On Thu, 19 Apr 2007 01:58:38 -0600
"Eric W. Biederman" <ebiederm@xmission.com> wrote:

quoted text
> From: Eric W. Biederman <ebiederm@xmission.com>
> 
> This patch modifies the sas scsi host thread startup
> to use kthread_run not kernel_thread and deamonize.
> kthread_run is slightly simpler and more maintainable.
> 

Again, I'll rename this to "partially convert...".  This driver should be
using kthread_should_stop() and kthread_stop() rather than the
apparently-unnecessary ->queue_thread_kill thing.

This driver was merged two and a half years after the kthread API was
available.   Our coding-vs-reviewing effort is out of balance.


quoted text
> ---
>  drivers/scsi/libsas/sas_scsi_host.c |   11 ++++++-----
>  1 files changed, 6 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/scsi/libsas/sas_scsi_host.c b/drivers/scsi/libsas/sas_scsi_host.c
> index 46ba3a7..7a38ac5 100644
> --- a/drivers/scsi/libsas/sas_scsi_host.c
> +++ b/drivers/scsi/libsas/sas_scsi_host.c
> @@ -40,6 +40,7 @@
>  #include <linux/blkdev.h>
>  #include <linux/scatterlist.h>
>  #include <linux/freezer.h>
> +#include <linux/kthread.h>
>  
>  /* ---------- SCSI Host glue ---------- */
>  
> @@ -870,7 +871,6 @@ static int sas_queue_thread(void *_sas_ha)
>  	struct sas_ha_struct *sas_ha = _sas_ha;
>  	struct scsi_core *core = &sas_ha->core;
>  
> -	daemonize("sas_queue_%d", core->shost->host_no);
>  	current->flags |= PF_NOFREEZE;
>  
>  	complete(&queue_th_comp);
> @@ -891,19 +891,20 @@ static int sas_queue_thread(void *_sas_ha)
>  
>  int sas_init_queue(struct sas_ha_struct *sas_ha)
>  {
> -	int res;
>  	struct scsi_core *core = &sas_ha->core;
> +	struct task_struct *task;
>  
>  	spin_lock_init(&core->task_queue_lock);
>  	core->task_queue_size = 0;
>  	INIT_LIST_HEAD(&core->task_queue);
>  	init_MUTEX_LOCKED(&core->queue_thread_sema);
>  
> -	res = kernel_thread(sas_queue_thread, sas_ha, 0);
> -	if (res >= 0)
> +	task = kthread_run(sas_queue_thread, sas_ha,
> +			   "sas_queue_%d", core->shost->host_no);
> +	if (!IS_ERR(task))
>  		wait_for_completion(&queue_th_comp);
>  
> -	return res < 0 ? res : 0;
> +	return IS_ERR(task) ? PTR_ERR(task) : 0;

Does that wait_for_completion(&queue_th_comp) actually do anything useful?

If so, what is serialising access to the single queue_th_comp?
-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/