Re: [2.6.28-rc] Sata soft reset filling log

[Tejun Heo <tj@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 903 bytes --]

Hello,

Justin Madru wrote:
>> Ah.. strange.  The command protocol is ATAPI_PROT_NODATA and the drive
>> should be setting DRQ to receive CDB but it's not doing that.
>> Strange.  It could be that the drive is a little bit tardy with DRQ -
>> ie. setting DRQ after clearing BUSY and ata_piix is now successfully
>> hitting the small window for some reason.
>>
>> Does the attached patch make any difference?
>>
>> Thanks.
>>
>>   
> Ok, I applied your patch. Unfortunately, it didn't fix my problem.
> I now get a different, but similar message; not sure if it happens less
> frequently - still often though.

Aiee...

> What changes happened between .27 and .28 could've caused this?
> Anyways, below is what I get now (with your patch), including the
> frequency of occurrence.

I have no idea and I'm afraid we'll need a bisection here. :-( Can you
please try the attached patch?

Thanks.

-- 
tejun

[-- Attachment #2: DRQ-debug-1.patch --]
[-- Type: text/x-patch, Size: 668 bytes --]

diff --git a/drivers/ata/libata-sff.c b/drivers/ata/libata-sff.c
index 4b47394..fcd7d08 100644
--- a/drivers/ata/libata-sff.c
+++ b/drivers/ata/libata-sff.c
@@ -1120,6 +1120,22 @@ fsm_start:
 		 */
 		poll_next = (qc->tf.flags & ATA_TFLAG_POLLING);
 
+		{
+			int cnt = 0;
+
+			while (!(status & (ATA_DRQ | ATA_ERR | ATA_DF)) &&
+			       cnt < 100) {
+				status = ap->ops->sff_check_status(ap);
+				cnt++;
+				udelay(10);
+			}
+
+			if (cnt)
+				ata_port_printk(ap, KERN_INFO,
+						"XXX status=%02x after %d tries\n",
+						status, cnt);
+		}
+
 		/* check device status */
 		if (unlikely((status & ATA_DRQ) == 0)) {
 			/* handle BSY=0, DRQ=0 as error */