Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 18 Jun 2004 20:45:44 +0200
From:      Radek Kozlowski <radek@raadradd.com>
To:        current@freebsd.org
Subject:   Transfer mode of my ad0 no longer recognized correctly
Message-ID:  <40D33858.50300@raadradd.com>

next in thread | raw e-mail | index | archive | help
Hi,

when using a freshly built -CURRENT kernel (acpi enabled) the transfer 
mode of my hard disk is set to PIO4 during system init (ad0: 38154MB 
<IC25N040ATMR04-0> [77520/16/63] at ata0-master PIO4), whereas with a 
kernel from 28th of May (acpi enabled) it is recognized corretly and set 
to UDMA100 (ad0: 38154MB <IC25N040ATMR04-0> [77520/16/63] at ata0-master 
UDMA100). My system's performance is very poor when the disk works in 
PIO mode, so I'm using my older kernel for now.

My ata controller:

atapci0@pci0:16:0:      class=0x0101b0 card=0x0024103c chip=0x522910b9 
rev=0xc4 hdr=0x00
     vendor   = 'Acer Labs Incorporated (ALi)'
     device   = 'M1543 Southbridge EIDE Controller'
     class    = mass storage
     subclass = ATA

I haven't changed anything in my kernel config file since this last 
working kernel. Also, when I boot my recent kernel with acpi disabled, 
the transfer rate is being correctly set to UDMA100.

When I boot with acpi enabled and try to change the transfer mode manually:

# atacontrol mode 0 udma100 xxx
Master = UDMA100
Slave  = BIOSPIO

the transfer mode is changed, but I get a kernel panic (fatal trap 12) 
immediately after that. I wanted to capture a core dump to later on use 
it with gdb, but I'm unable to produce one. I did everything according 
to developer's handbook but it's not working:

# grep dump /etc/rc.conf
dumpdev="/dev/ad0s1b"
dumpdir="/usr/crash"

# swapctl -l
Device:       1024-blocks     Used:
/dev/ad0s1b     1048576         0

# sysctl kern | grep dump
kern.sugid_coredump: 0
kern.coredump: 1

One interesting thing is that there's no such oid as kern.dumpdev 
(sysctl: unknown oid 'kern.dumpdev') and according to man sysctl there 
should be one.

I'll probably take a photo of the output of the trace command in ddb 
when I get home if I can't get a crash dump.

Should I provide more info about my system?

Thanks for any help.

-Radek



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?40D33858.50300>