Bug 84370 - (SCSI AIC79XX)SCSI corruption with aic79xx driver versions 1.0.0 and 1.1.0
(SCSI AIC79XX)SCSI corruption with aic79xx driver versions 1.0.0 and 1.1.0
Status: CLOSED WONTFIX
Product: Red Hat Linux
Classification: Retired
Component: kernel (Show other bugs)
8.0
athlon Linux
medium Severity high
: ---
: ---
Assigned To: Dave Jones
Brian Brock
:
Depends On:
Blocks:
  Show dependency treegraph
 
Reported: 2003-02-14 19:15 EST by Stephen Walton
Modified: 2015-01-04 17:02 EST (History)
5 users (show)

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Environment:
Last Closed: 2004-05-27 10:10:54 EDT
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---


Attachments (Terms of Use)

  None (edit)
Description Stephen Walton 2003-02-14 19:15:08 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20021003

Description of problem:
I have been seeing persistent data corruption when running I/O intensive
operations, such as a kernel rebuild, on a system with the following specs: 
Tyan S2469UGN motherboard, two AMD Athlon 2200+ CPU's, three Western Digital 160
GB IDE disks, and two Seagate Cheetah 36.7 GB LVD disks.  The vendor, Monarch
Computer, shipped me a driver disk with version 1.1.0 of the aic79xx driver.

Updating to version 1.3.1 of the aic79xx driver seems to have fixed my problems;
 this driver can be downloaded from http://people.freebsd.org/~gibbs/linux .


Version-Release number of selected component (if applicable):
2.4.18-14smp

How reproducible:
Always

Steps to Reproduce:
1.  Buy or build a dual Athlon system with an Adaptec U320 controller.
2.  Install the kernel-source RPM and build a kernel.  OR
3.  Download and install the ATLAS subroutine library from
math-atlas.sourceforge.net.
4   Building on the same system on an IDE disk should work normally.
    

Actual Results:  Data corruption:  oddball syntax errors from the kernel build

Expected Results:  No corruption

Additional info:

RedHat should in my opinion absolutely ship version 1.3.1 of Adaptec's aic79xx
driver with its kernels.
Comment 1 Nate Faerber 2003-02-18 15:09:38 EST
I second this opinion.  We have been having major issues with anything below the
1.3.0 version of this driver.  Although I haven't gotten my hands on 1.3.1, it
should be included with this all new kernels including 8.1.
Comment 2 Nate Faerber 2003-02-18 15:20:28 EST
Sorry, I realize my previous comment was painfully unhelpful.

We have experienced serious stability issues with regards to the 1.0.0 and 1.1.0
drivers on dual Xeon based motherboards.  These are Tyan S2665 and S2721. 
Basically, lots of disk IO and/or system stress can and will cause kernel panics
and system lockups.

One easily reproducible test should be to create a RAID 0 stripe with 3 drives
and run a disk benchmark like bonnie++.
Comment 3 Samuel Flory 2003-04-01 14:18:31 EST
I've seen similar issues on the intel SE7501wv2.   Under both Red Hat's latest
eratta kernel, and 2.4.21-pre5-ac1.  I saw file system corruption.  Updating to
the 1.3.5 driver seems to have fixed the issue.
http://people.freebsd.org/~gibbs/linux/

More info on my issues here:
http://marc.theaimsgroup.com/?l=linux-kernel&m=104916069915618&w=2
Comment 4 Nate Faerber 2003-04-01 14:31:14 EST
Yes, 1.3.5 has helped immensely with our issues.  The point right now is to get
Red Hat to recognize the problem with 1.0.0 drivers and update the kernel to
include _reliable_ versions of the aic79xx driver.
Comment 5 Samuel Flory 2003-04-01 15:11:34 EST
   Hmmm I think I gave the wrong version.  I could swear Justin said 1.3.5, but
the driver I'm using claims it's 1.3.4.
Comment 6 Nate Faerber 2003-04-01 15:33:54 EST
It may be a snapshot of the 1.3.5 driver that was released a week before the
official 1.3.5 driver.  This snapshot still had the 1.3.4 version number in it.
 The snapshot was dated 20030318 while the driver properly label 1.3.5 is 20030325.
Comment 7 Alan Cox 2003-06-05 11:06:07 EDT
Yes. 1.3.5 has been running in the -ac tree for a while and it seems way way better
Comment 8 Nate Faerber 2003-06-05 16:42:56 EDT
1.3.8 has fixed even more problems for us.  Based on my testing, I wouldn't
recommend anything less than 1.3.8 on systems with heavy disk I/O.

Alan, do you know what's holding Red Hat back from updating this driver in the
errata kernels?
Comment 9 Alan Cox 2003-06-08 11:13:45 EDT
Well founded paranoia of changes breaking things mostly.

Note You need to log in before you can comment on or make changes to this bug.