Date: Thu, 12 Jan 2006 21:43:34 GMT From: Miles Nordin <carton@Ivy.NET> To: freebsd-gnats-submit@FreeBSD.org Subject: kern/91723: IDE corruption on ICH controller, affects the channel disk was plugged into at boot only Message-ID: <200601122143.k0CLhYtf089460@www.freebsd.org> Resent-Message-ID: <200601122150.k0CLo5GU027028@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 91723
>Category: kern
>Synopsis: IDE corruption on ICH controller, affects the channel disk was plugged into at boot only
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Thu Jan 12 21:50:04 GMT 2006
>Closed-Date:
>Last-Modified:
>Originator: Miles Nordin
>Release: 6.0
>Organization:
>Environment:
FreeBSD cortez 6.0-RELEASE FreeBSD 6.0-RELEASE #1: Tue Nov 8 20:23:54 UTC 2005 carton@cortez:/usr/src/sys/i386/compile/CORTEZ i386
>Description:
I'm booting this system over PXE so I can more effectively diagnose this IDE problem. The system isn't in production, so I'm happy to test things.
If I boot up the system normally, I see repeatable data corruption writing to the disk by executing the following command:
bsdlabel -w -B ad0; bsdlabel ad0
bsdlabel: /dev/ad0: no valid label found
If I put the disk into PIO4 mode:
atacontrol mode ad0 pio4
then the corruption goes away. If I put it back into UDMA33 mode, the corruption returns. Corruption seen with that bsdlabel write-then-read command is repeatable every time---prints the error in UDMA33, works in PIO4.
If I detach and retach the ata0 bus:
atacontrol detach ata0; atacontrol attach ata0
the corruption stays. If I detach both busses:
atacontrol detach ata0; atacontrol detach ata1
[hot-swap the drive to the secondary controller]
atacontrol attach ata1
bsdlabel -w -B ad2; bsdlabel ad2
then the disk _works fine_. Even in UDMA33 mode.
If I reboot the machine with the disk plugged into the secondary IDE controller, then hot-swap it to the primary controller using the above procedure, then the primary controller works fine, and the secondary is broken in UDMA33 mode.
If I boot the machine with the hard disk's power connector unplugged, then plug in the power connector, then 'atacontrol detach ata0; atacontrol attach ata0', then the disk also works fine in UDMA33 mode.
I did try swapping the cable in one of the broken configurations, and the problem stays even with a fresh cable. However, I'd think the fact that I can make the problem manifest or go away, repeatably, by following a certain simple script of actions proves that the cable is good, and there's something wrong with the way the BIOS and the FreeBSD driver are setting up the IDE controller.
This is FreeBSD 6.0 with an Intel ICH IDE controller and an old 8GB western digital drive.
>How-To-Repeat:
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200601122143.k0CLhYtf089460>