New Year, New Laptop, New Observations

Published on: 2025-01-01

Welcome to 2025

So I never publish blogs when the publication date actually says in Lektor, they appear afterwards. This time, however, I start on the 1st of January and aim to have this done by the time I'm off of my paternity leave. Software engineering mixed with child rearing, this is the hippest place to be in 2025!

That said, it offers an outlet for my ramblings that isn't my ever patient partner. Which is good, because the content of this and many other articles are just not their bag. However, it is in my nature to want to write about things, so I'm going to try and chart what I'm doing on the day to day, week to week, month to month a bit better so that my technical frustrations and explorations have an outlet.

New Laptop

Lenovo ThinkPad T15g 4K Xeon W-10885M 64GB 1TB SSD NVIDIA GeForce 2080 Laptop
Lenovo ThinkPad T15g 4K Xeon W-10885M 64GB 1TB SSD NVIDIA GeForce 2080 Laptop (image courtesy of Newandusedlaptops4u)

I've just gotten hold of a dedicated coding machine. I've again gotten it from a refurb seller on ebay who I've used previously for many Thinkpad: two T410's, a T430, a T430s, that have tided me and members of my family over brilliantly. My T430s is still going, serving as a general purpose laptop, having spent eight years as my primary "home" machine. I don't know why people don't use refurbed machines more often. The only exceptions I've made are for servers, which I go for as tower units.

That said, this is the "newest" machine I've ever bought and by far the most powerful. It's going to be a dedicated experimentation and code machine (plus maybe the odd game with some friends given its power). This was justified simply as everything else I own is now in use elsewhere and isn't powerful enough for local model development and, ahem, Age of Empires. More genuinely, I find that not having a dedicated code machine has resulted in a dangerous conflation of purpose with my work machine and that causes me concentration problems.

Fedora

I've been using CentOS for server and desktop, and Fedora for desktop, for easily over a decade now. CentOS disappearing has left me saddened and given work have shifted to Rocky 9, that's what I'm replacing trusted CentOS with for the moment.

I really am keen to stick with Fedora. But the new laptop is presenting two challenges: the WIFI is unstable and XFCE dislikes any 16:9 resolutions that aren't 4k!

WIFI problems, a blast from the past

The T15g is not new enough that it's bleeding edge and I've noticed circumstantially the following:

  • Fedora 39 (out of support) - it seems to work fine on power up, not on reboot
  • Fedora 40 (about to go out of support) - it seems to work fine on power up, not on reboot
  • Fedora 41 (current version) - not working

This thread on Fedora discussion appears to have similar issues as the error code probe of 0000:00:14.3 failed with error -110 is exactly what I'm experiencing. But there are key differences and that thread doesn't conclude with specific solutions to the problem. Other than that I've found little else to help, so I'm going to undertake some diagnostics based on that to bring me up to speed before progressing to raising a discussion item!

I suspect the issue comes from changing firmware versions loaded by iwlwifi between kernels in the respective OS's, but given we're fighting the sleeping patterns of a newborn I want to take the opportunity to investigate and get this new laptop working so I can code near the sleeping baba!

My steps to investigate are:

  1. Find out exactly what equipment is here (which is good to know) and investigate how it's failing. I don't think is the same device code as that mentioned above and firmware support suggest it should work, so best to make sure it isn't.
  2. Double check there's no existing issue that explains it (such as this one, I appear to be very non-linear writing this, because I did a quick check before anything else).
  3. Raise it in the appropriate place such as the Linux kernel Bugzilla.

This is very unusual though, as I've not had any issues like this for a good long while with Fedora or similar, so it is interesting that I've ended up in a situation of diagnosing hardware issues like it's two decades ago. Eeek! I so wanted to be moving quickly into developing some toys as well!!!

So these steps are useful to go through, it's been ages since I did any real investigation of Linux systems: I've not worked in sysadmin or devops for over four years and not seriously for eight or so. Therefore, these might be interesting steps to take for others, especially if you're looking at iwlwifi...

1. Work out what's happening

hw-probe is new to me, but running it on my new laptop is reasonable so I have done that to add to the collective consciousness of the internet. What a cool project, I hadn't come across it before. It was easily found in the Fedora repositories and uploaded. I'd be a lot more concerned about privacy and looking in depth if it wasn't a completely fresh laptop and install. ;)

$ sudo -E hw-probe -all -upload
Probe for hardware ... Ok
Reading logs ... Ok
Uploaded to DB, Thank you!

Probe URL: https://linux-hardware.org/?probe=c5dc058f75

The probe includes the error I have in dmesg, with nmcli and rfkill from the same probe both showing that there is no wifi network apparently present from the Comet Lake PCH CNVi WiFi listed. Interestingly, the 8086:06f0 device code doesn't match the Fedora thread I saw previously so that quickly tells me to largely ignore that (it seemed to get rambly after a while anyway) and "The device is supported by kernel versions 5.7 and newer according to the LKDDb".

Confirming some details, there is definitely a device and firmware available, but I would perhaps expect there to be

# inxi -Nxx
Network:
  Device-1: Intel Comet Lake PCH CNVi WiFi driver: N/A bus-ID: 00:14.3 chip-ID: 8086:06f0
  Device-2: Intel Ethernet I219-LM vendor: Lenovo driver: e1000e v: kernel port: N/A
    bus-ID: 00:1f.6 chip-ID: 8086:0d4c

# lsmod | grep wifi
iwlwifi               557056  0
cfg80211             1404928  1 iwlwifi

# modinfo iwlwifi
... lots of output, including plenty of available firmware ...

So why with dmesg do we see [ 17.780760] iwlwifi 0000:00:14.3: probe with driver iwlwifi failed with error -110?

Stab in the dark turning off UEFI network stack: it's always good fun idea to try something random, for a bit of excitement. New to the software engineering with parenting game, I'm trying to keep myself excited somehow. It turns out UEFI secure boot was already off, so unchecking the stack options (a) probably had no effect and (b) didn't fix anything even if they did. It was nice to try and live dangerously for a moment!!! :P

2. Investigate

So next, assuming both the intel hardware support and device information pages are correct, the card should be working. Before I start debugging iwlwifi, perhaps looking at the kernel bugzilla is a good shout, rather than the Fedora forums, as it obviously isn't a Fedora issue if it's in the kernel logs.

The best candidate I came up with from a specific search for the error code was this issue, which matches the error code to a different type of adapter, though it is from intel. Interestingly, there's mention of "Disabling Windows Fast Startup fixed the issue", as well as this being linked to the hibernate / shutdown of the adapter in Windows. This I confirm is mentioned in the iwlwifi documentation and so seems like a good candidate.

I still have the active Windows partition issued with the unit, so will see again if it works (it has every time so far!)

No luck. Sad face... :(

Raising this as an issue seems like the only option, so have done so in the kernel bugzilla.

Graphic problems, I'm probably stuck in the past

Why on earth is a resolution listed that doesn't seem to do anything!? If I select 16:10 resolutions or 4:3 resolutions, it all works fine. If I go for any 16:9 that isn't 3840x2160 it doesn't work.

Until I change the refresh rate from 60.01 to <60Hz!

Then it works fine.

That's something at least.

Takeaways

  1. Sometimes there seems like too much information, especially with all the distros cross-fertilising in advice terms. The truth is, think logically about where the problem is and it gets a lot easier.
  2. It would be nice if there was a simple algorithm painted for diagnosing hardware failures and understanding as a semi-pro linux user with no time (thanks parenthood!) Perhaps that could be a subject for a future post.
  3. A nice map of all the most poignant sites to uncover information about the interconnected world of linux and software would be nice. A bit like this fancy open infrastructure map I just found as I checked it didn't already exist!

Comments

Please do leave a comment. I'm moderating them manually for the moment and the Isso project I'm finding slightly experimental, but AMAZING nonetheless. I won't reset the comments database now though, so feedback will be valued!


Tags:

software  linux