[PATCH] Fix dma timeout bugs

From Alexander Atanasov <alex@ssi.bg>

Fix DMA I/O and state machine fixes, error recovery

drivers/ide/ide-dma.c

@@ -442,9 +442,10 @@ static int config_drive_for_dma (ide_drive_t *drive)
  *	the driver to resolve the problem, if a DMA transfer is still
  *	in progress we continue to wait (arguably we need to add a 
  *	secondary 'I don't care what the drive thinks' timeout here)
- *	Finally if we have an interrupt but for some reason got the
- *	timeout first we complete the I/O. This can occur if an 
- *	interrupt is lost or due to bugs.
+ *	Finally if we have an interrupt we let it complete the I/O.
+ *	But only one time - we clear expiry and if it's still not
+ *	completed after WAIT_CMD, we error and retry in PIO.
+ *	This can occur if an interrupt is lost or due to hang or bugs.
  */
  
 static int dma_timer_expiry (ide_drive_t *drive)
@@ -461,19 +462,16 @@ static int dma_timer_expiry (ide_drive_t *drive)
 	HWGROUP(drive)->expiry = NULL;	/* one free ride for now */
 
 	/* 1 dmaing, 2 error, 4 intr */
+	if (dma_stat & 2)	/* ERROR */
+		return -1;
 
-	if (dma_stat & 2) {	/* ERROR */
-		(void) hwif->ide_dma_end(drive);
-		return DRIVER(drive)->error(drive,
-			"dma_timer_expiry", hwif->INB(IDE_STATUS_REG));
-	}
 	if (dma_stat & 1)	/* DMAing */
 		return WAIT_CMD;
 
 	if (dma_stat & 4)	/* Got an Interrupt */
-		HWGROUP(drive)->handler(drive);
+		return WAIT_CMD;
 
-	return 0;
+	return 0;	/* Status is unknown -- reset the bus */
 }
 
 /**

drivers/ide/ide-io.c

@@ -980,21 +980,25 @@ void do_ide_request(request_queue_t *q)
  * retry the current request in pio mode instead of risking tossing it
  * all away
  */
-void ide_dma_timeout_retry(ide_drive_t *drive)
+static ide_startstop_t ide_dma_timeout_retry(ide_drive_t *drive, int error)
 {
 	ide_hwif_t *hwif = HWIF(drive);
 	struct request *rq;
+	ide_startstop_t ret = ide_stopped;
 
 	/*
 	 * end current dma transaction
 	 */
-	(void) hwif->ide_dma_end(drive);
 
-	/*
-	 * complain a little, later we might remove some of this verbosity
-	 */
-	printk(KERN_WARNING "%s: timeout waiting for DMA\n", drive->name);
+	if (error < 0) {
+		printk(KERN_WARNING "%s: DMA timeout error\n", drive->name);
+		(void)HWIF(drive)->ide_dma_end(drive);
+		ret = DRIVER(drive)->error(drive, "dma timeout error",
+						hwif->INB(IDE_STATUS_REG));
+	} else {
+		printk(KERN_WARNING "%s: DMA timeout retry\n", drive->name);
 		(void) hwif->ide_dma_timeout(drive);
+	}
 
 	/*
 	 * disable dma for now, but remember that we did so because of
@@ -1018,9 +1022,9 @@ void ide_dma_timeout_retry(ide_drive_t *drive)
 	rq->hard_cur_sectors = rq->current_nr_sectors;
 	if (rq->bio)
 		rq->buffer = NULL;
-}
 
-EXPORT_SYMBOL(ide_dma_timeout_retry);
+	return ret;
+}
 
 /**
  *	ide_timer_expiry	-	handle lack of an IDE interrupt
@@ -1042,10 +1046,9 @@ void ide_timer_expiry (unsigned long data)
 	ide_handler_t	*handler;
 	ide_expiry_t	*expiry;
 	unsigned long	flags;
-	unsigned long	wait;
+	unsigned long	wait = -1;
 
 	spin_lock_irqsave(&ide_lock, flags);
-	del_timer(&hwgroup->timer);
 
 	if ((handler = hwgroup->handler) == NULL) {
 		/*
@@ -1072,7 +1075,7 @@ void ide_timer_expiry (unsigned long data)
 			}
 			if ((expiry = hwgroup->expiry) != NULL) {
 				/* continue */
-				if ((wait = expiry(drive)) != 0) {
+				if ((wait = expiry(drive)) > 0) {
 					/* reset timer */
 					hwgroup->timer.expires  = jiffies + wait;
 					add_timer(&hwgroup->timer);
@@ -1107,15 +1110,15 @@ void ide_timer_expiry (unsigned long data)
 				startstop = handler(drive);
 			} else {
 				if (drive->waiting_for_dma) {
-					startstop = ide_stopped;
-					ide_dma_timeout_retry(drive);
+					startstop = ide_dma_timeout_retry(drive, wait);
 				} else
-					startstop = DRIVER(drive)->error(drive, "irq timeout", hwif->INB(IDE_STATUS_REG));
+					startstop =
+					DRIVER(drive)->error(drive, "irq timeout", hwif->INB(IDE_STATUS_REG));
 			}
 			set_recovery_timer(hwif);
 			drive->service_time = jiffies - drive->service_start;
-			enable_irq(hwif->irq);
 			spin_lock_irq(&ide_lock);
+			enable_irq(hwif->irq);
 			if (startstop == ide_stopped)
 				hwgroup->busy = 0;
 		}