search

rockowitz / ddcutil

Control monitor settings using DDC/CI and USB
http://www.ddcutil.com
GNU General Public License v2.0
948 stars 39 forks source link

Ddcutil broken on 2 monitors after 1.4.0 release #293

Closed Hekzory closed 1 year ago

Hekzory commented 1 year ago

My setup: 3 monitors, before the update i could control monitor brightness on any of them. Now it works only on the primary.

I'm running Wayland+Kde+Nvidia Propietary drivers (which is known to be a bad combo, but worked on a previous version of ddcutil, right?)

Here's the output of ddcutil detect -v on 1.3.0: Even though it says something about errors at first few lines, it actually works.

❯ ddcutil detect --show-table -v
Unable to open directory /sys/bus/i2c/devices/i2c--1: No such file or directory
Device /dev/i2c-255 does not exist. Error = ENOENT(2): No such file or directory
/sys/bus/i2c buses without /dev/i2c-N devices: /sys/bus/i2c/devices/i2c-255
Driver i2c_dev must be loaded or builtin
See https://www.ddcutil.com/kernel_module
Display 1
   I2C bus:  /dev/i2c-3
      DRM connector:                      card0-HDMI-A-1
      Driver:                             nvidia
      I2C address 0x50 (EDID) responsive: true 
      Is eDP device:                      false
      Is LVDS device:                     false
      /sys/bus/i2c/devices/i2c-3/name     NVIDIA i2c adapter 5 at 1:00.0
      PCI device path:                    /sys/devices/pci0000:00/0000:00:01.1/0000:01:00.0/i2c-3
   EDID synopsis:
      Mfg id:               GSM - Goldstar Company Ltd (LG)
      Model:                LG IPS FULLHD
      Product code:         23224  (0x5ab8)
      Serial number:        
      Binary serial number: 16843009 (0x01010101)
      Manufacture year:     2014,  Week: 1
      EDID version:         1.3
      Extra descriptor:        
      Video input definition:    0x80 - Digital Input
      Supported features:
         DPMS standby
         DPMS suspend
         DPMS active-off
         Digital display type: RGB 4:4:4
         Standard sRGB color space: False
      White x,y:        0.313, 0.329
      Red   x,y:        0.645, 0.335
      Green x,y:        0.305, 0.630
      Blue  x,y:        0.148, 0.050
      Extension blocks: 1
   EDID source: I2C
   EDID hex dump:
              +0          +4          +8          +c            0   4   8   c   
      +0000   00 ff ff ff ff ff ff 00 1e 6d b8 5a 01 01 01 01   .........m.Z....
      +0010   01 18 01 03 80 30 1b 78 ea 31 35 a5 55 4e a1 26   .....0.x.15.UN.&
      +0020   0c 50 54 a5 4b 00 71 4f 81 80 95 00 b3 00 a9 c0   .PT.K.qO........
      +0030   81 00 81 c0 90 40 02 3a 80 18 71 38 2d 40 58 2c   .....@.:..q8-@X,
      +0040   45 00 e0 0e 11 00 00 1e 00 00 00 fd 00 38 4b 1e   E............8K.
      +0050   53 0f 00 0a 20 20 20 20 20 20 00 00 00 fc 00 4c   S...      .....L
      +0060   47 20 49 50 53 20 46 55 4c 4c 48 44 00 00 00 ff   G IPS FULLHD....
      +0070   00 0a 20 20 20 20 20 20 20 20 20 20 20 20 01 64   ..            .d
   VCP version:         2.1
   Controller mfg:      Novatek
   Firmware version:    3.2
   Monitor returns DDC Null Response for unsupported features: false

Display 2
   I2C bus:  /dev/i2c-4
      DRM connector:                      card0-DP-1
      Driver:                             nvidia
      I2C address 0x50 (EDID) responsive: true 
      Is eDP device:                      false
      Is LVDS device:                     false
      /sys/bus/i2c/devices/i2c-4/name     NVIDIA i2c adapter 6 at 1:00.0
      PCI device path:                    /sys/devices/pci0000:00/0000:00:01.1/0000:01:00.0/i2c-4
   EDID synopsis:
      Mfg id:               SAM - Samsung Electric Company
      Model:                S24F350
      Product code:         3360  (0x0d20)
      Serial number:        H4ZN908407
      Binary serial number: 1129793347 (0x43574743)
      Manufacture year:     2020,  Week: 38
      EDID version:         1.3
      Extra descriptor:        
      Video input definition:    0x80 - Digital Input
      Supported features:
         DPMS active-off
         Digital display type: RGB 4:4:4
         Standard sRGB color space: False
      White x,y:        0.313, 0.329
      Red   x,y:        0.650, 0.333
      Green x,y:        0.316, 0.612
      Blue  x,y:        0.152, 0.067
      Extension blocks: 1
   EDID source: I2C
   EDID hex dump:
              +0          +4          +8          +c            0   4   8   c   
      +0000   00 ff ff ff ff ff ff 00 4c 2d 20 0d 43 47 57 43   ........L- .CGWC
      +0010   26 1e 01 03 80 34 1d 78 2a 93 15 a6 55 51 9c 27   &....4.x*...UQ.'
      +0020   11 50 54 bb 8c 00 b3 00 81 c0 81 00 81 80 95 00   .PT.............
      +0030   a9 c0 01 01 01 01 02 3a 80 18 71 38 2d 40 58 2c   .......:..q8-@X,
      +0040   45 00 09 25 21 00 00 1e 00 00 00 fd 00 30 48 1e   E..%!........0H.
      +0050   52 16 00 0a 20 20 20 20 20 20 00 00 00 fc 00 53   R...      .....S
      +0060   32 34 46 33 35 30 0a 20 20 20 20 20 00 00 00 ff   24F350.     ....
      +0070   00 48 34 5a 4e 39 30 38 34 30 37 0a 20 20 01 8a   .H4ZN908407.  ..
   VCP version:         2.0
   Controller mfg:      Mstar
   Firmware version:    0.1
   Monitor returns DDC Null Response for unsupported features: false

Display 3
   I2C bus:  /dev/i2c-5
      DRM connector:                      card0-DP-2
      Driver:                             nvidia
      I2C address 0x50 (EDID) responsive: true 
      Is eDP device:                      false
      Is LVDS device:                     false
      /sys/bus/i2c/devices/i2c-5/name     NVIDIA i2c adapter 7 at 1:00.0
      PCI device path:                    /sys/devices/pci0000:00/0000:00:01.1/0000:01:00.0/i2c-5
   EDID synopsis:
      Mfg id:               GSM - Goldstar Company Ltd (LG)
      Model:                LG Ultra HD
      Product code:         23305  (0x5b09)
      Serial number:        
      Binary serial number: 237432 (0x00039f78)
      Manufacture year:     2018,  Week: 9
      EDID version:         1.4
      Extra descriptor:        
      Video input definition:    0xb5 - Digital Input (DisplayPort), Bit depth: 10
      Supported features:
         DPMS standby
         Digital display type: RGB 4:4:4 + YCrCb 4:2:2
         Standard sRGB color space: True
      White x,y:        0.313, 0.329
      Red   x,y:        0.652, 0.335
      Green x,y:        0.305, 0.637
      Blue  x,y:        0.148, 0.062
      Extension blocks: 1
   EDID source: I2C
   EDID hex dump:
              +0          +4          +8          +c            0   4   8   c   
      +0000   00 ff ff ff ff ff ff 00 1e 6d 09 5b 78 9f 03 00   .........m.[x...
      +0010   09 1c 01 04 b5 3c 22 78 9f 30 35 a7 55 4e a3 26   .....<"x.05.UN.&
      +0020   0f 50 54 21 08 00 71 40 81 80 81 c0 a9 c0 d1 c0   .PT!..q@........
      +0030   81 00 01 01 01 01 4d d0 00 a0 f0 70 3e 80 30 20   ......M....p>.0 
      +0040   65 0c 58 54 21 00 00 1a 28 68 00 a0 f0 70 3e 80   e.XT!...(h...p>.
      +0050   08 90 65 0c 58 54 21 00 00 1a 00 00 00 fd 00 28   ..e.XT!........(
      +0060   3d 87 87 38 01 0a 20 20 20 20 20 20 00 00 00 fc   =..8..      ....
      +0070   00 4c 47 20 55 6c 74 72 61 20 48 44 0a 20 01 21   .LG Ultra HD. .!
   VCP version:         2.1
   Controller mfg:      Mstar
   Firmware version:    3.23
   Monitor returns DDC Null Response for unsupported features: false`

Here's the output after switching to 1.4.0: Even though no error lines at start of the output, now actual displays wont response to brightness commands

❯ ddcutil detect --show-table -v
Invalid display
   I2C bus:  /dev/i2c-3
      DRM connector:                      card0-HDMI-A-1
      Driver:                             nvidia
      I2C address 0x50 (EDID) responsive: true 
      Is eDP device:                      false
      Is LVDS device:                     false
      /sys/bus/i2c/devices/i2c-3/name     NVIDIA i2c adapter 5 at 1:00.0
      PCI device path:                    /sys/devices/pci0000:00/0000:00:01.1/0000:01:00.0/i2c-3
   EDID synopsis:
      Mfg id:               GSM - Goldstar Company Ltd (LG)
      Model:                LG IPS FULLHD
      Product code:         23224  (0x5ab8)
      Serial number:        
      Binary serial number: 16843009 (0x01010101)
      Manufacture year:     2014,  Week: 1
      EDID version:         1.3
      Extra descriptor:        
      Video input definition:    0x80 - Digital Input
      Supported features:
         DPMS standby
         DPMS suspend
         DPMS active-off
         Digital display type: RGB 4:4:4 + YCrCb 4:4:4
         Standard sRGB color space: False
      White x,y:        0.313, 0.329
      Red   x,y:        0.645, 0.335
      Green x,y:        0.305, 0.630
      Blue  x,y:        0.148, 0.050
      Extension blocks: 1
   EDID source: I2C
   EDID hex dump:
              +0          +4          +8          +c            0   4   8   c   
      +0000   00 ff ff ff ff ff ff 00 1e 6d b8 5a 01 01 01 01   .........m.Z....
      +0010   01 18 01 03 80 30 1b 78 ea 31 35 a5 55 4e a1 26   .....0.x.15.UN.&
      +0020   0c 50 54 a5 4b 00 71 4f 81 80 95 00 b3 00 a9 c0   .PT.K.qO........
      +0030   81 00 81 c0 90 40 02 3a 80 18 71 38 2d 40 58 2c   .....@.:..q8-@X,
      +0040   45 00 e0 0e 11 00 00 1e 00 00 00 fd 00 38 4b 1e   E............8K.
      +0050   53 0f 00 0a 20 20 20 20 20 20 00 00 00 fc 00 4c   S...      .....L
      +0060   47 20 49 50 53 20 46 55 4c 4c 48 44 00 00 00 ff   G IPS FULLHD....
      +0070   00 0a 20 20 20 20 20 20 20 20 20 20 20 20 01 64   ..            .d
   DDC communication failed

Invalid display
   I2C bus:  /dev/i2c-4
      DRM connector:                      card0-DP-1
      Driver:                             nvidia
      I2C address 0x50 (EDID) responsive: true 
      Is eDP device:                      false
      Is LVDS device:                     false
      /sys/bus/i2c/devices/i2c-4/name     NVIDIA i2c adapter 6 at 1:00.0
      PCI device path:                    /sys/devices/pci0000:00/0000:00:01.1/0000:01:00.0/i2c-4
   EDID synopsis:
      Mfg id:               SAM - Samsung Electric Company
      Model:                S24F350
      Product code:         3360  (0x0d20)
      Serial number:        H4ZN908407
      Binary serial number: 1129793347 (0x43574743)
      Manufacture year:     2020,  Week: 38
      EDID version:         1.3
      Extra descriptor:        
      Video input definition:    0x80 - Digital Input
      Supported features:
         DPMS active-off
         Digital display type: RGB 4:4:4 + YCrCb 4:4:4
         Standard sRGB color space: False
      White x,y:        0.313, 0.329
      Red   x,y:        0.650, 0.333
      Green x,y:        0.316, 0.612
      Blue  x,y:        0.152, 0.067
      Extension blocks: 1
   EDID source: I2C
   EDID hex dump:
              +0          +4          +8          +c            0   4   8   c   
      +0000   00 ff ff ff ff ff ff 00 4c 2d 20 0d 43 47 57 43   ........L- .CGWC
      +0010   26 1e 01 03 80 34 1d 78 2a 93 15 a6 55 51 9c 27   &....4.x*...UQ.'
      +0020   11 50 54 bb 8c 00 b3 00 81 c0 81 00 81 80 95 00   .PT.............
      +0030   a9 c0 01 01 01 01 02 3a 80 18 71 38 2d 40 58 2c   .......:..q8-@X,
      +0040   45 00 09 25 21 00 00 1e 00 00 00 fd 00 30 48 1e   E..%!........0H.
      +0050   52 16 00 0a 20 20 20 20 20 20 00 00 00 fc 00 53   R...      .....S
      +0060   32 34 46 33 35 30 0a 20 20 20 20 20 00 00 00 ff   24F350.     ....
      +0070   00 48 34 5a 4e 39 30 38 34 30 37 0a 20 20 01 8a   .H4ZN908407.  ..
   DDC communication failed

Display 1
   I2C bus:  /dev/i2c-5
      DRM connector:                      card0-DP-2
      Driver:                             nvidia
      I2C address 0x50 (EDID) responsive: true 
      Is eDP device:                      false
      Is LVDS device:                     false
      /sys/bus/i2c/devices/i2c-5/name     NVIDIA i2c adapter 7 at 1:00.0
      PCI device path:                    /sys/devices/pci0000:00/0000:00:01.1/0000:01:00.0/i2c-5
   EDID synopsis:
      Mfg id:               GSM - Goldstar Company Ltd (LG)
      Model:                LG Ultra HD
      Product code:         23305  (0x5b09)
      Serial number:        
      Binary serial number: 237432 (0x00039f78)
      Manufacture year:     2018,  Week: 9
      EDID version:         1.4
      Extra descriptor:        
      Video input definition:    0xb5 - Digital Input (DisplayPort), Bit depth: 10
      Supported features:
         DPMS standby
         Digital display type: RGB 4:4:4 + YCrCb 4:4:4 + YCrCb 4:2:2
         Standard sRGB color space: True
      White x,y:        0.313, 0.329
      Red   x,y:        0.652, 0.335
      Green x,y:        0.305, 0.637
      Blue  x,y:        0.148, 0.062
      Extension blocks: 1
   EDID source: I2C
   EDID hex dump:
              +0          +4          +8          +c            0   4   8   c   
      +0000   00 ff ff ff ff ff ff 00 1e 6d 09 5b 78 9f 03 00   .........m.[x...
      +0010   09 1c 01 04 b5 3c 22 78 9f 30 35 a7 55 4e a3 26   .....<"x.05.UN.&
      +0020   0f 50 54 21 08 00 71 40 81 80 81 c0 a9 c0 d1 c0   .PT!..q@........
      +0030   81 00 01 01 01 01 4d d0 00 a0 f0 70 3e 80 30 20   ......M....p>.0 
      +0040   65 0c 58 54 21 00 00 1a 28 68 00 a0 f0 70 3e 80   e.XT!...(h...p>.
      +0050   08 90 65 0c 58 54 21 00 00 1a 00 00 00 fd 00 28   ..e.XT!........(
      +0060   3d 87 87 38 01 0a 20 20 20 20 20 20 00 00 00 fc   =..8..      ....
      +0070   00 4c 47 20 55 6c 74 72 61 20 48 44 0a 20 01 21   .LG Ultra HD. .!
   VCP version:         2.1
Extended delay as recovery from DDC Null Response...
   Controller mfg:      Mstar
Extended delay as recovery from DDC Null Response...
   Firmware version:    3.23
   Monitor returns DDC Null Response for unsupported features: false

When trying to change brightness on previously working displays it either says

Display not found

or

VCP (aka MCCS) version for display is undetected or less than 2.0. Output may not be accurate.
Setting value failed for feature 10, rc=DDCRC_RETRIES(-3007): maximum retries exceeded
    Try errors: DDCRC_RETRIES

I hope that info helps, I am ready to provide more if you need something else, thanks in advance!

rockowitz commented 1 year ago

First, please run ddcutil interrogate and send the output. That preempts lots of back and forth on basic questions.

The most obvious suspect is an error in the new code to workaround a bug in the way the nvidia driver is coded. Some options to experiment with: --use-file-io, --use-ioctl-io. Any effect?

The fact that communication with the LG monitor succeeds, but with error recovery, suggests that the problem is happening at the I2C level. Let's look at the communication layer in detail. Execute ddcutil detect --verbose --ddc --trcfile i2c_strategy_dispatcher.c --trcfile i2c_execute.c. Does using option --sleep-multiplier with a value greater than 1, e.g. --sleep-multiplier 2, have any effect?

Are you using the Special Nvidia Driver Settings?

@fionnb has reported what appears to be a similar regression as an additional comment on issue #273.

Please submit long output as files of some sort, e.g. as attachments, instead of inline.

Thank you for your help in diagnosing this situation. I'd like to quickly resolve the problem. The cutoffs for including ddcutil release 1.4 in the upcoming Debian release (bookworm) and in Ubuntu 23.04 are less than a month away.

fionnb commented 1 year ago

Quick update from my side: Problem is with an LG monitor too. But X11 here and vanilla intel-mesa. So probably not an issue with the GFX drivers. Will gather debug data tomorrow and send another update.

Hekzory commented 1 year ago

OK, so here are the outputs: Quickly looking at them, it looks like appending --sleep-multiplier 2 for the last command actually seems to improve the situation, resulting in "Invalid display" header to disappear for them and become just "Display 1", "Display 2", "Display 3", after that i tried ddcutil setvcp 10 for all 3 monitors with --sleep-multiplier 2 and all of them worked. However, is there an option to change that setting by default in config or as an environment variable?

interrogate.log detect.log detect_with_sleep_2.log

rockowitz commented 1 year ago

What I'm seeing is corrupt DDC data packets, which is resolved by longer waits between I2C read and write operations. Why this effect only appears with release 1.4.0 is unclear. I'll likely get back to you with additional tests.

A default value for --sleep-multiplier can be set using a Configuration File.

I suggest experimenting to see if a sleep multiplier value less than 2, e.g. 1.5, works for you. Options --ddc and --stats can be helpful in that they'll show the extent that errors are recovered using retry.

Hekzory commented 1 year ago

Well, after playing with parameters I got that it's pretty unstable if sleep value is <=0.3, so I chose 0.5, which is quite stable for me(no errors after many retries). Changing config worked.

However, sadly, changing config resolved issue with cli only, but not with powerdevil-ddcutil, which uses ddcutil to help with automating lowering brightness when AFK on KDE. I think powerdevil ignores/overrides config parameters, so I'll stay on 1.3.0 for now

rockowitz commented 1 year ago

@copysiper I'm a bit confused here. Your original post indicated that ddcutil failed on release 1.4.0, and it turned out that increasing sleep time using --sleep-multiplier 2 resolved the problem. There's no indication that you by default were using a sleep multiplier value other than 1. If a value of 2 succeeds but 1 fails, it makes no sense for .5 to work. Can you clarify?

Re powerdevil, it uses shared library libddcutil, not command line program ddcutil. To pass options to the shared library, use the [libddcutil] section of configuration file ddcutilrc.

Hekzory commented 1 year ago

Actually, for me it behaves like if --sleep-multiplier 0 is set by default if nothing specified in arguments or config, so when i add --sleep-multiplier 1 in config or arguments it starts working too

rockowitz commented 1 year ago

The default sleep-multiplier value was being set to 0, not 1, in certain circumstances. This has been fixed in branch 1.4.1-dev. Please build from that branch and verify that both command line ddcutil and shared library libddcutil (called by PowerDevil) are both working properly without explicit --sleep-multiplier. (Depending on your configuration, you may have to perform some additional steps to ensure that the new version of libddcutil is used. There should be messages in the system log indicating which version of libddcutil is being used.)

Hekzory commented 1 year ago

Ok, so i cleared the config and updated my package to 1.4.1 (https://github.com/rockowitz/ddcutil/commit/0dec3a4e9714e2f8ee4b93ed629c00ea10979d82), ddcutil cli definitely works, sleep multiplier now behaves like if it is 1 by default, no errors appear when changing brightness on any display. powerdevil is able to manage brightness correctly too, here is the log of the correct version of the lib being initialized:

libddcutil[3749]: Initializing.  ddcutil version 1.4.1
libddcutil[3749]: Library initialization complete.

It looks like the issue is resolved now, thank you for such fast support!

hesdago commented 1 year ago

Hello, and thanks for supporting on this. I'm seeing similar issues after upgrading from 1.3 to 1.4, and the mentioned parameter for sleep-multiplier does help: --sleep-multiplier 2

➜ dago@dago-pc-amd64  ~  sudo ddcutil detect
Display 1
   I2C bus:  /dev/i2c-1
   DRM connector:           card0-HDMI-A-1
   EDID synopsis:
      Mfg id:               DEL - Dell Inc.
      Model:                DELL U2715H
      Product code:         53351  (0xd067)
      Serial number:        GH85D7CM0T8S
      Binary serial number: 810825811 (0x30543853)
      Manufacture year:     2017,  Week: 51
   VCP version:         Detection failed

Display 2
   I2C bus:  /dev/i2c-3
   DRM connector:           card0-DP-1
   EDID synopsis:
      Mfg id:               DEL - Dell Inc.
      Model:                DELL U2715H
      Product code:         53349  (0xd065)
      Serial number:        GH85D81801PL
      Binary serial number: 808538188 (0x3031504c)
      Manufacture year:     2018,  Week: 2
   VCP version:         2.1

➜ dago@dago-pc-amd64  ~  sudo ddcutil detect --sleep-multiplier 2
Display 1
   I2C bus:  /dev/i2c-1
   DRM connector:           card0-HDMI-A-1
   EDID synopsis:
      Mfg id:               DEL - Dell Inc.
      Model:                DELL U2715H
      Product code:         53351  (0xd067)
      Serial number:        GH85D7CM0T8S
      Binary serial number: 810825811 (0x30543853)
      Manufacture year:     2017,  Week: 51
   VCP version:         2.1

Display 2
   I2C bus:  /dev/i2c-3
   DRM connector:           card0-DP-1
   EDID synopsis:
      Mfg id:               DEL - Dell Inc.
      Model:                DELL U2715H
      Product code:         53349  (0xd065)
      Serial number:        GH85D81801PL
      Binary serial number: 808538188 (0x3031504c)
      Manufacture year:     2018,  Week: 2
   VCP version:         2.1
fionnb commented 1 year ago

I confirm that the "Verification failed" issue has gone away with 1.4.1-dev