All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] scsi: core: Micro-optimize the hot path
@ 2026-09-25 18:09 Bart Van Assche
  2026-09-28  8:58 ` John Garry
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: Bart Van Assche @ 2026-09-25 18:09 UTC (permalink / raw)
  To: Martin K . Petersen
  Cc: linux-scsi, Bart Van Assche, John Garry, Hannes Reinecke

Every code path that inserts scmd->eh_entry into a list uses
list_add_tail() (see also scsi_abort_command() and scsi_eh_scmd_add()).
list_add_tail() overwrites the next and prev pointers. Hence, the only
code that depends on the initialization of scmd->eh_entry is the
BUG_ON(!list_empty(&scmd->eh_entry)) statement in scsi_abort_command().
Remove this statement because scsi_timeout() already prevents
scsi_abort_command() from being called twice for the same command by
setting SCMD_STATE_COMPLETE. Remove the scmd->eh_entry initialization
because the only statement that relied on this initialization has been
removed.

Cc: John Garry <john.garry@linux.dev>
Cc: Hannes Reinecke <hare@kernel.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/scsi/scsi_error.c | 1 -
 drivers/scsi/scsi_lib.c   | 3 ---
 2 files changed, 4 deletions(-)

diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
index 2e0b1909dc69..75442d60c30e 100644
--- a/drivers/scsi/scsi_error.c
+++ b/drivers/scsi/scsi_error.c
@@ -270,7 +270,6 @@ scsi_abort_command(struct scsi_cmnd *scmd)
 	spin_lock_irqsave(&shost->host_lock, flags);
 	if (shost->eh_deadline != -1 && !shost->last_reset)
 		shost->last_reset = jiffies;
-	BUG_ON(!list_empty(&scmd->eh_entry));
 	list_add_tail(&scmd->eh_entry, &shost->eh_abort_list);
 	spin_unlock_irqrestore(&shost->host_lock, flags);
 
diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
index db64470f5229..f695a6605815 100644
--- a/drivers/scsi/scsi_lib.c
+++ b/drivers/scsi/scsi_lib.c
@@ -1322,7 +1322,6 @@ void scsi_init_command(struct scsi_device *dev, struct scsi_cmnd *cmd)
 	}
 
 	cmd->device = dev;
-	INIT_LIST_HEAD(&cmd->eh_entry);
 	INIT_DELAYED_WORK(&cmd->abort_work, scmd_eh_abort_handler);
 }
 
@@ -1590,8 +1589,6 @@ static void scsi_complete(struct request *rq)
 		return;
 	}
 
-	INIT_LIST_HEAD(&cmd->eh_entry);
-
 	atomic_inc(&cmd->device->iodone_cnt);
 	if (cmd->result)
 		atomic_inc(&cmd->device->ioerr_cnt);

^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH] scsi: core: Micro-optimize the hot path
  2026-09-25 18:09 [PATCH] scsi: core: Micro-optimize the hot path Bart Van Assche
@ 2026-09-28  8:58 ` John Garry
  2026-10-03  9:16 ` Hannes Reinecke
  2026-10-03 13:49 ` Martin K. Petersen (Oracle)
  2 siblings, 0 replies; 4+ messages in thread
From: John Garry @ 2026-09-28  8:58 UTC (permalink / raw)
  To: Bart Van Assche, Martin K . Petersen; +Cc: linux-scsi, Hannes Reinecke

On 9/25/26 19:09, Bart Van Assche wrote:

nit: I think that the subject can be more specific, like "don't 
unnecessarily init scmd->eh_entry"

> Every code path that inserts scmd->eh_entry into a list uses
> list_add_tail() (see also scsi_abort_command() and scsi_eh_scmd_add()).
> list_add_tail() overwrites the next and prev pointers. Hence, the only
> code that depends on the initialization of scmd->eh_entry is the
> BUG_ON(!list_empty(&scmd->eh_entry)) statement in scsi_abort_command().
> Remove this statement because scsi_timeout() already prevents
> scsi_abort_command() from being called twice for the same command by
> setting SCMD_STATE_COMPLETE. Remove the scmd->eh_entry initialization
> because the only statement that relied on this initialization has been
> removed.
> 
> Cc: John Garry <john.garry@linux.dev>
> Cc: Hannes Reinecke <hare@kernel.org>
> Signed-off-by: Bart Van Assche <bvanassche@acm.org>

Seems fine, so:
Reviewed-by: John Garry <john.garry@linux.dev>

> ---
>   drivers/scsi/scsi_error.c | 1 -
>   drivers/scsi/scsi_lib.c   | 3 ---
>   2 files changed, 4 deletions(-)
> 
> diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
> index 2e0b1909dc69..75442d60c30e 100644
> --- a/drivers/scsi/scsi_error.c
> +++ b/drivers/scsi/scsi_error.c
> @@ -270,7 +270,6 @@ scsi_abort_command(struct scsi_cmnd *scmd)
>   	spin_lock_irqsave(&shost->host_lock, flags);
>   	if (shost->eh_deadline != -1 && !shost->last_reset)
>   		shost->last_reset = jiffies;
> -	BUG_ON(!list_empty(&scmd->eh_entry));
>   	list_add_tail(&scmd->eh_entry, &shost->eh_abort_list);
>   	spin_unlock_irqrestore(&shost->host_lock, flags);
>   
> diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
> index db64470f5229..f695a6605815 100644
> --- a/drivers/scsi/scsi_lib.c
> +++ b/drivers/scsi/scsi_lib.c
> @@ -1322,7 +1322,6 @@ void scsi_init_command(struct scsi_device *dev, struct scsi_cmnd *cmd)
>   	}
>   
>   	cmd->device = dev;
> -	INIT_LIST_HEAD(&cmd->eh_entry);
>   	INIT_DELAYED_WORK(&cmd->abort_work, scmd_eh_abort_handler);
>   }
>   
> @@ -1590,8 +1589,6 @@ static void scsi_complete(struct request *rq)
>   		return;
>   	}
>   
> -	INIT_LIST_HEAD(&cmd->eh_entry);
> -
>   	atomic_inc(&cmd->device->iodone_cnt);
>   	if (cmd->result)
>   		atomic_inc(&cmd->device->ioerr_cnt);


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] scsi: core: Micro-optimize the hot path
  2026-09-25 18:09 [PATCH] scsi: core: Micro-optimize the hot path Bart Van Assche
  2026-09-28  8:58 ` John Garry
@ 2026-10-03  9:16 ` Hannes Reinecke
  2026-10-03 13:49 ` Martin K. Petersen (Oracle)
  2 siblings, 0 replies; 4+ messages in thread
From: Hannes Reinecke @ 2026-10-03  9:16 UTC (permalink / raw)
  To: Bart Van Assche, Martin K . Petersen
  Cc: linux-scsi, John Garry, Hannes Reinecke

On 9/25/26 8:09 PM, Bart Van Assche wrote:
> Every code path that inserts scmd->eh_entry into a list uses
> list_add_tail() (see also scsi_abort_command() and scsi_eh_scmd_add()).
> list_add_tail() overwrites the next and prev pointers. Hence, the only
> code that depends on the initialization of scmd->eh_entry is the
> BUG_ON(!list_empty(&scmd->eh_entry)) statement in scsi_abort_command().
> Remove this statement because scsi_timeout() already prevents
> scsi_abort_command() from being called twice for the same command by
> setting SCMD_STATE_COMPLETE. Remove the scmd->eh_entry initialization
> because the only statement that relied on this initialization has been
> removed.
> 
> Cc: John Garry <john.garry@linux.dev>
> Cc: Hannes Reinecke <hare@kernel.org>
> Signed-off-by: Bart Van Assche <bvanassche@acm.org>
> ---
>   drivers/scsi/scsi_error.c | 1 -
>   drivers/scsi/scsi_lib.c   | 3 ---
>   2 files changed, 4 deletions(-)
> 
> diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
> index 2e0b1909dc69..75442d60c30e 100644
> --- a/drivers/scsi/scsi_error.c
> +++ b/drivers/scsi/scsi_error.c
> @@ -270,7 +270,6 @@ scsi_abort_command(struct scsi_cmnd *scmd)
>   	spin_lock_irqsave(&shost->host_lock, flags);
>   	if (shost->eh_deadline != -1 && !shost->last_reset)
>   		shost->last_reset = jiffies;
> -	BUG_ON(!list_empty(&scmd->eh_entry));
>   	list_add_tail(&scmd->eh_entry, &shost->eh_abort_list);
>   	spin_unlock_irqrestore(&shost->host_lock, flags);
>   
> diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
> index db64470f5229..f695a6605815 100644
> --- a/drivers/scsi/scsi_lib.c
> +++ b/drivers/scsi/scsi_lib.c
> @@ -1322,7 +1322,6 @@ void scsi_init_command(struct scsi_device *dev, struct scsi_cmnd *cmd)
>   	}
>   
>   	cmd->device = dev;
> -	INIT_LIST_HEAD(&cmd->eh_entry);
>   	INIT_DELAYED_WORK(&cmd->abort_work, scmd_eh_abort_handler);
>   }
>   
> @@ -1590,8 +1589,6 @@ static void scsi_complete(struct request *rq)
>   		return;
>   	}
>   
> -	INIT_LIST_HEAD(&cmd->eh_entry);
> -
>   	atomic_inc(&cmd->device->iodone_cnt);
>   	if (cmd->result)
>   		atomic_inc(&cmd->device->ioerr_cnt);

Reviewed-by: Hannes Reinecke <hare@kernel.org>

Cheers,

Hannes
-- 
Dr. Hannes Reinecke                  Kernel Storage Architect
hare@suse.de                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] scsi: core: Micro-optimize the hot path
  2026-09-25 18:09 [PATCH] scsi: core: Micro-optimize the hot path Bart Van Assche
  2026-09-28  8:58 ` John Garry
  2026-10-03  9:16 ` Hannes Reinecke
@ 2026-10-03 13:49 ` Martin K. Petersen (Oracle)
  2 siblings, 0 replies; 4+ messages in thread
From: Martin K. Petersen (Oracle) @ 2026-10-03 13:49 UTC (permalink / raw)
  To: Bart Van Assche
  Cc: Martin K . Petersen, linux-scsi, John Garry, Hannes Reinecke


Bart,

> Every code path that inserts scmd->eh_entry into a list uses
> list_add_tail() (see also scsi_abort_command() and scsi_eh_scmd_add()).
> list_add_tail() overwrites the next and prev pointers. Hence, the only
> code that depends on the initialization of scmd->eh_entry is the
> BUG_ON(!list_empty(&scmd->eh_entry)) statement in scsi_abort_command().

Applied to 7.4/scsi-staging, thanks!

-- 
Martin K. Petersen

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2026-10-03 13:49 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-09-25 18:09 [PATCH] scsi: core: Micro-optimize the hot path Bart Van Assche
2026-09-28  8:58 ` John Garry
2026-10-03  9:16 ` Hannes Reinecke
2026-10-03 13:49 ` Martin K. Petersen (Oracle)

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.