Patchwork [06/27] UBUNTU: [Upstream] SCSI: AACRAID: Relax the tight timeout loop on fib commands

login
register
mail settings
Submitter Benjamin Collins
Date June 11, 2012, 6:44 p.m.
Message ID <ac221a34df5a45aacc29b2cdc156c4e778506fce.1339455421.git.bcollins@ubuntu.com>
Download mbox | patch
Permalink /patch/164265/
State New
Headers show

Comments

Benjamin Collins - June 11, 2012, 6:44 p.m.
The loop that waited for syncronous fib commands was causing a CPU stall
when a timeout actually occured.

1) Switch to using a more accurate timeout mechanism.
2) Do not pace the loop with udelay(). Use cpu_relax() to allow for
   scheduling to occur.

Signed-off-by: Ben Collins <bcollins@ubuntu.com>
Cc: Adaptec OEM Raid Solutions <aacraid@adaptec.com
Cc: linux-scsi@vger.kernel.org
---
 drivers/scsi/aacraid/commsup.c |    9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

Patch

diff --git a/drivers/scsi/aacraid/commsup.c b/drivers/scsi/aacraid/commsup.c
index 4b32ca4..906a501 100644
--- a/drivers/scsi/aacraid/commsup.c
+++ b/drivers/scsi/aacraid/commsup.c
@@ -564,10 +564,10 @@  int aac_fib_send(u16 command, struct fib *fibptr, unsigned long size,
 			 * functioning because an interrupt routing or other
 			 * hardware failure has occurred.
 			 */
-			unsigned long count = 36000000L; /* 3 minutes */
+			unsigned long timeout = jiffies + (180 * HZ); /* 3 minutes */
 			while (down_trylock(&fibptr->event_wait)) {
 				int blink;
-				if (--count == 0) {
+				if (time_is_before_eq_jiffies(timeout)) {
 					struct aac_queue * q = &dev->queues->queue[AdapNormCmdQueue];
 					spin_lock_irqsave(q->lock, qflags);
 					q->numpending--;
@@ -588,7 +588,10 @@  int aac_fib_send(u16 command, struct fib *fibptr, unsigned long size,
 					}
 					return -EFAULT;
 				}
-				udelay(5);
+				/* We used to udelay() here but that absorbed
+				 * a CPU when a timeout occured. Not very
+				 * useful. */
+				cpu_relax();
 			}
 		} else if (down_interruptible(&fibptr->event_wait)) {
 			/* Do nothing ... satisfy