Bug 199853 - nouveau cannot find monitor attached to DP
Summary: nouveau cannot find monitor attached to DP
Status: NEW
Alias: None
Product: Drivers
Classification: Unclassified
Component: Console/Framebuffers (show other bugs)
Hardware: x86-64 Linux
: P1 normal
Assignee: James Simmons
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2018-05-27 15:23 UTC by Alexandru N. Barloiu
Modified: 2018-07-27 14:35 UTC (History)
0 users

See Also:
Kernel Version: 4.12.11
Subsystem:
Regression: No
Bisected commit-id:


Attachments
lspci dump for the board (1.76 KB, text/plain)
2018-05-27 15:23 UTC, Alexandru N. Barloiu
Details
dmesg before the patch (3.38 KB, text/plain)
2018-05-27 15:38 UTC, Alexandru N. Barloiu
Details
dmesg after the patch (3.39 KB, text/plain)
2018-05-27 15:40 UTC, Alexandru N. Barloiu
Details
dmesg before the patch WITH debug on (3.88 MB, text/plain)
2018-05-27 16:12 UTC, Alexandru N. Barloiu
Details
dmesg after the patch WITH debug on (306.50 KB, text/plain)
2018-05-27 16:14 UTC, Alexandru N. Barloiu
Details
dmesg results for 4.18.0-rc5 (67.16 KB, text/plain)
2018-07-27 14:34 UTC, Alexandru N. Barloiu
Details
Xorg.log complaining of the same thing as the kernel (28.55 KB, text/plain)
2018-07-27 14:35 UTC, Alexandru N. Barloiu
Details

Description Alexandru N. Barloiu 2018-05-27 15:23:52 UTC
Created attachment 276223 [details]
lspci dump for the board

Hello. 

The system in question has a Gigabyte 960 GTX board with 4Gb of ram, detected as GM206. The card works fine with nouveau drivers (with only one monitor) with all kernels below 4.12.10 (including 4.9.103/4.10.17/4.11.12). During the 4.12 line it broke with this commit:

8f5d65ad997efdded1ac08eeb0ca3f6190ff5f7 is the first bad commit
commit 38f5d65ad997efdded1ac08eeb0ca3f6190ff5f7
Author: Ben Skeggs <bskeggs@redhat.com>
Date:   Wed Jul 19 16:49:59 2017 +1000

    drm/nouveau/i2c/gf119-: add support for address-only transactions
    
    commit 13a86519202c5d119d83640d6f781f3181205d2c upstream.
    
    Since switching the I2C-over-AUX helpers, there have been regressions on
    some display combinations due to us not having support for "address only"
    transactions.
    
    This commits enables support for them for GF119 and newer.
    
    Earlier GPUs have been reverted to a custom I2C-over-AUX algorithm.
    
    Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
    Cc: Ilia Mirkin <imirkin@alum.mit.edu>
    Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

:040000 040000 4540f172f019291963bd1aea3e2b0ece64698edb 51b8e7f3995b243cce54cfd8643f28dead604a9d M	drivers

This particular commit this is what dmesg is reporting about nouveau:

May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: NVIDIA GM206 (126010a1)
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: bios: version 84.06.14.00.fe
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: disp: dcb 15 type 8 unknown
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: fb: 4096 MiB GDDR5
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: bus: MMIO write of 8000014a FAULT at 10eb14 [ IBUS ]
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: VRAM: 4096 MiB
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: GART: 1048576 MiB
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: TMDS table version 2.0
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB version 4.1
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 00: 01000f02 00020030
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 01: 02000f00 00000000
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 02: 02811f76 04400020
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 03: 02011f72 00020020
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 04: 04822f86 04400010
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 05: 04022f82 00020010
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 06: 04833f96 04400020
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 07: 04033f92 00020020
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 08: 02044f62 00020010
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB outp 15: 01df5ff8 00000000
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB conn 00: 00001030
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB conn 01: 00020146
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB conn 02: 01000246
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB conn 03: 02000346
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB conn 04: 00010461
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: DCB conn 05: 00000570
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: Pointer to flat panel table invalid
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: unknown connector type 70
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: failed to create encoder 1/8/0: -19
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: Unknown-1 has no encoders, removing
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: hwmon_device_register() is deprecated. Please convert the driver to use hwmon_device_register_with_info().
May 27 17:48:29 magdalina kernel: nouveau 0000:02:00.0: DRM: MM: using COPY for buffer copies
May 27 17:48:30 magdalina kernel: nouveau 0000:02:00.0: DRM: DDC responded, but no EDID for DP-2
May 27 17:48:30 magdalina kernel: nouveau 0000:02:00.0: DRM: allocated 1024x2304 fb: 0x60000, bo ffff88078c314000
May 27 17:48:30 magdalina kernel: fbcon: nouveaufb (fb0) is primary device
May 27 17:48:30 magdalina kernel: nouveau 0000:02:00.0: DRM: DDC responded, but no EDID for DP-2
May 27 17:48:30 magdalina kernel: nouveau 0000:02:00.0: DRM: DDC responded, but no EDID for DP-2
May 27 17:48:30 magdalina kernel: nouveau 0000:02:00.0: fb0: nouveaufb frame buffer device
May 27 17:48:30 magdalina kernel: [drm] Initialized nouveau 1.3.1 20120801 for 0000:02:00.0 on minor 0

The same error is present in ALL kernels above 4.12.11, including in lines 4.13, 4.14, 4.15 and 4.16. 

I hope I didn't forget anything.
Comment 1 Alexandru N. Barloiu 2018-05-27 15:38:23 UTC
Created attachment 276225 [details]
dmesg before the patch

this is what dmesg is reporting about nouveau before the patch.
Comment 2 Alexandru N. Barloiu 2018-05-27 15:40:18 UTC
Created attachment 276227 [details]
dmesg after the patch

this is what dmesg is reporting after the patch. basically it stops detecting the digital ports. tried different ports, different monitors. before this patch, all ports and all monitors work just fine. just as long as you don't connect more than one.
Comment 3 Alexandru N. Barloiu 2018-05-27 16:12:08 UTC
Created attachment 276231 [details]
dmesg before the patch WITH debug on

I also went ahead and compiled a stock vanilla version of 4.12.10 with debug set to 5 on module nouveau. So this is the log for the working 4.12.10.
Comment 4 Alexandru N. Barloiu 2018-05-27 16:14:14 UTC
Created attachment 276233 [details]
dmesg after the patch WITH debug on

This is the output for stock vanilla kernel 4.12.11, with nouveau debug turned to 5. Don't know how much these will help. I have complete 24/7 access to this machine and would be happy to provide and further needed information.
Comment 5 Alexandru N. Barloiu 2018-07-27 14:34:30 UTC
Created attachment 277577 [details]
dmesg results for 4.18.0-rc5

after looking a bit on the internet for this specific commit, i can see  most people say it was fixed for them by kernel 4.13. for me the problem persists in the exact same manner. I'll attach a log from X as well.
Comment 6 Alexandru N. Barloiu 2018-07-27 14:35:51 UTC
Created attachment 277579 [details]
Xorg.log complaining of the same thing as the kernel

I thought this might help.

Note You need to log in before you can comment on or make changes to this bug.