(Solved) RocketLake-S GT1 [UHD Graphics 750] not working on LinuxMint 20.3

[blockhead47]

SMG your are fantastic!! I very much appreciate you not giving up on this problem.

So your research proved to be excellent. System booted fine with 5.14-oem kernel with one monitor attached. The ASUS Prime Z590-A has 2 video ports. One "DisplayPort" and one HDMI Port. It did not matter which port combination I used (DisplayPort or HDMI), as long as I only connected one monitor. The system would boot up just fine. After the system booted with one monitor I could plug the other monitor in, the system would recognize it and so far everything is working.

Thank you so much for your assistance.

I am marking this as solved. The may not be the ultimate resolution but it solves the immediate problem.

[SMG]

SMG your are fantastic!! I very much appreciate you not giving up on this problem.

So your research proved to be excellent. System booted fine with 5.14-oem kernel with one monitor attached. The ASUS Prime Z590-A has 2 video ports. One "DisplayPort" and one HDMI Port. It did not matter which port combination I used (DisplayPort or HDMI), as long as I only connected one monitor. The system would boot up just fine. After the system booted with one monitor I could plug the other monitor in, the system would recognize it and so far everything is working.

Thanks for describing what you tried as that may help others. That is such a bizarre fix, but I do not argue with what works.

Thank you so much for your assistance.

I am marking this as solved. The may not be the ultimate resolution but it solves the immediate problem.

You're welcome. The 5.15 kernel will become available around the April time-frame for Mint users. Maybe that will be a more permanent solution.

I was going back through all the firmware posts when I saw you marked this as solved, so I will add what I found in case it might be needed in the future. (So I won't have to hunt for it again. )

This was posted back in July i915: Add v2.03 DMC for RKL which I would have thought would have already made it to your system by now either in the kernel or firmware upgrades. (RKL = Rocket Lake, DMC=display microcontroller) The change is marked as "This version has fixes for timeout issues."

I do not know if that relates at all to the issue at hand, but the log indicated you have the prior version, 2.02.

Code: Select all

Jan 23 12:54:26 sys76 kernel: i915 0000:00:02.0: [drm] Finished loading DMC firmware i915/rkl_dmc_ver2_02.bin (v2.2)

[blockhead47]

I do not know if that relates at all to the issue at hand, but the log indicated you have the prior version, 2.02.

So I have installed 5.16 generic, just because it is easy to install using TuxInvader PPA. I noticed when I installed 5.16 during the update-initramfs command it complained about "Possible missing firmware i915 yada..yada. On reboot with both monitors connected, no joy. Still a hang. I installed all the missing i915 firmware (v2.03 now) and update-initramfs no longer complained about missing firmware. Still no boot with both monitors connected.

Also just for completeness if some else finds this thread, Windows 10 boots just fine with both monitors connected. Boots fine with Windows Update drivers and the ASUS provided drivers.

It is curious that kernel 5.11 boots ok with both monitors connected. There is difference in the output of

Code: Select all

lshw -C display

between the two versions.

Code: Select all

5.16  sudo lshw -C display
  *-display                 
       description: VGA compatible controller
       product: RocketLake-S GT1 [UHD Graphics 750]
       vendor: Intel Corporation
       physical id: 2
       bus info: pci@0000:00:02.0
       logical name: /dev/fb0
       version: 04
       width: 64 bits
       clock: 33MHz
       capabilities: pciexpress msi pm vga_controller bus_master cap_list fb
       configuration: depth=32 driver=i915 latency=0 mode=1920x1080 visual=truecolor xres=1920 yres=1080
       resources: iomemory:600-5ff iomemory:400-3ff irq:130 memory:6000000000-6000ffffff memory:4000000000-400fffffff ioport:3000(size=64) memory:c0000-dffff

Notice the configuration line above on 5.16 with both monitors plugged in and working only one monitor shows.

Code: Select all

5.11 sudo lshw -C display
  *-display                 
       description: VGA compatible controller
       product: RocketLake-S GT1 [UHD Graphics 750]
       vendor: Intel Corporation
       physical id: 2
       bus info: pci@0000:00:02.0
       logical name: /dev/fb0
       version: 04
       width: 64 bits
       clock: 33MHz
       capabilities: pciexpress msi pm vga_controller bus_master cap_list rom fb
       configuration: depth=32 driver=i915 latency=0 mode=1280x1024 visual=truecolor xres=1280 yres=1024
       resources: iomemory:600-5ff iomemory:400-3ff irq:128 memory:6000000000-6000ffffff memory:4000000000-400fffffff ioport:3000(size=64) memory:c0000-dffff
 

Now with the 5.11, again looking at the configuration line, both monitors are shown.

I don't know it that is significant or not, but is is a difference between the 2 kernel verions.

[SMG]

It is curious that kernel 5.11 boots ok with both monitors connected. There is difference in the output of

Code: Select all

lshw -C display

between the two versions.

I'm not seeing the difference you mentioned.

I removed the lines which were the same and there is only one line which is different. I do not see where either shows two monitors.

5.16 sudo lshw -C display
configuration: depth=32 driver=i915 latency=0 mode=1920x1080 visual=truecolor xres=1920 yres=1080

5.11 sudo lshw -C display
configuration: depth=32 driver=i915 latency=0 mode=1280x1024 visual=truecolor xres=1280 yres=1024


If you would like to boot into the 5.16 or the 5.14-oem kernel (using the method you described which worked) and then run the following:
journalctl -b | nc termbin.com 9999
I will take a look and see if there are any messages which might give a clue as to what is happening.

[blockhead47]

Logs from the 5.14 kernel.

Code: Select all

https://termbin.com/gkr87

[SMG]

During the boot process, the root partition is mounted so the system can start up. It is then remounted in read-only status so nothing can accidentally (or otherwise) mess with it while the operating system is being used. Your root partition as listed in your original inxi output was sdb5.

Code: Select all

Partition:
  ID-1: / size: 36.42 GiB used: 12.63 GiB (34.7%) fs: ext4 dev: /dev/sdb5
  ID-2: /boot/efi size: 96 MiB used: 30.2 MiB (31.5%) fs: vfat
    dev: /dev/sdb1
  ID-3: /home size: 28.71 GiB used: 2.9 GiB (10.1%) fs: ext4 dev: /dev/sda1

In the 5.14-oem log, this does happen.

Code: Select all

Jan 23 21:26:44 sys76 kernel: EXT4-fs (sdb5): re-mounted. Opts: errors=remount-ro. Quota mode: none.

This is the line in the 5.11 log. It mentions sdc5. There is no mention of sdb5 in the 5.11 log.

Code: Select all

Jan 23 12:54:26 sys76 kernel: EXT4-fs (sdc5): re-mounted. Opts: errors=remount-ro. Quota mode: none.

What is sdc5? There's no mention of it in the 5.14-oem log.

There's a couple of other things I want to check, but it's getting very late for me.

[blockhead47]

I have an external USB drive that normally boots up as /dev/sde.
For some reason in the pastebin log for the 5.11 kernel it got mounted as /dev/sda. That pushed all the other drive letters down.
5.11 -> 5.16
sde -> sda
sda -> sdb
sdb -> sdc
sdc -> sdd
sdd -> sde
Is that ok? Otherwise I can boot into 5.11 and send new logs?

[SMG]

Is that ok? Otherwise I can boot into 5.11 and send new logs?

You have an explanation for it, so no reason to redo the logs. I had noticed errors with some of the drives attaching, but it hadn't dawned on me the drive letters would change.

Will be going through the logs again in a bit and will let you know if I see anything which might be helpful.

[SMG]

Perhaps the drives spun up differently because only one monitor was attached. The UUID of the root partition was also different in the two logs.

I did a search on the general topic of a system not booting with both monitors attached, but booting with one and being able to attach the second. Found people experiencing it with Windows-only systems. It was considered an unsolved mystery in the places I found it.

For some reason, the firmware was still showing as version 2.02 in the 5.14-oem log.

Code: Select all

Jan 23 21:26:44 sys76 kernel: i915 0000:00:02.0: [drm] Finished loading DMC firmware i915/rkl_dmc_ver2_02.bin (v2.2)

Core temps were slightly higher with the 5.14-oem than the 5.11, but temps changing with kernel versions has happened to others as well.

The info change you noticed regarding the sudo lshw -C display also was in the logs. First two lines are 5.11 and second is 5.14-oem.

Code: Select all

Jan 23 12:54:26 sys76 kernel: efifb: framebuffer at 0x4000000000, using 3072k, total 3072k
Jan 23 12:54:26 sys76 kernel: efifb: mode is 1024x768x32, linelength=4096, pages=1

Jan 23 21:26:44 sys76 kernel: efifb: framebuffer at 0x4000000000, using 8100k, total 8100k
Jan 23 21:26:44 sys76 kernel: efifb: mode is 1920x1080x32, linelength=7680, pages=1

TLDR: I didn't really find anything to indicate why the change happened.

[blockhead47]

Thanks for all the followup.
Let's close this thread and wait until the official 5.11 replacement comes out. I think you said this would be sometime in the April 2022 time-frame.

[SMG]

Let's close this thread and wait until the official 5.11 replacement comes out.

The official replacement is the 5.13 kernel and it is already out. We've had quite a few people report on the forum that Update Manager is automatically giving them 5.13 to replace 5.11.

The next kernel after 5.13 will be 5.15 which will become available in April timeframe.