scsi: hisi_sas: work around build failure in suspend function
Subsystems (2)
PATCH 1/1 scsi: hisi_sas: work around build failure in suspend function
From: Arnd Bergmann <[email protected]>
The suspend/resume functions in this driver seem to have multiple problems, the latest one just got introduced by a bugfix:
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c: In function '_suspend_v3_hw':drivers/scsi/hisi_sas/hisi_sas_v3_hw.c:5142:39: error: 'struct dev_pm_info' has no member named 'usage_count'
5142 | if (atomic_read(&device->power.usage_count)) {drivers/scsi/hisi_sas/hisi_sas_v3_hw.c: In function '_suspend_v3_hw':drivers/scsi/hisi_sas/hisi_sas_v3_hw.c:5142:39: error: 'struct dev_pm_info' has no member named 'usage_count'
5142 | if (atomic_read(&device->power.usage_count)) {
As far as I can tell, the 'usage_count' is not meant to be accessed by device drivers at all, though I don't know what the driver is supposed to do instead.
Another problem is the use of the deprecated UNIVERSAL_DEV_PM_OPS(), and marking functions as __maybe_unused to avoid warnings about unused functions. This should probably be changed to using DEFINE_RUNTIME_DEV_PM_OPS().
Both changes require actually understanding what the driver needs to do, and being able to test this, so instead here is the simplest patch to make it pass the randconfig builds instead.
---
Maintainers: If possible, please revisit this to do a proper fix. If that takes too much time, this patch can be applied as a workaround in the meantime, and might also help in case the e368d38cb952 patch gets backported to stable kernels.
--- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c +4 -0
@@ -5139,11 +5139,13 @@ static int _suspend_v3_hw(struct device *device)flush_workqueue (hisi_hba->wq);interrupt_disable_v3_hw (hisi_hba); +#ifdef CONFIG_PM if (atomic_read (&device->power.usage_count)) {dev_err (dev,"PM suspend: host status cannot be suspended\n" ); rc = -EBUSY ; goto err_out; } +#endif rc =disable_host_v3_hw (hisi_hba);if (rc) {
@@ -5162,7 +5164,9 @@ static int _suspend_v3_hw(struct device *device) err_out_recover_host:enable_host_v3_hw (hisi_hba); +#ifdef CONFIG_PM err_out: +#endif interrupt_enable_v3_hw (hisi_hba);clear_bit (HISI_SAS_REJECT_CMD_BIT , &hisi_hba->flags);clear_bit (HISI_SAS_RESETTING_BIT , &hisi_hba->flags);
Signature · 1 line
2.39.2
Hi Arnd,
Quoted 12 lines · Arnd Bergmann
From: Arnd Bergmann <[email protected]>
The suspend/resume functions in this driver seem to have multiple problems, the latest one just got introduced by a bugfix:
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c: In function '_suspend_v3_hw':drivers/scsi/hisi_sas/hisi_sas_v3_hw.c:5142:39: error: 'structdev_pm_info' has no member named 'usage_count'
5142 | if (atomic_read(&device->power.usage_count)) {drivers/scsi/hisi_sas/hisi_sas_v3_hw.c: In function '_suspend_v3_hw':drivers/scsi/hisi_sas/hisi_sas_v3_hw.c:5142:39: error: 'structdev_pm_info' has no member named 'usage_count'
5142 | if (atomic_read(&device->power.usage_count)) {
Quoted 3 lines
As far as I can tell, the '
usage_count' is not meant to be accessed by device drivers at all, though I don't know what the driver is supposed to do instead.
Thank you for reporting the issue.
There is a extreme situation that hisi_sas driver tries to resume controller when it is in the process of suspend, which will cause a deadlock. So we check usage_count of controller, and if usage_count > 0, failed to suspend to avoid the issue. But there is no common function defined in pm_runtime.h which check the usage_count of device, so use it directly (i saw the check also be used in other drivers). But i didn't realize that member usage_count is defined only under CONFIG_PM=y.
Quoted 5 lines
Another problem is the use of the deprecated
UNIVERSAL_DEV_PM_OPS(), and marking functions as__maybe_unusedto avoid warnings about unused functions. This should probably be changed to usingDEFINE_RUNTIME_DEV_PM_OPS().
We use UNIVERSAL_DEV_PM_OPS() just because runtime callbacks runtime_{suspend|resume} and system callbacks {suspend|resume} use the same operations in the driver, otherwise, we need to use DEFINE_RUNTIME_DEV_PM_OPS() to define runtime callbacks and use DEFINE_SIMPLE_DEV_PM_OPS() to define system callbacks.
Quoted 5 lines
Both changes require actually understanding what the driver needs to do, and being able to test this, so instead here is the simplest patch to make it pass the randconfig builds instead.
Quoted 7 lines
Fixes e368d38cb952 ("scsi: hisi_sas: Exit suspend state when usage count is greater than 0")Signed-off-by Arnd Bergmann <[email protected]>---
Maintainers: If possible, please revisit this to do a proper fix. If that takes too much time, this patch can be applied as a workaround in the meantime, and might also help in case thee368d38cb952patch gets backported to stable kernels.
It is ok to apply the patch as a workaround as soon as possible.
Thanks!
Quoted 32 lines
--- drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 4 ++++ 1 file changed, 4 insertions(+)drivers/scsi/hisi_sas/hisi_sas_v3_hw.c +4
-0@@ -5139,11 +5139,13 @@ static int _suspend_v3_hw(struct device *device)flush_workqueue (hisi_hba->wq);interrupt_disable_v3_hw (hisi_hba); +#ifdef CONFIG_PM if (atomic_read (&device->power.usage_count)) {dev_err (dev,"PM suspend: host status cannot be suspended\n" ); rc = -EBUSY ; goto err_out; } +#endif rc =disable_host_v3_hw (hisi_hba);if (rc) {@@ -5162,7 +5164,9 @@ static int _suspend_v3_hw(struct device *device) err_out_recover_host:enable_host_v3_hw (hisi_hba); +#ifdef CONFIG_PM err_out: +#endif interrupt_enable_v3_hw (hisi_hba);clear_bit (HISI_SAS_REJECT_CMD_BIT , &hisi_hba->flags);clear_bit (HISI_SAS_RESETTING_BIT , &hisi_hba->flags);
Martin K. Petersen
AppliedArnd,
Quoted 2 lines
The suspend/resume functions in this driver seem to have multiple problems, the latest one just got introduced by a bugfix:
Applied to 6.4/scsi-staging, thanks!
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering
Martin K. Petersen
AppliedQuoted 11 lines · Arnd Bergmann
The suspend/resume functions in this driver seem to have multiple problems, the latest one just got introduced by a bugfix:
drivers/scsi/hisi_sas/hisi_sas_v3_hw.c: In function '_suspend_v3_hw':drivers/scsi/hisi_sas/hisi_sas_v3_hw.c:5142:39: error: 'structdev_pm_info' has no member named 'usage_count'
5142 | if (atomic_read(&device->power.usage_count)) {drivers/scsi/hisi_sas/hisi_sas_v3_hw.c: In function '_suspend_v3_hw':drivers/scsi/hisi_sas/hisi_sas_v3_hw.c:5142:39: error: 'structdev_pm_info' has no member named 'usage_count'
5142 | if (atomic_read(&device->power.usage_count)) {[...]
Applied to 6.4/scsi-queue, thanks!
[1/1] scsi: hisi_sas: work around build failure in suspend function
https://git.kernel.org/mkp/scsi/c/e01e2290f094
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering