Bug 10759

Created attachment 9738 [details]
the log

Created attachment 9739 [details]
conf

It always defaults to 1280x768. I can get the native resolution from VGA output, but that has other issues; it could hang after a server restart, or start with a wrong resolution...

I was able to change the resolution via xrandr with a previous version of this driver (can't remember which one), but now it crashes the server (1.2.0).

Oh, forgot to mention that this test was on Ubuntu 7.04 with intel-2.0.0

Weird.  Can you try a more recent driver version?  I've been using large resolutions on my 965 with an add-in DVI card for awhile now, no problems.

Still the same with xserver 1.4.1pre and intel 2.1.99. Now the log doesn't even show any probed values.

Created attachment 12598 [details]
log with 2.1.99

(In reply to comment #7)
> Created an attachment (id=12598) [details]
> log with 2.1.99
> 

Would you please turn on the modedebug option in the device section of your xorg.conf and repost the log?

Thanks,
Hong

Created attachment 12665 [details]
log wit modedebug

Sure, here you go.

(In reply to comment #9)
> Created an attachment (id=12665) [details]
> log wit modedebug
> 
> Sure, here you go.
> 

It's weird I can't see any EDID data from your log file.

(II) intel(0): Not using default mode "1680x1050" (hsync out of range)
It seems your "1680x1050" is filtered out, but I can't find any hsync range info from your log.

Please try to use the latest driver and
1. move Option "monitor-TMDS-1" "Generic Monitor" to the Device section if you want to use the manually set sync range.
2. add Option "PreferredMode" "1680x1050" to your monitor section.

Thanks,
Hong

Just adding the PreferredMode and monitor-TMDS-1 is not enough, I have to add the modeline to the Monitor-section. Then it works.

Duplicate of Bug #10625 ?

Could be, I'll build a new server and test.

Actually no, since that patch is already in the Debian/Ubuntu xorg-server package.

Seaching in google, it seems the IBM T221 monitor is a very special one.
Only certain kinds of cards can work with this monitor (and IBM provides drivers for these cards).

It may explain that we can't probe the EDID data from this monitor.

Since you can workaround it with some xorg.conf hack, I am closing it now :)

Thanks,
Hong

Oops, sorry.. this is a D221, not the über-expensive T221! :)

Also, the monitor provides valid EDID over the d-sub cable, so why does the DVI addon-card fail to do so?

I'll reopen this because the original information was not accurate. If you still think that the driver can't ever configure this monitor automatically, please close it again.

Tried 'xresprobe' on ubuntu, and somehow it manages to read the edid:

 # xresprobe intel
id: LEN D221 Wid
res: 1680x1680 1680x1050 1280x1024 1152x864 1024x768 832x624 800x600 720x400 640x480
freq: 30-83 50-76
disptype:

(In reply to comment #19)
> I'll reopen this because the original information was not accurate. If you
> still think that the driver can't ever configure this monitor automatically,
> please close it again.
> 
> Tried 'xresprobe' on ubuntu, and somehow it manages to read the edid:
> 
>  # xresprobe intel
> id: LEN D221 Wid
> res: 1680x1680 1680x1050 1280x1024 1152x864 1024x768 832x624 800x600 720x400
> 640x480
> freq: 30-83 50-76
> disptype:
> 

It seems xresprobe calls ddcprobe for probe, and ddcprobe uses video bios service to get edid data. Would you please attach the ddcprobe output?

I am not sure now why we can't get it from DVI add-on card :(

Timo, would you please attach ddcprobe output?

Created attachment 12925 [details]
ddcprobe

Sorry, I was away for a couple of days. Here you go.

Created attachment 13502 [details]
Sam 243T & intel driver log

I am seeing this same problem when attaching my Samsung 243T (1920x1200) monitor to my Think pad T61 (with 965GM). It is stuck at max of 1280x1024 @ 75. It appears for some crazy reason it cannot read the EDID based on the xlogs. This only happens with the intel driver. Using the same monitor with the "nv" driver I have no problems.

I will attach my two logs, perhaps this can help in this issue.

xrandr_intel_samsung243T_BAD.txt <-- intel driver log
xrandr_nv_samsung_243T_WORKS.txt <-- nv driver log

Created attachment 13503 [details]
Sam 243T & nv driver log

(In reply to comment #23)
> Created an attachment (id=13502) [details]
> Sam 243T & intel driver log
> 

Would you please turn on the Modedebug option in Device section of your xorg.conf file and attach Xorg log and xorg.conf ?

Thanks,
Hong

Created attachment 13703 [details] [review]
correct setting dvi target input

Hi, Timo
Sorry for the long latency.

Would you please try this patch, and attach the xorg log with modedebug turned on?

Thanks,
Hong

Created attachment 13706 [details]
logfile with modedebug

I added the patch on top of 2.2.0, but it still fails. See the log. I added HorizSync and VertRefresh values for the monitor, but they didn't help.

Thanks for trying the test patch.

Several questions:
1. Would you please try with other monitors to see if our driver can get edid data from these monitors. If can, at least our driver can work correctly with your addin SDVO card. It may be some specific timing issue with your IBM D221 monitor.

2. what is your daughter card? or where can I find the info for this card?

3. From the xorg log, 
(II) intel(0): I2C device "SDVOB DDC Bus:ddc2" registered at address 0xA0.
(II) intel(0): SDVO: W: 7A 02 (SDVO_CMD_SET_CONTROL_BUS_SWITCH)
(II) intel(0): SDVO: W: 7A 02 (SDVO_CMD_SET_CONTROL_BUS_SWITCH)
(II) intel(0): SDVO: W: 7A 02 (SDVO_CMD_SET_CONTROL_BUS_SWITCH)
(II) intel(0): SDVO: W: 7A 02 (SDVO_CMD_SET_CONTROL_BUS_SWITCH)
(II) intel(0): I2C device "SDVOB DDC Bus:ddc2" removed.

This is when our driver is trying to retrive the EDID data block for your monitor, but it fails for 4 times, and then driver gives up.
You can do some test to see if adding retry times can help?
(The code is in xserver/hw/xfree86/ddc/xf86DDC.c function: DDCRead_DDC2, you can modify the macro RETRIES in this file).

Thanks,
Hong

(In reply to comment #29)
> Thanks for trying the test patch.
> 
> Several questions:
> 1. Would you please try with other monitors to see if our driver can get edid
> data from these monitors. If can, at least our driver can work correctly with
> your addin SDVO card. It may be some specific timing issue with your IBM D221
> monitor.

I've tried with Sun L194RH, and I do get a picture with it, but it's the same 1280x768 which is wrong, since it's a 4/3 monitor with 1280x1024 native resolution.

> 2. what is your daughter card? or where can I find the info for this card?

Seems to be Lenovo 73P2516 with a Conexant CX25904-12Z chip.

> 3. From the xorg log, 
> (II) intel(0): I2C device "SDVOB DDC Bus:ddc2" registered at address 0xA0.
> (II) intel(0): SDVO: W: 7A 02 (SDVO_CMD_SET_CONTROL_BUS_SWITCH)
> (II) intel(0): SDVO: W: 7A 02 (SDVO_CMD_SET_CONTROL_BUS_SWITCH)
> (II) intel(0): SDVO: W: 7A 02 (SDVO_CMD_SET_CONTROL_BUS_SWITCH)
> (II) intel(0): SDVO: W: 7A 02 (SDVO_CMD_SET_CONTROL_BUS_SWITCH)
> (II) intel(0): I2C device "SDVOB DDC Bus:ddc2" removed.
> 
> This is when our driver is trying to retrive the EDID data block for your
> monitor, but it fails for 4 times, and then driver gives up.
> You can do some test to see if adding retry times can help?
> (The code is in xserver/hw/xfree86/ddc/xf86DDC.c function: DDCRead_DDC2, you
> can modify the macro RETRIES in this file).

Ok, I'll try this later.

Created attachment 13875 [details] [review]
i2c bit output fix

Another test patch, would you please have a try?

Created attachment 14008 [details]
logfile with the latest patch

finally got the machine up and running again, but it's still the same. Log attached..

(In reply to comment #32)
> Created an attachment (id=14008) [details]
> logfile with the latest patch
> 
> finally got the machine up and running again, but it's still the same. Log
> attached..
> 

Still got no luck :(

So what's the result of increasing the retry times I suggested in comment #29?

Thanks,
Hong

I changed RETRIES to 10, but it didn't do any good :/

Created attachment 14779 [details] [review]
fix sdvo ddc bus selection

Hi, Timo

Would you please try this patch? Thanks for the help.

Hong

thanks, tried it but still the same as before..

Would you please attach the xorg log with modedebug turned on?

Thanks,
Hong

Created attachment 14785 [details]
new logfile

Timo, are you able to test the monitor on another machine? It sounds like an issue on the Lenovo 73P2516...

(In reply to comment #39)
> Timo, are you able to test the monitor on another machine? It sounds like an
> issue on the Lenovo 73P2516...
> 
sorry, what I mean is to test on a machine with intel graphic ( thus use intel graphic driver ).. it's better to be using the DVI interface as well.

I'll try next week. There should be some Fujitsu-Siemens box that I'm able to test with. The monitors they have are working fine.

Ok, the monitor works fine on a Fujitsu Esprimo E5915, which has a 965 chip in it, and some DVI addon card. Since the Sun monitor fails to work with the Lenovo addon card, I'd say the card is buggy. The mode can be forced via xorg.conf, so maybe just close this bug as invalid?

thanks for your efforts!

ok. thanks, Timo. I also got some confirm from HW vendor. 

Mark this bug as wont fix, as it's HW addon bug issue.