kern/39275: HEAD amd64 kernel hang at uhub5 at usb5 vendor 0x1002 EHCI ...
re...@reedmedia.net
>Category: kern
>Synopsis: HEAD amd64 kernel hang at uhub5 at usb5 vendor 0x1002 EHCI ...
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: kern-bug-people
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Sun Aug 03 00:25:00 +0000 2008
>Originator: re...@reedmedia.net
>Release: NetBSD
>Organization:
Jeremy C. Reed
>Environment:
Architecture: x86_64
Machine: amd64
>Description:
I built today's amd64 GENERIC and it hung during boot at:
uhub5 at usb5 vendor 0x1002 EHCI ...
I tried "disable acpi" and that booted, but then I couldn't type. I had a
similar experience with not typing (on different system), so I also tried
"disable ohci" and that hung at:
audio0 at azalia0: full duplex ...
uhub0 at usb0: vendor 0x1002 ...
So I booted again with my previous kernel NetBSD 4.99.55 from March.
I had mentioned it before, but it is very slow to boot before here
(maybe three minutes?):
uhub5: port 6
cursor stays there for a maybe another minute until same line shows:
uhub5: port 6, set config at addr 3 failed
uhub5: device problem disabling port 6
So I guess this is all related (new kernel not booting, old kernel
delayed)
My old kernel dmesg is here
http://reedmedia.net/~reed/tmp-nb73459yt/dmesg.boot
It is a Toshiba Satellite A215-S7422.
I received a reply where anotheruser had same problem since June:
http://mail-index.netbsd.org/port-amd64/2008/08/02/msg000376.html
>How-To-Repeat:
>Fix:
>Unformatted:
--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-...@muc.de
Joerg Sonnenberger
From: Joerg Sonnenberger <jo...@britannica.bec.de>
To: "Jeremy C. Reed" <re...@reedmedia.net>
Cc: gnats...@netbsd.org, dhol...@netbsd.org, bou...@antioche.eu.org
Subject: Re: kern/39275
Date: Wed, 6 Aug 2008 23:13:47 +0200
On Wed, Aug 06, 2008 at 03:40:45PM -0500, Jeremy C. Reed wrote:
> After many tests, I found if I reverted src/sys/arch/x86/x86/intr.c back
> to 1.49 I can boot again.
As I said, can you provide me the MPVERBOSE output and additionally call
ioapic_dump? If it hangs, try to add a breakpoint to wd's attach
routing. Also, acpidump -o output (the binary file) will be helpful.
Joerg
jo...@netbsd.org
Responsible-Changed-From-To: kern-bug-people->joerg
Responsible-Changed-By: jo...@NetBSD.org
Responsible-Changed-When: Wed, 06 Aug 2008 21:15:43 +0000
Responsible-Changed-Why:
Take it.
Joerg Sonnenberger
From: Joerg Sonnenberger <jo...@britannica.bec.de>
To: gnats...@NetBSD.org
Cc:
Subject: Re: kern/39275
Date: Thu, 7 Aug 2008 00:24:17 +0200
On Wed, Aug 06, 2008 at 10:20:03PM +0000, Jeremy C. Reed wrote:
> > As I said, can you provide me the MPVERBOSE output and additionally call
> > ioapic_dump?
>
> First of all -- thank you for taking my PR. I don't recall any suggestion
> for MPVERBOSE and, sorry, I don't know anything about calling ioapic_dump.
> I can google for this, but if you or anyone can teach me the steps, it
> would be appreciated.
It is in one of the threads you quoted :)
> I can't find MPVERBOSE documented on my system. I will add "options
> MPVERBOSE" to my GENERIC of source from the 13th of May. I will reply with
> that info later.
Either that or use boot -d to set mp_verbose to 1. ioapic_dump can be
accessed with "call ioapic_dump" in ddb as well.
Joerg
Jeremy C. Reed
From: "Jeremy C. Reed" <re...@reedmedia.net>
To: gnats...@NetBSD.org
Cc:
Subject: Re: kern/39275
Date: Wed, 6 Aug 2008 17:41:33 -0500 (CDT)
The dmesg with MPVERBOSE is now at
http://reedmedia.net/~reed/tmp-nb73459yt/dmesg.boot (that is before the
May 13 change).
By the way, I also tried to boot yesterday's kernel with "disable ioapic"
and it got further, but then had messages like:
ixpide0:0:0 lost interrupt
type: ata tc_bcount: 512 tc_skip: 0
uhub5: device problem, disabling port 2
ixpide0:0:0 lost interrupt
type: ata tc_bcount: 512 tc_skip: 0
uhub5: device problem, disabling port 6
It didn't seem to boot.
Later I can try the "call ioapic_dump".
David Holland
From: David Holland <dhol...@netbsd.org>
To: "Jeremy C. Reed" <re...@reedmedia.net>
Cc: gnats...@netbsd.org, dhol...@netbsd.org, jo...@britannica.bec.de,
bou...@antioche.eu.org
Subject: Re: kern/39275
Date: Thu, 7 Aug 2008 04:04:12 +0000
On Wed, Aug 06, 2008 at 03:40:45PM -0500, Jeremy C. Reed wrote:
> I cc'd David and Joerg and Manuel because of thread at
> http://mail-index.netbsd.org/source-changes/2008/05/14/msg006381.html
>
> After many tests, I found if I reverted src/sys/arch/x86/x86/intr.c back
> to 1.49 I can boot again.
The issue there turned out to be (at least in part) that nvidia mcp65+
IDE controllers are not really viaide. I don't think that's the
problem you're having, since the PR describes problems with ixpide and
usb. More likely it's yet another ACPI interrupt routing issue, in
which case you should listen to Joerg.
--
David A. Holland
dhol...@netbsd.org
Jeremy C. Reed
From: "Jeremy C. Reed" <re...@reedmedia.net>
To: gnats...@netbsd.org
Cc: dhol...@netbsd.org, jo...@britannica.bec.de, bou...@antioche.eu.org
Subject: Re: kern/39275
Date: Wed, 6 Aug 2008 15:40:45 -0500 (CDT)
I cc'd David and Joerg and Manuel because of thread at
http://mail-index.netbsd.org/source-changes/2008/05/14/msg006381.html
After many tests, I found if I reverted src/sys/arch/x86/x86/intr.c back
to 1.49 I can boot again.
I have not tried Joerg's suggestions in
http://mail-index.netbsd.org/source-changes/2008/05/15/msg006390.html yet
I also haven't looked in the BIOS settings (if that is what David was
talking about).
The booted kernel is at
http://reedmedia.net/~reed/tmp-nb73459yt/dmesg.boot.new
Running ident on the kernel is at
http://www.reedmedia.net/~reed/tmp-nb73459yt/ident
More information:
/netbsd.20080317 booted
/netbsd.20080327 booted
/netbsd.20080330 from Launey, booted
/netbsd.20080406 booted 4.99.58
/netbsd.20080410 booted 4.99.58
/netbsd.20080412 from Steve, did not boot (but was a
custom kernel)
/netbsd.20080413 booted 4.99.59
/netbsd.20080505 booted 4.99.62
/netbsd.20080507 from Launey, hangs at cpu0: ITLB
/netbsd.20080510 booted, but ral0 didn't work:
ral0: could not read MAX register: TIMEOUT
/netbsd.20080512 from Launey, booted 4.99.62 (ral0 working again)
/netbsd.20080513 booted 4.99.62
/netbsd.200805132230b booted! this is same kernel as below after:
cvs up -dP -r 1.49 arch/x86/x86/intr.c
/netbsd.200805132230 did not boot, so something broken between
20080513050000 UTC and 20080513223000 UTC
/netbsd.20080514 did not boot
/netbsd.20080515 from Launey, did not boot 4.99.62
/netbsd.20080520 did not boot 4.99.63
/netbsd.20080525 did not boot
/netbsd.20080605 did not boot
/netbsd.20080804 did not boot (don't have anymore)
/netbsd.20080805 did not boot
That sure took a long time to test all these!
Jeremy C. Reed
From: "Jeremy C. Reed" <re...@reedmedia.net>
To: gnats...@gnats.netbsd.org
Cc:
Subject: Re: kern/39275
Date: Fri, 31 Oct 2008 08:21:21 -0500 (CDT)
Trying again. This time with "Re:" added to subject ...
---------- Forwarded message ----------
Date: Thu, 21 Aug 2008 15:08:24 -0500 (CDT)
From: Jeremy C. Reed <re...@reedmedia.net>
To: gnats...@netbsd.org
Subject: kern/39275
Updated src today and booted new kernel.
Hung as before.
In ddb, I ran "call ioapic_dump"
ioapic0: dump1 61<vector=61,delmode,dest=0> 0<target=0>
ioapic0: dump9 a060<vector=60,delmode,actlo,level,dest=0> 0<target=0>
ioapic0: dump12 62<vector=62,delmode,dest=0> 0<target=0>
ioapic0: dump14 68<vector=68,delmode,dest=0> 0<target=0>
ioapic0: dump15 69<vector=69,delmode,dest=0> 0<target=0>
ioapic0: dump16 a065<vector=65,delmode,actlo,level,dest=0> 0<target=0>
ioapic0: dump17 a063<vector=63,delmode,actlo,level,dest=0> 0<target=0>
ioapic0: dump18 a067<vector=67,delmode,actlo,level,dest=0> 0<target=0>
ioapic0: dump20 a06a<vector=6a,delmode,actlo,level,dest=0> 0<target=0>
ioapic0: dump22 a064<vector=64,delmode,actlo,level,dest=0> 0<target=0>
0xffff800001cebc00
Also booting same kernel with disable ioapic fails with several wd0d
device timeouts, uhub4 device problems, uhub5 disabling ports, and ixoide0
lost interrupts ... and hangs.
Couldn't build new kernel with old sys/arch/x86/x86/intr.c
revision 1.49
So I am back to running old May kernel.
Anything else I can provide?
Eric Schnoebelen
From: er...@cirr.com (Eric Schnoebelen)
To: gnats...@NetBSD.org
Cc:
Subject: Re: kern/39275
Date: Wed, 11 Feb 2009 23:07:27 -0600
I played with a 5.99.7 kernel on my Toshiba Satellite
A215-S7422 the other night.
I found that if I disabled ACPI, the system would boot, but the
keyboard is unresponsive. (the system is otherwise fully
responsive, I was able to log in over the network..)
If ACPI is enabled, the boot hangs after the "last" USB bus
probe.
I've gathered as much info as I could think of, for both the
working kernel (4.99.62 from May 2008), and a -current kernel
(5.99.7 from 02 Feb 2009). I've put it all up at
http://www.cirr.com/~eric/Toshiba-A215-S7422/.
I hope this proves useful to someone. If there is any more
information needed, tell me what it is, and how to aquire it
(remembering that the keyboard isn't working.), and I'll attach
it to the PR.
Thank you,
Eric Schnoebelen
--
Eric Schnoebelen er...@cirr.com http://www.cirr.com
When I grow up I want to be a little boy.
--Joseph Heller, Something Happened, 1974
Jeremy C. Reed
From: "Jeremy C. Reed" <re...@reedmedia.net>
To: gnats...@gnats.NetBSD.org
Cc:
Subject: Re: kern/39275
Date: Thu, 20 Aug 2009 15:06:16 -0500 (CDT)
Problem continues...
Tried both:
netbsd-5/200908190000Z/amd64/binary/kernel/netbsd-GENERIC.gz
and
HEAD/200908190000Z/amd64/binary/kernel/netbsd-GENERIC.gz
Both hang at
uhub5 at usb5: vendor 0x1002 EHCI root hub, class 9/0, rev 2.00/1.00, addr 1
Debugger backtrace shows:
Xintr_legacy() at netbsd:Xintr_legacy1+0x103
--- interrupt ---
x86_stihlt() at netbsd:x86_stihlt+0x86
My old kernel also hangs there for about 15 seconds but finally continues.
I tried also with changing BIOS "Legacy USB Support" from enabled to
disabled.
Others have reported same problem. I am stuck using kernel from May 2008.
Anyone have suggestions?
Jeremy C. Reed
From: "Jeremy C. Reed" <re...@reedmedia.net>
To: gnats...@gnats.netbsd.org
Cc: schn...@netbsd.org
Subject: Re: kern/39275
Date: Tue, 22 Dec 2009 16:18:52 -0600 (CST)
Latest 5.99.22 kernel still hangs.
http://nyftp.netbsd.org/pub/NetBSD-daily/HEAD/200912220000Z/amd64/binary/kernel/netbsd-GENERIC.gz
So I boot with "disable acpi0". Keyboard doesn't work. But at least
networking came up so I can ssh in.
Eric: do you still have problems with this same laptop (Toshiba
Satellite A215-S7422)?
Eric Schnoebelen
From: er...@cirr.com (Eric Schnoebelen)
To: "Jeremy C. Reed" <re...@reedmedia.net>
Cc: gnats...@gnats.netbsd.org, schn...@netbsd.org
Subject: Re: kern/39275
Date: Tue, 22 Dec 2009 16:42:12 -0600
"Jeremy C. Reed" writes:
- Latest 5.99.22 kernel still hangs.
- http://nyftp.netbsd.org/pub/NetBSD-daily/HEAD/200912220000Z/amd64/binary/ke
- rnel/netbsd-GENERIC.gz
-
- So I boot with "disable acpi0". Keyboard doesn't work. But at least
- networking came up so I can ssh in.
-
- Eric: do you still have problems with this same laptop (Toshiba
- Satellite A215-S7422)?
Yes, I still have problems running a native kernel.. But I've
been running a Xen kernel on it with some success. A bit of
keyboard bounce, and some extra sensitivity on the track pad,
but it works..
But every Xen kernel complains about the system being over temp
at boot.
--
Eric Schnoebelen er...@cirr.com http://www.cirr.com
Users - the biological test suites.
Jeremy C. Reed
From: "Jeremy C. Reed" <re...@reedmedia.net>
To: gnats...@gnats.netbsd.org
Cc:
Subject: Re: kern/39275
Date: Mon, 23 Aug 2010 16:56:41 -0500 (CDT)
"disable acpi" made it so my machdep.powernow.frequency.target tunable
didn't exist (so I couldn't lower it and my laptop would overheat).
Now I am using "disable pckbc" to use the system with NetBSD 4.99.62
kernel.
I got that hint from http://gnats.netbsd.org/39092
Later I will try "pckbc at isa" and see if my keyboard works.
sk...@netbsd.org
State-Changed-From-To: open->feedback
State-Changed-By: sk...@NetBSD.org
State-Changed-When: Tue, 21 Jul 2015 23:50:09 +0000
State-Changed-Why:
Can you try recent -current, please?