Re: [PATCH v2 0/5] scsi: ufs: Fix device management timeouts and runtime PM errors

[Peter Wang <peter.wang@mediatek.com>]

On Thu, 2026-10-29 at 15:08 +0800, Stanley Jhu wrote:
> This series fixes two problems in UFS core entry points that issue
> device management commands or touch host registers.
> 
> Patch 1/5 restores the millisecond to jiffies conversion of device
> management command timeouts. Since the timeout moved to rq->timeout,
> a 1500 ms query timeout waits 6 s with CONFIG_HZ=250.
> 
> Patches 2/5 to 5/5 handle runtime PM resume errors. Several UFS core,
> sysfs, BSG, hwmon, and debugfs entry points call
> ufshcd_rpm_get_sync()
> before accessing host registers or issuing device management and UIC
> commands, but ignore its return value.
> 
> ufshcd_rpm_get_sync() wraps pm_runtime_get_sync() on the UFS device
> W-LUN and returns a negative errno when runtime resume of the W-LUN
> or
> host controller fails, leaving the device or controller in
> RPM_SUSPENDED.
> 
> Continuing after such a failure can access host registers or issue
> commands while the device or controller is not known to be active,
> while checking pm_runtime_get_sync() directly would leave
> power.usage_count elevated unless balanced with
> pm_runtime_put_noidle().
> 
> Add ufshcd_rpm_resume_and_get() around pm_runtime_resume_and_get()
> and
> convert the 20 external and helper call sites to abort and propagate
> the
> error when resuming the W-LUN fails. The three lifecycle call sites
> in
> ufshcd_rpm_dev_flush_recheck_work(), ufshcd_wl_shutdown(), and
> ufshcd_remove(), plus ufshcd_err_handling_prepare(), are
> intentionally
> left unchanged.
> 
> Tested on QEMU ARM64 UFS (SDB and MCQ) with W-LUN runtime resume and
> suspend failures via sysfs offline (no behavior change on suspend
> failure) and GDB power.disable_depth=1 to make ufshcd_rpm_get_sync()
> fail, and on SDB with a timed out query request.
> 
> Changes since v1:
> - Drop the incorrect claim that a failed W-LUN runtime suspend
>   latches power.runtime_error from the cover letter and commit
>   messages. The PM core latches a runtime suspend error only if it
>   is neither -EBUSY nor -EAGAIN, and in-tree UFS drivers fail W-LUN
>   runtime suspend only with -EBUSY or -EAGAIN (Bart).
> - Say in the cover letter how the series was tested.
> - Describe each failing sequence in patches 2-5 step by step in
>   time order, and drop the runtime PM disabled (-EACCES) scenario.
> 
> Link:
> https://lore.kernel.org/linux-scsi/20261025013446.345974-1-stanleyjhu@google.com/
>  
> 
> Stanley Jhu (5):
>   scsi: ufs: core: Restore jiffies conversion of device management
>     timeouts
>   scsi: ufs: core: Introduce ufshcd_rpm_resume_and_get() helper
>   scsi: ufs: sysfs: Propagate runtime PM resume errors
>   scsi: ufs: bsg: Abort ufs_bsg_request() when runtime resume fails
>   scsi: ufs: hwmon, debugfs: Propagate runtime PM resume errors
> 
>  drivers/ufs/core/ufs-debugfs.c |  8 +++-
>  drivers/ufs/core/ufs-hwmon.c   | 12 +++++-
>  drivers/ufs/core/ufs-sysfs.c   | 67 +++++++++++++++++++++++++++-----
> --
>  drivers/ufs/core/ufs_bsg.c     |  4 +-
>  drivers/ufs/core/ufshcd-priv.h |  5 +++
>  drivers/ufs/core/ufshcd.c      |  9 +++--
>  6 files changed, 84 insertions(+), 21 deletions(-)
> 
> 
> base-commit: 503e6172fa74da6b4944ece458f5534681e51d21

For the series:
Reviewed-by: Peter Wang <peter.wang@mediatek.com>