*** DONE Resolved Issue: Lenovo x13 2-in-1 With Wrong 34" TFT Screen Resolution :issues:blog:
CLOSED: [2025-09-14 Sun 17:13] SCHEDULED: <2025-09-14 Sun>
:PROPERTIES:
:CREATED: [2025-09-14 Sun 12:15]
:ID: 2025-09-14-fix-wrong-34-TFT-resolution
:END:
:LOGBOOK:
- State "DONE" from "DONE" [2025-10-05 Sun 17:43]
- State "DONE" from "DONE" [2025-10-04 Sat 13:35]
- State "DONE" from "STARTED" [2025-09-14 Sun 17:13]
:END:
# - [2025-09-14 Sun 12:15] ich dachte, ich hätte schon irgendwo einen Task dazu, finde ihn aber nicht.
- Updates
- 2025-10-04:
- Test If Issue Relates to HW or SW → issue sticks to SSD when switched → *it's a software bug*
- done: gdctl comparison
- done: Compare BIOS settings
- done: Compare GRUB settings
- still: *different "Capabilities"* listed by =lspci -v= are probably the hottest candidate to the solution.
- 2025-10-05: found and solve the issue
Problem description: Lenovo x13 2-in-1 Gen 5 with Debian 13 GNOME 48:
I can't select the native screen resolution for the external 34" TFT
which is 2560x1080. Hostname = Cosmo.
In the GNOME screen configuration dialog, I may only choose a
resolution (among others) that is similar but not correct: 2560x1440
2025-10-05: TL;DR: For some particular package, I needed
testing/unstable Debian sources on Jackson and pinned this package to
testing or unstable. *Somehow, Jackson took all packages from Debian
unstable*. So I got a newer kernel than Debian 13 stable/trixie on
Jackson without realizing but not on Cosmo. It seems to be the case
that the stable kernel from Debian 13 is not able to properly
recognize the external TFT. By installing the newer kernel to Cosmo,
*the external TFT now works with its native resolution*. I still need
to figure out how to define the unstable package exception for the
kernel only. This could have been much easier when I would have
checked the kernel version right away.
**** DONE Testing GRUB parameter
CLOSED: [2025-10-04 Sat 14:20]
:PROPERTIES:
:END:
:LOGBOOK:
- State "DONE" from "STARTED" [2025-10-04 Sat 14:20]
- State "STARTED" from "DONE" [2025-09-17 Wed 18:35]
- State "DONE" from [2025-09-17 Wed 18:11]
:END:
I modified =/tmp/etc/default/grub= and added a video-parameter that did not change anythig:
: GRUB_CMDLINE_LINUX_DEFAULT="quiet video=eDP-1:2560x1080@30"
**** Testing Another, Almost Identical Device
Testing a different Lenovo x13 2-in-1 Gen 5 (hostname = Jackson; same generation,
different battery, i7 CPU instead of i5) with the same OS + GNOME:
display resolution is detected and selected correctly!
- [X] do a firmware-update
- did install but did not fix the issue.
- Firmware is now identical on both devices.
***** Compare output of =lspci -v=
Cosmo: (wrong resolution)
#+BEGIN_EXAMPLE
00:02.0 VGA compatible controller: Intel Corporation Meteor Lake-P [Intel Graphics] (rev 08) (prog-if 00 [VGA controller])
Subsystem: Lenovo Device 232a
Flags: bus master, fast devsel, latency 0, IRQ 192, IOMMU group 0
Memory at 4058000000 (64-bit, prefetchable) [size=16M]
Memory at 4000000000 (64-bit, prefetchable) [size=256M]
Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
Capabilities: [40] Vendor Specific Information: Len=0c
Capabilities: [70] Express Root Complex Integrated Endpoint, IntMsgNum 0
Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable+ 64bit+
Capabilities: [d0] Power Management version 3
Capabilities: [100] Null
Capabilities: [110] Process Address Space ID (PASID)
Capabilities: [200] Address Translation Service (ATS)
Capabilities: [420] Physical Resizable BAR
Capabilities: [320] Single Root I/O Virtualization (SR-IOV)
Capabilities: [400] Latency Tolerance Reporting
Kernel driver in use: i915
Kernel modules: i915, xe
#+END_EXAMPLE
jackson: (resolution OK)
#+BEGIN_EXAMPLE
00:02.0 VGA compatible controller: Intel Corporation Meteor Lake-P [Intel Graphics] (rev 08) (prog-if 00 [VGA controller])
Subsystem: Lenovo Device 232a
Flags: bus master, fast devsel, latency 0, IRQ 192, IOMMU group 0
Memory at 4058000000 (64-bit, prefetchable) [size=16M]
Memory at 4000000000 (64-bit, prefetchable) [size=256M]
Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
Capabilities: [40] Vendor Specific Information: Intel Capabilities v1
CapA: Peg60Dis- Peg12Dis- Peg11Dis- Peg10Dis- PeLWUDis- DmiWidth=x4
EccDis- ForceEccEn- VTdDis- DmiG2Dis- PegG2Dis- DDRMaxSize=Unlimited
1NDis- CDDis- DDPCDis- X2APICEn- PDCDis- IGDis- CDID=0 CRID=0
DDROCCAP+ OCEn- DDRWrtVrefEn+ DDR3LEn+
CapB: ImguDis- OCbySSKUCap- OCbySSKUEn- SMTCap- CacheSzCap 0x0
SoftBinCap- DDR3MaxFreqWithRef100=Disabled PegG3Dis-
PkgTyp- AddGfxEn- AddGfxCap- PegX16Dis- DmiG3Dis- GmmDis-
DDR3MaxFreq=2932MHz LPDDR3En-
Capabilities: [70] Express Root Complex Integrated Endpoint, IntMsgNum 0
Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable+ 64bit+
Capabilities: [d0] Power Management version 3
Capabilities: [100] Null
Capabilities: [110] Process Address Space ID (PASID)
Capabilities: [200] Address Translation Service (ATS)
Capabilities: [420] Physical Resizable BAR
Capabilities: [320] Single Root I/O Virtualization (SR-IOV)
Capabilities: [400] Latency Tolerance Reporting
Kernel driver in use: i915
Kernel modules: i915, xe
#+END_EXAMPLE
So I've got the same graphics hardware, same kernel driver, same version.
However, the difference is:
#+BEGIN_EXAMPLE
non-working host:
Capabilities: [40] Vendor Specific Information: Len=0c
vs. working host:
Capabilities: [40] Vendor Specific Information: Intel Capabilities v1
CapA: Peg60Dis- Peg12Dis- Peg11Dis- Peg10Dis- PeLWUDis- DmiWidth=x4
EccDis- ForceEccEn- VTdDis- DmiG2Dis- PegG2Dis- DDRMaxSize=Unlimited
1NDis- CDDis- DDPCDis- X2APICEn- PDCDis- IGDis- CDID=0 CRID=0
DDROCCAP+ OCEn- DDRWrtVrefEn+ DDR3LEn+
CapB: ImguDis- OCbySSKUCap- OCbySSKUEn- SMTCap- CacheSzCap 0x0
SoftBinCap- DDR3MaxFreqWithRef100=Disabled PegG3Dis-
PkgTyp- AddGfxEn- AddGfxCap- PegX16Dis- DmiG3Dis- GmmDis-
DDR3MaxFreq=2932MHz LPDDR3En-
#+END_EXAMPLE
*This seems to be the most important hint, why those two devices
differ*. However, I have no idea what to do next.
***** STARTED Find Out What This Difference Means and How to Get Cosmo to the Same Capabilities
:PROPERTIES:
:END:
Any idea? Drop me a line.
***** DONE Test If Issue Relates to HW or SW
CLOSED: [2025-10-04 Sat 13:35]
:PROPERTIES:
:END:
:LOGBOOK:
- State "DONE" from "TODO" [2025-10-04 Sat 13:35]
:END:
I switched SSDs between the two x13 devices I've got.
The wrongly identified native screen resolution sticks to the SSD, not
the actual hardware device.
In other words: The issue does not relate to the hardware, *it relates
to the booted operating system and its configuration*.
***** DONE Compare /etc between the two hosts
CLOSED: [2025-09-17 Wed 18:36]
:PROPERTIES:
:END:
:LOGBOOK:
- State "DONE" from "TODO" [2025-09-17 Wed 18:36]
:END:
I did a comparison using "meld" of the two =/etc= hierarchies and did
not find any obvious difference that could explain the situation.
***** DONE Using gdctl
CLOSED: [2025-10-04 Sat 14:28]
:PROPERTIES:
:END:
:LOGBOOK:
- State "DONE" from "STARTED" [2025-10-04 Sat 14:28]
:END:
GNOME 48 has a new tool as an in-place replacement of =(x)randr= named =gdctl=.
Here's the output for cosmo which shows the wrong "Current mode" for HDMI-1:
#+BEGIN_EXAMPLE
gdctl show
Monitors:
├──Monitor HDMI-1 (LG Electronics 34")
│ ├──Vendor: GSM
│ ├──Product: LG ULTRAWIDE
│ ├──Serial: 0x0002de64
│ ├──Current mode
│ │ └──2560x1440@59.951
│ └──Preferences:
│ └──Luminances:
│ ├──default ⇒ 100.0 (default) (current)
│ └──bt2100 ⇒ 100.0 (default)
└──Monitor eDP-1 (Built-in display)
├──Vendor: AUO
├──Product: 0x6693
├──Serial: 0x00000000
├──Current mode
│ └──1920x1200@60.026
└──Preferences:
└──Luminances:
└──default ⇒ 100.0 (default) (current)
Logical monitors:
├──Logical monitor #1
│ ├──Position: (2560, 480)
│ ├──Scale: 1.25
│ ├──Transform: 180
│ ├──Primary: no
│ └──Monitors: (1)
│ └──eDP-1 (Built-in display)
└──Logical monitor #2
├──Position: (0, 0)
├──Scale: 1.0
├──Transform: normal
├──Primary: yes
└──Monitors: (1)
└──HDMI-1 (LG Electronics 34")
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@cosmo ~ % gdctl show -m
Monitors:
├──Monitor HDMI-1 (LG Electronics 34")
│ ├──Vendor: GSM
│ ├──Product: LG ULTRAWIDE
│ ├──Serial: 0x0002de64
│ └──Modes (23)
│ ├──3840x2160@30.000
│ ├──3840x2160@29.970
│ ├──3840x2160@25.000
│ ├──3840x2160@24.000
│ ├──3840x2160@23.976
│ ├──2560x1440@59.951
│ ├──1920x1080@60.000
│ ├──1920x1080@59.940
│ ├──1920x1080@50.000
│ ├──1680x1050@59.883
│ ├──1600x900@60.000
│ ├──1280x1024@75.025
│ ├──1280x1024@60.020
│ ├──1280x800@59.910
│ ├──1280x720@60.000
│ ├──1280x720@59.940
│ ├──1280x720@50.000
│ ├──1024x768@75.029
│ ├──1024x768@60.004
│ ├──832x624@74.551
│ ├──800x600@75.000
│ ├──800x600@60.317
│ └──720x576@50.000
│ └──Preferences:
[...]
#+END_EXAMPLE
- [X] try setting via gdctl:
#+BEGIN_EXAMPLE
user@cosmo ~ % gdctl set -LpM HDMI-1 --mode 2560x1080@60
Failed to create configuration: No mode 2560x1080@60 available for HDMI-1
1 user@cosmo ~ %
#+END_EXAMPLE
→ fails.
- [X] Compare with output of host jackson when connected to same TFT
#+BEGIN_EXAMPLE
user@jackson ~ % gdctl show
Monitors:
├──Monitor HDMI-1 (LG Electronics 34")
│ ├──Vendor: GSM
│ ├──Product: LG ULTRAWIDE
│ ├──Serial: 0x0002de64
│ ├──Current mode
│ │ └──2560x1080@59.978
│ └──Preferences:
│ └──Luminances:
│ ├──default ⇒ 100.0 (default) (current)
│ └──bt2100 ⇒ 100.0 (default)
└──Monitor eDP-1 (Built-in display)
├──Vendor: AUO
├──Product: 0x6693
├──Serial: 0x00000000
├──Current mode
│ └──1920x1200@60.026
└──Preferences:
└──Luminances:
└──default ⇒ 100.0 (default) (current)
Logical monitors:
├──Logical monitor #1
│ ├──Position: (0, 0)
│ ├──Scale: 1.25
│ ├──Transform: normal
│ ├──Primary: yes
│ └──Monitors: (1)
│ └──eDP-1 (Built-in display)
└──Logical monitor #2
├──Position: (1536, 0)
├──Scale: 1.0
├──Transform: normal
├──Primary: no
└──Monitors: (1)
└──HDMI-1 (LG Electronics 34")
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % gdctl show -m
Monitors:
├──Monitor HDMI-1 (LG Electronics 34")
│ ├──Vendor: GSM
│ ├──Product: LG ULTRAWIDE
│ ├──Serial: 0x0002de64
│ └──Modes (28)
│ ├──3840x2160@30.000
│ ├──3840x2160@29.970
│ ├──3840x2160@25.000
│ ├──3840x2160@24.000
│ ├──3840x2160@23.976
│ ├──2560x1440@59.951
│ ├──2560x1080@74.991
│ ├──2560x1080@59.978
│ ├──2560x1080@50.000
│ ├──1920x1080@74.986
│ ├──1920x1080@60.000
│ ├──1920x1080@59.940
│ ├──1920x1080@50.000
│ ├──1680x1050@59.883
│ ├──1600x900@60.000
│ ├──1280x1024@75.025
│ ├──1280x1024@60.020
│ ├──1280x800@59.910
│ ├──1280x720@60.000
│ ├──1280x720@59.940
│ ├──1280x720@50.000
│ ├──1152x864@59.967
│ ├──1024x768@75.029
│ ├──1024x768@60.004
│ ├──832x624@74.551
│ ├──800x600@75.000
│ ├──800x600@60.317
│ └──720x576@50.000
│ └──Preferences:
│ └──Luminances:
│ ├──default ⇒ 100.0 (default) (current)
│ └──bt2100 ⇒ 100.0 (default)
└──Monitor eDP-1 (Built-in display)
[...]
Logical monitors:
├──Logical monitor #1
│ ├──Position: (0, 0)
│ ├──Scale: 1.25
│ ├──Transform: normal
│ ├──Primary: yes
│ └──Monitors: (1)
│ └──eDP-1 (Built-in display)
└──Logical monitor #2
├──Position: (1536, 0)
├──Scale: 1.0
├──Transform: normal
├──Primary: no
└──Monitors: (1)
└──HDMI-1 (LG Electronics 34")
#+END_EXAMPLE
So it seems as if everything is exactly the same except the listed Modes/resolutions.
***** DONE Compare BIOS settings
CLOSED: [2025-10-04 Sat 14:05]
:PROPERTIES:
:END:
:LOGBOOK:
- State "DONE" from "TODO" [2025-10-04 Sat 14:05]
:END:
- Why? https://graz.social/@alienghic@timeloop.cafe/115203647620325001
- uh. something different about the ACPI? Not that I could find how to check or update that.
- Do they both have the same grub settings? There's some settings that can change how the kernel detects hardware.
- It seems like one system is able to detecting more information about the
- Oh did you go into the bios settings and compare them? Maybe something is set slightly differently?
| What? | Working installation | Non-working installation | Notes |
|-----------------------------+----------------------+--------------------------+-----------------------------------------------|
| UEFI BIOS Version | N45ET25W (1.15) | same | still most current as of 2025-10-04 |
| Embedded Controller Version | N45HT20W (1.11) | same | |
| ME Firmware Version | 18.0.5.2155 | *18.0.15.2515* | actually, it's newer on the non-working setup |
| Machine Type Model | 21LWCTO1WW | same | |
- [[https://pcsupport.lenovo.com/at/de/products/laptops-and-netbooks/thinkpad-x-series-laptops/thinkpad-x13-2-in-1-gen-5-type-21lw-21lx/21lw/21lwcto1ww/pf5mcsty/downloads/ds568074-bios-update-utility-bootable-cd-for-windows-11-10-64-bit-thinkpad-x13-2-in-1-gen-5-type-21lw-21lx-x13-gen-5-type-21lu-21lv?category=BIOS][(German) Lenovo page with most current BIOS Update versions]]
- [[https://en.wikipedia.org/wiki/Intel_ME][Intel ME Firmware]] should not cause this issue, I would guess.
Comparing the BIOS settings, I also got same settings for:
- Config: (everything the same)
- Network
- USB
- Keyboard/Mouse
- Display
- Power
- Beep and Alarm
- CPU
- Thunderbolt 4
- Intel AMT
- Security
- (everything the same)
***** DONE Compare GRUB settings
CLOSED: [2025-10-04 Sat 14:18]
:PROPERTIES:
:END:
:LOGBOOK:
- State "DONE" from "TODO" [2025-10-04 Sat 14:18]
:END:
- Why? https://graz.social/@alienghic@timeloop.cafe/115203647620325001
- uh. something different about the ACPI? Not that I could find how to check or update that.
- Do they both have the same grub settings? There's some settings that can change how the kernel detects hardware.
- It seems like one system is able to detecting more information about the
- Oh did you go into the bios settings and compare them? Maybe something is set slightly differently?
The info in =/etc/default/grub= was identical on both machines.
: GRUB_CMDLINE_LINUX_DEFAULT="quiet"
I once tried following adaption on Cosmo:
: GRUB_CMDLINE_LINUX_DEFAULT="quiet video=HDMI-1:2560x1080@60"
Unfornately, this did not help as mentioned above.
**** My Cry For Help
:PROPERTIES:
:END:
- [2025-09-14 Sun 17:41] Mastodon-thread: https://graz.social/@publicvoit/115203409039346508
**** DONE Check Driver
CLOSED: [2025-10-05 Sun 17:50] SCHEDULED: <2025-09-15 Mon>
:PROPERTIES:
:END:
:LOGBOOK:
- State "DONE" from "STARTED" [2025-10-05 Sun 17:50]
:END:
- https://graz.social/@gnomon@mastodon.social/115203465545303106
- since you're using the i915 driver instead of the xe driver per
your dmesg output, what is the value of your 'enable_guc' kernel
parameter, and what are the specific versions of the firmware
blobs at /lib/firmware/i915/mtl_* ?
- I'm not sure that this is necessarily your problem but it's the
first thing I would check after looking at the info you provided.
#+BEGIN_EXAMPLE
user@cosmo ~ % l /lib/firmware/i915/mtl_*
-rw-r--r-- 1 root root 52712 Apr 24 18:35 /lib/firmware/i915/mtl_dmc.bin
-rw-r--r-- 1 root root 1142784 Apr 24 18:35 /lib/firmware/i915/mtl_gsc_1.bin
-rw-r--r-- 1 root root 312064 Apr 24 18:35 /lib/firmware/i915/mtl_guc_70.bin
-rw-r--r-- 1 root root 561152 Apr 24 18:35 /lib/firmware/i915/mtl_huc_gsc.bin
user@cosmo ~ %
#+END_EXAMPLE
... which is identical (file name and size) on jackson, the host where it works.
Not sure how to determine my 'enable_guc' kernel parameter:
#+BEGIN_EXAMPLE
user@cosmo /etc/modprobe.d % l
total 20
drwxr-xr-x 2 root root 4096 Aug 7 19:06 .
drwxr-xr-x 134 root root 12288 Sep 17 15:23 ..
-rw-r--r-- 1 root root 154 May 17 06:35 intel-microcode-blacklist.conf
user@cosmo /etc/modprobe.d % cat intel-microcode-blacklist.conf
# The microcode module attempts to apply a microcode update when
# it autoloads. This is not always safe, so we block it by default.
blacklist microcode
user@cosmo /etc/modprobe.d %
#+END_EXAMPLE
**** DONE Ask Claude.ai
CLOSED: [2025-10-05 Sun 17:50]
:PROPERTIES:
:END:
:LOGBOOK:
- State "DONE" from "STARTED" [2025-10-05 Sun 17:50]
:END:
I've got access to Sonnet 4.5 and Opus 4.1 from Athropic.
Here's my prompt:
#+BEGIN_VERSE
I've got two Lenovo x13 2-in-1 Gen 5 with Debian 13 GNOME 48 with hostnames Cosmo and Jackson.
On Cosmo, I can't select the native screen resolution for the external LG 34" TFT which is 2560x1080. In the GNOME screen configuration dialog, I may only choose a resolution (among others) that is similar but not correct: 2560x1440.
On Jackson, the native screen resolution is recognized out of the box.
Both machines have same:
- graphics hardware
- Linux driver
- BIOS firmware: "UEFI BIOS Version", "Embedded Controller Version", "Machine Type Model"
- BIOS settings
- operating system
- GRUB config with: GRUB_CMDLINE_LINUX_DEFAULT="quiet"
- I also tried with no success: GRUB_CMDLINE_LINUX_DEFAULT="quiet video=eDP-1:2560x1080@30"
Different is ME Firmware Version with "18.0.5.2155" on Jackson and 18.0.15.2515 on Cosmo although I think that this has no effect on the issue.
Also same on both machines:
```
user@cosmo ~ % l /lib/firmware/i915/mtl_*
-rw-r--r-- 1 root root 52712 Apr 24 18:35 /lib/firmware/i915/mtl_dmc.bin
-rw-r--r-- 1 root root 1142784 Apr 24 18:35 /lib/firmware/i915/mtl_gsc_1.bin
-rw-r--r-- 1 root root 312064 Apr 24 18:35 /lib/firmware/i915/mtl_guc_70.bin
-rw-r--r-- 1 root root 561152 Apr 24 18:35 /lib/firmware/i915/mtl_huc_gsc.bin
user@cosmo ~ %
```
I switched the SSDs between Cosmo and Jackson and the issue with the wrong resoltion sticks with the SSD. So it's not an hardware issue.
--------------------
The output of "gdctl show" is:
On Cosmo:
```
Monitors:
├──Monitor HDMI-1 (LG Electronics 34")
│ ├──Vendor: GSM
│ ├──Product: LG ULTRAWIDE
│ ├──Serial: 0x0002de64
│ ├──Current mode
│ │ └──2560x1440@59.951
│ └──Preferences:
│ └──Luminances:
│ ├──default ⇒ 100.0 (default) (current)
│ └──bt2100 ⇒ 100.0 (default)
[...]
Logical monitors:
├──Logical monitor #1
│ ├──Position: (2560, 480)
│ ├──Scale: 1.25
│ ├──Transform: 180
│ ├──Primary: no
│ └──Monitors: (1)
│ └──eDP-1 (Built-in display)
└──Logical monitor #2
├──Position: (0, 0)
├──Scale: 1.0
├──Transform: normal
├──Primary: yes
└──Monitors: (1)
└──HDMI-1 (LG Electronics 34")
```
On Jackson:
```
Monitors:
├──Monitor HDMI-1 (LG Electronics 34")
│ ├──Vendor: GSM
│ ├──Product: LG ULTRAWIDE
│ ├──Serial: 0x0002de64
│ ├──Current mode
│ │ └──2560x1080@59.978
│ └──Preferences:
│ └──Luminances:
│ ├──default ⇒ 100.0 (default) (current)
│ └──bt2100 ⇒ 100.0 (default)
[...]
Logical monitors:
├──Logical monitor #1
│ ├──Position: (0, 0)
│ ├──Scale: 1.25
│ ├──Transform: normal
│ ├──Primary: yes
│ └──Monitors: (1)
│ └──eDP-1 (Built-in display)
└──Logical monitor #2
├──Position: (1536, 0)
├──Scale: 1.0
├──Transform: normal
├──Primary: no
└──Monitors: (1)
└──HDMI-1 (LG Electronics 34")
```
--------------------
The output of "gdctl show -m":
Cosmo:
```
Monitors:
├──Monitor HDMI-1 (LG Electronics 34")
│ ├──Vendor: GSM
│ ├──Product: LG ULTRAWIDE
│ ├──Serial: 0x0002de64
│ └──Modes (23)
│ ├──3840x2160@30.000
│ ├──3840x2160@29.970
│ ├──3840x2160@25.000
│ ├──3840x2160@24.000
│ ├──3840x2160@23.976
│ ├──2560x1440@59.951
│ ├──1920x1080@60.000
│ ├──1920x1080@59.940
│ ├──1920x1080@50.000
│ ├──1680x1050@59.883
│ ├──1600x900@60.000
│ ├──1280x1024@75.025
│ ├──1280x1024@60.020
│ ├──1280x800@59.910
│ ├──1280x720@60.000
│ ├──1280x720@59.940
│ ├──1280x720@50.000
│ ├──1024x768@75.029
│ ├──1024x768@60.004
│ ├──832x624@74.551
│ ├──800x600@75.000
│ ├──800x600@60.317
│ └──720x576@50.000
│ └──Preferences:
[...]
```
--------------------
Jackson:
```
Monitors:
├──Monitor HDMI-1 (LG Electronics 34")
│ ├──Vendor: GSM
│ ├──Product: LG ULTRAWIDE
│ ├──Serial: 0x0002de64
│ └──Modes (28)
│ ├──3840x2160@30.000
│ ├──3840x2160@29.970
│ ├──3840x2160@25.000
│ ├──3840x2160@24.000
│ ├──3840x2160@23.976
│ ├──2560x1440@59.951
│ ├──2560x1080@74.991
│ ├──2560x1080@59.978
│ ├──2560x1080@50.000
│ ├──1920x1080@74.986
│ ├──1920x1080@60.000
│ ├──1920x1080@59.940
│ ├──1920x1080@50.000
│ ├──1680x1050@59.883
│ ├──1600x900@60.000
│ ├──1280x1024@75.025
│ ├──1280x1024@60.020
│ ├──1280x800@59.910
│ ├──1280x720@60.000
│ ├──1280x720@59.940
│ ├──1280x720@50.000
│ ├──1152x864@59.967
│ ├──1024x768@75.029
│ ├──1024x768@60.004
│ ├──832x624@74.551
│ ├──800x600@75.000
│ ├──800x600@60.317
│ └──720x576@50.000
│ └──Preferences:
│ └──Luminances:
│ ├──default ⇒ 100.0 (default) (current)
│ └──bt2100 ⇒ 100.0 (default)
└──Monitor eDP-1 (Built-in display)
[...]
Logical monitors:
├──Logical monitor #1
│ ├──Position: (0, 0)
│ ├──Scale: 1.25
│ ├──Transform: normal
│ ├──Primary: yes
│ └──Monitors: (1)
│ └──eDP-1 (Built-in display)
└──Logical monitor #2
├──Position: (1536, 0)
├──Scale: 1.0
├──Transform: normal
├──Primary: no
└──Monitors: (1)
└──HDMI-1 (LG Electronics 34")
```
--------------------
I tried to set the resolution but that failed as well:
```
user@cosmo ~ % gdctl set -LpM HDMI-1 --mode 2560x1080@60
Failed to create configuration: No mode 2560x1080@60 available for HDMI-1
1 user@cosmo ~ %
```
--------------------
"lspci -v" shows different output though:
Cosmo:
```
00:02.0 VGA compatible controller: Intel Corporation Meteor Lake-P [Intel Graphics] (rev 08) (prog-if 00 [VGA controller])
Subsystem: Lenovo Device 232a
Flags: bus master, fast devsel, latency 0, IRQ 192, IOMMU group 0
Memory at 4058000000 (64-bit, prefetchable) [size=16M]
Memory at 4000000000 (64-bit, prefetchable) [size=256M]
Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
Capabilities: [40] Vendor Specific Information: Len=0c
Capabilities: [70] Express Root Complex Integrated Endpoint, IntMsgNum 0
Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable+ 64bit+
Capabilities: [d0] Power Management version 3
Capabilities: [100] Null
Capabilities: [110] Process Address Space ID (PASID)
Capabilities: [200] Address Translation Service (ATS)
Capabilities: [420] Physical Resizable BAR
Capabilities: [320] Single Root I/O Virtualization (SR-IOV)
Capabilities: [400] Latency Tolerance Reporting
Kernel driver in use: i915
Kernel modules: i915, xe
```
Jackson:
```
00:02.0 VGA compatible controller: Intel Corporation Meteor Lake-P [Intel Graphics] (rev 08) (prog-if 00 [VGA controller])
Subsystem: Lenovo Device 232a
Flags: bus master, fast devsel, latency 0, IRQ 192, IOMMU group 0
Memory at 4058000000 (64-bit, prefetchable) [size=16M]
Memory at 4000000000 (64-bit, prefetchable) [size=256M]
Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
Capabilities: [40] Vendor Specific Information: Intel Capabilities v1
CapA: Peg60Dis- Peg12Dis- Peg11Dis- Peg10Dis- PeLWUDis- DmiWidth=x4
EccDis- ForceEccEn- VTdDis- DmiG2Dis- PegG2Dis- DDRMaxSize=Unlimited
1NDis- CDDis- DDPCDis- X2APICEn- PDCDis- IGDis- CDID=0 CRID=0
DDROCCAP+ OCEn- DDRWrtVrefEn+ DDR3LEn+
CapB: ImguDis- OCbySSKUCap- OCbySSKUEn- SMTCap- CacheSzCap 0x0
SoftBinCap- DDR3MaxFreqWithRef100=Disabled PegG3Dis-
PkgTyp- AddGfxEn- AddGfxCap- PegX16Dis- DmiG3Dis- GmmDis-
DDR3MaxFreq=2932MHz LPDDR3En-
Capabilities: [70] Express Root Complex Integrated Endpoint, IntMsgNum 0
Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable+ 64bit+
Capabilities: [d0] Power Management version 3
Capabilities: [100] Null
Capabilities: [110] Process Address Space ID (PASID)
Capabilities: [200] Address Translation Service (ATS)
Capabilities: [420] Physical Resizable BAR
Capabilities: [320] Single Root I/O Virtualization (SR-IOV)
Capabilities: [400] Latency Tolerance Reporting
Kernel driver in use: i915
Kernel modules: i915, xe
```
I don't know why there is a difference in this output and what to do in order to get the same "Capabilities" on both machines.
Please analyse the information above carefully and tell me what I may check or change so that Cosmo gets the native screen resolution for the external TFT as well.
#+END_VERSE
I had to add:
#+BEGIN_QUOTE
Related to your xrandr tips: Debian 13 GNOME 48 runs Wayland. I guess
that xrandr and probably other mentioned X11 related stuff won't work
that way.
#+END_QUOTE
***** Opus 4.1 Answer
You're absolutely right! I apologize for the X11-centric suggestions.
With Wayland and GNOME 48, the approach is quite different. Let me
provide Wayland-appropriate solutions:
****** 1. Check EDID under Wayland
#+begin_src sh
# Check what the kernel/DRM sees
sudo cat /sys/class/drm/card*-HDMI-A-1/edid | edid-decode
# or
sudo cat /sys/class/drm/card*-HDMI-A-1/modes
# Check current mode
cat /sys/class/drm/card*-HDMI-A-1/modes | grep 2560x1080
# See what mutter/GNOME detects
journalctl -b | grep -i "2560x1080"
journalctl -b | grep -i edid
#+end_src
****** 2. Clear GNOME/Mutter Display Cache (Wayland-specific)
#+begin_src sh
# Remove GNOME's monitor configurations
rm ~/.config/monitors.xml
rm ~/.config/monitors.xml~
# If using GDM
sudo rm /var/lib/gdm3/.config/monitors.xml
sudo rm /var/lib/gdm3/.config/monitors.xml~
# Clear mutter's cache
rm -rf ~/.cache/mutter/
# Restart GNOME Shell (on Wayland)
# Either logout/login or:
systemctl --user restart gnome-shell
#+end_src
****** 3. Force Mode via Kernel/KMS
Since Wayland relies on KMS (Kernel Mode Setting), add the mode at the
kernel level:
#+begin_src sh
# In /etc/default/grub
GRUB_CMDLINE_LINUX_DEFAULT="quiet video=HDMI-A-1:2560x1080@60"
# Update and reboot
sudo update-grub
sudo reboot
#+end_src
****** 4. Create Custom EDID
Since Jackson reads the EDID correctly, you can extract and force it on
Cosmo:
*On Jackson:*
#+begin_src sh
# Extract working EDID
sudo cat /sys/class/drm/card*-HDMI-A-1/edid > ~/lg_monitor.edid
# Copy this file to Cosmo
#+end_src
*On Cosmo:*
#+begin_src sh
# Place the EDID file
sudo mkdir -p /lib/firmware/edid
sudo cp lg_monitor.edid /lib/firmware/edid/
# Force kernel to use this EDID
# In /etc/default/grub
GRUB_CMDLINE_LINUX_DEFAULT="quiet drm.edid_firmware=HDMI-A-1:edid/lg_monitor.edid"
# Update and reboot
sudo update-grub
sudo reboot
#+end_src
****** 5. Use Mutter Debug Settings
Create a debug configuration for mutter:
#+begin_src sh
# Create mutter debug config
mkdir -p ~/.config/mutter
# Enable experimental features that might help with display detection
gsettings set org.gnome.mutter experimental-features "['scale-monitor-framebuffer']"
#+end_src
****** 6. Check Mutter/GNOME Logs
#+begin_src sh
# Check what mutter sees
MUTTER_DEBUG=kms journalctl -f
# In another terminal, restart GNOME
systemctl --user restart gnome-shell
# Look for messages about available modes
#+end_src
****** 7. Compare Working System Configuration
*On Jackson*, gather the working configuration:
#+begin_src sh
# Get the exact mode timings from DRM
cat /sys/class/drm/card*-HDMI-A-1/modes
cat /sys/class/drm/card*-HDMI-A-1/mode
# Get monitor info as seen by mutter
gdbus call --session --dest org.gnome.Mutter.DisplayConfig \
--object-path /org/gnome/Mutter/DisplayConfig \
--method org.gnome.Mutter.DisplayConfig.GetCurrentState
# Export current monitor.xml if it has correct config
cat ~/.config/monitors.xml
#+end_src
****** 8. Manual Configuration via monitors.xml
If GNOME saved a working configuration on Jackson, you can copy its
=~/.config/monitors.xml= to Cosmo. Make sure to adjust any
system-specific identifiers.
****** 9. Try Forcing with =gdctl= Directly
Since you mentioned =gdctl= recognizes the monitor but not the mode:
#+begin_src sh
# First check if the mode is detected at kernel level
cat /sys/class/drm/card*-HDMI-A-1/modes | grep 2560x1080
# If it shows up there but not in gdctl, try:
# Force a display reconfiguration
gdctl apply
#+end_src
****** Most Likely Solution for Wayland
Given that: - The issue follows the SSD (not hardware) - Jackson shows
more modes (28 vs 23) - The same monitor and cable work fine on Jackson
I'd prioritize: 1. *Extracting and forcing the EDID from Jackson* (#4) -
this is the most reliable solution 2. *Clearing all monitor
configurations* (#2) and letting GNOME re-detect 3. *Using kernel
parameters* to force the mode (#3)
The EDID solution is particularly good because it completely bypasses
any detection issues and forces the kernel to use the known-good EDID
data from Jackson.
***** Trying Opus 4.1 Suggestions
:PROPERTIES:
:END:
****** 4. Create Custom EDID
This resulted in a weird situation on Cosmo: desktop background image
was shown on external TFT only and just in the upper right hand
corner. Opened windows are not all visible. I had to revert this.
I re-tried this a second time and this time, the weird effects were
gone but the wrong resolution still persisted.
****** 1. Check EDID under Wayland
Had to install dependencies:
: sudo apt install edid-decode
Then I executed:
: user@jackson ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/edid | edid-decode > jackson-edid-decode.log
#+BEGIN_EXAMPLE
user@jackson ~ % cat /sys/class/drm/card*-HDMI-A-1/modes | grep 2560x1080
2560x1080
2560x1080
2560x1080
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % sudo journalctl -b | grep -i "2560x1080"
1 user@jackson ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % sudo journalctl -b | grep -i edid
Oct 05 14:21:43 jackson sudo[6209]: user : TTY=pts/1 ; PWD=/home/user ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
Oct 05 14:23:31 jackson sudo[6316]: user : TTY=pts/1 ; PWD=/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-1 ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
Oct 05 14:24:04 jackson sudo[6330]: user : TTY=pts/1 ; PWD=/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-1 ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
Oct 05 14:24:36 jackson sudo[6340]: user : TTY=pts/1 ; PWD=/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-1 ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
Oct 05 14:34:13 jackson sudo[6548]: user : TTY=pts/1 ; PWD=/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-1 ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
Oct 05 14:34:34 jackson sudo[6566]: user : TTY=pts/1 ; PWD=/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-1 ; USER=root ; COMMAND=/usr/bin/apt install edid-decode
Oct 05 14:34:49 jackson sudo[6726]: user : TTY=pts/1 ; PWD=/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-1 ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
Oct 05 14:34:52 jackson sudo[6736]: user : TTY=pts/1 ; PWD=/sys/devices/pci0000:00/0000:00:02.0/drm/card0/card0-HDMI-A-1 ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
Oct 05 14:35:50 jackson sudo[6778]: user : TTY=pts/1 ; PWD=/home/shared/shares/sting/abzulegen ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
#+END_EXAMPLE
And now compare with Cosmo:
#+BEGIN_EXAMPLE
user@cosmo ~ % cat /sys/class/drm/card*-HDMI-A-1/modes | grep 2560x1080
1 user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo journalctl -b | grep -i "2560x1080"
Oct 05 14:33:28 cosmo kernel: Command line: BOOT_IMAGE=/vmlinuz-6.12.48+deb13-amd64 root=/dev/mapper/cosmo--vg-root ro quiet video=HDMI-1:2560x1080@60
Oct 05 14:33:28 cosmo kernel: Kernel command line: BOOT_IMAGE=/vmlinuz-6.12.48+deb13-amd64 root=/dev/mapper/cosmo--vg-root ro quiet video=HDMI-1:2560x1080@60
user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo journalctl -b | grep -i edid
Oct 05 14:40:06 cosmo sudo[2787]: user : TTY=pts/0 ; PWD=/home/user ; USER=root ; COMMAND=/usr/bin/apt install edid-decode
Oct 05 14:40:22 cosmo sudo[2934]: user : TTY=pts/0 ; PWD=/home/user ; USER=root ; COMMAND=/usr/bin/cat /sys/class/drm/card0-HDMI-A-1/edid
user@cosmo ~ %
#+END_EXAMPLE
Well, there are differences but as it looks to my eye, this actually
looks better on Cosmo, doesn't it?
Comparing the =edid-decode.log= files:
#+BEGIN_EXAMPLE
user@sting ~ % diff jackson-edid-decode.log cosmo-edid-decode.log
user@sting ~ %
#+END_EXAMPLE
They're same.
****** 2. Clear GNOME/Mutter Display Cache
Situation before:
#+BEGIN_EXAMPLE
user@cosmo ~ % cat ~/.config/monitors.xml
logical
2560
480
1.25
eDP-1
AUO
0x6693
0x00000000
1920
1200
60.026
0
0
1
yes
HDMI-1
GSM
LG ULTRAWIDE
0x0002de64
2560
1440
59.951
user@cosmo ~ % cat /home/user/.config/monitors.xml~
logical
1920
480
1.25
eDP-1
AUO
0x6693
0x00000000
1920
1200
60.026
0
0
1
yes
HDMI-1
GSM
LG ULTRAWIDE
0x0002de64
1920
1080
60.000
user@cosmo ~ %
#+END_EXAMPLE
No =/var/lib/gdm3/.config/monitors.xm*= found:
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo ls -la /var/lib/gdm3/.config/monitors.xm*
zsh: no matches found: /var/lib/gdm3/.config/monitors.xm*
1 user@cosmo ~ %
#+END_EXAMPLE
OK, let's remove the files:
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo rm ~/.config/monitors.x*
user@cosmo ~ % sudo rm -rf ~/.cache/mutter/
user@cosmo ~ % sudo reboot
#+END_EXAMPLE
No change of the issue after reboot finishes. :-(
****** 3. Force Mode via Kernel/KMS
Well, this is similar to what I've already tried but let's re-try
this.
#+BEGIN_EXAMPLE
vim /etc/default/grub
# GRUB_CMDLINE_LINUX_DEFAULT="quiet video=HDMI-A-1:2560x1080@60"
# Update and reboot
sudo update-grub
sudo reboot
#+END_EXAMPLE
Did not fix the issue.
****** 9. Try Forcing with =gdctl= Directly
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/modes | grep 2560x1080
1 user@cosmo ~ %
#+END_EXAMPLE
No luck here: not detected on kernel level.
****** 5. Use Mutter Debug Settings
I didn't try this since this should be the same on both X13 machines anyway.
****** 7. Compare Working System Configuration
On Jackson:
#+BEGIN_EXAMPLE
user@jackson ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/modes
2560x1080
3840x2160
3840x2160
3840x2160
3840x2160
3840x2160
2560x1440
2560x1080
2560x1080
1920x1080
1920x1080
1920x1080
1920x1080
1920x1080
1680x1050
1600x900
1280x1024
1280x1024
1280x800
1152x864
1280x720
1280x720
1280x720
1280x720
1024x768
1024x768
832x624
800x600
800x600
720x576
720x480
720x480
640x480
640x480
640x480
640x480
user@jackson ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/mode
zsh: no matches found: /sys/class/drm/card*-HDMI-A-1/mode
1 user@jackson ~ %
#+END_EXAMPLE
Something's wrong with this call:
#+BEGIN_EXAMPLE
user@jackson ~ % sudo gdbus call --session --dest org.gnome.Mutter.DisplayConfig \
--object-path /org/gnome/Mutter/DisplayConfig \
--method org.gnome.Mutter.DisplayConfig.GetCurrentState
Error connecting: Cannot autolaunch D-Bus without X11 $DISPLAY
1 user@jackson ~ %
#+END_EXAMPLE
****** Copy ~/.config/monitors.xml
:PROPERTIES:
:END:
On Jackson, I fetch =~/.config/monitors.xml= and copy to Cosmo.
#+BEGIN_EXAMPLE
user@cosmo ~ % mv ~/.config/monitors.xml ~/.config/monitors.xml_OLD
user@cosmo ~ % mv jackson_monitors.xml ~/.config/monitors.xml
user@cosmo ~ %
#+END_EXAMPLE
Reboot and test.
No change. :-(
***** Sending Back My Notes To Opus 4.1 And Getting Answer From Opus 4.l
:PROPERTIES:
:END:
Excellent debugging work! The key finding is that *Cosmo doesn't detect
2560x1080 at the kernel DRM level* - it's not in
=/sys/class/drm/card*-HDMI-A-1/modes=. This means the issue is happening
before GNOME/Mutter even gets involved.
Let's focus on why the kernel isn't detecting the mode properly:
****** 1. Compare EDID Raw Data
Since the kernel isn't seeing the 2560x1080 mode, let's compare the raw
EDID data:
*On both machines:*
#+begin_src sh
# Get raw EDID in hex
sudo cat /sys/class/drm/card*-HDMI-A-1/edid | xxd > edid-raw.txt
# Get decoded EDID
sudo cat /sys/class/drm/card*-HDMI-A-1/edid | edid-decode > edid-decoded.txt
# Get MD5 hash to quickly check if they're identical
sudo cat /sys/class/drm/card*-HDMI-A-1/edid | md5sum
#+end_src
Compare these files between Jackson and Cosmo. If they're different, the
monitor is sending different EDID data to each machine.
****** 2. Check i915 Driver Debug Info
The kernel driver might be filtering modes for some reason:
#+begin_src sh
# Check i915 parameters on both machines
sudo cat /sys/module/i915/parameters/* > i915-params.txt
# Enable verbose DRM debugging temporarily
echo 0x1ff | sudo tee /sys/module/drm/parameters/debug
# Reconnect the monitor (unplug/replug HDMI)
# Then check kernel messages
sudo dmesg | grep -i "2560x1080"
sudo dmesg | grep -i "pruning\|reject\|filter"
# Disable debug mode after
echo 0 | sudo tee /sys/module/drm/parameters/debug
#+end_src
****** 3. Check Display Port/HDMI Bandwidth Limits
The i915 driver might be rejecting modes due to perceived bandwidth
limitations:
#+begin_src sh
# Check if there are different link rates
sudo cat /sys/kernel/debug/dri/0/i915_display_info | grep -A10 HDMI
# Check connector status
sudo cat /sys/class/drm/card*-HDMI-A-1/status
sudo cat /sys/class/drm/card*-HDMI-A-1/enabled
#+end_src
****** 4. Force EDID via Kernel (Different Approach)
Since the previous EDID forcing had issues, try a different method:
*On Jackson:*
#+begin_src sh
# Save the EDID
sudo cat /sys/class/drm/card*-HDMI-A-1/edid > ~/lg-good.bin
# Copy to Cosmo
#+end_src
*On Cosmo:*
#+begin_src sh
# Place in firmware directory
sudo cp lg-good.bin /lib/firmware/edid/lg-ultrawide.bin
# Use different kernel parameter syntax
# In /etc/default/grub:
GRUB_CMDLINE_LINUX_DEFAULT="quiet drm_kms_helper.edid_firmware=HDMI-A-1:edid/lg-ultrawide.bin video=HDMI-A-1:e"
sudo update-grub
sudo reboot
#+end_src
The =video=HDMI-A-1:e= forces the connector to be enabled.
****** 5. Check HDMI Cable/Port Version
This is interesting - could there be a difference in HDMI capability
detection?
#+begin_src sh
# Check what the driver thinks about HDMI version
sudo dmesg | grep -i "hdmi\|tmds"
# Look for maximum pixel clock
sudo cat /sys/kernel/debug/dri/*/i915_display_info | grep -i "max\|clock\|tmds"
#+end_src
****** 6. Try i915 Module Options
Create =/etc/modprobe.d/i915.conf=:
#+begin_src sh
# Try disabling some automatic features
options i915 enable_dp_mst=0
options i915 enable_psr=0
options i915 enable_fbc=0
#+end_src
Then rebuild initramfs:
#+begin_src sh
sudo update-initramfs -u
sudo reboot
#+end_src
****** 7. Most Revealing Test - Boot with drm.debug
Add to GRUB for one boot:
#+begin_src sh
# At boot, edit GRUB entry (press 'e' at GRUB menu)
# Add to kernel line:
drm.debug=0x1e log_buf_len=16M
# Boot and immediately after login, save the log:
sudo dmesg > /tmp/drm-debug.log
# Search for why 2560x1080 is rejected:
grep -C5 "2560x1080\|2560 x 1080" /tmp/drm-debug.log
grep -C5 "pruning\|removing\|reject" /tmp/drm-debug.log
#+end_src
****** Most Likely Cause
Given that: - The kernel doesn't see 2560x1080 in available modes - The
same hardware with Jackson's SSD works fine - The GRUB parameter
=video=HDMI-A-1:2560x1080@60= doesn't help
This points to either: 1. *Different EDID being received* (monitor
sending different data) 2. *i915 driver filtering the mode* due to some
configuration 3. *A stored bad EDID* somewhere that persists across
reboots
The raw EDID comparison (#1) will tell us definitively if the monitor is
sending different data. The debug output (#7) will show why the kernel
is rejecting the mode if the EDID is the same.
***** Trying Opus 4.1 Suggestions Again
OK, so Opus thinks that with the additional information, the issue
does not involve Wayland/Mutter. So let's do that.
****** 1. Compare EDID Raw Data
Requires:
: sudo apt install xxd
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/edid | xxd > ~/share/sting/abzulegen/cosmo-edid-raw.txt
user@cosmo ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/edid | edid-decode > ~/share/sting/abzulegen/cosmo-edid-decoded.txt
user@cosmo ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/edid | md5sum > ~/share/sting/abzulegen/cosmo-edid-md5sum.txt
user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/edid | xxd > ~/share/sting/abzulegen/jackson-edid-raw.txt
user@jackson ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/edid | edid-decode > ~/share/sting/abzulegen/jackson-edid-decoded.txt
user@jackson ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/edid | md5sum > ~/share/sting/abzulegen/jackson-edid-md5sum.txt
#+END_EXAMPLE
Comparing:
#+BEGIN_EXAMPLE
user@sting ~/share/business/abzulegen % diff jackson-edid-raw.txt ~/share/user/abzulegen/cosmo-edid-raw.txt
user@sting ~/share/business/abzulegen % diff jackson-edid-decoded.txt ~/share/user/abzulegen/cosmo-edid-decoded.txt
user@sting ~/share/business/abzulegen % diff jackson-edid-md5sum.txt ~/share/user/abzulegen/cosmo-edid-md5sum.txt
user@sting ~/share/business/abzulegen %
#+END_EXAMPLE
... they are the same on both machines.
****** 2. Check i915 Driver Debug Info
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo cat /sys/module/i915/parameters/* > ~/share/sting/abzulegen/cosmo-i915-parameters.log
user@cosmo ~
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % sudo cat /sys/module/i915/parameters/* > ~/share/sting/abzulegen/jackson-i915-parameters.log
user@jackson ~ %
#+END_EXAMPLE
Comparing:
#+BEGIN_EXAMPLE
user@sting ~/share/business/abzulegen % diff jackson-i915-parameters.log ~/share/user/abzulegen/cosmo-i915-parameters.log
6c6
< -1
---
> N
1 user@sting ~/share/business/abzulegen %
#+END_EXAMPLE
Yes, there are differences.
****** 3. Check Display Port/HDMI Bandwidth Limits
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo cat /sys/kernel/debug/dri/0/i915_display_info | grep -A10 HDMI
[CONNECTOR:250:HDMI-A-1]: status: connected
physical dimensions: 800x340mm
subpixel order: Unknown
CEA rev: 3
audio support: yes
HDCP version: HDCP1.4 HDCP2.2
max bpc: 12
modes:
"3840x2160": 30 297000 3840 4016 4104 4400 2160 2168 2178 2250 0x40 0x5
"3840x2160": 30 296703 3840 4016 4104 4400 2160 2168 2178 2250 0x40 0x5
"3840x2160": 25 297000 3840 4896 4984 5280 2160 2168 2178 2250 0x40 0x5
user@cosmo ~ % sudo cat /sys/kernel/debug/dri/0/i915_display_info | grep -A10 HDMI > ~/share/sting/abzulegen/cosmo--link-rates.log
user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % sudo cat /sys/kernel/debug/dri/0/i915_display_info | grep -A10 HDMI
[CONNECTOR:271:HDMI-A-1]
[PLANE:90:plane 1B]: type=PRI
uapi: [FB:323] XR30 little-endian (0x30335258),0x10000000000000f,2560x1080, visible=visible, src=2560.000000x1080.000000+0.000000+0.000000, dst=2560x1080+0+0, rotation=0 (0x00000001)
hw: [FB:323] XR30 little-endian (0x30335258),0x10000000000000f,2560x1080, visible=yes, src=2560.000000x1080.000000+0.000000+0.000000, dst=2560x1080+0+0, rotation=0 (0x00000001)
[PLANE:100:plane 2B]: type=OVL
uapi: [FB:0] n/a,0x0,0x0,, visible=hidden, src=0.000000x0.000000+0.000000+0.000000, dst=0x0+0+0, rotation=0 (0x00000001)
[PLANE:110:plane 3B]: type=OVL
uapi: [FB:0] n/a,0x0,0x0,, visible=hidden, src=0.000000x0.000000+0.000000+0.000000, dst=0x0+0+0, rotation=0 (0x00000001)
[PLANE:120:plane 4B]: type=OVL
uapi: [FB:0] n/a,0x0,0x0,, visible=hidden, src=0.000000x0.000000+0.000000+0.000000, dst=0x0+0+0, rotation=0 (0x00000001)
[PLANE:130:plane 5B]: type=OVL
--
[CONNECTOR:271:HDMI-A-1]: status: connected
physical dimensions: 800x340mm
subpixel order: Unknown
CEA rev: 3
audio support: yes
HDCP version: HDCP1.4 HDCP2.2
max bpc: 12
modes:
"2560x1080": 60 181250 2560 2608 2640 2720 1080 1083 1093 1111 0x48 0x9
"3840x2160": 30 297000 3840 4016 4104 4400 2160 2168 2178 2250 0x40 0x5
"3840x2160": 30 296703 3840 4016 4104 4400 2160 2168 2178 2250 0x40 0x5
user@jackson ~ % sudo cat /sys/kernel/debug/dri/0/i915_display_info | grep -A10 HDMI > ~/share/sting/abzulegen/jackson-link-rates.log
user@jackson ~ %
#+END_EXAMPLE
Oh, there are differences!
****** 4. Force EDID via Kernel (Different Approach)
#+BEGIN_EXAMPLE
user@jackson ~ % sudo cat /sys/class/drm/card*-HDMI-A-1/edid > ~/share/sting/abzulegen/jackson-lg-good.bin
#+END_EXAMPLE
Apply on Cosmo:
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo cp ~/share/sting/abzulegen/jackson-lg-good.bin /lib/firmware/edid/lg-ultrawide.bin
user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
# Use different kernel parameter syntax
# In /etc/default/grub:
GRUB_CMDLINE_LINUX_DEFAULT="quiet drm_kms_helper.edid_firmware=HDMI-A-1:edid/lg-ultrawide.bin video=HDMI-A-1:e"
sudo update-grub
sudo reboot
#+END_EXAMPLE
This resulted in showing 1024x768 on the external TFT which was
labeled as "Unknown" in GNOME display settings. No higher resolution
was available.
I reverted those changes again.
****** 5. Check HDMI Cable/Port Version
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo dmesg | grep -i "hdmi\|tmds"
[ 28.446711] skl_hda_dsp_generic skl_hda_dsp_generic: hda_dsp_hdmi_build_controls: no PCM in topology for HDMI converter 3
[ 28.463438] input: sof-hda-dsp HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:1f.3/skl_hda_dsp_generic/sound/card0/input31
[ 28.463470] input: sof-hda-dsp HDMI/DP,pcm=4 as /devices/pci0000:00/0000:00:1f.3/skl_hda_dsp_generic/sound/card0/input32
[ 28.463499] input: sof-hda-dsp HDMI/DP,pcm=5 as /devices/pci0000:00/0000:00:1f.3/skl_hda_dsp_generic/sound/card0/input33
user@cosmo ~ % sudo cat /sys/kernel/debug/dri/*/i915_display_info | grep -i "max\|clock\|tmds"
zsh: no matches found: /sys/kernel/debug/dri/*/i915_display_info
user@cosmo ~ % sudo find /sys/kernel/debug -name 'i915_display_info'
/sys/kernel/debug/dri/0000:00:02.0/i915_display_info
1 user@cosmo ~ % sudo cat /sys/kernel/debug/dri/0000:00:02.0/i915_display_info | grep -i "max\|clock\|tmds"
max bpc: 8
max bpc: 12
user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % sudo dmesg | grep -i "hdmi\|tmds"
[ 12.490709] skl_hda_dsp_generic skl_hda_dsp_generic: hda_dsp_hdmi_build_controls: no PCM in topology for HDMI converter 3
[ 12.507576] input: sof-hda-dsp HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:1f.3/skl_hda_dsp_generic/sound/card0/input26
[ 12.507615] input: sof-hda-dsp HDMI/DP,pcm=4 as /devices/pci0000:00/0000:00:1f.3/skl_hda_dsp_generic/sound/card0/input27
[ 12.507652] input: sof-hda-dsp HDMI/DP,pcm=5 as /devices/pci0000:00/0000:00:1f.3/skl_hda_dsp_generic/sound/card0/input28
user@jackson ~ % sudo cat /sys/kernel/debug/dri/0000:00:02.0/i915_display_info | grep -i "max\|clock\|tmds"
port_clock=0, lane_count=0
port_clock=0, lane_count=0
port_clock=0, lane_count=0
port_clock=0, lane_count=0
max bpc: 8
max bpc: 12
user@jackson ~ %
#+END_EXAMPLE
There are differences but I don't know what this means.
****** 6. Try i915 Module Options
#+BEGIN_EXAMPLE
user@cosmo ~ % l /etc/modprobe.d/i915.conf
ls: cannot access '/etc/modprobe.d/i915.conf': No such file or directory
2 user@cosmo ~ %
#+END_EXAMPLE
So the config file does not exist on both machines. I will skip this
because there doesn't seem to be any difference with respect to those options.
****** 7. Most Revealing Test - Boot with drm.debug
I'm not sure if I was able to add the kernel line:
: drm.debug=0x1e log_buf_len=16M
#+BEGIN_EXAMPLE
user@sting ~/share/user/abzulegen % grep -C5 "2560x1080\|2560" cosmo-dmesg
1 user@sting ~/share/user/abzulegen % grep -C5 "pruning\|removing\|reject" cosmo-dmesg
1 user@sting ~/share/user/abzulegen %
#+END_EXAMPLE
No match for that.
***** Sending Back My Notes To Opus 4.1 And Getting Answer From Opus 4.l
:PROPERTIES:
:END:
Excellent detective work! The smoking gun is in the i915_display_info
output. On Jackson, it shows:
#+begin_example
"2560x1080": 60 181250 2560 2608 2640 2720 1080 1083 1093 1111 0x48 0x9
#+end_example
But on Cosmo, this mode is *completely missing* from the kernel's mode
list, even though *the EDID data is identical*!
This means the i915 driver on Cosmo is rejecting/filtering the 2560x1080
mode for some reason.
****** The Critical Difference Found
Looking at the i915 parameters comparison:
#+begin_example
< -1
---
> N
#+end_example
This suggests there's a parameter difference. Let's investigate this
more carefully:
****** 1. Compare i915 Parameters Properly
#+begin_src sh
# On both machines, get a more readable comparison:
for param in /sys/module/i915/parameters/*; do
echo "$(basename $param): $(cat $param)"
done | sort > ~/i915-params-sorted.txt
#+end_src
Compare these files to see which specific parameter differs.
****** 2. Try the Nuclear Reset for i915
Since the EDID is the same but the driver behaves differently, let's
reset the i915 driver state:
#+begin_src sh
# On Cosmo, completely reset i915 module parameters
sudo rm -f /etc/modprobe.d/i915*
sudo rm -f /etc/modprobe.d/*i915*
# Check if there are any kernel command line i915 parameters
cat /proc/cmdline | grep i915
# Remove any i915 state files
sudo rm -rf /var/cache/*i915*
sudo rm -rf /var/lib/*i915*
#+end_src
****** 3. Force Mode Addition via i915 Debugfs
Since we know the exact modeline from Jackson, let's try to force it
through debugfs:
#+begin_src sh
# First, check current state
sudo cat /sys/kernel/debug/dri/0/i915_display_info | grep -A20 "HDMI-A-1"
# Try to force the mode (this is a long shot but worth trying)
echo "2560x1080@60" | sudo tee /sys/class/drm/card0-HDMI-A-1/mode
#+end_src
****** 4. Compare Kernel Versions and i915 Driver Versions
#+begin_src sh
# On both machines
uname -a
modinfo i915 | head -20
dmesg | grep -i "i915.*firmware"
#+end_src
****** 5. The Real Debug - Why i915 Rejects the Mode
We need better debugging. Let's try again with more specific debug
flags:
#+begin_src sh
# Add to /etc/default/grub on Cosmo:
GRUB_CMDLINE_LINUX_DEFAULT="quiet drm.debug=0x1ff i915.verbose_state_checks=1 log_buf_len=16M"
sudo update-grub
sudo reboot
# After reboot, immediately:
sudo dmesg > /tmp/full-debug.log
# Look for mode validation
grep -i "mode\|2560\|1080\|valid\|reject\|prune\|filter" /tmp/full-debug.log > /tmp/mode-debug.log
# Look for HDMI specific issues
grep -i "hdmi\|tmds\|clock\|bandwidth" /tmp/full-debug.log > /tmp/hdmi-debug.log
#+end_src
****** 6. Check for BIOS/UEFI Display Settings
Even though you mentioned BIOS settings are the same, double-check: -
Look for any display-related settings like "HDMI Max Resolution" - Check
for "Display Pre-allocation" or "DVMT" settings - Look for "IGD" or
"Graphics" settings
****** 7. Try Different i915 Enable Mode
Create =/etc/modprobe.d/i915-test.conf= on Cosmo:
#+begin_src sh
# Force enable all outputs
options i915 enable_dc=-1
options i915 enable_dp_mst=1
options i915 enable_hdcp=1
options i915 force_probe=!xe
#+end_src
Then:
#+begin_src sh
sudo update-initramfs -u
sudo reboot
#+end_src
****** 8. Most Important - Detailed Parameter Comparison
Since we found a difference in the parameters, let's see the actual
values:
#+begin_src sh
# On Cosmo
echo "=== Cosmo i915 Parameters ===" > ~/cosmo-i915-detailed.txt
for p in /sys/module/i915/parameters/*; do
echo "$p: $(cat $p 2>/dev/null || echo 'ERROR')" >> ~/cosmo-i915-detailed.txt
done
# On Jackson
echo "=== Jackson i915 Parameters ===" > ~/jackson-i915-detailed.txt
for p in /sys/module/i915/parameters/*; do
echo "$p: $(cat $p 2>/dev/null || echo 'ERROR')" >> ~/jackson-i915-detailed.txt
done
#+end_src
****** The Mystery
The fact that: 1. Same EDID data 2. Same hardware 3. Different modes
detected by kernel
Points to either: - A different i915 module parameter (most likely given
your finding) - A different kernel/driver version - Some persistent
state that survives reboots
The parameter difference you found (=-1= vs =N=) is the key. We need to
identify which parameter this is and correct it.
***** Trying Opus 4.1 Suggestions Again
:PROPERTIES:
:END:
OK, so Opus thinks that it narrowed down the issue further. Let's go
with its flow.
****** 1. Compare i915 Parameters Properly
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo bash
root@cosmo:/home/user# for param in /sys/module/i915/parameters/*; do echo "$(basename $param): $(cat $param)"; done | sort > ~user/share/sting/abzulegen/cosmo-i915-parameters-sorted.log
root@cosmo:/home/user#
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % sudo bash
[sudo] password for user:
root@jackson:/home/user# for param in /sys/module/i915/parameters/*; do echo "$(basename $param): $(cat $param)"; done | sort > ~user/share/sting/abzulegen/jackson-i915-parameters-sorted.log
root@jackson:/home/user#
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@sting ~/share/user/abzulegen % diff cosmo-i915-parameters-sorted.log ~/share/business/abzulegen/jackson-i915-parameters-sorted.log
6c6
< enable_dmc_wl: N
---
> enable_dmc_wl: -1
1 user@sting ~/share/user/abzulegen %
#+END_EXAMPLE
So this is the different setting: on cosmo, this is "N". On jackson,
this is "-1".
Could this be the source of the issue?
***** Sending Back My Notes To Opus 4.1 And Getting Answer From Opus 4.l
:PROPERTIES:
:END:
YES! This could absolutely be the source of the issue! The
=enable_dmc_wl= parameter controls Display Microcontroller (DMC)
workarounds/features, and this difference is very significant.
The values mean:
- =-1= = auto/default (let the driver decide)
- =N= = disabled/no
This parameter affects display power management and mode detection. The
fact that Jackson (working) has =-1= and Cosmo (broken) has =N= is
highly suspicious.
****** Fix This Immediately
*Option 1: Runtime change (temporary test):*
#+begin_src sh
# On Cosmo, as root:
echo -1 | sudo tee /sys/module/i915/parameters/enable_dmc_wl
# Then unplug and replug the HDMI cable
# Check if 2560x1080 appears:
cat /sys/class/drm/card*-HDMI-A-1/modes | grep 2560x1080
#+end_src
*Option 2: Make it permanent:*
#+begin_src sh
# Create/edit /etc/modprobe.d/i915.conf on Cosmo:
echo "options i915 enable_dmc_wl=-1" | sudo tee /etc/modprobe.d/i915.conf
# Rebuild initramfs
sudo update-initramfs -u
# Reboot
sudo reboot
#+end_src
****** Why This Matters
The DMC (Display Microcontroller) in Intel graphics handles: - Display
power states - Mode detection and validation - Display timing generation
When =enable_dmc_wl= is set to =N= (disabled), it can cause: - Missing
display modes - Incorrect EDID parsing - Problems with certain
resolutions
****** How Did This Happen?
Check where this setting comes from on Cosmo:
#+begin_src sh
# Check for any configs that might set this
grep -r "enable_dmc_wl" /etc/modprobe.d/
grep -r "enable_dmc_wl" /etc/default/
grep "enable_dmc_wl" /proc/cmdline
#+end_src
This setting might have been added by: - A previous troubleshooting
attempt - A package/driver update - A system configuration tool
****** Verify the Fix
After changing the parameter and rebooting:
#+begin_src sh
# Confirm the parameter is set correctly
cat /sys/module/i915/parameters/enable_dmc_wl # Should show -1
# Check available modes
cat /sys/class/drm/card*-HDMI-A-1/modes | grep 2560x1080
# Use gdctl to see if the mode is available
gdctl show -m | grep 2560x1080
#+end_src
This is very likely the root cause. The DMC workaround being disabled on
Cosmo is preventing proper mode detection for your ultrawide monitor!
***** Trying Opus 4.1 Suggestions Again
:PROPERTIES:
:END:
How can I set the property?
#+BEGIN_EXAMPLE
user@cosmo ~ % echo -1 | sudo tee /sys/module/i915/parameters/enable_dmc_wl
tee: /sys/module/i915/parameters/enable_dmc_wl: Permission denied
-1
1 user@cosmo ~ % sudo echo -1 > /sys/module/i915/parameters/enable_dmc_wl
zsh: permission denied: /sys/module/i915/parameters/enable_dmc_wl
1 user@cosmo ~ %
#+END_EXAMPLE
OK, this seems to be read-only.
Let's do the permanent fix.
#+BEGIN_EXAMPLE
echo "options i915 enable_dmc_wl=-1" | sudo tee /etc/modprobe.d/i915.conf
sudo update-initramfs -u
sudo reboot
#+END_EXAMPLE
Now, the external TFT is not recognized at all! Furthermore, the GNOME
panel is not shown as before. Instead, it looks like the default of
the GNOME extension I've installed.
While checking the parameter, I realized that the whole sub-hierarchie
is missing:
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo ls -la /sys/module/i915/
ls: cannot access '/sys/module/i915/': No such file or directory
2 user@cosmo ~ %
#+END_EXAMPLE
Checking the reasons:
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo grep -r "enable_dmc_wl" /etc/modprobe.d/
/etc/modprobe.d/i915.conf:options i915 enable_dmc_wl=-1
user@cosmo ~ % sudo grep -r "enable_dmc_wl" /etc/default/
1 user@cosmo ~ % sudo grep "enable_dmc_wl" /proc/cmdline
1 user@cosmo ~ %
#+END_EXAMPLE
OK, I do the alternative version using kernel command line.
I revert the change:
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo rm /etc/modprobe.d/i915.conf
user@cosmo ~ % sudo update-initramfs -u
#+END_EXAMPLE
#+BEGIN_EXAMPLE
# Edit /etc/default/grub
sudo nano /etc/default/grub
# Add to GRUB_CMDLINE_LINUX_DEFAULT:
GRUB_CMDLINE_LINUX_DEFAULT="quiet i915.enable_dmc_wl=-1"
# Update grub
sudo update-grub
# Reboot
sudo reboot
#+END_EXAMPLE
Again: the external TFT is not recognized at all. Furthermore, the
GNOME panel is not shown as before. Instead, it looks like the default
of the GNOME extension I've installed.
It looks as if parts of GNOME aren't startet. The keyring doesn't ask
for its passphrase.
***** Sending Back My Notes To Opus 4.1 And Getting Answer From Opus 4.l
Oh no! The i915 module is failing to load entirely! This is a critical
issue - the parameter value might be invalid or causing a conflict.
Let's fix this immediately:
****** Emergency Fix - Get Display Working Again
#+begin_src sh
# Remove the problematic kernel parameter
sudo nano /etc/default/grub
# Change back to just:
GRUB_CMDLINE_LINUX_DEFAULT="quiet"
# Update grub
sudo update-grub
# Ensure no modprobe config remains
sudo rm -f /etc/modprobe.d/i915.conf
# Rebuild initramfs
sudo update-initramfs -u
# Reboot
sudo reboot
#+end_src
****** After Getting Display Back
Once your display is working again, let's investigate why this parameter
causes the module to fail:
#+begin_src sh
# Check if i915 is loaded
lsmod | grep i915
# Check for driver errors
sudo dmesg | grep -i "i915\|error\|fail"
# Check the actual parameter syntax on Jackson
cat /sys/module/i915/parameters/enable_dmc_wl
#+end_src
****** Important Discovery
The fact that setting =enable_dmc_wl=-1= causes the i915 module to fail
to load entirely suggests:
1. This parameter might not exist in your kernel version
2. The syntax might be different
3. There might be a different underlying issue
****** Let's Check Kernel and Driver Versions
On *both* machines:
#+begin_src sh
# Kernel version
uname -r
# i915 driver info
modinfo i915 | grep -E "version|filename"
# Check if the parameter even exists
ls /sys/module/i915/parameters/ | grep dmc
#+end_src
****** Alternative Theory
The parameter difference might be a *symptom*, not the cause. The real
issue might be:
1. Different kernel versions between the systems
2. Different firmware files
3. A corrupted driver state on Cosmo
****** Check for Firmware Differences
#+begin_src sh
# On both machines
ls -la /lib/firmware/i915/ | grep -i dmc
md5sum /lib/firmware/i915/*dmc*
#+end_src
****** Safe Test Approach
Once you have your display back, let's try a safer approach:
#+begin_src sh
# First, check what parameters actually exist
ls /sys/module/i915/parameters/
# Try setting a different, safer parameter
echo "options i915 enable_psr=0" | sudo tee /etc/modprobe.d/i915-test.conf
sudo update-initramfs -u
#+end_src
The fact that the i915 module completely fails to load with that
parameter is very revealing - it suggests the parameter might not be
valid for your kernel version, or there's something else going on. Let's
get your system working first, then investigate more carefully.
***** Trying Opus 4.1 Suggestions Again
I already reverted the changes and got the original setup back. So no
further rollback necessary.
Let's jump in to the fix:
****** Let's Check Kernel and Driver Versions
#+BEGIN_EXAMPLE
user@cosmo ~ % uname -r
6.12.48+deb13-amd64
user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % uname -r
6.16.9+deb14-amd64
user@jackson ~ %
#+END_EXAMPLE
Wow, there's a different kernel! :-O
Comparing the =modinfo=:
#+BEGIN_EXAMPLE
user@cosmo ~ % sudo modinfo i915 | grep -E "version|filename"
filename: /lib/modules/6.12.48+deb13-amd64/kernel/drivers/gpu/drm/i915/i915.ko.xz
vermagic: 6.12.48+deb13-amd64 SMP preempt mod_unload modversions
parm: invert_brightness:Invert backlight brightness (-1 force normal, 0 machine defaults, 1 force inversion), please report PCI device ID, subsystem vendor and subsystem device ID to dri-devel@lists.freedesktop.org, if your machine needs it. It will then be included in an upcoming module version. (int)
user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % sudo modinfo i915 | grep -E "version|filename"
filename: /lib/modules/6.16.9+deb14-amd64/kernel/drivers/gpu/drm/i915/i915.ko.xz
vermagic: 6.16.9+deb14-amd64 SMP preempt mod_unload modversions
parm: invert_brightness:Invert backlight brightness (-1 force normal, 0 machine defaults, 1 force inversion), please report PCI device ID, subsystem vendor and subsystem device ID to dri-devel@lists.freedesktop.org, if your machine needs it. It will then be included in an upcoming module version. (int)
user@jackson ~ %
#+END_EXAMPLE
Yes, there are difference.
Comparing the parameters:
#+BEGIN_EXAMPLE
user@cosmo ~ % ls /sys/module/i915/parameters/ | grep dmc
dmc_firmware_path
enable_dmc_wl
user@cosmo ~ %
#+END_EXAMPLE
#+BEGIN_EXAMPLE
user@jackson ~ % ls /sys/module/i915/parameters/ | grep dmc
dmc_firmware_path enable_psr2_sel_fetch memtest
enable_dmc_wl force_probe modeset
user@jackson ~ %
#+END_EXAMPLE
Also some differences.
Weird, I set up both machines using the Debian 13 defaults without any
manual package/kernel overrides. Must have been happened by mistake or
via dependencies I'm not aware of.
So how to fix that?
I should have checked the kernel earlier. :-(
****** Getting the Newer Kernel on Cosmo
:PROPERTIES:
:END:
#+BEGIN_EXAMPLE
user@jackson ~ % apt policy linux-image-6.16.9+deb14-amd64
linux-image-6.16.9+deb14-amd64:
Installed: 6.16.9-1
Candidate: 6.16.9-1
Version table:
*** 6.16.9-1 500
500 http://deb.debian.org/debian unstable/main amd64 Packages
100 /var/lib/dpkg/status
user@jackson ~ %
#+END_EXAMPLE
OK, jackson has its kernel from unstable. For whatever reason.
Jackson has no =/etc/apt/apt.conf.d/99defaultrelease= file. → I needed
some packages from sid/testing and this must have gone wrong somehow.
I now set the default release to "trixie" on both machines:
#+BEGIN_EXAMPLE
echo 'APT::Default-Release "trixie";' | sudo tee /etc/apt/apt.conf.d/99defaultrelease
#+END_EXAMPLE
On Cosmo, I add unstable sources:
#+BEGIN_EXAMPLE
echo "deb http://deb.debian.org/debian testing main contrib non-free-firmware" | sudo tee /etc/apt/sources.list.d/testing.list
#+END_EXAMPLE
After =sudo apt update=:
#+BEGIN_EXAMPLE
[...]
41 packages can be upgraded. Run 'apt list --upgradable' to see them.
user@cosmo ~ % sudo apt list --upgradable
cups-bsd/testing 2.4.10-4 amd64 [upgradable from: 2.4.10-3+deb13u1]
cups-client/testing 2.4.10-4 amd64 [upgradable from: 2.4.10-3+deb13u1]
cups-common/testing 2.4.10-4 all [upgradable from: 2.4.10-3+deb13u1]
cups-core-drivers/testing 2.4.10-4 amd64 [upgradable from: 2.4.10-3+deb13u1]
cups-daemon/testing 2.4.10-4 amd64 [upgradable from: 2.4.10-3+deb13u1]
cups-ipp-utils/testing 2.4.10-4 amd64 [upgradable from: 2.4.10-3+deb13u1]
cups-ppdc/testing 2.4.10-4 amd64 [upgradable from: 2.4.10-3+deb13u1]
cups-server-common/testing 2.4.10-4 all [upgradable from: 2.4.10-3+deb13u1]
cups/testing 2.4.10-4 amd64 [upgradable from: 2.4.10-3+deb13u1]
firefox-esr/testing 140.3.1esr-2 amd64 [upgradable from: 140.3.1esr-1~deb13u1]
imagemagick-7-common/testing 8:7.1.2.3+dfsg1-1 all [upgradable from: 8:7.1.1.43+dfsg1-1+deb13u2]
imagemagick-7.q16/testing 8:7.1.2.3+dfsg1-1 amd64 [upgradable from: 8:7.1.1.43+dfsg1-1+deb13u2]
imagemagick/testing 8:7.1.2.3+dfsg1-1 amd64 [upgradable from: 8:7.1.1.43+dfsg1-1+deb13u2]
libavcodec61/testing 7:7.1.2-1+b1 amd64 [upgradable from: 7:7.1.2-0+deb13u1]
libavfilter10/testing 7:7.1.2-1+b1 amd64 [upgradable from: 7:7.1.2-0+deb13u1]
libavformat61/testing 7:7.1.2-1+b1 amd64 [upgradable from: 7:7.1.2-0+deb13u1]
libavutil59/testing 7:7.1.2-1+b1 amd64 [upgradable from: 7:7.1.2-0+deb13u1]
libcups2t64/testing 2.4.10-4 amd64 [upgradable from: 2.4.10-3+deb13u1]
libjson-xs-perl/testing 4.040-1 amd64 [upgradable from: 4.040-1~deb13u1]
libmagickcore-7.q16-10-extra/testing 8:7.1.2.3+dfsg1-1 amd64 [upgradable from: 8:7.1.1.43+dfsg1-1+deb13u2]
libmagickcore-7.q16-10/testing 8:7.1.2.3+dfsg1-1 amd64 [upgradable from: 8:7.1.1.43+dfsg1-1+deb13u2]
libmagickwand-7.q16-10/testing 8:7.1.2.3+dfsg1-1 amd64 [upgradable from: 8:7.1.1.43+dfsg1-1+deb13u2]
libnss-myhostname/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
libnss-systemd/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
libpam-systemd/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
libpostproc58/testing 7:7.1.2-1+b1 amd64 [upgradable from: 7:7.1.2-0+deb13u1]
libssl3t64/testing 3.5.4-1 amd64 [upgradable from: 3.5.1-1+deb13u1]
libswresample5/testing 7:7.1.2-1+b1 amd64 [upgradable from: 7:7.1.2-0+deb13u1]
libswscale8/testing 7:7.1.2-1+b1 amd64 [upgradable from: 7:7.1.2-0+deb13u1]
libsystemd-shared/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
libsystemd0/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
libudev1/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
libxslt1.1/testing 1.1.43-0.2 amd64 [upgradable from: 1.1.35-1.2+deb13u2]
linux-image-amd64/testing 6.16.9-1 amd64 [upgradable from: 6.12.48-1]
openssl-provider-legacy/testing 3.5.4-1 amd64 [upgradable from: 3.5.1-1+deb13u1]
openssl/testing 3.5.4-1 amd64 [upgradable from: 3.5.1-1+deb13u1]
systemd-cryptsetup/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
systemd-sysv/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
systemd-timesyncd/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
systemd/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
udev/testing 258-1 amd64 [upgradable from: 257.8-1~deb13u2]
user@cosmo ~ %
#+END_EXAMPLE
This defaultrelease settings doesn't seem to have much influence.
So how to keep trixie/stable as the default for all packages except
those who are installed with "-t sid" or similar?
Set APT Pinning for Extra Safety: To ensure packages don't
accidentally get pulled from unstable, create
=/etc/apt/preferences.d/priority= (file did not exist on both hosts):
#+BEGIN_EXAMPLE
cat << 'EOF' | sudo tee /etc/apt/preferences.d/priority
Package: *
Pin: release a=stable
Pin-Priority: 900
Package: *
Pin: release a=trixie
Pin-Priority: 900
Package: *
Pin: release a=unstable
Pin-Priority: 100
Package: linux-image-* linux-headers-*
Pin: release a=unstable
Pin-Priority: 500
EOF
#+END_EXAMPLE
Still got packages from unstable or testing when I do =sudo apt
upgrade --dry-run= → this is a separate issue.