Skip to content
Snippets Groups Projects

Blank internal laptop screen from boot after kernel update 6.3.9->6.4.1

    • Closed Issue created by ¯\_(ツ)_/¯ @beryl

    Internal laptop screen goes blank as soon as systemd boots from the intrd image with 6.4 kernel. Laptop has worked fine with all previous kernels to 6.3.9. Only error hint in the non-debug logs is "Cannot find any crtc or sizes". Laptop does boot into X behind the screen but has to be used blind. No joy connecting external monitors via DP or HDMI. Exact same problem booting into 6.4 with DRM-Tip.
    System Architecture: x86-64
    Kernel Versions: 6.4.1-arch1-1 and 6.4.0-1-drm-tip-git-g52c775d5fac7
    Distro: Archlinux
    Machine: Coffee Lake i7-8750H iGPU 630
    Connector: eDP (and others)
    Debug drm dmesg: http://ix.io/4zZO

    Edited
    To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information

    Child items ...

    • Activity

    • ¯\_(ツ)_/¯ changed the description

      changed the description

    • Suresh added Community label

      added Community label

    • Suresh added platform: CFL label

      added platform: CFL label

    • Suresh
      Suresh @suresh_kurmi ·
      Reporter

      @beryl Please attach the detailed i915 logs in a file : drm.debug=0x1e log_buf_len=4M https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs Cc : @vsyrjala @armurthy @jani

      Edited by Suresh
    • Ville Syrjälä
      Ville Syrjälä @vsyrjala ·
      Developer 
      [ 2.596798] i915 0000:00:02.0: [drm:do_detailed_mode] [CONNECTOR:95:eDP-1] Analog composite sync!
[ 2.596811] i915 0000:00:02.0: [drm:do_detailed_mode] [CONNECTOR:95:eDP-1] Analog composite sync!
[ 2.596823] [drm:drm_mode_debug_printmodeline] Modeline "1920x1080": 144 368140 1920 1968 2000 2102 1080 1090 1095 1216 0x48 0x140
[ 2.596827] [drm:drm_mode_prune_invalid] Not using 1920x1080 mode: HSYNC
[ 2.596830] [drm:drm_mode_debug_printmodeline] Modeline "1920x1080": 60 368140 1920 1968 2000 2102 1080 1090 1095 2915 0x40 0x140
[ 2.596833] [drm:drm_mode_prune_invalid] Not using 1920x1080 mode: HSYNC

      Looks like the eDP EDID is garbage. Probably need to quirk it. Please attach a copy of /sys/class/drm/card0-eDP-1/edid

      That doesn't really explain why external displays wouldn't work though...

      • ¯\_(ツ)_/¯
        ¯\_(ツ)_/¯ @beryl ·
        Author

        hexdump -C /sys/class/drm/card0-eDP-1/edid

        00000000  00 ff ff ff ff ff ff 00  06 af ed 80 00 00 00 00  |................|  
00000010  10 1b 01 04 a5 22 13 78  02 6a 75 a4 56 52 9c 27  |.....".x.ju.VR.'|  
00000020  0b 50 54 00 00 00 01 01  01 01 01 01 01 01 01 01  |.PT.............|  
00000030  01 01 01 01 01 01 ce 8f  80 b6 70 38 88 40 30 20  |..........p8.@0 |  
00000040  a5 00 58 c1 10 00 00 00  ce 8f 80 b6 70 38 2b 47  |..X.........p8+G|  
00000050  30 20 a5 00 58 c1 10 00  00 00 00 00 00 fe 00 41  |0 ..X..........A|  
00000060  55 4f 0a 20 20 20 20 20  20 20 20 20 00 00 00 fe  |UO.         ....|  
00000070  00 42 31 35 36 48 41 4e  30 38 2e 30 20 0a 00 17  |.B156HAN08.0 ...|  
00000080

        Does it explain why the EDID is corrupt for the 6.4.1 version of the driver and not 6.3.9 driver?

        Edited by ¯\_(ツ)_/¯
      • Jani Nikula
        Jani Nikula @jani ·
        Owner

        Please attach /sys/class/drm/card0-eDP-1/edid verbatim. It's binary.

        Please also attach full dmesg from boot with drm.debug=0xe for both working and non-working cases. Early part of the dmesg is missing. Please add the attachments here on fdo gitlab.

        It's a bit tedious, but if you can compile your own kernels, the quickest way to pinpoint the problem is probably a git bisect between the working and non-working versions.

      • Jani Nikula
        Jani Nikula @jani ·
        Owner

        A shot in the dark would be to try to revert commit ca62297b2085 ("drm/edid: Fix csync detailed mode parsing") first to see if it makes a difference.

      • Suresh
        Suresh @suresh_kurmi ·
        Reporter

        @beryl Any update if you have tried reverting the commit suggested by Jani.

      • Gerd Richard
        Gerd Richard @Gori ·

        Hi, I'm having the same issue (black screen at boot) with my "MSI GS65 Stealth Thin RE8" notebook. And did some investigations in the opensuse bug tracker: Issue 1213693

        We did try a test kernel 6.5.0-rc3-4.g651715d with the revert of change ca62297b2085 and found, that it is running afterwards.

        So please go ahead with your investigations.

      • Please register or sign in to reply
    • ¯\_(ツ)_/¯ changed the description

      changed the description

    • ¯\_(ツ)_/¯
      ¯\_(ツ)_/¯ @beryl ·
      Author

      for the avoidance of doubt i've been running this laptop hardware as my daily driver with frequent kernel updates for several years with no issues. does the edid change between kernels? isn't the crtc error more concerning?

    • slamhk mentioned in issue #8930 (closed)

      mentioned in issue #8930 (closed)

    • Dan Arena
      Dan Arena @dan39 ·

      I seem to be having a similar or the same issue myself. I am also on a laptop, mine is an MSI GS65 8RE with intel + nvidia gtx 1060 hybrid graphics.

      attached journalctl output of drm-tip with the i915 debug on

      my edid looks the same:

      # hexdump -C /sys/class/drm/card0-eDP-1/edid
00000000  00 ff ff ff ff ff ff 00  06 af ed 80 00 00 00 00  |................|
00000010  10 1b 01 04 a5 22 13 78  02 6a 75 a4 56 52 9c 27  |.....".x.ju.VR.'|
00000020  0b 50 54 00 00 00 01 01  01 01 01 01 01 01 01 01  |.PT.............|
00000030  01 01 01 01 01 01 ce 8f  80 b6 70 38 88 40 30 20  |..........p8.@0 |
00000040  a5 00 58 c1 10 00 00 00  ce 8f 80 b6 70 38 2b 47  |..X.........p8+G|
00000050  30 20 a5 00 58 c1 10 00  00 00 00 00 00 fe 00 41  |0 ..X..........A|
00000060  55 4f 0a 20 20 20 20 20  20 20 20 20 00 00 00 fe  |UO.         ....|
00000070  00 42 31 35 36 48 41 4e  30 38 2e 30 20 0a 00 17  |.B156HAN08.0 ...|
00000080

      I also see "kernel: i915 0000:00:02.0: [drm] Cannot find any crtc or sizes" when the issue occurs.

      I am currently switched to the 6.1.38-2-lts kernel to get a bootable system. I am compiling the drm-tip now. I will also try reverting commit ca62297b2085 after that.

      Even on this LTS kernel though, I see lots of these errors in dmesg, although everything seems to work fine:

      [  713.990567] i915 0000:00:02.0: [drm] *ERROR* Atomic update failure on pipe A (start=102470 end=102471) time 1028 us, min 1062, max 1079, scanline start 987, end 1142
[  932.989922] i915 0000:00:02.0: [drm] *ERROR* Atomic update failure on pipe A (start=134012 end=134013) time 241 us, min 1062, max 1079, scanline start 1043, end 1085

      i915-debug.txt

      Edited by Dan Arena
    • Dan Arena
      Dan Arena @dan39 ·

      Good news, reverting commit ca62297b2085 did indeed fix the issue for me.

      Attaching updated dmesg debug log with the commit reverted.

      fixed-i915-debug-dmesg.txt

      Whatever those atomic update failures are, I think they might be unrelated, but they don't cause any harm that I know of.

      Thanks

    • ¯\_(ツ)_/¯
      ¯\_(ツ)_/¯ @beryl ·
      Author

      that's good news - i'm a msi gs65 too. unfort haven't been able to experiment with reverting the commit yet since it's my daily driver... thanks for your efforts @dan39 !

      Edited by ¯\_(ツ)_/¯
    • ¯\_(ツ)_/¯
      ¯\_(ツ)_/¯ @beryl ·
      Author

      @jani your shot in the dark was spot on, i can confirm with @dan39 that using drm-tip with ca62297b2085 reverted gets the laptop back. @suresh_kurmi is the edid a red herring or is it all related?

    • Suresh
      Suresh @suresh_kurmi ·
      Reporter

      @vsyrjala any input on the next steps whether this need a quirk or revert of the above commit ?

    • Abigaliz
      Abigaliz @Abigaliz ·

      Experiencing a similar (maybe the same?) issue with a laptop Asus ROG Strix Hero II. Using Arch Linux, LTS kernel works fine but mainline kernel 6.4.x doesn't recognize eDP monitor. It recognizes it as a generic monitor, and doesn't allow it to go further than 60hz (laptop's monitor is 144hz).

    • Michael Burge
      Michael Burge @acolablab ·

      Greetings,

      I were directed here from https://bugzilla.suse.com/show_bug.cgi?id=1213693

      Summary:

      ASUS GU502GV, Intel UHD Graphics 630, 1920 x 1080 144Hz

      kernel 6.4 reports [drm] Cannot find any crtc or sizes

      Xorg logs show assumed or wrongly interpreted modeset(0): Modelines and requests an unsupported initial mode which causes the screen to turn off:

      [ 16.390] (II) modeset(0): Output eDP-1 using initial mode 15360x8640 +0+0

      Have tried

      • kernel 6.4 .2/.3/.4, 6.5-rc3 -> issue present
      • a custom 6.4.7 kernel, provided by Takashi Iwai, w/ ca62297b2085 reverted -> works
    • Suresh
      Suresh @suresh_kurmi ·
      Reporter

      Cc : @jani @armurthy

    • Jani Nikula
      Jani Nikula @jani ·
      Owner

      Please attach /sys/class/drm/card0-eDP-1/edid. It's a binary file, please attach as binary.

      Possible dupes #8789 (closed) #8930 (closed) #9044 (closed)

    • Jani Nikula mentioned in issue #9044 (closed)

      mentioned in issue #9044 (closed)

    • Jani Nikula added feature: display/eDP label

      added feature: display/eDP label

    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    Please register or sign in to reply