| Message ID | 20231025070117.464903-1-dlemoal@kernel.org (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | scsi: sd: Introduce manage_shutdown device flag | expand |
On 10/25/23 16:01, Damien Le Moal wrote: > Commit aa3998dbeb3a ("ata: libata-scsi: Disable scsi device > manage_system_start_stop") change setting the manage_system_start_stop > flag to false for libata managed disks to enable libata internal > management of disk suspend/resume. However, a side effect of this change > is that on system shutdown, disks are no longer being stopped (set to > standby mode with the heads unloaded). While this is not a critical > issue, this unclean shutdown is not recommended and shows up with > increased smart counters (e.g. the unexpected power loss counter > "Unexpect_Power_Loss_Ct"). > > Instead of defining a shutdown driver method for all ATA adapter > drivers (not all of them define that operation), this patch resolves > this issue by further refining the sd driver start/stop control of disks > using the new flag manage_shutdown. If set to true, the function > sd_shutdown() will issue a START STOP UNIT command with the start > argument set to 0 when a disk is shutdown on system power off > (system_state == SYSTEM_POWER_OFF). > > Fixes: aa3998dbeb3a ("ata: libata-scsi: Disable scsi device manage_system_start_stop") > Cc: stable@vger.kernel.org > Closes: https://bugzilla.kernel.org/show_bug.cgi?id=218038 > Link: https://lore.kernel.org/all/cd397c88-bf53-4768-9ab8-9d107df9e613@gmail.com/ > Signed-off-by: Damien Le Moal <dlemoal@kernel.org> Martin, I can take this patch through libata fixes branch if you prefer (as I suspect it may not apply to your tree without rebasing first). > --- > drivers/ata/libata-scsi.c | 5 +++-- > drivers/firewire/sbp2.c | 1 + > drivers/scsi/sd.c | 6 ++++-- > include/scsi/scsi_device.h | 1 + > 4 files changed, 9 insertions(+), 4 deletions(-) > > diff --git a/drivers/ata/libata-scsi.c b/drivers/ata/libata-scsi.c > index a371b497035e..3a957c4da409 100644 > --- a/drivers/ata/libata-scsi.c > +++ b/drivers/ata/libata-scsi.c > @@ -1053,10 +1053,11 @@ int ata_scsi_dev_config(struct scsi_device *sdev, struct ata_device *dev) > > /* > * Ask the sd driver to issue START STOP UNIT on runtime suspend > - * and resume only. For system level suspend/resume, devices > - * power state is handled directly by libata EH. > + * and resume and shutdown only. For system level suspend/resume, > + * devices power state is handled directly by libata EH. > */ > sdev->manage_runtime_start_stop = true; > + sdev->manage_shutdown = true; > } > > /* > diff --git a/drivers/firewire/sbp2.c b/drivers/firewire/sbp2.c > index 749868b9e80d..7edf2c95282f 100644 > --- a/drivers/firewire/sbp2.c > +++ b/drivers/firewire/sbp2.c > @@ -1521,6 +1521,7 @@ static int sbp2_scsi_slave_configure(struct scsi_device *sdev) > if (sbp2_param_exclusive_login) { > sdev->manage_system_start_stop = true; > sdev->manage_runtime_start_stop = true; > + sdev->manage_shutdown = true; > } > > if (sdev->type == TYPE_ROM) > diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c > index 83b6a3f3863b..52fa266d976f 100644 > --- a/drivers/scsi/sd.c > +++ b/drivers/scsi/sd.c > @@ -3819,8 +3819,10 @@ static void sd_shutdown(struct device *dev) > sd_sync_cache(sdkp, NULL); > } > > - if (system_state != SYSTEM_RESTART && > - sdkp->device->manage_system_start_stop) { > + if ((system_state != SYSTEM_RESTART && > + sdkp->device->manage_system_start_stop) || > + (system_state == SYSTEM_POWER_OFF && > + sdkp->device->manage_shutdown)) { > sd_printk(KERN_NOTICE, sdkp, "Stopping disk\n"); > sd_start_stop_device(sdkp, 0); > } > diff --git a/include/scsi/scsi_device.h b/include/scsi/scsi_device.h > index fd41fdac0a8e..7edefb73bf69 100644 > --- a/include/scsi/scsi_device.h > +++ b/include/scsi/scsi_device.h > @@ -164,6 +164,7 @@ struct scsi_device { > > bool manage_system_start_stop; /* Let HLD (sd) manage system start/stop */ > bool manage_runtime_start_stop; /* Let HLD (sd) manage runtime start/stop */ > + bool manage_shutdown; /* Let HLD (sd) manage shutdown */ > > unsigned removable:1; > unsigned changed:1; /* Data invalid due to media change */
On Wed, 2023-10-25 at 16:01 +0900, Damien Le Moal wrote: > +++ b/include/scsi/scsi_device.h > @@ -164,6 +164,7 @@ struct scsi_device { > > bool manage_system_start_stop; /* Let HLD (sd) manage system > start/stop */ > bool manage_runtime_start_stop; /* Let HLD (sd) manage > runtime start/stop */ > + bool manage_shutdown; /* Let HLD (sd) manage shutdown */ > I think at least 85% of the world gets confused about the difference between runtime/system start/stop and shutdown. Could we at least point to a doc explaining it in a comment here? James
On 10/25/23 20:57, James Bottomley wrote: > On Wed, 2023-10-25 at 16:01 +0900, Damien Le Moal wrote: >> +++ b/include/scsi/scsi_device.h >> @@ -164,6 +164,7 @@ struct scsi_device { >> >> bool manage_system_start_stop; /* Let HLD (sd) manage system >> start/stop */ >> bool manage_runtime_start_stop; /* Let HLD (sd) manage >> runtime start/stop */ >> + bool manage_shutdown; /* Let HLD (sd) manage shutdown */ >> > > I think at least 85% of the world gets confused about the difference > between runtime/system start/stop and shutdown. Could we at least > point to a doc explaining it in a comment here? Would improving the comments here be enough ? E.g. something like: /* Let the HLD (sd) manage system suspend (start) and resume (stop). * This applies to both suspend to RAM and suspend to disk * (hybernation). */ bool manage_system_start_stop; /* * Let the HLD (sd) manage device runtime suspend (stop) and * resume (start). */ bool manage_runtime_start_stop; /* Let the HLD (sd) manage system power-off (shutdown) */ bool manage_shutdown;
On Thu, 2023-10-26 at 06:30 +0900, Damien Le Moal wrote: > On 10/25/23 20:57, James Bottomley wrote: > > On Wed, 2023-10-25 at 16:01 +0900, Damien Le Moal wrote: > > > +++ b/include/scsi/scsi_device.h > > > @@ -164,6 +164,7 @@ struct scsi_device { > > > > > > bool manage_system_start_stop; /* Let HLD (sd) manage > > > system > > > start/stop */ > > > bool manage_runtime_start_stop; /* Let HLD (sd) manage > > > runtime start/stop */ > > > + bool manage_shutdown; /* Let HLD (sd) manage shutdown > > > */ > > > > > > > I think at least 85% of the world gets confused about the > > difference > > between runtime/system start/stop and shutdown. Could we at least > > point to a doc explaining it in a comment here? > > Would improving the comments here be enough ? E.g. something like: > > /* Let the HLD (sd) manage system suspend (start) and resume > (stop). > * This applies to both suspend to RAM and suspend to disk > * (hybernation). > */ > bool manage_system_start_stop; > > /* > * Let the HLD (sd) manage device runtime suspend (stop) and > * resume (start). > */ > bool manage_runtime_start_stop; > > /* Let the HLD (sd) manage system power-off (shutdown) */ > bool manage_shutdown; Heh, well, I was going to say we should still point to the doc, but I simply can't find it, so the above is perhaps the best we can do, thanks! James
On 10/26/23 05:01, James Bottomley wrote: > Heh, well, I was going to say we should still point to the doc, but I > simply can't find it, so the above is perhaps the best we can do, > thanks! I think this should be documented in the Documentation/power directory. After having taken another look at that directory, I see that there is only detailed documentation and no overview documentation. Maybe I overlooked something but I couldn't find an explanation of the system suspend/resume nor of the runtime power management concepts in that directory. My understanding is that system suspend/resume is about system-wide power state changes (hibernation and suspend-to-RAM) and also that runtime power management is about changing the power state of a single device or bus if no activity has happened within a certain time. Bart.
On 10/27/23 06:36, Bart Van Assche wrote: > On 10/26/23 05:01, James Bottomley wrote: >> Heh, well, I was going to say we should still point to the doc, but I >> simply can't find it, so the above is perhaps the best we can do, >> thanks! > > I think this should be documented in the Documentation/power directory. > After having taken another look at that directory, I see that there > is only detailed documentation and no overview documentation. Maybe I > overlooked something but I couldn't find an explanation of the system > suspend/resume nor of the runtime power management concepts in that > directory. My understanding is that system suspend/resume is about > system-wide power state changes (hibernation and suspend-to-RAM) and > also that runtime power management is about changing the power state of > a single device or bus if no activity has happened within a certain > time. I actually thought that James wanted a reference to scsi sysfs attributes documentation, which is also not in the best of shape, to say the least... In any case, I would like to push this fix for 6.6-final as this is a tracked regression. Martin, James, are you OK with this patch ?
On Fri, 2023-10-27 at 09:26 +0900, Damien Le Moal wrote: > On 10/27/23 06:36, Bart Van Assche wrote: > > On 10/26/23 05:01, James Bottomley wrote: > > > Heh, well, I was going to say we should still point to the doc, > > > but I simply can't find it, so the above is perhaps the best we > > > can do, thanks! > > > > I think this should be documented in the Documentation/power > > directory. After having taken another look at that directory, I see > > that there is only detailed documentation and no overview > > documentation. Maybe I overlooked something but I couldn't find an > > explanation of the system suspend/resume nor of the runtime power > > management concepts in that directory. My understanding is that > > system suspend/resume is about system-wide power state changes > > (hibernation and suspend-to-RAM) and also that runtime power > > management is about changing the power state of a single device or > > bus if no activity has happened within a certain time. > > I actually thought that James wanted a reference to scsi sysfs > attributes documentation, which is also not in the best of shape, to > say the least... > > In any case, I would like to push this fix for 6.6-final as this is a > tracked regression. Martin, James, are you OK with this patch ? Yes, works for me, you can add my Reviewed-by. James
Damien, > In any case, I would like to push this fix for 6.6-final as this is a > tracked regression. Martin, James, are you OK with this patch ? Yep. Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
diff --git a/drivers/ata/libata-scsi.c b/drivers/ata/libata-scsi.c index a371b497035e..3a957c4da409 100644 --- a/drivers/ata/libata-scsi.c +++ b/drivers/ata/libata-scsi.c @@ -1053,10 +1053,11 @@ int ata_scsi_dev_config(struct scsi_device *sdev, struct ata_device *dev) /* * Ask the sd driver to issue START STOP UNIT on runtime suspend - * and resume only. For system level suspend/resume, devices - * power state is handled directly by libata EH. + * and resume and shutdown only. For system level suspend/resume, + * devices power state is handled directly by libata EH. */ sdev->manage_runtime_start_stop = true; + sdev->manage_shutdown = true; } /* diff --git a/drivers/firewire/sbp2.c b/drivers/firewire/sbp2.c index 749868b9e80d..7edf2c95282f 100644 --- a/drivers/firewire/sbp2.c +++ b/drivers/firewire/sbp2.c @@ -1521,6 +1521,7 @@ static int sbp2_scsi_slave_configure(struct scsi_device *sdev) if (sbp2_param_exclusive_login) { sdev->manage_system_start_stop = true; sdev->manage_runtime_start_stop = true; + sdev->manage_shutdown = true; } if (sdev->type == TYPE_ROM) diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c index 83b6a3f3863b..52fa266d976f 100644 --- a/drivers/scsi/sd.c +++ b/drivers/scsi/sd.c @@ -3819,8 +3819,10 @@ static void sd_shutdown(struct device *dev) sd_sync_cache(sdkp, NULL); } - if (system_state != SYSTEM_RESTART && - sdkp->device->manage_system_start_stop) { + if ((system_state != SYSTEM_RESTART && + sdkp->device->manage_system_start_stop) || + (system_state == SYSTEM_POWER_OFF && + sdkp->device->manage_shutdown)) { sd_printk(KERN_NOTICE, sdkp, "Stopping disk\n"); sd_start_stop_device(sdkp, 0); } diff --git a/include/scsi/scsi_device.h b/include/scsi/scsi_device.h index fd41fdac0a8e..7edefb73bf69 100644 --- a/include/scsi/scsi_device.h +++ b/include/scsi/scsi_device.h @@ -164,6 +164,7 @@ struct scsi_device { bool manage_system_start_stop; /* Let HLD (sd) manage system start/stop */ bool manage_runtime_start_stop; /* Let HLD (sd) manage runtime start/stop */ + bool manage_shutdown; /* Let HLD (sd) manage shutdown */ unsigned removable:1; unsigned changed:1; /* Data invalid due to media change */
Commit aa3998dbeb3a ("ata: libata-scsi: Disable scsi device manage_system_start_stop") change setting the manage_system_start_stop flag to false for libata managed disks to enable libata internal management of disk suspend/resume. However, a side effect of this change is that on system shutdown, disks are no longer being stopped (set to standby mode with the heads unloaded). While this is not a critical issue, this unclean shutdown is not recommended and shows up with increased smart counters (e.g. the unexpected power loss counter "Unexpect_Power_Loss_Ct"). Instead of defining a shutdown driver method for all ATA adapter drivers (not all of them define that operation), this patch resolves this issue by further refining the sd driver start/stop control of disks using the new flag manage_shutdown. If set to true, the function sd_shutdown() will issue a START STOP UNIT command with the start argument set to 0 when a disk is shutdown on system power off (system_state == SYSTEM_POWER_OFF). Fixes: aa3998dbeb3a ("ata: libata-scsi: Disable scsi device manage_system_start_stop") Cc: stable@vger.kernel.org Closes: https://bugzilla.kernel.org/show_bug.cgi?id=218038 Link: https://lore.kernel.org/all/cd397c88-bf53-4768-9ab8-9d107df9e613@gmail.com/ Signed-off-by: Damien Le Moal <dlemoal@kernel.org> --- drivers/ata/libata-scsi.c | 5 +++-- drivers/firewire/sbp2.c | 1 + drivers/scsi/sd.c | 6 ++++-- include/scsi/scsi_device.h | 1 + 4 files changed, 9 insertions(+), 4 deletions(-)