Re: [PATCH v2 8/8] scsi: target: tcmu: Make TMR notification optional

[Mike Christie <michael.christie@oracle.com>]

On 7/17/20 11:12 AM, Bodo Stroesser wrote:
> Add "tmr_notification" configFS attribute to tcmu devices.
> If default value 0 of the attribute is used, tcmu only
> removes aborted commands from qfull_queue.
> If user changes tmr_notification to 1, additionally
> TMR notifications will be written to the cmd ring.
> 
> Signed-off-by: Bodo Stroesser <bstroesser@ts.fujitsu.com>
> ---
>  drivers/target/target_core_user.c | 39 +++++++++++++++++++++++++++++++++++++++
>  1 file changed, 39 insertions(+)
> 
> diff --git a/drivers/target/target_core_user.c b/drivers/target/target_core_user.c
> index 92b7a2e84e64..95e66b707373 100644
> --- a/drivers/target/target_core_user.c
> +++ b/drivers/target/target_core_user.c
> @@ -118,6 +118,7 @@ struct tcmu_dev {
>  #define TCMU_DEV_BIT_OPEN 0
>  #define TCMU_DEV_BIT_BROKEN 1
>  #define TCMU_DEV_BIT_BLOCKED 2
> +#define TCMU_DEV_BIT_TMR_NOTIFY 3
>  	unsigned long flags;
>  
>  	struct uio_info uio_info;
> @@ -1260,6 +1261,9 @@ tcmu_tmr_notify(struct se_device *se_dev, enum tcm_tmreq_table tmf,
>  	if (unqueued)
>  		tcmu_set_next_deadline(&dev->qfull_queue, &dev->qfull_timer);
>  
> +	if (!test_bit(TCMU_DEV_BIT_TMR_NOTIFY, &dev->flags))
> +		goto unlock;
> +
>  	pr_debug("TMR event %d on dev %s, aborted cmds %d, afflicted cmd_ids %d\n",
>  		 tcmu_tmr_type(tmf), dev->name, i, cmd_cnt);
>  
> @@ -2706,6 +2710,40 @@ static ssize_t tcmu_emulate_write_cache_store(struct config_item *item,
>  }
>  CONFIGFS_ATTR(tcmu_, emulate_write_cache);
>  
> +static ssize_t tcmu_tmr_notification_show(struct config_item *item,
> +					       char *page)

Sorry about this. Just some nits.

The spacing above got messed up a little. I think you only need 2 spaces, but it looks like we got some extras.


> +{
> +	struct se_dev_attrib *da = container_of(to_config_group(item),
> +					struct se_dev_attrib, da_group);
> +	struct tcmu_dev *dev = TCMU_DEV(da->da_dev);


Could you use udev or tcmu_dev for the name. Sorry for being a broken record on this one. We use dev or se_dev for the se_device struct already and it throws me off when scanning the code.

I think patch 5 and 7 need the same fix up.