Re: [PATCH] ufs: core: Use link recovery when the h8 exit failure during runtime resume

[Bean Huo <huobean@gmail.com>]

On Tue, 2025-07-15 at 11:23 +0900, 이승희 wrote:
> 2> if eh_in_progress, err EBUSY return in ufshcd_runtime_suspend to guarantee the error handling done.
> -> It doesn't work as well.
> 
> --- a/common/drivers/ufs/core/ufshcd.c
> +++ b/common/drivers/ufs/core/ufshcd.c
> @@ -10371,6 +10371,9 @@ int ufshcd_runtime_suspend(struct device *dev)
>         int ret;
>         ktime_t start = ktime_get();
>  
> +       if (ufshcd_eh_in_progress(hba))
> +               return -EBUSY;
> +
>         ret = ufshcd_suspend(hba);
>  
>         trace_ufshcd_runtime_suspend(hba, ret,
> 
> [   63.010841] [4:    kworker/4:0:   52] ufshcd-qcom 1d84000.ufshc: ufshcd_uic_hibern8_exit: hibern8 exit failed. ret = -110
> [   63.010844] [4:    kworker/4:0:   52] ufshcd-qcom 1d84000.ufshc: __ufshcd_wl_resume: hibern8 exit failed -110
> 
> [   63.010845] [0: kworker/u32:10:13604] ufshcd-qcom 1d84000.ufshc: ufshcd_err_handler started; HBA state eh_fatal; powered 1; shutting down 0; saved_err = 0; saved_uic_err = 0; force_reset = 0;
> link is broken
> 
> [   63.011430] [4:    kworker/4:0:   52] ufs_device_wlun 0:0:0:49488: ufshcd_wl_runtime_resume failed: -110
> [   63.011433] [4:    kworker/4:0:   52] ufs_device_wlun 0:0:0:49488: ufshcd_wl_runtime_resume: 574917us, pwr_mode(1), link state(3)
> -> ufshcd_wl_runtime_resume failed done.
>    -> ufshcd_rpm_get_sync() in ufshcd_err_handling_prepare()
> 
> [   63.011457] [4:    kworker/4:0:   52] ufshcd-qcom 1d84000.ufshc: ufshcd_runtime_suspend: eh_in_progress
> -> EBUSY return in ufshcd_runtime_suspend due to eh_in_progress
> 
> [   63.011464]I[0: kworker/u32:10:13604] ufshcd-qcom 1d84000.ufshc: ufshcd_check_errors: Auto Hibern8 Exit failed - status: 0x00000020, upmcrs: 0x00000001
> [   63.011468]I[0: kworker/u32:10:13604] ufshcd-qcom 1d84000.ufshc: ufshcd_check_errors: saved_err 0x20 saved_uic_err 0x0
> 
> [   63.039824] [0: kworker/u32:10:13604] ufshcd-qcom 1d84000.ufshc: ufs_qcom_device_reset: Waiting for device internal cache flush
> 
> [   65.084604] [0: kworker/u32:10:13604] ufshcd-qcom 1d84000.ufshc: ESI configured
> [   65.084728] [0: kworker/u32:10:13604] ufshcd-qcom 1d84000.ufshc: MCQ configured, nr_queues=9, io_queues=8, read_queue=0, poll_queues=1, queue_depth=64
> -> ufshcd_reset_and_restore() works well.
> 
> [   65.105186] [1: kworker/u32:10:13604] ufs_device_wlun 0:0:0:49488: runtime PM trying to activate child device 0:0:0:49488 but parent (target0:0:0) is not active
> -> ufschd_recover_pm_err()
>    -> Because of this error, pm_request_resume doesn't call here.
>    
> [   65.105305] [1: kworker/u32:10:13604] ufshcd-qcom 1d84000.ufshc: ufshcd_err_handler finished; HBA state operational
> [   65.105310] [1: kworker/u32:10:13604] ufshcd-qcom 1d84000.ufshc: ufshcd_err_handler started; HBA state operational; powered 1; shutting down 0; saved_err = 0; saved_uic_err = 0; force_reset = 


Thanks for updating the logs — I can now see the issue clearly.
Please add the appropriate Fixes tag to the patch.

By the way, I prefer the second approach as well and would rather handle recovery in a single interface.
It’s strange that it's not working as expected.
However, I don’t see ufschd_recover_pm_err() being invoked in my code either.

kind regards,
Bean