ATA still broken

[Lefteris Chatzibarbas]

On Thu, Jan 29, 2004 at 12:37:35PM +0100, S?ren Schmidt wrote:
> Lefteris Chatzibarbas wrote:
> >After the "sos 2004/01/11 14:08:35 PST" commit mentionted in my previous
> >mail, system hangs during boot and I have already sent a more detailed
> >message (hardware, verbose boot, etc.) about this on freebsd-current@ a
> >couple of days ago (some threads above this):
> >
> >  Date: Sun, 25 Jan 2004 13:17:43 +0200
> >  Subject: System still hangs with "ata1-master: TIMEOUT - SETFEATURES SET 
> >  TRANSFER MODE"
> >
> >  http://lists.freebsd.org/pipermail/freebsd-current/2004-January/019497.html
> 
> Well I have no idea whats causing this except those drives does 
> something they shouldn't.
> 
> I'll try to get ahold of one of those faulty devices, but until then 
> your best bet is to dive in yourself and try to figure out what goes 
> wrong. I think we have a Liteon at work that I'll hopefully be able to 
> borrow late next week, of that shows the troublesome behavior I'll work 
> on a solution..

I've been able to boot and have all my devices correctly probed with the
attached patch.  It seems that it is this change that you made to
sys/dev/ata/ata-lowlevel.c, that my system does not like.


-------------- next part --------------
--- sys/dev/ata/ata-lowlevel.c~	Sun Feb  1 14:45:41 2004
+++ sys/dev/ata/ata-lowlevel.c	Mon Feb  2 16:47:14 2004
@@ -645,10 +645,6 @@
 	DELAY(100000);
     }	
 
-    /* enable interrupt */
-    DELAY(10);
-    ATA_IDX_OUTB(ch, ATA_ALTSTAT, ATA_A_4BIT);
-
     if (stat0 & ATA_S_BUSY)
 	mask &= ~0x01;
     if (stat1 & ATA_S_BUSY)
@@ -732,6 +728,9 @@
 	ata_prtdev(atadev, "timeout sending command=%02x\n", command);
 	return -1;
     }
+
+    /* enable interrupt */
+    ATA_IDX_OUTB(atadev->channel, ATA_ALTSTAT, ATA_A_4BIT);
 
     /* only use 48bit addressing if needed (avoid bugs and overhead) */
     if ((lba > 268435455 || count > 256) && atadev->param &&