| Message ID | 20221216032936.17841-1-yangxingui@huawei.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | scsi: libsas: Directly kick-off EH when ATA device fell off | expand |
Hi Xingui, On 2022/12/16 11:29, Xingui Yang wrote: > If the ATA device fell off, call sas_ata_device_link_abort() directly and > mark all outstanding QCs as failed and kick-off EH Immediately. This avoids > having to wait for block layer timeouts. > Why does ATA device need this special operation? SAS device does not have to wait for block layer timeouts? Thanks, Jason > Signed-off-by: Xingui Yang <yangxingui@huawei.com> > --- > drivers/scsi/libsas/sas_discover.c | 5 +++++ > 1 file changed, 5 insertions(+) > > diff --git a/drivers/scsi/libsas/sas_discover.c b/drivers/scsi/libsas/sas_discover.c > index d5bc1314c341..bd22741daa99 100644 > --- a/drivers/scsi/libsas/sas_discover.c > +++ b/drivers/scsi/libsas/sas_discover.c > @@ -362,6 +362,11 @@ static void sas_destruct_ports(struct asd_sas_port *port) > > void sas_unregister_dev(struct asd_sas_port *port, struct domain_device *dev) > { > + if (test_bit(SAS_DEV_GONE, &dev->state) && > + (dev->dev_type == SAS_SATA_DEV || > + (dev->tproto & SAS_PROTOCOL_STP))) > + sas_ata_device_link_abort(dev, false); > + > if (!test_bit(SAS_DEV_DESTROY, &dev->state) && > !list_empty(&dev->disco_list_node)) { > /* this rphy never saw sas_rphy_add */ >
On 2022/12/16 11:53, Jason Yan wrote: > Hi Xingui, > > On 2022/12/16 11:29, Xingui Yang wrote: >> If the ATA device fell off, call sas_ata_device_link_abort() directly and >> mark all outstanding QCs as failed and kick-off EH Immediately. This >> avoids >> having to wait for block layer timeouts. >> > > Why does ATA device need this special operation? SAS device does not > have to wait for block layer timeouts? Hi Jason, Applications that depend on I/O return may be blocked for 30 seconds, and this can be optimized for SATA disks. Different from SATA disks, I/Os on SAS disks can still be returned if they are quickly plugged in after the SAS disks fall off. So I'm not sure if this is appropriate for a sas disk. BTW, I wish the sas disk did the same. Thanks, Xingui > > Thanks, > Jason > >> Signed-off-by: Xingui Yang <yangxingui@huawei.com> >> --- >> drivers/scsi/libsas/sas_discover.c | 5 +++++ >> 1 file changed, 5 insertions(+) >> >> diff --git a/drivers/scsi/libsas/sas_discover.c >> b/drivers/scsi/libsas/sas_discover.c >> index d5bc1314c341..bd22741daa99 100644 >> --- a/drivers/scsi/libsas/sas_discover.c >> +++ b/drivers/scsi/libsas/sas_discover.c >> @@ -362,6 +362,11 @@ static void sas_destruct_ports(struct >> asd_sas_port *port) >> void sas_unregister_dev(struct asd_sas_port *port, struct >> domain_device *dev) >> { >> + if (test_bit(SAS_DEV_GONE, &dev->state) && >> + (dev->dev_type == SAS_SATA_DEV || >> + (dev->tproto & SAS_PROTOCOL_STP))) >> + sas_ata_device_link_abort(dev, false); >> + >> if (!test_bit(SAS_DEV_DESTROY, &dev->state) && >> !list_empty(&dev->disco_list_node)) { >> /* this rphy never saw sas_rphy_add */ >> > .
On 2022/12/16 11:29, Xingui Yang wrote: > If the ATA device fell off, call sas_ata_device_link_abort() directly and > mark all outstanding QCs as failed and kick-off EH Immediately. This avoids > having to wait for block layer timeouts. > > Signed-off-by: Xingui Yang <yangxingui@huawei.com> > --- > drivers/scsi/libsas/sas_discover.c | 5 +++++ > 1 file changed, 5 insertions(+) > > diff --git a/drivers/scsi/libsas/sas_discover.c b/drivers/scsi/libsas/sas_discover.c > index d5bc1314c341..bd22741daa99 100644 > --- a/drivers/scsi/libsas/sas_discover.c > +++ b/drivers/scsi/libsas/sas_discover.c > @@ -362,6 +362,11 @@ static void sas_destruct_ports(struct asd_sas_port *port) > > void sas_unregister_dev(struct asd_sas_port *port, struct domain_device *dev) > { > + if (test_bit(SAS_DEV_GONE, &dev->state) && > + (dev->dev_type == SAS_SATA_DEV || > + (dev->tproto & SAS_PROTOCOL_STP))) dev_is_sata() would be better here. Thanks, Jason > + sas_ata_device_link_abort(dev, false); > + > if (!test_bit(SAS_DEV_DESTROY, &dev->state) && > !list_empty(&dev->disco_list_node)) { > /* this rphy never saw sas_rphy_add */ >
On 2022/12/16 17:29, Jason Yan wrote: > On 2022/12/16 11:29, Xingui Yang wrote: >> If the ATA device fell off, call sas_ata_device_link_abort() directly and >> mark all outstanding QCs as failed and kick-off EH Immediately. This >> avoids >> having to wait for block layer timeouts. >> >> Signed-off-by: Xingui Yang <yangxingui@huawei.com> >> --- >> drivers/scsi/libsas/sas_discover.c | 5 +++++ >> 1 file changed, 5 insertions(+) >> >> diff --git a/drivers/scsi/libsas/sas_discover.c >> b/drivers/scsi/libsas/sas_discover.c >> index d5bc1314c341..bd22741daa99 100644 >> --- a/drivers/scsi/libsas/sas_discover.c >> +++ b/drivers/scsi/libsas/sas_discover.c >> @@ -362,6 +362,11 @@ static void sas_destruct_ports(struct >> asd_sas_port *port) >> void sas_unregister_dev(struct asd_sas_port *port, struct >> domain_device *dev) >> { >> + if (test_bit(SAS_DEV_GONE, &dev->state) && >> + (dev->dev_type == SAS_SATA_DEV || >> + (dev->tproto & SAS_PROTOCOL_STP))) > > dev_is_sata() would be better here. ok, Thanks for your advice. Thanks, Xingui > > Thanks, > Jason > >> + sas_ata_device_link_abort(dev, false); >> + >> if (!test_bit(SAS_DEV_DESTROY, &dev->state) && >> !list_empty(&dev->disco_list_node)) { >> /* this rphy never saw sas_rphy_add */ >> > .
diff --git a/drivers/scsi/libsas/sas_discover.c b/drivers/scsi/libsas/sas_discover.c index d5bc1314c341..bd22741daa99 100644 --- a/drivers/scsi/libsas/sas_discover.c +++ b/drivers/scsi/libsas/sas_discover.c @@ -362,6 +362,11 @@ static void sas_destruct_ports(struct asd_sas_port *port) void sas_unregister_dev(struct asd_sas_port *port, struct domain_device *dev) { + if (test_bit(SAS_DEV_GONE, &dev->state) && + (dev->dev_type == SAS_SATA_DEV || + (dev->tproto & SAS_PROTOCOL_STP))) + sas_ata_device_link_abort(dev, false); + if (!test_bit(SAS_DEV_DESTROY, &dev->state) && !list_empty(&dev->disco_list_node)) { /* this rphy never saw sas_rphy_add */
If the ATA device fell off, call sas_ata_device_link_abort() directly and mark all outstanding QCs as failed and kick-off EH Immediately. This avoids having to wait for block layer timeouts. Signed-off-by: Xingui Yang <yangxingui@huawei.com> --- drivers/scsi/libsas/sas_discover.c | 5 +++++ 1 file changed, 5 insertions(+)