[PATCH 2/2] libata: Remove 'case ATA_PROT_ATAPI' in ata_host_intr()

[Albert Lee <albertcc@tw.ibm.com>]

[-- Attachment #1: Type: text/plain, Size: 773 bytes --]

Hi Jeff,

This is patch 2/2.
Problem:
  Given ATA_PROT_ATAPI is handled with interrupt off.
=> If any interrupt occurs with protocol ATA_PROT_ATAPI, it is not our interrupt and should be ignored.
(This problem is previously indicated by Bart as in:
http://marc.theaimsgroup.com/?l=linux-ide&m=111822488916620&w=2 )

  In ata_pio_complete(), we have the following code:
	ata_irq_on(ap);
	// interrupt here might cause double completion of the command.
	ata_qc_complete(qc, drv_stat);
So, there might be race condition between ata_pio_complete() and interrupt handler.

Changes:
- Remove 'case ATA_PROT_ATAPI' in ata_host_intr().

Attached please find the patch against the linux-2.6.git tree for your review. Thanks.

Albert

Signed-off-by: Albert Lee <albertcc@tw.ibm.com>

[-- Attachment #2: patch2.diff --]
[-- Type: text/plain, Size: 382 bytes --]

--- linux/drivers/scsi/libata-core.c	2005-06-23 15:57:16.000000000 +0800
+++ linux-intr/drivers/scsi/libata-core.c	2005-06-23 16:06:48.000000000 +0800
@@ -3506,7 +3506,6 @@
 
 	case ATA_PROT_DMA:
 	case ATA_PROT_ATAPI_DMA:
-	case ATA_PROT_ATAPI:
 		/* check status of DMA engine */
 		host_stat = ap->ops->bmdma_status(ap);
 		VPRINTK("ata%u: host_stat 0x%X\n", ap->id, host_stat);