Re: [RFC PATCH v1 1/2] scsi: ufs: introduce vendor isr

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

On 8/5/21 11:34 PM, Kiwoong Kim wrote:
> This patch is to activate some interrupt sources
> that aren't defined in UFSHCI specifications. Those
> purpose could be error handling, workaround or whatever.
> 
> Signed-off-by: Kiwoong Kim <kwmad.kim@samsung.com>
> ---
>   drivers/scsi/ufs/ufshcd.c | 10 ++++++++++
>   drivers/scsi/ufs/ufshcd.h |  8 ++++++++
>   2 files changed, 18 insertions(+)
> 
> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> index 05495c34a2b7..f85a9b335e0b 100644
> --- a/drivers/scsi/ufs/ufshcd.c
> +++ b/drivers/scsi/ufs/ufshcd.c
> @@ -6428,6 +6428,16 @@ static irqreturn_t ufshcd_tmc_handler(struct ufs_hba *hba)
>   static irqreturn_t ufshcd_sl_intr(struct ufs_hba *hba, u32 intr_status)
>   {
>   	irqreturn_t retval = IRQ_NONE;
> +	int res = 0;
> +	unsigned long flags;
> +
> +	retval = ufshcd_vops_intr(hba, &res);
> +	if (res) {
> +		spin_lock_irqsave(hba->host->host_lock, flags);
> +		hba->force_reset = true;
> +		ufshcd_schedule_eh_work(hba);
> +		spin_unlock_irqrestore(hba->host->host_lock, flags);
> +	}

How can a non-standard extension have error handling code in common 
code? Please move the code under if (res) into the vendor code.

>   	if (intr_status & UFSHCD_UIC_MASK)
>   		retval |= ufshcd_uic_cmd_compl(hba, intr_status);
> diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> index 971cfabc4a1e..1ed0a911f864 100644
> --- a/drivers/scsi/ufs/ufshcd.h
> +++ b/drivers/scsi/ufs/ufshcd.h
> @@ -356,6 +356,7 @@ struct ufs_hba_variant_ops {
>   			       const union ufs_crypto_cfg_entry *cfg, int slot);
>   	void	(*event_notify)(struct ufs_hba *hba,
>   				enum ufs_event_type evt, void *data);
> +	irqreturn_t	(*intr)(struct ufs_hba *hba, int *res);
>   };
>   
>   /* clock gating state  */
> @@ -1296,6 +1297,13 @@ static inline void ufshcd_vops_config_scaling_param(struct ufs_hba *hba,
>   		hba->vops->config_scaling_param(hba, profile, data);
>   }
>   
> +static inline irqreturn_t ufshcd_vops_intr(struct ufs_hba *hba, int *res)
> +{
> +	if (hba->vops && hba->vops->intr)
> +		return hba->vops->intr(hba, res);
> +	return IRQ_NONE;
> +}
> +
>   extern struct ufs_pm_lvl_states ufs_pm_lvl_states[];

So this code adds an indirect function call in the interrupt handler? 
This will have a negative impact on performance, especially on a kernel 
with security mitigations enabled. See also 
https://lwn.net/Articles/774743/.

Thanks,

Bart.