Re: [PATCH v3 1/2] scsi: ufs: core: Introduce a new clock_gating lock

[Bean Huo <huobean@gmail.com>]

On Tue, 2024-11-05 at 13:25 +0200, Avri Altman wrote:
> -       spin_lock_irqsave(hba->host->host_lock, flags);
> -       /*
> -        * In case you are here to cancel this work the gating state
> -        * would be marked as REQ_CLKS_ON. In this case save time by
> -        * skipping the gating work and exit after changing the clock
> -        * state to CLKS_ON.
> -        */
> -       if (hba->clk_gating.is_suspended ||
> -               (hba->clk_gating.state != REQ_CLKS_OFF)) {
> -               hba->clk_gating.state = CLKS_ON;
> -               trace_ufshcd_clk_gating(dev_name(hba->dev),
> -                                       hba->clk_gating.state);
> -               goto rel_lock;
> +       scoped_guard(spinlock_irqsave, &hba->clk_gating.lock)
> +       {
> +               /*
> +                * In case you are here to cancel this work the
> gating state
> +                * would be marked as REQ_CLKS_ON. In this case save
> time by
> +                * skipping the gating work and exit after changing
> the clock
> +                * state to CLKS_ON.
> +                */
> +               if (hba->clk_gating.is_suspended ||
> +                   hba->clk_gating.state != REQ_CLKS_OFF) {
> +                       hba->clk_gating.state = CLKS_ON;
> +                       trace_ufshcd_clk_gating(dev_name(hba->dev),
> +                                               hba-
> >clk_gating.state);
> +                       return;
> +               }
> +               if (ufshcd_is_ufs_dev_busy(hba) ||
> +                   hba->ufshcd_state != UFSHCD_STATE_OPERATIONAL)
> +                       return;
>         }

I'm wondering if it would be safe to replace host_lock with gating.lock
or scaling.lock. For instance, in above context, ufshcd_state needs to
be checked, but it's currently serialized by host_lock.

King regards,
Bean