Re: [PATCH 8/8] scsi: ufs: Check for completion from the timeout handler

[Bart Van Assche <bvanassche@acm.org>]

On 6/25/24 3:04 AM, Peter Wang (王信友) wrote:
> When eh_timed_out() called concurrently with the regular completion
> handler.
> Both process use test_and_set_bit(SCMD_STATE_COMPLETE, &cmd->state) to
> set SCMD_STATE_COMPLETE flag.
> test_and_set_bit should be atomice operation? and only one can set this
> bit than run forward?

Yes, only one of the two test_and_set_bit(SCMD_STATE_COMPLETE,
&cmd->state) calls will set the SCMD_STATE_COMPLETE bit and only
one of these two calls will return the value 'true'.

> BTW, I still don't understand if both eh_timed_out and regular
> completion handler set SCMD_STATE_COMPLETE,
> what is the relation between SCMD_STATE_COMPLETE and
> ufshcd_queuecommand null pointer?

While ufshcd_eh_timed_out() does not set the SCMD_STATE_COMPLETE bit,
its caller may set that bit after ufshcd_eh_timed_out() has returned.
Hence, a command may be completed while ufshcd_eh_timed_out() is in
progress.

Thanks,

Bart.