Booted kde 3beta3 livedvd for the first time on a new HDMI connected monitor with nvidia GT220 graphics. It looked stretched and too big so loaded kranrd which says it is currently 1024x768 @ 75hz rather than 1920x1080 @ 60hz. Reproducible: Steps to Reproduce:
Confirmed the nvidia kernel module is loaded krandr also shows Auto against 1920x1080 despite it being 1024x768 It is trivial then to switch to the correct resolution but it should be detected on boot.
Whiteboard: (none) => 3beta3
Hmm seems it may be the monitors fault.. from Xorg.0.log [ 59.482] (--) NVIDIA(0): LG Electronics W2361 (DFP-1): 165.0 MHz maximum pixel clock [ 59.482] (--) NVIDIA(0): LG Electronics W2361 (DFP-1): Internal Single Link TMDS [ 59.482] (**) NVIDIA(0): Using HorizSync/VertRefresh ranges from the EDID for display [ 59.482] (**) NVIDIA(0): device LG Electronics W2361 (DFP-1) (Using EDID [ 59.482] (**) NVIDIA(0): frequencies has been enabled on all display devices.) [ 59.484] (WW) NVIDIA(GPU-0): The EDID for LG Electronics W2361 (DFP-1) contradicts itself: [ 59.484] (WW) NVIDIA(GPU-0): mode "1920x1080" is specified in the EDID; however, the [ 59.484] (WW) NVIDIA(GPU-0): EDID's valid HorizSync range (30.000-83.000 kHz) would [ 59.484] (WW) NVIDIA(GPU-0): exclude this mode's HorizSync (28.1 kHz); ignoring [ 59.484] (WW) NVIDIA(GPU-0): HorizSync check for mode "1920x1080". repeats then [ 59.488] (II) NVIDIA(0): Validated MetaModes: [ 59.488] (II) NVIDIA(0): "DFP-1:1024x768" [ 59.488] (II) NVIDIA(0): "DFP-1:800x600" [ 59.488] (II) NVIDIA(0): "DFP-1:640x480" [ 59.488] (II) NVIDIA(0): Virtual screen size determined to be 1024 x 768 [ 59.510] (--) NVIDIA(0): DPI set to (50, 67); computed from "UseEdidDpi" X config [ 59.510] (--) NVIDIA(0): option [ 59.510] (**) NVIDIA(0): Enabling 32-bit ARGB GLX visuals. [ 59.510] (--) Depth 24 pixmap format is 32 bpp [ 59.510] (II) NVIDIA: Using 768.00 MB of virtual memory for indirect memory access. [ 59.514] (II) NVIDIA(0): Setting mode "DFP-1:1024x768"
Comparing the same monitor with mga2 [ 34.716] (WW) NVIDIA(0): The EDID for LG Electronics W2361 (DFP-1) contradicts itself: [ 34.716] (WW) NVIDIA(0): mode "1024x768" is specified in the EDID; however, the [ 34.716] (WW) NVIDIA(0): EDID's valid VertRefresh range (56.000-61.000 Hz) would [ 34.716] (WW) NVIDIA(0): exclude this mode's VertRefresh (75.0 Hz); ignoring [ 34.716] (WW) NVIDIA(0): VertRefresh check for mode "1024x768". [ 34.762] (II) NVIDIA(0): Assigned Display Device: DFP-1 [ 34.762] (==) NVIDIA(0): [ 34.762] (==) NVIDIA(0): No modes were requested; the default mode "nvidia-auto-select" [ 34.762] (==) NVIDIA(0): will be used as the requested mode. [ 34.762] (==) NVIDIA(0): [ 34.762] (II) NVIDIA(0): Validated modes: [ 34.762] (II) NVIDIA(0): "nvidia-auto-select" [ 34.762] (II) NVIDIA(0): Virtual screen size determined to be 1920 x 1080 [ 34.789] (--) NVIDIA(0): DPI set to (95, 94); computed from "UseEdidDpi" X config [ 34.789] (--) NVIDIA(0): option [ 34.789] (**) NVIDIA(0): Enabling 32-bit ARGB GLX visuals. [ 34.789] (--) Depth 24 pixmap format is 32 bpp [ 34.789] (II) NVIDIA: Using 768.00 MB of virtual memory for indirect memory access. [ 34.795] (II) NVIDIA(0): Setting mode "nvidia-auto-select" [ 34.832] (II) Loading extension NV-GLX
valid 3beta4
Whiteboard: 3beta3 => 3beta3 3beta4
valid 3RC HDMI monitor at 1024x768 looks bad and a regression from mga2. Setting release blocker as it affects live isos. I'll test with classic also.
Priority: Normal => release_blockerWhiteboard: 3beta3 3beta4 => 3beta3 3beta4 3RC
See also bug 9934
See Also: (none) => https://bugs.mageia.org/show_bug.cgi?id=9934
@claire: what resolutions are listed in the monitor section? and what happens when you put in 1920x1080 ?
CC: (none) => alien
I'll attach some screenshots. When first booted MCC reports monitor and resolution at 1024 x 768. If I manually select Monitor to be Flat Panel 1920 x 1080 the resolution still doesn't offer it, only up to 1400 x 1050 or Automatic. Selecting Automatic and hitting Test does appear to choose the correct reolution but when it returns from the test it asks to reboot which is no good on a Live DVD. Using KRandRTray instead, it lists the Monitor as HDMI-0 at 1024 x 768 @ 75Hz but in the Size's it offers 1920 x 1080 is listed as Auto. Selecting that and applying it does set the resolution to 1920x1080 at 59.9Hz. After doing though the wallpaper appears stretched.
Created attachment 3868 [details] mcc when first booted
Created attachment 3869 [details] mcc offered monitor sizes
Created attachment 3870 [details] mcc offered resolutions after setting screen to Flat Panel 1920 x 1080
Created attachment 3871 [details] mcc showing 1024 x 768 still selected after setting monitor to 1920 x 1080
Created attachment 3872 [details] krandrtray showing 1024x768 but 1920x1080 (Auto)
Created attachment 3873 [details] krandrtray after applying 1920x1080 (Auto)
Created attachment 3874 [details] shows the stretched wallpaper when done (shrunk for bugzilla)
what about the /etc/X11/xorg.conf file? (immediately after boot)
see comment 2
no, the *conf* file, not the log file plz
Created attachment 3875 [details] xorg.conf just after boot oops, sorry.
ok, it looks like it's the same issue as my own one: you can see: Modes "1024x768" "832x624" "800x600" "640x480" "480x360" "320x240" in the config, add "1920x1080" to all 4 entries and relogout/relogin or restart prefdm service. if that fixes it, then this might be a commit by Anssi that tv referred to in that other bug report.
Confirmed that adding "1920x1080" to the Modes lines and restarting prefdm fixes this in live mode on kde livedvd. It even then selects the appropriate background wallpaper. Adding Anssi just in case.
CC: (none) => anssi.hannula
i can confirm that this is the same issue that i had. thierry alluded that a (recent) commit of Anssi in harddrake might be responsible. i think that at boot time, before X gets started, harddrake tries to determine the max resolution of monitor and does so wrongly, even though EDID should be working fine for card and monitor. perhaps it doesn't take into account if the primary one is unused and HDMI is used instead?
CC: (none) => thierry.vignaudSeverity: normal => critical
Please provide the output of: monitor-edid -v IMHO we really should not try to probe monitor resolutions anymore and just rely on the autodetection, but it is a bit too late for mga3... Maybe for mga4.
This is in mga2, i'll repeat in mga3 to see if it is any different.. # monitor-edid -v probing EDID using VBE (port 0) VBE version: 3.0, oem version = 112.22 Memory: 14336k OEM name: NVIDIA Vendor name: NVIDIA Corporation Product name: BIOS-P/N@N5601 Product revision: Chip Rev Port 0: DDC1 not supported DDC2 not supported Screen not blanked during data transfer Time to transfer one EDID block: 0 sec (rounded up) DDC (0x4f15:00): DDC not supported, not continuing probing EDID using VBE (port 1) Port 1: DDC1 not supported DDC2 not supported Screen not blanked during data transfer Time to transfer one EDID block: 0 sec (rounded up) DDC (0x4f15:00): DDC not supported, not continuing probing EDID using VBE (port 2) Port 2: DDC1 not supported DDC2 supported Screen not blanked during data transfer Time to transfer one EDID block: 1 sec (rounded up) EDID: Error (0x4f15:01): 0x014f
It's identical mga3
Try this as well: monitor-edid -v --try-in-console And post "lspcidrake -v" output as well. Both preferably in mga3.
Gah! I just restarted MrB's VM. If it helps, this is mga2. If not i'll try mga3 again in the morning. # monitor-edid -v --try-in-console probing EDID using VBE (port 0) VBE version: 3.0, oem version = 112.22 Memory: 14336k OEM name: NVIDIA Vendor name: NVIDIA Corporation Product name: BIOS-P/N@N5601 Product revision: Chip Rev Port 0: DDC1 not supported DDC2 not supported Screen not blanked during data transfer Time to transfer one EDID block: 0 sec (rounded up) DDC (0x4f15:00): DDC not supported, not continuing probing EDID using VBE (port 1) Port 1: DDC1 not supported DDC2 not supported Screen not blanked during data transfer Time to transfer one EDID block: 0 sec (rounded up) DDC (0x4f15:00): DDC not supported, not continuing probing EDID using VBE (port 2) Port 2: DDC1 not supported DDC2 supported Screen not blanked during data transfer Time to transfer one EDID block: 1 sec (rounded up) EDID: Error (0x4f15:01): 0x014f
Created attachment 3880 [details] lspcidrake.txt again from mga2
Can you switch to a text console (ctrl+alt+f2) and then run: monitor-edid -v mga2 is ok.
Couldn't seem to get it to save it to a file in a console so took a pic.. https://dl.dropboxusercontent.com/u/4147101/mga3rc/monitor-edid-tty2.jpg
Thanks, try this in text console: monitor-edid --max-vbe-port 15 -v mga2 is ok. You can use this to store output to file.txt: monitor-edid --max-vbe-port 15 -v &> file.txt
from me if it makes a difference: https://gist.github.com/5513742
or with "monitor-edid --max-vbe-port 15 -v" https://gist.github.com/anonymous/5513865
for reference, i had Modes "640x480" "480x360" "320x240" in the my config file after the cauldron upgrade (+- 2months)
I have committed a fix for this in drakx-kbd-mouse-x11 (i.e. if monitor detection fails, do not assume 1024x768), and soon-to-be-submitted 0.108 should fix this. However, I'd still like the details asked for in comment #31, in case there is another issue.
Created attachment 3882 [details] monitor-edid --max-vbe-port 15 -v &> file.txt Thanks Anssi, please find file.txt attached.
Created attachment 3883 [details] EDID test script for Claire's system Thanks. The output is nice (the probing succeeds), though a bit unexpected. To confirm it was not a fluke, please run the attached script in a text console and provide the output: sh edid-test-claire.sh &> edid-test.log
Created attachment 3884 [details] edid-test-claire.sh &> edid-test.log from tty2
OK, thanks. I opened bug #9976 about the failing probe issue, which I think is not high-priority and safe to fix enough for mga3. This bug, though, should be fixed by relying on X server automatics as per comment #35. Please confirm this on mga3 by removing or backing up /etc/X11/xorg.conf, and then running "XFdrake --auto", and then (re)starting the X server (or reboot). The display resolution should now be correct.
I installed from the livedvd onto hdd so it's easier to test. Wasn't sure if it needed to be but I did the rest in tty2. Deleted /etc/X11/xorg.conf and ran XFdrake --auto, which blanked the screen for a few seconds. After rebooting the resolution is still the same, 1024x768@75Hz.
Did you install the drakx-kbd-mouse-x11-0.108-1.mga3 uploaded today?
Ahh, no, oops. Repeating the test :)
Fixed \o/ Thanks for your work on this Anssi. I think it's something which could potentially have affected alot of people these days and would give a bad first impression.
Now running at 1920x1080@59.9Hz
Great, closing as fixed :)
Status: NEW => RESOLVEDResolution: (none) => FIXED