[Intel-xe] [PATCH V5 3/6] drm/xe: Add job timeout engine property to sysfs

[Matthew Brost]

On Thu, Jul 20, 2023 at 07:55:07PM +0530, Tejas Upadhyay wrote:
> The time after which a job is removed from the scheduler.
> Add sysfs entry to provide user defined job timeout to
> scheduler.
> 
> The job timeout can be adjusted per-engine class using,
> 
> /sys/class/drm/cardX/device/tileN/gtN/engines/ccs/job_timeout_ms
> 
> V3:
>   - Rebase
>   - Update commit message to reflect tile update
> V2:
>   - Use sysfs_create_files as part of this patch
> 
> Signed-off-by: Tejas Upadhyay <tejas.upadhyay at intel.com>
> ---
>  drivers/gpu/drm/xe/xe_engine_class_sysfs.c | 42 ++++++++++++++++++++++
>  1 file changed, 42 insertions(+)
> 
> diff --git a/drivers/gpu/drm/xe/xe_engine_class_sysfs.c b/drivers/gpu/drm/xe/xe_engine_class_sysfs.c
> index 7804cc27bd05..7f4ecef4ef81 100644
> --- a/drivers/gpu/drm/xe/xe_engine_class_sysfs.c
> +++ b/drivers/gpu/drm/xe/xe_engine_class_sysfs.c
> @@ -39,6 +39,37 @@ kobj_xe_engine(struct kobject *parent, char *name)
>  	return khwe;
>  }
>  
> +static ssize_t job_timeout_store(struct kobject *kobj,
> +				 struct kobj_attribute *attr,
> +				 const char *buf, size_t count)
> +{
> +	struct xe_hw_engine *hwe = kobj_to_hwe(kobj);
> +	u32 timeout;
> +	int err;
> +
> +	err = kstrtou32(buf, 0, &timeout);
> +	if (err)
> +		return err;
> +
> +	if (timeout > jiffies_to_usecs(MAX_SCHEDULE_TIMEOUT))

We need a better cap here, this allows a user to break the dma-fence
rules creating endless fences.

> +		return -EINVAL;
> +
> +	WRITE_ONCE(hwe->sched_props.job_timeout_ms, timeout);

Again as I said in the previous patch this really needs to write a per
class structure, not a hwe structure.

Matt

> +
> +	return count;
> +}
> +
> +static ssize_t job_timeout_show(struct kobject *kobj,
> +				struct kobj_attribute *attr, char *buf)
> +{
> +	struct xe_hw_engine *hwe = kobj_to_hwe(kobj);
> +
> +	return sprintf(buf, "%u\n", hwe->sched_props.job_timeout_ms);
> +}
> +
> +static struct kobj_attribute job_timeout_attr =
> +__ATTR(job_timeout_ms, 0644, job_timeout_show, job_timeout_store);
> +
>  static ssize_t job_timeout_default(struct kobject *kobj,
>  				   struct kobj_attribute *attr, char *buf)
>  {
> @@ -80,6 +111,11 @@ static const struct attribute *defaults[] = {
>  	NULL
>  };
>  
> +static const struct attribute *files[] = {
> +	&job_timeout_attr.attr,
> +	NULL
> +};
> +
>  static int xe_add_engine_class_defaults(struct kobject *parent)
>  {
>  	struct kobject *kobj;
> @@ -188,6 +224,12 @@ int xe_engine_class_sysfs_init(struct xe_gt *gt)
>  			return err;
>  		}
>  
> +		err = sysfs_create_files(&khwe->base, files);
> +		if (err) {
> +			kobject_put(kobj);
> +			return err;
> +		}
>  	}
> +
>  	return err;
>  }
> -- 
> 2.25.1
>