|Summary:||"Set VBE Mode failed" with dualhead and different resolutions|
|Product:||xorg||Reporter:||Osma Ahvenlampi <oa>|
|Component:||Driver/intel||Assignee:||Alan Hourihane <alanh>|
|Status:||RESOLVED DUPLICATE||QA Contact:|
|i915 platform:||i915 features:|
Description Osma Ahvenlampi 2005-08-28 09:37:50 UTC
On an Acer TravelMate 382TMi laptop, with i855GM graphics, use of dualhead has recently started to work with Alan's test drivers. However, it still fails if different resolution is desired on the external screen. This is what would happen if possible modes are autodetected (LFP is 1024x768, and DDC will detect 1280x1024 capability on the VGA-port attached LCD monitor). Apparently this might be a bug in Acer's VideoBIOS - only one BIOS version is available for this machine (v 1.07).
Comment 1 Osma Ahvenlampi 2005-08-28 09:39:46 UTC
Created attachment 3081 [details] xorg.conf used FixedPipe, DevicePresence, DisplayInfo etc options do not change results. Problem can be worked around by forcing 1024x768 resolution on both displays.
Comment 3 Osma Ahvenlampi 2005-08-28 09:43:06 UTC
lspci information: 00:02.0 VGA compatible controller: Intel Corporation 82852/855GM Integrated Graphics Device (rev 02) (prog-if 00 [VGA]) Subsystem: Acer Incorporated [ALI]: Unknown device 0037 Flags: bus master, fast devsel, latency 0, IRQ 11 Memory at e8000000 (32-bit, prefetchable) [size=128M] Memory at e0000000 (32-bit, non-prefetchable) [size=512K] I/O ports at 1800 [size=8] Capabilities: [d0] Power Management version 1 00:02.1 Display controller: Intel Corporation 82852/855GM Integrated Graphics Device (rev 02) Subsystem: Acer Incorporated [ALI]: Unknown device 0037 Flags: fast devsel Memory at f0000000 (32-bit, prefetchable) [size=128M] Memory at e0080000 (32-bit, non-prefetchable) [size=512K] Capabilities: [d0] Power Management version 1
Comment 4 Alan Hourihane 2005-10-03 02:51:54 UTC
The fact is, that the Acer's VideoBIOS is restricting the modes when in dual head. Probably related to bandwidth issues. The ModeAttributes field shows as 0x9a when a mode cannot be supported. 0x9b is what is expected when a mode is possible. You'll notice that these values change in dual head, so Acer's BIOS is restricting you to 1024x768 on both heads.
Comment 5 Alan Hourihane 2005-10-03 02:53:23 UTC
Whoops wrong report. Sorry about that.
Comment 6 Alan Hourihane 2005-10-03 03:00:09 UTC
Just uploaded 1.5.157 to see if I can fix this in the driver. Can you try it and let me know ?
Comment 7 Alan Hourihane 2005-10-06 02:11:01 UTC
Can you try my latest test driver and upload a new log file ??
Comment 8 Osma Ahvenlampi 2005-10-27 05:00:09 UTC
The Acer is currently in the shop for fan repairs. I'll test when I get it back. However, I can state that on my substitute machine, a Fujitsu Siemens laptop also based on Centrino platform, I can use dual head with 1400x1050 on the laptop display and 1280x1024 on the external display (via the VGA port) without problems, even with an older driver build. As for bandwidth, under Windows the Acer is capable of different modes on the two displays.
Comment 9 Alan Hourihane 2005-10-27 08:37:30 UTC
Right. I know (or at least have an idea) where the problem is, so I'm just trying to find if it's possible to work around the problems in the driver.
Comment 10 Osma Ahvenlampi 2005-11-25 05:06:25 UTC
Created attachment 3893 [details] log Got the Acer back. With a test driver downloaded two hours ago, the machine freezes on virtual console change if VBERestore, DevicePresence and CheckDevices are turned on. With them commented out, the machine is stable. However, I can not get 1280x1024, or in fact a "good" 1024x768 on the external display - I only get a flattened 1024x768 display and my monitor complains "Input signal out of range". In single head mode, I believe the driver thinks its programming 1280x1024 on, but the display remains blank.
Comment 11 Osma Ahvenlampi 2005-11-25 05:49:38 UTC
Created attachment 3894 [details] multihead log file Found that by commenting out VBERestore and DevicePresence, but retaining the CheckDevices option, I can get a stable 1280x1024 on a single head (the external screen). However, dualhead still fails. Hopefully a trained eye can read the log file better than I (disregard the earlier log).
Comment 12 Alan Hourihane 2005-11-25 23:43:57 UTC
O.k. Can you try this for the MonitorLayout command and post another log. Option "MonitorLayout" "LFP,CRT"
Comment 13 Osma Ahvenlampi 2005-11-26 02:35:12 UTC
Created attachment 3906 [details] log with monitorlayout switched Won't start at all with MonitorLayout "LFP,CRT".
Comment 14 Alan Hourihane 2005-11-26 02:48:58 UTC
o.k. put it back to CRT,LFP and add Option "FixedPipe" "B" and post another log.
Comment 15 Osma Ahvenlampi 2005-11-27 17:10:10 UTC
Created attachment 3918 [details] log with fixedpipe not sure which head was supposed to be FixedPipe "B", but here's one with it different from the default. Both ways it dies with the same error if the modes list for the external head includes 1280x1024 or I leave the modes list for autodetection. Initializes correctly if I force modes to 1024x768 only. My home monitor doesn't mind the refresh rates set by this driver, and the image on the external display has correct proportions with the same config file that creates a flattened image on my work monitor.
Comment 16 Jos Delbar 2006-01-23 12:17:24 UTC
I have a similar problem with my video BIOS restricting a dual head setup with a CRT resolution of 1280x1024 and a LFP resolution of 1024x768. Dual head with both screens at 1024x768 works fine. As always, in Windows this setup works using the Intel driver. If you're still attempting to work around this in the i810 driver, I'm available for testing. It would be great if you could fix this "bug".
Comment 17 Alan Hourihane 2006-05-08 20:51:58 UTC
Marking as duplicate of 643 because the new modesetting branch should deal with all vbios resolution issues. *** This bug has been marked as a duplicate of 643 ***
Comment 18 Osma Ahvenlampi 2006-05-08 21:45:28 UTC
ooohh.. by any chance could you make a build of the modesetting branch available somewhere for xorg 7.0? I'd be more than happy to test it, but not sure I can build it.. never had that much luck building xfree and never tried to with xorg. I've nearly given up on being able to utilize dualhead with this machine..