Bug 8536 - i915 BIOS fails when restarting Xserver.
Summary: i915 BIOS fails when restarting Xserver.
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: 6.8.99.901 (6.9 RC1)
Hardware: x86 (IA32) Linux (All)
: high normal
Assignee: Alan Hourihane
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-10-06 16:39 UTC by Carl Michal
Modified: 2007-02-13 22:07 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
Xorg log on failed startup attempt (19.97 KB, text/plain)
2006-10-06 16:41 UTC, Carl Michal
no flags Details
First startup-successful (104.30 KB, text/plain)
2006-10-10 16:29 UTC, Carl Michal
no flags Details
2nd start-successful (104.30 KB, text/plain)
2006-10-10 16:29 UTC, Carl Michal
no flags Details
3rd start, fails (19.92 KB, text/plain)
2006-10-10 16:29 UTC, Carl Michal
no flags Details
First startup-successful (104.30 KB, text/plain)
2006-10-11 09:38 UTC, Carl Michal
no flags Details
2nd start-successful (104.30 KB, text/plain)
2006-10-11 09:38 UTC, Carl Michal
no flags Details
3rd start, fails (19.92 KB, text/plain)
2006-10-11 09:38 UTC, Carl Michal
no flags Details
first try, successful (105.07 KB, text/plain)
2006-10-11 14:52 UTC, Carl Michal
no flags Details
2nd start successful (105.07 KB, text/plain)
2006-10-11 14:52 UTC, Carl Michal
no flags Details
3rd start fails (20.02 KB, text/plain)
2006-10-11 14:52 UTC, Carl Michal
no flags Details
3rd try fails (21.80 KB, text/plain)
2006-10-11 18:25 UTC, Carl Michal
no flags Details
1st try (195.60 KB, text/plain)
2006-10-12 08:47 UTC, Carl Michal
no flags Details
2nd try (171.30 KB, text/plain)
2006-10-12 08:47 UTC, Carl Michal
no flags Details
3rd try fails (21.80 KB, text/plain)
2006-10-12 08:48 UTC, Carl Michal
no flags Details
Here's my xorg.conf (with dual head disabled right now) (1.77 KB, text/plain)
2006-11-09 10:16 UTC, Rich Landers
no flags Details
Here's my Xorg.0.log upon 3rd-try failure. (19.06 KB, text/plain)
2006-11-09 10:31 UTC, Rich Landers
no flags Details
Here's my lspci output. (1.66 KB, text/plain)
2006-11-09 10:32 UTC, Rich Landers
no flags Details

Description Carl Michal 2006-10-06 16:39:43 UTC
Using xf86-video-i810-1.7, I've noticed a couple of problems with dual head. 
I'll file the separately.

After the server quits, it won't start again.  The second time you try, the
error is:

> (EE) I810:(0): unknown reason for exception
> (EE) I810:(0): cannot continue
> (EE) I810:(0): VBE initialization failed.
> 
> Backtrace:
> 0: X(cf86SigHandler+0x88) [0x80c2f38]
> 
> Fatal server error:
> Caught signal 11. Server aborting

the intel-agp module can't be removed from the kernel at this point (although
i915 and drm can be).

It seems a reboot is needed to clear this up.

This is an i915 in a Dell Inspiron.  I'll attach the log from the failed
startup.  There is mesage: Bad V_BIOS checksum that occurs even when all is
well, I presume this comes from using 855resolution to get support for the
laptop native screen resolution (1280x800).
Comment 1 Carl Michal 2006-10-06 16:41:01 UTC
Created attachment 7275 [details]
Xorg log on failed startup attempt

Here's the log from a failed startup attempt following a successful dual-head 
session.
Comment 2 Alan Hourihane 2006-10-08 13:58:00 UTC
Did this work with any previous driver correctly ? If not, check if there's a
BIOS upgrade for your machine.
Comment 3 Alan Hourihane 2006-10-08 14:23:46 UTC
Can you also try a test driver from http://www.fairlite.demon.co.uk/intel.html
Comment 4 Carl Michal 2006-10-10 16:28:09 UTC
There was in fact a new BIOS (MXC051-A04), that I have upgraded to, and the
behaviour seems a little different now.

It starts up fine the first time, starts up fine the second time, but always
fails on the third time.  I'll attach logs from all three.

I tried to use the test driver, but the server complains that it is ABI version
(0), while the server is ABI version (1).
Comment 5 Carl Michal 2006-10-10 16:29:02 UTC
Created attachment 7342 [details]
First startup-successful
Comment 6 Carl Michal 2006-10-10 16:29:24 UTC
Created attachment 7343 [details]
2nd start-successful
Comment 7 Carl Michal 2006-10-10 16:29:48 UTC
Created attachment 7344 [details]
3rd start, fails
Comment 8 Alan Hourihane 2006-10-11 00:39:03 UTC
Use the -ignoreABI switch to the Xserver which will get past the ABI error and
try again.
Comment 9 Alan Hourihane 2006-10-11 08:55:03 UTC
I've also uploaded a new driver which emits some extra debug information for me.

Can you try it, and upload a new log.
Comment 10 Carl Michal 2006-10-11 09:38:07 UTC
Created attachment 7355 [details]
First startup-successful

Similar behaviour with the test driver.  This is the first log with a
successful start.
Comment 11 Carl Michal 2006-10-11 09:38:35 UTC
Created attachment 7356 [details]
2nd start-successful
Comment 12 Carl Michal 2006-10-11 09:38:53 UTC
Created attachment 7357 [details]
3rd start, fails
Comment 13 Carl Michal 2006-10-11 14:52:02 UTC
Created attachment 7364 [details]
first try, successful

Sorry, I seem to have uploaded the wrong logs earlier...
Comment 14 Carl Michal 2006-10-11 14:52:26 UTC
Created attachment 7365 [details]
2nd start successful
Comment 15 Carl Michal 2006-10-11 14:52:45 UTC
Created attachment 7366 [details]
3rd start fails
Comment 16 Alan Hourihane 2006-10-11 15:34:48 UTC
Uploaded another test driver to try for this.
Comment 17 Carl Michal 2006-10-11 18:22:47 UTC
ok, so that's a little different.

Now when it fails on the third try, I get dumped back into the console, but the
console is on the second head.  The server now says:

(EE) I810(0): VBE initialization failed.
TRYING REPOST
(EE) I810(0): unknown reason for exception
(EE) I810(0): cannot continue
(EE) I810(0): unknown reason for exception
(EE) I810(0): cannot continue

Backtrace:
0: X(xf86SigHandler+0x84) [0x80b85e4]
1: [0xffffe420]
2: X(InitOutput+0x9b1) [0x80a0421]
3: X(main+0x276) [0x806e3f6]
4: /lib/libc.so.6(__libc_start_main+0xd8) [0xb7cf7878]
5: X(FontFileCompleteXLFD+0xad) [0x806d931]

and I'll attach the log.
Comment 18 Carl Michal 2006-10-11 18:25:55 UTC
Created attachment 7369 [details]
3rd try fails
Comment 19 Alan Hourihane 2006-10-12 01:14:53 UTC
Did this work correctly in any previous driver ?
Comment 20 Alan Hourihane 2006-10-12 01:29:16 UTC
O.k. I've enabled some quite extensive debug in a new test driver.

Can you try and upload the three logs again ?
Comment 21 Carl Michal 2006-10-12 08:47:17 UTC
Created attachment 7382 [details]
1st try
Comment 22 Carl Michal 2006-10-12 08:47:32 UTC
Created attachment 7383 [details]
2nd try
Comment 23 Carl Michal 2006-10-12 08:48:07 UTC
Created attachment 7384 [details]
3rd try fails
Comment 24 Carl Michal 2006-10-12 08:49:17 UTC
I'm certain I set up dual-head non-xinerama and used it with a previous driver.
 I can't promise that I quit and restarted it 3 times between reboots...
Comment 25 Alan Hourihane 2006-10-12 09:09:50 UTC
Unfortunately this looks like a buggy BIOS. The modesetting branch will remove
the BIOS from the equation and sort this kind of problem out.
Comment 26 Alan Hourihane 2006-10-12 09:17:14 UTC
I've changed the title because intel-agp can never be unloaded. It's been like
this for quite some time regardless of this bug.
Comment 27 Rich Landers 2006-11-09 10:14:47 UTC
I am having this exact problem, including the 3rd-time-only crash, on an Acer
Aspire 3620, dual head setup. I have just installed Fedora Core 6 with the
included Xorg 7.1.1. server.  Everything works fine with a single head setup.
The dual head works fine for the first two restarts, and again after rebooting.
 I too get a warning about a bad VBIOS checksum.

An additional clue, perhaps: after startx, I see a garbled screen until the
server gets going, even on successful starts.


Comment 28 Rich Landers 2006-11-09 10:16:46 UTC
Created attachment 7714 [details]
Here's my xorg.conf (with dual head disabled right now)

Dual head setup on Acer Aspire 3620.
Comment 29 Rich Landers 2006-11-09 10:31:41 UTC
Created attachment 7715 [details]
Here's my Xorg.0.log upon 3rd-try failure.
Comment 30 Rich Landers 2006-11-09 10:32:26 UTC
Created attachment 7716 [details]
Here's my lspci output.
Comment 31 Marcin Zajaczkowski 2006-12-09 03:33:17 UTC
(In reply to comment #17)
> ok, so that's a little different.
> 
> Now when it fails on the third try, I get dumped back into the console, but the
> console is on the second head.  The server now says:
> 
> (EE) I810(0): VBE initialization failed.
> TRYING REPOST
> (EE) I810(0): unknown reason for exception
> (EE) I810(0): cannot continue
> (EE) I810(0): unknown reason for exception
> (EE) I810(0): cannot continue
> 
> Backtrace:
> 0: X(xf86SigHandler+0x84) [0x80b85e4]
> 1: [0xffffe420]
> 2: X(InitOutput+0x9b1) [0x80a0421]
> 3: X(main+0x276) [0x806e3f6]
> 4: /lib/libc.so.6(__libc_start_main+0xd8) [0xb7cf7878]
> 5: X(FontFileCompleteXLFD+0xad) [0x806d931]

I have similar behaviour (and similar stack trace) with Amilo Pro V8010 (Intel
Mobile 915GM Express) with enabled dual head view.

Fedora Core 6, x.org 7.1,
xorg-x11-drv-i810-1.6.5-10.fc6 (also with 1.7.3 compiled by hand).

I can attach configuration and logs.
Comment 32 Alan Hourihane 2007-01-05 08:48:47 UTC
O.k. I've found the cause of this and I'll be committing a fix soon with a new
release of driver.
Comment 33 Alan Hourihane 2007-01-09 06:02:30 UTC
Committed. Closing.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.