NetBSD Problem Report #24094
Received: (qmail 19058 invoked by uid 605); 14 Jan 2004 18:46:09 -0000
Message-Id: <20040114184606.C7F9D11155@narn.netbsd.org>
Date: Wed, 14 Jan 2004 18:46:06 +0000 (UTC)
From: swilde@sha-bang.de
Sender: gnats-bugs-owner@NetBSD.org
Reply-To: swilde@sha-bang.de
To: gnats-bugs@gnats.NetBSD.org
Subject: pciide broken on Compaq Armada 1110
X-Send-Pr-Version: www-1.0
>Number: 24094
>Category: kern
>Synopsis: pciide broken on Compaq Armada 1110
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: kern-bug-people
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Jan 14 18:47:00 +0000 2004
>Closed-Date:
>Last-Modified: Sun Jan 03 14:25:01 +0000 2010
>Originator: Sascha Wilde
>Release: NetBSD 1.6.2_RC3
>Organization:
>Environment:
NetBSD schleppy 1.6.2_RC3 NetBSD 1.6.2_RC3 (SCHLEPPY) #0: Wed Jan 7 21:10:35 CE T 2004 wilde@kenny:/usr/src/sys/arch/i386/compile/SCHLEPPY i386
>Description:
Starting with NetBSD 1.5.x the pciide driver stoped working on my Compaq Armada 1110, the last working Version i tried was NetBSD 1.4.2, dmesg snippet:
| pci0 at mainbus0 bus 0: configuration mode 1
| pci0: i/o enabled, memory enabled
| pchb0 at pci0 dev 0 function 0
| pchb0: Opti 82C557 (rev. 0x14)
| pcib0 at pci0 dev 1 function 0
| pcib0: Opti 82C558 (rev. 0x02)
| vga1 at pci0 dev 2 function 0: Cirrus Logic CL-GD7543 (rev. 0x00)
| wsdisplay0 at vga1: console (80x25, vt100 emulation)
| pciide0 at pci0 dev 20 function 0: Opti 82C621 (rev. 0x12)
| pciide0: bus-master DMA support present, but unused (no driver
| support)
| pciide0: primary channel wired to compatibility mode
| pciide0: primary channel interrupting at irq 14
| wd0 at pciide0 channel 0 drive 0: <ST9810AG>
| wd0: drive supports 16-sector pio transfers, lba addressing
| wd0: 773MB, 1572 cyl, 16 head, 63 sec, 512 bytes/sect x 1584968
| sectors
| wd0: drive supports PIO mode 4, DMA mode 2
In all later Versions (including NetBSD 1.6.2 RC3) the pciide looses
its interrupt, detects an error, trys to "repair it" and repeats so
for nearly each and every block it trys to read. (This is a dump from
Memory). This misbehavior damages the file-system unrecoverable
(thats why I have no logs, sorry) -- I had to reinstall the system
from scratch.
>How-To-Repeat:
Find an old Armada 1110... ;-)
I don't know if the problem occures on any other Hardware.
>Fix:
No fix, but workaround: a Kernel with only isa support (like NetBSD 1.6.2 RC3 GENERIC_TINY) works without problems.
>Release-Note:
>Audit-Trail:
From: Steve Woodford <scw@netbsd.org>
To: swilde@sha-bang.de, gnats-bugs@gnats.NetBSD.org
Cc:
Subject: Re: kern/24094: pciide broken on Compaq Armada 1110
Date: Thu, 15 Jan 2004 09:42:02 +0000
On Wednesday 14 January 2004 6:46 pm, swilde@sha-bang.de wrote:
> Starting with NetBSD 1.5.x the pciide driver stoped working on my
> Compaq Armada 1110, the last working Version i tried was NetBSD 1.4.2.
That pretty much coincides with when I added native bus-master DMA
support for the Opti chipset found in some Armada notebooks. The Opti
chipset was one of the first batch of PCI bus-master IDE chipsets to
exist, and there are several variants with subtle differences.
I got feedback from a few people, and for the most part (after a few
iterations), the driver worked for everyone at the time.
I still have my Armada 1520, but it's no longer running NetBSD, so it's
possible that Opti support has bit-rotted since then. I'll attempt to
boot NetBSD-current from floppy sometime soon to how it copes.
In the meantime, if you can (cross-) compile a kernel, you can comment
out the optiide driver in the config file to make the kernel fallback
to non-DMA mode. If this works for you, I think we should consider
disabling optiide by default in the i386 kernels...
Cheers, Steve
From: Sascha Wilde <wilde@sha-bang.de>
To: gnats-bugs@gnats.NetBSD.org
Cc: Steve Woodford <scw@netbsd.org>
Subject: Re: kern/24094: pciide broken on Compaq Armada 1110
Date: Thu, 15 Jan 2004 20:06:35 +0100
Steve Woodford wrote on Thu, 15 Jan 2004 09:42:02 +0000
> On Wednesday 14 January 2004 6:46 pm, swilde@sha-bang.de wrote:
> > Starting with NetBSD 1.5.x the pciide driver stoped working on my
> > Compaq Armada 1110, the last working Version i tried was NetBSD
> > 1.4.2.
> That pretty much coincides with when I added native bus-master DMA
> support for the Opti chipset found in some Armada notebooks.
Thats what found in the Changelogs, so I tried to disable DMA by
setting the pciide flag to 0x0002:
pciide* at pci? dev ? function ? flags 0x0002
anyway, that did not help.
> In the meantime, if you can (cross-) compile a kernel, you can
> comment out the optiide driver in the config file to make the kernel
> fallback to non-DMA mode. If this works for you, I think we should
> consider disabling optiide by default in the i386 kernels...
normaly I use NetBSD 1.6.x which seems not to support this finegrained
configuration there is only pciide in the GENERIC file, so that's what
I used.
If it's really necessary I'll give current a try, using your suggested
configuration. I reinstalled NetBSD at least seven times on the Armada
because of that problem and I would be happy to keep the system
working for now.
cheers
--
Sascha Wilde : xedit? Das sieht zwar wie vi aus als könne es nix, aber im
: Gegensatz zu vi kann es wirklich nix und nix ist noch geschönt!
: (Michael Core in dafc)
State-Changed-From-To: open->feedback
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Thu, 31 Dec 2009 19:00:40 +0000
State-Changed-Why:
Has this improved since 1.6?
From: Sascha Wilde <wilde@sha-bang.de>
To: gnats-bugs@NetBSD.org
Cc: kern-bug-people@netbsd.org, netbsd-bugs@netbsd.org, gnats-admin@netbsd.org, dholland@NetBSD.org, swilde@sha-bang.de
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Fri, 01 Jan 2010 15:51:13 +0100
dholland@NetBSD.org wrote:
> Synopsis: pciide broken on Compaq Armada 1110
>
> State-Changed-From-To: open->feedback
> State-Changed-By: dholland@NetBSD.org
> State-Changed-When: Thu, 31 Dec 2009 19:00:40 +0000
> State-Changed-Why:
> Has this improved since 1.6?
Good question... :-)
I'll try to investigate this question, but this isn't that easy to do,
as the Laptop which shows the problem has only 4 MiB[1] of RAM so trying
to boot a recent standard GENERIC kernel (which is about 10 MiB) is no
option.
So I'll have to try to build an _small_ kernel, including PCI support to
test if the problem still shows up.
I'm willing to do so[2] -- but it will take some time. Does it make
any sense trying to get 5.0.1 running on this old system or should I try
some older version? (which?)
cheers
sascha
[1] Yes, this about one thousandth of what people carry around in there
notebooks nowadays... :-/
[2] though I'm a bit hesitating to risk the destruction of my currently
installed system
--
Sascha Wilde
Wer HTML postet oder gepostetes HTML quotet oder sich gepostetes oder
gequotetes HTML beschafft, um es in Verkehr zu bringen, wird geplonkt.
From: Sascha Wilde <wilde@sha-bang.de>
To: gnats-bugs@NetBSD.org
Cc: kern-bug-people@netbsd.org, netbsd-bugs@netbsd.org, gnats-admin@netbsd.org, dholland@NetBSD.org, swilde@sha-bang.de
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sat, 02 Jan 2010 21:30:07 +0100
dholland@NetBSD.org wrote:
> Synopsis: pciide broken on Compaq Armada 1110
>
> State-Changed-From-To: open->feedback
> State-Changed-By: dholland@NetBSD.org
> State-Changed-When: Thu, 31 Dec 2009 19:00:40 +0000
> State-Changed-Why:
> Has this improved since 1.6?
Ok, test done: I did build a boiled down NetBSD 5 kernel with PCI
support (configuration attached for reference) and booted my Armada from
it -- with the old catastrophic results: the symptoms are exactly the
same as described in my original six year old report... :-(
So, no, it has not improved since 1.6!
Cheers
sascha, facing the tedious task of reinstalling the prehistoric
monster -- wondering if any more recent version of netbsd than 1.6 might
be appropriate...
--
Sascha Wilde
"Liebet eure Feinde, vielleicht schadet das ihrem Ruf"
(Stanislaw Jerzy Lec)
From: Martin Husemann <martin@duskware.de>
To: Sascha Wilde <wilde@sha-bang.de>
Cc: gnats-bugs@NetBSD.org, dholland@NetBSD.org, swilde@sha-bang.de
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sat, 2 Jan 2010 21:32:30 +0100
On Sat, Jan 02, 2010 at 09:30:07PM +0100, Sascha Wilde wrote:
> Ok, test done: I did build a boiled down NetBSD 5 kernel with PCI
> support (configuration attached for reference) and booted my Armada from
> it -- with the old catastrophic results: the symptoms are exactly the
> same as described in my original six year old report... :-(
The configuration was missing ;-)
Could you please make sure it has
optiide* at pci? dev ? function ?
in it?
Martin
From: David Holland <dholland-bugs@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc: swilde@sha-bang.de
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sat, 2 Jan 2010 20:53:21 +0000
On Sat, Jan 02, 2010 at 08:30:03PM +0000, Sascha Wilde wrote:
> Ok, test done: I did build a boiled down NetBSD 5 kernel with PCI
> support (configuration attached for reference) and booted my Armada from
> it -- with the old catastrophic results: the symptoms are exactly the
> same as described in my original six year old report... :-(
Oog. Sorry to hear that...
> sascha, facing the tedious task of reinstalling the prehistoric
> monster -- wondering if any more recent version of netbsd than 1.6 might
> be appropriate...
There's a distinct possibility that you might be able to just rewrite
the partition table/disklabel, fsck using an alternate superblock, and
then patch up whatever else got trashed, depending on just how far it
went.
I wouldn't think it would have got much further than attempting to
read the partition table and disklabel, and it's unreasonable for it
to have trashed anything it didn't try reading. Then again, it's
unreasonable for this to happen at all, so who knows... :-/
--
David A. Holland
dholland@netbsd.org
State-Changed-From-To: feedback->open
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Sat, 02 Jan 2010 21:13:22 +0000
State-Changed-Why:
Still totally broken. sigh.
From: David Holland <dholland-bugs@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sat, 2 Jan 2010 21:23:32 +0000
On Fri, Jan 01, 2010 at 04:10:05PM +0000, Sascha Wilde wrote:
> I'm willing to do so[2] -- but it will take some time. Does it make
> any sense trying to get 5.0.1 running on this old system or should I try
> some older version? (which?)
Unless you want to try mouse@'s branch of 1.4T, 5.0 (or 5.1, which is
due out Real Soon Now) is probably as good as any. It ought to manage
to fit if you build a small kernel without all the bloaty drivers, and
it is likely faster than 4.0. 3.1 is probably the next fallback choice.
--
David A. Holland
dholland@netbsd.org
From: Sascha Wilde <wilde@sha-bang.de>
To: Martin Husemann <martin@duskware.de>
Cc: gnats-bugs@NetBSD.org, dholland@NetBSD.org, swilde@sha-bang.de
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sat, 02 Jan 2010 22:30:55 +0100
--=-=-=
Martin Husemann <martin@duskware.de> wrote:
> On Sat, Jan 02, 2010 at 09:30:07PM +0100, Sascha Wilde wrote:
>> Ok, test done: I did build a boiled down NetBSD 5 kernel with PCI
>> support (configuration attached for reference) and booted my Armada from
>> it -- with the old catastrophic results: the symptoms are exactly the
>> same as described in my original six year old report... :-(
>
> The configuration was missing ;-)
Ooops, sorry -- attached, this time for real...
> Could you please make sure it has
>
> optiide* at pci? dev ? function ?
yes, it has.
sascha
--
Sascha Wilde
"If you were young again, would you start writing TeX again or would
you use Microsoft Word, or another word processor?" - "I hope to die
before I *have* to use Microsoft Word." -- Prof. Donald E. Knuth
--=-=-=
Content-Disposition: inline; filename=SCHLEPPY
include "arch/i386/conf/std.i386"
maxusers 8 # estimated number of users
makeoptions CPUFLAGS="-march=i486 -mtune=i586" # Assume at least 486,
# tune for more modern
makeoptions COPTS="-Os"
options MTRR # memory-type range register syscall support
options INSECURE # disable kernel security levels - X needs this
options RTC_OFFSET=0 # hardware clock is this many mins. west of GMT
options SYSVMSG # System V-like message queues
options SYSVSEM # System V-like semaphores
options SYSVSHM # System V-like memory sharing
options P1003_1B_SEMAPHORE # p1003.1b semaphore support
options COMPAT_16 # NetBSD 1.6
options COMPAT_20 # NetBSD 2.0
options COMPAT_30 # NetBSD 3.0
options COMPAT_30_PTHREAD # NetBSD 3.0 pthreads compatibility.
options COMPAT_40 # NetBSD 4.0
options COMPAT_43 # 4.3BSD, 386BSD, and BSDI
options COMPAT_BSDPTY # /dev/[pt]ty?? ptys.
file-system FFS # UFS
file-system NFS # Network File System client
file-system MSDOSFS # MS-DOS file system
file-system KERNFS # /kern
# immutable) behave as system flags.
options INET # IP + ICMP + TCP + UDP
options INET6 # IPV6
options WSEMUL_VT100 # VT100 / VT220 emulation
options WSDISPLAY_CUSTOM_OUTPUT # color customization from wsconsctl(8)
options WS_KERNEL_FG=WSCOL_GREEN
options WSDISPLAY_CUSTOM_BORDER # custom border colors via wsconsctl(8)
options WSDISPLAY_COMPAT_PCVT # emulate some ioctls
options WSDISPLAY_COMPAT_SYSCONS # emulate some ioctls
options WSDISPLAY_COMPAT_USL # VT handling
options WSDISPLAY_COMPAT_RAWKBD # can get raw scancodes
options PCDISPLAY_SOFTCURSOR
options WSDISPLAY_SCROLLSUPPORT
config netbsd root on ? type ffs
apm0 at mainbus0 # Advanced power management
options APM_V10_ONLY # Use only the APM 1.0 calls
options APM_POWER_PRINT # Print stats on the console
pci* at mainbus? bus ?
pci* at pchb? bus ?
# L2/L3 work, but should not be enabled
# on anything else.
# work, but should not be enabled
# on anything else.
# watchdog, SpeedStep and HPET
pchb* at pci? dev ? function ? # PCI-Host bridges
pcib* at pci? dev ? function ? # PCI-ISA bridges
isa0 at mainbus?
isa0 at pcib?
pcmcia* at pcic? controller ? socket ?
pcic0 at isa? port 0x3e0 iomem 0xd0000 iosiz 0x10000
pcic0 at pci? dev? function ?
cbb* at pci? dev ? function ?
cardslot* at cbb?
cardbus* at cardslot?
pcmcia* at cardslot?
npx0 at isa? port 0xf0 irq 13 # x86 math coprocessor
pckbc0 at isa? # pc keyboard controller
pckbd* at pckbc? # PC keyboard
pms* at pckbc? # PS/2 mouse for wsmouse
vga0 at isa?
vga* at pci? dev ? function ?
wsdisplay* at vga? console ?
wskbd* at pckbd? console ?
wsmouse* at pms? mux 0
com* at pcmcia? function ? # Modems and serial cards
com0 at isa? port 0x3f8 irq 4 # Standard PC serial ports
com1 at isa? port 0x2f8 irq 3
com2 at isa? port 0x3e8 irq 5
lpt0 at isa? port 0x378 irq 7 # standard PC parallel ports
pciide* at pci? dev ? function ? flags 0x0000 # GENERIC pciide driver
optiide* at pci? dev ? function ? # Opti IDE controllers
wdc0 at isa? port 0x1f0 irq 14 flags 0x00
wdc1 at isa? port 0x170 irq 15 flags 0x00
atabus* at ata?
options ATADEBUG
wd* at atabus? drive ? flags 0x0000
pseudo-device ataraid
fdc0 at isa? port 0x3f0 irq 6 drq 2 # standard PC floppy controllers
fd* at fdc? drive ? # the drives themselves
ep* at pcmcia? function ? # 3Com 3c589 and 3c562 Ethernet
ne* at pcmcia? function ? # NE2000-compatible Ethernet
wi* at pcmcia? function ? # Lucent/Intersil WaveLan IEEE (802.11)
pseudo-device crypto # /dev/crypto device
pseudo-device swcrypto # software crypto implementation
pseudo-device fss 4 # file system snapshot device
pseudo-device vnd # disk-like interface to files
options VND_COMPRESSION # compressed vnd(4)
pseudo-device bpfilter # Berkeley packet filter
pseudo-device loop # network loopback
pseudo-device tap # virtual Ethernet
pseudo-device tun # network tunneling over tty
pseudo-device pty # pseudo-terminals
pseudo-device rnd # /dev/random and in-kernel generator
pseudo-device clockctl # user control of clock subsystem
pseudo-device wsmux # mouse & keyboard multiplexor
pseudo-device wsfont
pseudo-device cmos
options VERIFIED_EXEC_FP_RMD160
options VERIFIED_EXEC_FP_SHA256
options VERIFIED_EXEC_FP_SHA384
options VERIFIED_EXEC_FP_SHA512
options VERIFIED_EXEC_FP_SHA1
options VERIFIED_EXEC_FP_MD5
options PAX_MPROTECT=0 # PaX mprotect(2) restrictions
options PAX_ASLR=0 # PaX Address Space Layout Randomization
--=-=-=--
From: Martin Husemann <martin@duskware.de>
To: Sascha Wilde <wilde@sha-bang.de>
Cc: gnats-bugs@NetBSD.org, dholland@NetBSD.org, swilde@sha-bang.de
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sat, 2 Jan 2010 22:41:36 +0100
On Sat, Jan 02, 2010 at 10:30:55PM +0100, Sascha Wilde wrote:
> > optiide* at pci? dev ? function ?
>
> yes, it has.
but it still attaches pciide?
Martin
From: Sascha Wilde <wilde@sha-bang.de>
To: David Holland <dholland-bugs@netbsd.org>
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sat, 02 Jan 2010 22:53:54 +0100
David Holland <dholland-bugs@netbsd.org> wrote:
> On Sat, Jan 02, 2010 at 08:30:03PM +0000, Sascha Wilde wrote:
> > sascha, facing the tedious task of reinstalling the prehistoric
> > monster -- wondering if any more recent version of netbsd than 1.6 might
> > be appropriate...
>
> There's a distinct possibility that you might be able to just rewrite
> the partition table/disklabel, fsck using an alternate superblock, and
> then patch up whatever else got trashed, depending on just how far it
> went.
Hmm, I did try simply to fsck (booting into single user with the "good"
kernel was still possible this time after 3 or 4 "lost interrupt"
iterations with the pci enabled kernel), but the kernel reports
something like:
wd0a: error reading fsbn 816 of 816-831 (wd0 bn 879; cn 0 tn 13 sn 60), retrying
wd0: (uncorrectable data error)
[repeated n times]
for many blocks, making fsck fail in many ways and resulting in
countless corrupted or removed files -- the fsck is still running while
I'm writing this message, but I doubt very much that it will result in
any kind of usable system... ;-/
> I wouldn't think it would have got much further than attempting to
> read the partition table and disklabel,
Funny (?!) enough, it did.
> and it's unreasonable for it to have trashed anything it didn't try
> reading. Then again, it's unreasonable for this to happen at all, so
> who knows... :-/
Its a shame that I cant post logs of the failed boot attempt -- maybe I
should have used a digicam to make a "screen shot" (the good ideas
always come to late) but its to late now...
cheers
sascha
--
Sascha Wilde
"C++ : an octopus made by nailing extra legs onto a dog"
From: Sascha Wilde <wilde@sha-bang.de>
To: Martin Husemann <martin@duskware.de>
Cc: gnats-bugs@NetBSD.org, dholland@NetBSD.org, swilde@sha-bang.de
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sun, 03 Jan 2010 15:10:13 +0100
Martin Husemann <martin@duskware.de> wrote:
> On Sat, Jan 02, 2010 at 10:30:55PM +0100, Sascha Wilde wrote:
>> > optiide* at pci? dev ? function ?
>>
>> yes, it has.
>
> but it still attaches pciide?
Uhm, good question. I have to admit, that I did not read the kernel
messages observant enough -- maybe I was a little nervous knowing what
might (and did) happen... :-(
Would you consider this information worth the destruction of my now
running 3.1.1[0] installation? In this case I might be willing to offer my
system a last time to the gods of debugging with a camera ready and
aimed at the display (hoping the results will be readable).
On the other hand I'm not sure if its really worth to go through all
this trouble just to get pci support on this prehistoric piece of
junk...
cheers
sascha
[0] The last version with a boot-tiny installation floppy.
And in addition I wasn't able to get a 5.x kernel below 2MiB -- and
even tough I was wrong stating having only 4MiB RAM, its really 8MiB,
wasting more than one fourth of memory on the kernel seems no good
to me...
--
>++++++[<+++++++++++>-]<+.>+++[<++++++>-]<.---.---------.++++++.++++.---------
-.+++++++++++.+++++.>+++++++[<-------->-]<-.>++++++[<+++++++>-]<+.--.+++..----
---.-.>++++++[<------>-]<.>++++[<+++++++++++++>-]<.------------.---.>++++++[<-
----->-]<-.>+++++[<+++++++>-]<.--.>+++[<++++++>-]<+.>++++++++[<--------->-]<--.
From: Martin Husemann <martin@duskware.de>
To: Sascha Wilde <wilde@sha-bang.de>
Cc: gnats-bugs@NetBSD.org, dholland@NetBSD.org, swilde@sha-bang.de
Subject: Re: kern/24094 (pciide broken on Compaq Armada 1110)
Date: Sun, 3 Jan 2010 15:22:05 +0100
On Sun, Jan 03, 2010 at 03:10:13PM +0100, Sascha Wilde wrote:
> Would you consider this information worth the destruction of my now
> running 3.1.1[0] installation?
No, certainly not. Something phishy is going on and we are all tapping
in the dark.
Thanks for the destructive testing you dared to do!
Martin
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2007
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.