash returns the failure to the > >upper layers to process. After processing the error, when the command is > >queued again, the private command structure will not be zeroed and the > >ioasc could be stale. Per the SISLite specification, the AFU only sets the > >ioasc in the presence of a failure. Thus, even though the original command > >succeeds the second time, the command is considered a failure due to stale > >ioasc. This cycle repeats indefinitely and can cause a hang or IO failure. > > > >To fix the issue, clear the ioasc before queuing any command. > > > >Fixes: 479ad8e9d48c ("scsi: cxlflash: Remove zeroing of private command > >data") > >Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com> > > Should this go to stable? Not a bad idea.
> On Jan 7, 2018, at 1:36 PM, Matthew R. Ochs <mrochs@linux.vnet.ibm.com> wrote: > > On Thu, Jan 04, 2018 at 05:33:48PM +1100, Andrew Donnellan wrote: >> On 04/01/18 09:54, Uma Krishnan wrote: >>> In the event of a command failure, cxlflash returns the failure to the >>> upper layers to process. After processing the error, when the command is >>> queued again, the private command structure will not be zeroed and the >>> ioasc could be stale. Per the SISLite specification, the AFU only sets the >>> ioasc in the presence of a failure. Thus, even though the original command >>> succeeds the second time, the command is considered a failure due to stale >>> ioasc. This cycle repeats indefinitely and can cause a hang or IO failure. >>> >>> To fix the issue, clear the ioasc before queuing any command. >>> >>> Fixes: 479ad8e9d48c ("scsi: cxlflash: Remove zeroing of private command >>> data") >>> Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com> >> >> Should this go to stable? > > Not a bad idea. > I will forward the patch to stable once the patch is in next/master. Unless there is a need to send a V2. Thanks for pointing out Andrew !!
diff --git a/drivers/scsi/cxlflash/main.c b/drivers/scsi/cxlflash/main.c index 38b3a9c..48d3663 100644 --- a/drivers/scsi/cxlflash/main.c +++ b/drivers/scsi/cxlflash/main.c @@ -620,6 +620,7 @@ static int cxlflash_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scp) cmd->parent = afu; cmd->hwq_index = hwq_index; + cmd->sa.ioasc = 0; cmd->rcb.ctx_id = hwq->ctx_hndl; cmd->rcb.msi = SISL_MSI_RRQ_UPDATED; cmd->rcb.port_sel = CHAN2PORTMASK(scp->device->channel);
In the event of a command failure, cxlflash returns the failure to the upper layers to process. After processing the error, when the command is queued again, the private command structure will not be zeroed and the ioasc could be stale. Per the SISLite specification, the AFU only sets the ioasc in the presence of a failure. Thus, even though the original command succeeds the second time, the command is considered a failure due to stale ioasc. This cycle repeats indefinitely and can cause a hang or IO failure. To fix the issue, clear the ioasc before queuing any command. Fixes: 479ad8e9d48c ("scsi: cxlflash: Remove zeroing of private command data") Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com> --- drivers/scsi/cxlflash/main.c | 1 + 1 file changed, 1 insertion(+)