From nobody@FreeBSD.ORG Tue Jun 29 09:25:13 1999
Return-Path: <nobody@FreeBSD.ORG>
Received: by hub.freebsd.org (Postfix, from userid 32767)
	id 655F91533B; Tue, 29 Jun 1999 09:25:13 -0700 (PDT)
Message-Id: <19990629162513.655F91533B@hub.freebsd.org>
Date: Tue, 29 Jun 1999 09:25:13 -0700 (PDT)
From: peter@gbg.netman.se
Sender: nobody@FreeBSD.ORG
To: freebsd-gnats-submit@freebsd.org
Subject: IBM -DTTA-351010 corrupts data when using DMA 32-bit & multi-block i/o
X-Send-Pr-Version: www-1.0

>Number:         12450
>Category:       kern
>Synopsis:       IBM -DTTA-351010 corrupts data when using DMA 32-bit & multi-block i/o
>Confidential:   no
>Severity:       serious
>Priority:       low
>Responsible:    freebsd-bugs
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Jun 29 09:30:00 PDT 1999
>Closed-Date:    Sun Aug 29 08:18:00 PDT 1999
>Last-Modified:  Sun Aug 29 08:18:53 PDT 1999
>Originator:     Peter Hkanson
>Release:        3.1 (cdrom kit)
>Organization:
Network management 
>Environment:
FreeBSD ture.home.se 3.1-RELEASE FreeBSD 3.1-RELEASE #9: Sun Jun 27 20:54:12 CEST 1999    
peter@ture.home.se:/usr/src/sys/compile/TURE  i386

>Description:
When enabling DMA using the flags paramater in config file
(controller  wdc1  at isa? port "IO_WD2" bio irq 15 flags 0xa0ffa0ff)
IBM-DTTA-351010 drive will occationly corrupt data during write.
Same disk works ok (but slow) if generic kernel or no flags given in
config file. Same system also houses QUANTUM FIREBALL SE3.2A that
works ok with this flag.

Board is chaintech 5agm2 using cyrix m-II (to be replaced!)
On-board is chip2: <VIA 82C586 PCI-ISA bridge> rev 0x47 and
<VIA 82C586x (Apollo) Bus-master IDE controller,rev 0x06 on pci0.7.1
 chip3: <VIA 82C586B ACPI interface> rev 0x10



>How-To-Repeat:
Write a sufficiently large file (> 100Mbyte) and examine with md5
subsequent reads of the same file gives same checksum, each
write will give different checksum when read
>Fix:
Work around : disable DMA 32bit & multi-block will remove problem,
but speed drops very much.

>Release-Note:
>Audit-Trail:

From: Steve Price <sprice@hiwaay.net>
To: freebsd-gnats-submit@freebsd.org
Cc:  
Subject: Re: kern/12450: IBM -DTTA-351010 corrupts data when using DMA 32-bit
 & multi-block i/o (fwd)
Date: Sun, 29 Aug 1999 09:59:50 -0500 (CDT)

 ---------- Forwarded message ----------
 Date: Tue, 20 Jul 1999 22:45:58 +0200 (CEST)
 From: Peter Hakanson <peter@netman.se>
 To: gnats-admin@FreeBSD.org, freebsd-bugs@FreeBSD.org
 Cc: peter@gbg.netman.se
 Subject: Re: kern/12450: IBM -DTTA-351010 corrupts data when using DMA 32-bit &
     multi-block i/o
 
 Dear "gnats",
 
 my problem is solved, it might be interesting for you to know the
 cause :
 
 The board i use (chaintech 5AGM2) has 2 possible "system frequencys"
 for Cyrix-MII, 66 X 3.5 and 75 X 3
 
 When using 75mhz the ibm disk cannot use DMA, it will
 corrupt data. Quantum does work however.
 
 When switching to 66 X 3.5 both IBM and quantum works ok.
 
 For the record, i use FreeBSD 3.1-RELEASE
 the disk(s) involved are :
 QUANTUM FIREBALL SE3.2A
 QUANTUM FIREBALL CR4.3A
 IBM-DTTA-351010
 
 
 
 
 On Tue, 29 Jun 1999 gnats-admin@FreeBSD.org wrote:
 
 > Thank you very much for your problem report.
 > It has the internal identification `kern/12450'.
 > The individual assigned to look at your
 > report is: freebsd-bugs. 
 > 
 > >Category:       kern
 > >Responsible:    freebsd-bugs
 > >Synopsis:       IBM -DTTA-351010 corrupts data when using DMA 32-bit & multi-block i/o
 > >Arrival-Date:   Tue Jun 29 09:30:00 PDT 1999
 > 
 
 
 
State-Changed-From-To: open->closed 
State-Changed-By: steve 
State-Changed-When: Sun Aug 29 08:18:00 PDT 1999 
State-Changed-Why:  
See followup from originator.  Seems he has resolved this problem. 
>Unformatted:
