NetBSD Problem Report #53795
From www@NetBSD.org Sun Dec 16 23:49:56 2018
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
(Client CN "mail.NetBSD.org", Issuer "mail.NetBSD.org CA" (not verified))
by mollari.NetBSD.org (Postfix) with ESMTPS id 8FE347A169
for <gnats-bugs@gnats.NetBSD.org>; Sun, 16 Dec 2018 23:49:56 +0000 (UTC)
Message-Id: <20181216234955.5B0617A219@mollari.NetBSD.org>
Date: Sun, 16 Dec 2018 23:49:55 +0000 (UTC)
From: elliott1787@gmail.com
Reply-To: elliott1787@gmail.com
To: gnats-bugs@NetBSD.org
Subject: i915drmkms causes crash on boot with Atom N270/945GSE netbook
X-Send-Pr-Version: www-1.0
>Number: 53795
>Category: kern
>Synopsis: i915drmkms causes crash on boot with Atom N270/945GSE netbook
>Confidential: no
>Severity: critical
>Priority: high
>Responsible: riastradh
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Sun Dec 16 23:50:00 +0000 2018
>Closed-Date: Thu Apr 04 17:10:49 +0000 2024
>Last-Modified: Thu Apr 04 17:10:49 +0000 2024
>Originator: Elliott Shue
>Release: NetBSD/i386-8.0
>Organization:
>Environment:
>Description:
I have an Atom N270 powered netbook with an Intel 945GSE chipset. It is a re-branded ECS E09EI6. It works fine in OpenBSD and Linux.
If I boot the GENERIC NetBSD kernel with i915drmkms enabled I get a hang with garbage on the screen right when intelfb should be initialized.
If I disable i915drmkms in userconf I can get the system to boot and complete the install. I've built a kernel without i915drmkms for now to play around. It runs, but X falls back to vesa.
The last lines before the screen goes blank and displays garbage are:
i915drmkms0: interrupting at ioapic0 pin 16 (i915)
drm: initialized overlay support
drm: GMBUS [i915 gmbus vga] timed out, falling back to bit-banging on pin 2
I had to take a video of the screen since a serial console is not possible. Here is a snapshot of the last frame before it goes black:
https://i.imgur.com/xnvurv3.jpg
I built a kernel with the legacy i915drm, it boots, but I don't get a proper framebuffer and also am stuck with vesa for X.
The problem exists in 7.x as well.
>How-To-Repeat:
Boot this netbook with i915drmkms enabled.
>Fix:
>Release-Note:
>Audit-Trail:
From: "David H. Gutteridge" <david@gutteridge.ca>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/53795: i915drmkms causes crash on boot with Atom
N270/945GSE netbook
Date: Mon, 17 Dec 2018 22:25:16 -0500
On Sun, 16 Dec 2018, at 23:50:00 +0000 (UTC), Elliott Shue wrote:
>I have an Atom N270 powered netbook with an Intel 945GSE chipset. It
>is a re-branded ECS E09EI6. It works fine in OpenBSD and Linux.
>
>If I boot the GENERIC NetBSD kernel with i915drmkms enabled I get a
>hang with garbage on the screen right when intelfb should be
>initialized.
I have an LG X110 that has the same CPU/GPU combo in it. I haven't
encountered this particular problem with it, curiously.
The first recommendation would be to try a current ("HEAD") kernel, as
the i915 DRM driver is significantly newer in it than what you'd be
getting with 8.0 (whether the formal release or 8.0_STABLE). You can
grab a kernel (and modules.tgz) from the releng builds[1], if you
don't want to build one yourself. Hopefully you'll get better results
with it.
Dave
1 http://nycdn.netbsd.org/pub/NetBSD-daily/HEAD/latest/i386/binary/sets
From: Elliott Shue <elliott1787@gmail.com>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/53795: i915drmkms causes crash on boot with Atom N270/945GSE netbook
Date: Tue, 18 Dec 2018 11:52:24 -0500
--0000000000007b7835057d4eb901
Content-Type: text/plain; charset="UTF-8"
>I have an LG X110 that has the same CPU/GPU combo in it. I haven't
>encountered this particular problem with it, curiously.
I have an Asus EeePC 901 with the same CPU/GPU, it boots fine with the
stock kernel as well.
It is a slightly weird netbook, but shouldn't be too far from the
basic Atom N270 system from the time.
> Hopefully you'll get better results with it.
Definitely a different result. Still garbage on the screen, but the
system actually boots.
Before the HD activity light would stop and any key press would cause a beep.
Now I've got ssh access and can maybe provide more useful information.
[ 6.8815928] kern.module.path=/stand/i386/8.99.27/modules
[ 6.8815928] kern info: [drm] Memory usable by graphics device = 256M
[ 6.8815928] kern info: [drm] Supports vblank timestamp caching Rev
2 (21.10.2013).
[ 6.8815928] kern info: [drm] Driver supports precise vblank timestamp query.
[ 6.8815928] kern info: [drm] failed to find VBIOS tables
[ 6.8815928] i915drmkms0: interrupting at ioapic0 pin 16 (i915drmkms0)
[ 6.9116124] kern info: [drm] initialized overlay support
[ 6.9816621] kern info: [drm] GMBUS [i915 gmbus vga] timed out,
falling back to bit banging on pin 2
[ 7.2418458] intelfb0 at i915drmkms0
[ 7.2418458] intelfb0: framebuffer at 0xdc167000, size 640x480,
depth 32, stride 2560
[ 7.9723633] kern error:
[drm:(/usr/src/sys/external/bsd/drm2/dist/drm/i915/intel_display.c:12659)intel_pipe_config_compare]
*ERROR* mismatch in base.adjusted_mode.flags(DRM_MODE_FLAG_PHSYNC)
(expected 0, found 1)
[ 7.9723633] kern error:
[drm:(/usr/src/sys/external/bsd/drm2/dist/drm/i915/intel_display.c:12663)intel_pipe_config_compare]
*ERROR* mismatch in base.adjusted_mode.flags(DRM_MODE_FLAG_PVSYNC)
(expected 0, found 4)
[ 7.9723633] warning:
/usr/src/sys/external/bsd/drm2/dist/drm/i915/intel_display.c:12899:
pipe state doesn't match!
[ 8.0123934] wsdisplay0 at intelfb0 kbdmux 1: console (default,
vt100 emulation), using wskbd0
[ 8.0338052] wsmux1: connecting to wsdisplay0
Complete dmesg here:
https://pastebin.com/K5k1p6JX
Any other information I could provide?
--0000000000007b7835057d4eb901
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
<div dir=3D"ltr"><div dir=3D"ltr"><div dir=3D"ltr"><div dir=3D"ltr"><pre> &=
gt;I have an LG X110 that has the same CPU/GPU combo in it. I haven't
>encountered this particular problem with it, curiously.<br><br></pre><=
pre>I have an Asus EeePC 901 with the same CPU/GPU, it boots fine with the =
stock kernel as well. <br></pre><pre>It is a slightly weird netbook, but sh=
ouldn't be too far from the basic Atom N270 system from the time.<br></=
pre><pre><br>> Hopefully you'll get better results with it.<br><br><=
/pre><pre>Definitely a different result. Still garbage on the screen, but t=
he system actually boots. <br></pre><pre>Before the HD activity light would=
stop and any key press would cause a beep. <br></pre><pre>Now I've got=
ssh access and can maybe provide more useful information.<br><br>[ 6.881=
5928] kern.module.path=3D/stand/i386/8.99.27/modules<br>[ 6.8815928] kern=
info: [drm] Memory usable by graphics device =3D 256M<br>[ 6.8815928] ke=
rn info: [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).<br>[ =
6.8815928] kern info: [drm] Driver supports precise vblank timestamp query=
.<br>[ 6.8815928] kern info: [drm] failed to find VBIOS tables<br>[ 6.8=
815928] i915drmkms0: interrupting at ioapic0 pin 16 (i915drmkms0)<br>[ 6.=
9116124] kern info: [drm] initialized overlay support<br>[ 6.9816621] ker=
n info: [drm] GMBUS [i915 gmbus vga] timed out, falling back to bit banging=
on pin 2<br>[ 7.2418458] intelfb0 at i915drmkms0<br>[ 7.2418458] intel=
fb0: framebuffer at 0xdc167000, size 640x480, depth 32, stride 2560<br>[ =
7.9723633] kern error: [drm:(/usr/src/sys/external/bsd/drm2/dist/drm/i915/i=
ntel_display.c:12659)intel_pipe_config_compare] *ERROR* mismatch in base.ad=
justed_mode.flags(DRM_MODE_FLAG_PHSYNC) (expected 0, found 1)<br>[ 7.9723=
633] kern error: [drm:(/usr/src/sys/external/bsd/drm2/dist/drm/i915/intel_d=
isplay.c:12663)intel_pipe_config_compare] *ERROR* mismatch in base.adjusted=
_mode.flags(DRM_MODE_FLAG_PVSYNC) (expected 0, found 4)<br>[ 7.9723633] w=
arning: /usr/src/sys/external/bsd/drm2/dist/drm/i915/intel_display.c:12899:=
pipe state doesn't match!<br>[ 8.0123934] wsdisplay0 at intelfb0 kbd=
mux 1: console (default, vt100 emulation), using wskbd0<br>[ 8.0338052] w=
smux1: connecting to wsdisplay0<br><br>Complete dmesg here:<br><a href=3D"h=
ttps://pastebin.com/K5k1p6JX">https://pastebin.com/K5k1p6JX</a><br><br></pr=
e><pre>Any other information I could provide?<br></pre><pre><br><br></pre><=
pre><br></pre></div></div></div></div>
--0000000000007b7835057d4eb901--
From: Elliott Shue <elliott1787@gmail.com>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/53795: i915drmkms causes crash on boot with Atom N270/945GSE netbook
Date: Tue, 18 Dec 2018 12:01:00 -0500
Forgot to switch to plain-text, sorry.
On Tue, Dec 18, 2018 at 11:52 AM Elliott Shue <elliott1787@gmail.com> wrote:
>
> >I have an LG X110 that has the same CPU/GPU combo in it. I haven't
> >encountered this particular problem with it, curiously.
>
> I have an Asus EeePC 901 with the same CPU/GPU, it boots fine with the stock kernel as well.
>
> It is a slightly weird netbook, but shouldn't be too far from the basic Atom N270 system from the time.
>
>
> > Hopefully you'll get better results with it.
>
> Definitely a different result. Still garbage on the screen, but the system actually boots.
>
> Before the HD activity light would stop and any key press would cause a beep.
>
> Now I've got ssh access and can maybe provide more useful information.
>
> [ 6.8815928] kern.module.path=/stand/i386/8.99.27/modules
> [ 6.8815928] kern info: [drm] Memory usable by graphics device = 256M
> [ 6.8815928] kern info: [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
> [ 6.8815928] kern info: [drm] Driver supports precise vblank timestamp query.
> [ 6.8815928] kern info: [drm] failed to find VBIOS tables
> [ 6.8815928] i915drmkms0: interrupting at ioapic0 pin 16 (i915drmkms0)
> [ 6.9116124] kern info: [drm] initialized overlay support
> [ 6.9816621] kern info: [drm] GMBUS [i915 gmbus vga] timed out, falling back to bit banging on pin 2
> [ 7.2418458] intelfb0 at i915drmkms0
> [ 7.2418458] intelfb0: framebuffer at 0xdc167000, size 640x480, depth 32, stride 2560
> [ 7.9723633] kern error: [drm:(/usr/src/sys/external/bsd/drm2/dist/drm/i915/intel_display.c:12659)intel_pipe_config_compare] *ERROR* mismatch in base.adjusted_mode.flags(DRM_MODE_FLAG_PHSYNC) (expected 0, found 1)
> [ 7.9723633] kern error: [drm:(/usr/src/sys/external/bsd/drm2/dist/drm/i915/intel_display.c:12663)intel_pipe_config_compare] *ERROR* mismatch in base.adjusted_mode.flags(DRM_MODE_FLAG_PVSYNC) (expected 0, found 4)
> [ 7.9723633] warning: /usr/src/sys/external/bsd/drm2/dist/drm/i915/intel_display.c:12899: pipe state doesn't match!
> [ 8.0123934] wsdisplay0 at intelfb0 kbdmux 1: console (default, vt100 emulation), using wskbd0
> [ 8.0338052] wsmux1: connecting to wsdisplay0
>
> Complete dmesg here:
> https://pastebin.com/K5k1p6JX
>
> Any other information I could provide?
>
>
>
>
From: "David H. Gutteridge" <david@gutteridge.ca>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/53795: i915drmkms causes crash on boot with Atom
N270/945GSE netbook
Date: Mon, 07 Jan 2019 21:16:06 -0500
Further questions and comments:
You didn't mention what make and model this machine is? I infer from
looking at the dmesg output that it's a 3rd generation Classmate by
Intel? On that basis, I tried to see if there were any quirks listed
for it in a recent version of the Linux i915 driver, but didn't note
anything.
Which specific Linux kernel versions does it work on? Same with
OpenBSD? NetBSD's i915 code is based on Linux 4.4, as is OpenBSD's
(for the most part). I tried comparing what had changed between 4.4
and 4.20 in Linux, but there's too much churn to be useful without
having more to go on to narrow it down. (And, indeed, it could be
the problem is somewhere in BSD-specific code, for all I know.)
Is there any more dmesg output than that? I'd expect if you ended up
with working ssh access there'd be more of it. The initial ACPI output
looks weird, i.e. I doubt all those video outputs exist on the machine.
I have no idea if that's anything more than a red herring, though.
You could enable debugging via the appropriate sysctl setting for more
verbose output (assuming it gets that far, per previous question). To
do so, you can set hw.drm2.drm_debug to a suitable value (e.g. 0xf) in
/etc/sysctl.conf. That might tell us something more.
Something else you could try -- if you're really curious -- would be to
enable the old i915drm driver instead, and see what happens there. (I
used to use the old driver on my LG X110.)
Dave
From: Elliott Shue <elliott1787@gmail.com>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/53795: i915drmkms causes crash on boot with Atom N270/945GSE netbook
Date: Tue, 8 Jan 2019 01:36:20 -0500
> You didn't mention what make and model this machine is?
I did in the original report, it is an ECS E09EI6, DakTech branded in
my case. It is slightly oddball, with a 1.8" ZIF drive and internal
USB based Wifi, but I don't think the graphics are anything that
weird.
http://www.ecs.com.tw/ECSWebSite/Product/Product_Overview/EN/Notebook/E09EI6,%20Intel%20classmate%20PC%20-%20Clamshell/E%20Series%20Intel-powered%20Classmate%20PC
I tried flashing it to the generic BIOS with no change.
> Which specific Linux kernel versions does it work on? Same with
> OpenBSD? NetBSD's i915 code is based on Linux 4.4, as is OpenBSD's
> (for the most part).
I ran OpenBSD 6.2 on it for a while. If I remember right the console
wasn't native resolution, but X was. I've never had an issue with any
version of Linux in the time I've had it, some before kernel 4.4, some
after. Just tried live Debian Stretch with kernel 4.9 and no issues
with console or X.
> Is there any more dmesg output than that? I'd expect if you ended up
> with working ssh access there'd be more of it.
That was everything I got with i915drmkms enabled. I get more with it disabled.
There is something weird going on with i915drmkms enabled because my
Wifi won't come up when enabled in rc.conf, I added it and sshd to
rc.local and that works. I should have mentioned that earlier, I am
not experienced enough with NetBSD to know what is going on but it
seems like something is halting the init.
i915drmkms disabled:
https://pastebin.com/Q4UQDUH6
Bad CMOS battery, ignore the date warning. That is with
hw.drm2.drm_debug=0xf, but I don't think it matters due to i915drmkms
being disabled.
> The initial ACPI output looks weird, i.e. I doubt all those video outputs exist on the machine.
> I have no idea if that's anything more than a red herring, though.
There is only one video output physically, the internal LCD, it
doesn't even have a VGA port. I would've tried an external monitor if
I could.
> You could enable debugging via the appropriate sysctl setting for more
> verbose output (assuming it gets that far, per previous question). To
> do so, you can set hw.drm2.drm_debug to a suitable value (e.g. 0xf) in
> /etc/sysctl.conf. That might tell us something more.
Done, i915drmkms enabled with hw.drm2.drm_debug=0xf
https://pastebin.com/Zbmxj5Zg
> Something else you could try -- if you're really curious -- would be to
> enable the old i915drm driver instead, and see what happens there. (I
> used to use the old driver on my LG X110.)
I tried that on 8.0, mentioned it in the original report. It booted,
but no native resolution in console or X. It leaves me about where I
am with the i915drmkms disabled.
I did try booting 8.0 and current on an Asus Eee 901 with the same
chipset, it didn't crash and I got native resolution at the console.
The system is in poor condition and I couldn't do much with it though.
From: "David H. Gutteridge" <david@gutteridge.ca>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/53795: i915drmkms causes crash on boot with Atom
N270/945GSE netbook
Date: Tue, 05 Feb 2019 21:17:43 -0500
On Tue, 8 Jan 2019, at 06:40:00 +0000 (UTC), Elliott Shue wrote:
>I did in the original report, it is an ECS E09EI6, DakTech branded in
>my case. It is slightly oddball, with a 1.8" ZIF drive and internal
>USB based Wifi, but I don't think the graphics are anything that
>weird.
So you did. My reading comprehension was a little off. I see I also
missed you mentioned you had tested i915drm already. My apologies.
>There is something weird going on with i915drmkms enabled because my
>Wifi won't come up when enabled in rc.conf, I added it and sshd to
>rc.local and that works. I should have mentioned that earlier, I am
>not experienced enough with NetBSD to know what is going on but it
>seems like something is halting the init.
In theory, in this example, you could be hitting two different bugs:
the DRM issue where you don't get any console output, and then an
unrelated USB bug that triggers a kernel panic. (I say that since your
WiFi is attached via USB.) I can sometimes reproduce a USB-related
panic on my X110 (http://gnats.netbsd.org/53895) while using a USB WiFi
device. I haven't isolated how to reproduce this consistently yet.
(Sometimes, it leaves trace information in the dmesg that's logged,
other times, nothing makes it.)
>> The initial ACPI output looks weird, i.e. I doubt all those video
>> outputs exist on the machine. I have no idea if that's anything more
>>than a red herring, though.
>There is only one video output physically, the internal LCD, it
>doesn't even have a VGA port. I would've tried an external monitor if
>I could.
For one thing, it's saying it can't get all the necessary data from the
BIOS about your display hardware, so it's guessing at what it needs to
find, but that's not uncommon. I see the same thing on my X110, not a
showstopper in itself. (There's another recent report of an issue that
could be related to yours, with no resolution for it, either:
http://mail-index.netbsd.org/current-users/2019/01/12/msg034921.html)
I was triaging this bug to help out, but it's probably beyond my
knowledge to help narrow further, so I'll assign it to our main DRM
maintainer.
Dave
Responsible-Changed-From-To: kern-bug-people->riastradh
Responsible-Changed-By: gutteridge@NetBSD.org
Responsible-Changed-When: Wed, 06 Feb 2019 02:29:31 +0000
Responsible-Changed-Why:
Over to main DRM maintainer.
From: "David H. Gutteridge" <david@gutteridge.ca>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/53795: i915drmkms causes crash on boot with Atom
N270/945GSE netbook
Date: Tue, 12 Mar 2019 23:37:50 -0400
FWIW, I don't think you're hitting the USB bug I filed as PR kern/53895
as part of your problem (with the WiFi not coming up). It seems too
difficult to reproduce for you to be encountering it consistently.
(Sorry I haven't been more help with the main DRM aspect.)
Dave
State-Changed-From-To: open->feedback
State-Changed-By: riastradh@NetBSD.org
State-Changed-When: Sun, 20 Aug 2023 06:23:30 +0000
State-Changed-Why:
Sorry to let this languish for so long. Any chance you could try with
a newer version of NetBSD? We've had a drm update and a lot of fixes
since then, including changes to fix finding the VBIOS.
From: Elliott Shue <elliott1787@gmail.com>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/53795 (i915drmkms causes crash on boot with Atom N270/945GSE netbook)
Date: Thu, 4 Apr 2024 12:52:48 -0400
I completely forgot about this until I came across the netbook while
cleaning. Drive in it is dead, but it does boot the 10.0 installer
fine and the resolution is good. I guess we can consider this
resolved.
On Sun, Aug 20, 2023 at 2:23=E2=80=AFAM <riastradh@netbsd.org> wrote:
>
> Synopsis: i915drmkms causes crash on boot with Atom N270/945GSE netbook
>
> State-Changed-From-To: open->feedback
> State-Changed-By: riastradh@NetBSD.org
> State-Changed-When: Sun, 20 Aug 2023 06:23:30 +0000
> State-Changed-Why:
> Sorry to let this languish for so long. Any chance you could try with
> a newer version of NetBSD? We've had a drm update and a lot of fixes
> since then, including changes to fix finding the VBIOS.
>
>
>
State-Changed-From-To: feedback->closed
State-Changed-By: riastradh@NetBSD.org
State-Changed-When: Thu, 04 Apr 2024 17:10:49 +0000
State-Changed-Why:
Great, thanks for testing!
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.47 2022/09/11 19:34:41 kim Exp $
$NetBSD: gnats_config.sh,v 1.9 2014/08/02 14:16:04 spz Exp $
Copyright © 1994-2024
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.