[ Original mail is on port-alpha ]

> Hi all, it appears the latest kernel from CVS is hanging during SCSI  
> enumeration on my AS1200.  The hang occurs immediately after finding  
> cd0 on siop, so I'm not sure if it's failing to read some SCSI-ready  
> messages from isp, or if it's crashing on siop...  

> NetBSD 3.99.18 (TERRAN.new) #0: Fri May  5 11:25:16 PDT 2006
>         bard@bsd.gesi.local:/raid/src/sys/arch/alpha/compile/ 
> obj.alpha/TERRAN.old
> AlphaServer 1200 5/533 4MB, 531MHz, s/n NI80901AEW

> isp0 at pci1 dev 3 function 0: QLogic 1020 Fast Wide SCSI HBA
> isp0: interrupting at kn300 irq 12

I think this is interaction of the isp driver with the change in revision
1.135 [*] of sys/dev/scsipi/scsipi_base.c.  My Sun Blade 100 with a QLogic
1040 hangs on boot without the attached patch.  With the patch, all seems
to be fine.  Can you check that it also fixes the problem on the Alpha?

Thanks,

J

[*] See:
  http://mail-index.NetBSD.org/source-changes/2006/04/17/0016.html
  http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/scsipi/scsipi_base.c.diff?r1=1.134&r2=1.135

  - - 8< - - - - - - - - - - - - - Cut here - - - - - - - - - - - - - >8 - -
--- src/sys/dev/ic/isp_netbsd.c.dist	2005-12-11 15:09:38.000000000 +0000
+++ src/sys/dev/ic/isp_netbsd.c	2006-05-06 19:06:26.000000000 +0100
@@ -640,7 +640,6 @@
 void
 isp_done(XS_T *xs)
 {
-	XS_CMD_S_DONE(xs);
 	if (XS_CMD_WDOG_P(xs) == 0) {
 		struct ispsoftc *isp = XS_ISP(xs);
 		callout_stop(&xs->xs_callout);
  - - 8< - - - - - - - - - - - - - Cut here - - - - - - - - - - - - - >8 - -

-- 
  My other computer also runs NetBSD    /        Sailing at Newbiggin
        http://www.netbsd.org/        /   http://www.newbigginsailingclub.org/