]> git.kernelconcepts.de Git - karo-tx-linux.git/commitdiff
[PATCH] libata: replace ap_lock w/ ap->lock in ata_scsi_error()
authorTejun Heo <htejun@gmail.com>
Sun, 2 Jul 2006 18:02:15 +0000 (03:02 +0900)
committerJeff Garzik <jeff@garzik.org>
Thu, 6 Jul 2006 02:05:34 +0000 (22:05 -0400)
ap_lock was used because &ap->host_set->lock was too long and used a
lot.  Now that &ap->host_set->lock is replaced with ap->lock, there's
no reason to keep ap_lock.

[ed. note: that's not entirely true.  ap_lock is a local variable,
caching the results of a de-ref.  In theory, if the compiler is smart
enough, this patch is cosmetic.  However, since this is not a fast
path (it is the error path), this patch is nonetheless acceptable,
even though it _may_ introduce a performance regression.]

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
drivers/scsi/libata-eh.c

index 4a670db9aa05691d8776efbd2006f18a7720297f..d19666c376ad3412efbbfdac2fa59e0338ec8a81 100644 (file)
@@ -190,7 +190,6 @@ enum scsi_eh_timer_return ata_scsi_timed_out(struct scsi_cmnd *cmd)
 void ata_scsi_error(struct Scsi_Host *host)
 {
        struct ata_port *ap = ata_shost_to_port(host);
-       spinlock_t *ap_lock = ap->lock;
        int i, repeat_cnt = ATA_EH_MAX_REPEAT;
        unsigned long flags;
 
@@ -217,7 +216,7 @@ void ata_scsi_error(struct Scsi_Host *host)
                struct scsi_cmnd *scmd, *tmp;
                int nr_timedout = 0;
 
-               spin_lock_irqsave(ap_lock, flags);
+               spin_lock_irqsave(ap->lock, flags);
 
                list_for_each_entry_safe(scmd, tmp, &host->eh_cmd_q, eh_entry) {
                        struct ata_queued_cmd *qc;
@@ -256,15 +255,15 @@ void ata_scsi_error(struct Scsi_Host *host)
                if (nr_timedout)
                        __ata_port_freeze(ap);
 
-               spin_unlock_irqrestore(ap_lock, flags);
+               spin_unlock_irqrestore(ap->lock, flags);
        } else
-               spin_unlock_wait(ap_lock);
+               spin_unlock_wait(ap->lock);
 
  repeat:
        /* invoke error handler */
        if (ap->ops->error_handler) {
                /* fetch & clear EH info */
-               spin_lock_irqsave(ap_lock, flags);
+               spin_lock_irqsave(ap->lock, flags);
 
                memset(&ap->eh_context, 0, sizeof(ap->eh_context));
                ap->eh_context.i = ap->eh_info;
@@ -273,7 +272,7 @@ void ata_scsi_error(struct Scsi_Host *host)
                ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
                ap->pflags &= ~ATA_PFLAG_EH_PENDING;
 
-               spin_unlock_irqrestore(ap_lock, flags);
+               spin_unlock_irqrestore(ap->lock, flags);
 
                /* invoke EH.  if unloading, just finish failed qcs */
                if (!(ap->pflags & ATA_PFLAG_UNLOADING))
@@ -285,14 +284,14 @@ void ata_scsi_error(struct Scsi_Host *host)
                 * recovered the port but before this point.  Repeat
                 * EH in such case.
                 */
-               spin_lock_irqsave(ap_lock, flags);
+               spin_lock_irqsave(ap->lock, flags);
 
                if (ap->pflags & ATA_PFLAG_EH_PENDING) {
                        if (--repeat_cnt) {
                                ata_port_printk(ap, KERN_INFO,
                                        "EH pending after completion, "
                                        "repeating EH (cnt=%d)\n", repeat_cnt);
-                               spin_unlock_irqrestore(ap_lock, flags);
+                               spin_unlock_irqrestore(ap->lock, flags);
                                goto repeat;
                        }
                        ata_port_printk(ap, KERN_ERR, "EH pending after %d "
@@ -302,14 +301,14 @@ void ata_scsi_error(struct Scsi_Host *host)
                /* this run is complete, make sure EH info is clear */
                memset(&ap->eh_info, 0, sizeof(ap->eh_info));
 
-               /* Clear host_eh_scheduled while holding ap_lock such
+               /* Clear host_eh_scheduled while holding ap->lock such
                 * that if exception occurs after this point but
                 * before EH completion, SCSI midlayer will
                 * re-initiate EH.
                 */
                host->host_eh_scheduled = 0;
 
-               spin_unlock_irqrestore(ap_lock, flags);
+               spin_unlock_irqrestore(ap->lock, flags);
        } else {
                WARN_ON(ata_qc_from_tag(ap, ap->active_tag) == NULL);
                ap->ops->eng_timeout(ap);
@@ -321,7 +320,7 @@ void ata_scsi_error(struct Scsi_Host *host)
        scsi_eh_flush_done_q(&ap->eh_done_q);
 
        /* clean up */
-       spin_lock_irqsave(ap_lock, flags);
+       spin_lock_irqsave(ap->lock, flags);
 
        if (ap->pflags & ATA_PFLAG_LOADING) {
                ap->pflags &= ~ATA_PFLAG_LOADING;
@@ -338,7 +337,7 @@ void ata_scsi_error(struct Scsi_Host *host)
        ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
        wake_up_all(&ap->eh_wait_q);
 
-       spin_unlock_irqrestore(ap_lock, flags);
+       spin_unlock_irqrestore(ap->lock, flags);
 
        DPRINTK("EXIT\n");
 }