Skip to main content


- power on the working FreeBSD 14.1 ThinkPad.
- log into X11, open a terminal and a web browser.
- read the "how to upgrade to 14.2" documentation.
- run the few recommended commands.
- reboot to apply kernel update.
- laptop wrecked, stuck in sddm screen with no possible input.
- (re)boot in single-user mode, disable sddm, finish booting.
- laptop wrecked, stuck on a console "Loading kernel modules".

Now go on, ask me again why I don't run #FreeBSD, The Power to Serve... others...

Unknown parent

Joel Carnat ♑ 🤪
@Tionisla nothing more than the defaults. That said, I disabled the GPU module while in single-user mode and could get a shell to finish the upgrade. Not like it is always the same drama with drm-kmod and release upgrade…
in reply to Joel Carnat ♑ 🤪

Its AGAIN the problem that FreeBSD project does not want to solve - for the short 3 months period (when 14.1 is still in support) the packages for 14.2 - including kernel related packages such as drm-kmod - are built against 14.1 kernel sources ... which give result such as yours.

The solutions are:

- do not upgrade to 14.2 when 14.1 is in support + 1 week so new packages in 'latest' branch gets build

- rebuild 'drm-kmod' packages from Ports:
# make -C /usr/ports/graphics/drm-kmod build deinstall install clean

- use GhostBSD as 'base' for desktop - I mean use GhostBSD repositories and kernel - they always build packages against the RIGHT sources.

I try to at least 'force' FreeBSD related people to switch 'latest' branch to always build against the latest FreeBSD - 14.2 now - and to keep 'quarterly' on older still supported for 3 months - 14.1 in that case ... but they would not listen to me.

in reply to vermaden

@vermaden
I'm not negligent in updating my systems by dragging my feet a couple weeks, I'm ensuring system-stability. 😉

@joel

in reply to Tim Chase

@gumnos @vermaden I've been bitten by these bugs in the past with FreeBSD. I heard about the 14.2 update, but then heard about others having the drm-kmod issues, so I'm holding off until things have stabilized for 14.2. I do the same for any other OS, really. I think OpenBSD is the only exception, but I'm running -current on that and expect bugs to happen. Still, I still believe this shouldn't be the case for releases. :flan_shrug:
in reply to ClaudioM

@claudiom

The notable saving grace I have with FreeBSD are the beadm/bectl boot environments. I've had upgrades go horribly sideways, but a quick switch back to the previous BE and I'm back up and running until I have the time to diagnose & remedy the issue.

My other notable FreeBSD issue is that packages occasionally go missing, so if I do a "hey, upgrade all my packages", if I miss that there's a REMOVED section that is removing, say Firefox or Chromium, I can end up in a state where it's anoying to research a fix and I just need to sit tight until it comes back (then manually re-add it)

But fortunately, OpenBSD updates are usually pretty uneventful.

@vermaden @joel

in reply to Ed Maste

@emaste While true, this isn't always the option for most users. While I know it's different, I run OpenBSD -curent on my laptop and desktop. However, other problems arise from there because it's a development branch, so the drm-kmod issue would likely not be the only issue encountered along the way, especially for normal users getting into FreeBSD. @gumnos @vermaden @joel
in reply to Ed Maste

@emaste @claudiom @gumnos
I probably would if:

- there would be binary updates - can be pkgbase.

- there would be debug disabled in kernel so I do not have to recompile it with every update

in reply to vermaden

you can use pkgbase to keep FreeBSD-current updated, and kernel packages are available with and without debugging options enabled
This entry was edited (1 month ago)
in reply to vermaden

@vermaden @Joel Carnat ♑ 🤪 There was a recent and sudden change in the primary release engineer for FreeBSD, so I think if there's problems with the release process it's due to a badly communicated process - we should probably specifically figure out what that error is?
in reply to silverwizard

@silverwizard
That would not happen if kernel related packages would be build against PROPER 14.2 version.
in reply to silverwizard

@silverwizard @vermaden This issue predates Colin's tenure as lead release engineer. It's not new, but is exacerbated by more frequent releases.
in reply to silverwizard

@silverwizard I should not portray this as a problem with the release process. It's certainly not related to team changes.

Please be aware of the relevant charter, and the scopes of the various teams.

Charter for the Release Engineering Team: <freebsd.org/releng/charter/>

The Primary Release Engineering Team in the context of FreeBSD Project Administration and Management: <freebsd.org/administration/#t-…>

@joel @vermaden

in reply to vermaden

@vermaden with all respect due to developers, this is the exact definition of shitty software (release process).

But, this is free software, no one put a gun on my head to force me using it. So I' ll endorse "if it doesn't you need, shut the fuck up and don't use it" 🤣

in reply to Joel Carnat ♑ 🤪

It also pisses me off a lot ... and I try to change that with every new release when this is broken.

For some reason they do not listen - maybe most of them still use macOS for their desktops and FreeBSD only on servers ...

in reply to vermaden

@vermaden
This is why you always see OpenBSD developers working with ThinkPads on Hackaton with current OpenBSD snapshot running.
This is why they have very good support mostly "out of the box"...
in reply to Pink Panther

@pinkopanterata @vermaden I switched from FreeBSD to OpenBSD exactly because of drm-kmod stuff. These breaks, and, it took FreeBSD 2 (3?) years to support my graphics card, an RX6700XT. By the time FreeBSD supported it, I was already an OpenBSD desktop user for I think a year. They track graphics drivers pretty closely to their release in Linux. A buddy and I would joked about how FreeBSD only ran on old computers, and OpenBSD ran on dinosaurs...I learned it is quite the opposite because of the graphics drivers
in reply to vermaden

@vermaden This is not the issue -- #FreeBSD developers are running FreeBSD on our laptops.

The problem is that for the most part developers are not running a release and drm-kmod from the package collection, and so bypass the issue.

I saw someone (on the forums or in a bug report) describe uninstalling and replacing drm-kmod without having a working console -- that process works, but is a terrible user experience.

in reply to Ed Maste

@emaste @vermaden Yeah, booting into single user, removing drm-kmod from rc.conf, continuing to multi-user, finishing the upgrade, then building drm-kmod from ports was a challenge.

I wouldn't wish it on anybody who wasn't an experience BSD admin.

in reply to vermaden

@vermaden There are like 210 kmod in FreeBSD ports. All that is needed is a separate pkg repository for them, which has just few new version slaves which build kmod packages on new system. It's not a rocket surgery.
in reply to Piotr Smyrak

@piero @vermaden This is also my preferred solution to the problem. The existing 14.x repo update process remains unchanged (on the latest supported 14.x release), but the kmod packages move to specific point release repos (14.1, 14.2).
in reply to Ed Maste

@emaste @vermaden I am glad to read this, as otherwise any laptop investment as announced recently would be just futile.
in reply to Piotr Smyrak

@piero @vermaden This specific issue likely gets solved in the 15.x timeframe by moving drm-kmod into the base system (but we still need to solve the general issue for other kernel modules, like VirtualBox).
in reply to Ed Maste

The solution is pretty straightforward:
To fix problems with RELEASEs in the future, all the packages from the "latest" branch should be built on periodically updated STABLE. That will guarantee that packages will be entirely tested, and ready for deployment when the new release is out. Some warnings that will show up for users following such a "latest" will be rather mild and only give a notion that the running system is not that fresh.
This entry was edited (1 month ago)
in reply to Marek Zarychta

@emaste @piero @vermaden
That's only a proposal for the "latest" branch.
When the build cycle for the packages from the "quarterly" branch remains untouched (they will be built on the oldest supported RELEASE), conservative users can still be happy.
in reply to Ed Maste

@emaste @piero @vermaden
The problem is one that we’ve discussed countless times and so I hope the solution is going be generic. The problem is that there are some things (pkg, Linux-derived drivers, and so on) that are logically part of the base system but need more frequent updates and so are not part of the src tree. We lack a place between src and ports that is for things that are developed by the project and officially supported as part of the OS but which have different update and support cycles. Currently, all of these things live in ports, which is far from ideal, and we have awful shims like pkg shim, which exists because working around our own policies was easier than modifying the policies to be sensible.
in reply to vermaden

@vermaden WTF.

Stop spreading jumped-up misinformation about the FreeBSD Project not wanting solutions.

If you want a pointer, to evidence, you could begin by showing some fucking respect.

Cc @joel

in reply to Graham Perrin

Maybe you stop spitting pointless hate first? @vermaden@bsd.cafe describes the problem correctly, and never said #FreeBSD "didn't want solutions", but only that they refuse his proposal. And to be fair, I'd refuse this as well, because I think it would just lead to more confusion.

To put this into perspective, we're talking about a rare corner case anyways. The idea of a stable branch is to never introduce breaking changes in the #ABI, and as it was meant, this should include in-kernel ABIs as well, so packages built for X.1 must always work on X.2 as well. Unfortunately, parts of the in-kernel ABI needed for some packages containing #kernel #modules are not stable in practice, and, also unfortunately, this includes a package almost everyone needs on a desktop installation: drm-kmod.

The obvious solution would be to make sure to keep the whole ABI stable as it was originally intended. I have doubts this will work out in practice.

So, you'd need to fiddle with the naming of ABI versions, including the minor release in them, to have distinct repositories for e.g. 14.1 and 14.2 to solve the issue for these rare corner cases. That would of course cost a lot, much more build time for ALL the packages, and also disk space. It would kind of spoil the idea of "stable".

There could be a more involved solution introducing a second "KBI" identifier, only attached to packages containing kernel modules, so only these would need extra builds. I personally think this would be the right thing to do, but it's a very intrusive change, touching the ports framework, the pkg utility, poudriere, the build and repository infrastructure, etc pp → in a nutshell, a damn lot of work for a complex change. 😞

CC: @joel@piou.foolbazar.eu

This entry was edited (1 month ago)
in reply to zirias (on snac)

@zirias @vermaden Vermaden did write "Its AGAIN the problem that FreeBSD project does not want to solve." This is absolutely not the case, and I object to framing it that way.

The reality is more "the problem that the FreeBSD project has been unable to solve" though, which isn't much better.

in reply to Ed Maste

Reading the whole post, I didn't understand that sentence as (intended) framing but more as a sloppy expression of frustration about the unsolved issue, given his proposal obviously isn't liked. In any case, a derogatory response with curse words is inappropriate. More appropriate IMHO:

- Explain how and why the suggested solution is bad
- Show possible alternative solutions, so they can be discussed, maybe refined and improved

CC: @vermaden@bsd.cafe @joel@piou.foolbazar.eu

in reply to Graham Perrin

profanity

Sensitive content

in reply to Graham Perrin

@vermaden

(2) Your ignorance of the Technology Roadmap for FreeBSD.

You promoted the map – as valuable information.

The FreeBSD Foundation promotes the map – prominently, and frequently.

Within the map:

― the first focus area describes release-specific overlays.

Your shouting on 4th December not only demonstrated ignorance, it also devalued and mis-portrayed the FreeBSD Project in a way that was guaranteed to offend any number of people.

Now: we can not guess how many people will read – and believe – your ignorant, offensive post. We can not guess how many people will read the boosts of your ignorant, offensive post.

There's more, however I find it difficult to compose this second post without a resurgence of anger. I have made countless mistakes in the past. Some mistakes are easier to correct than others, especially where pride is involved.

Please do the honourable thing.

#FreeBSD #upgrade #misinformation

in reply to Graham Perrin

@grahamperrin

If you really want another answer from my side - then here you are - but you are not gonna like it.

If it was known in the FreeBSD project that its gonna be broken again - why message such as this one was NOT added to Release Notes?

[WARNING:BEGIN]

In the last minutes of 14.2 release process we found a bug that fixing will temporarily break (for 3 months) all kernel related pkg(8) packages (like drm-kmod or virtualbox for example) because of the way FreeBSD project organizes its package building repositories and support cycles. FreeBSD project is working on a solution that hope to prevent such problems in the future.

[WARNING:END]

I did not saw anything like that in the Release Notes.

... and you are making more drama out of this then #Netflix could have ever dream of.

in reply to vermaden

@vermaden @grahamperrin
There's release errata (Open Issues section).
freebsd.org/releases/14.2R/err…
in reply to TomAoki

@TomAoki

Thanks, missed that.

Seem I need to start mention to everytime check ERRATA now after checking Release Notes.

in reply to vermaden

@vermaden
Usually, IIRC, release errata used to be a skeleton (space holder) at the beginning and updated as time goes by (with issues found and/or fixed by patch releases [-p*]).

But as users of graphics/[nvidia-]drm-[510|515|61]-kmod ports increased, the more screams are heard on point releases. So release engineering team were forced to write something about it.

And, as my prediction, release note is not a good place to do so, as this is clearly an issue with how pkgs are built and provided.

What was needed in release note would be, IMHO, pointing there's an open issue described in release errata. Preferrably at Abstract or at the top of Introduction, because Security and Errata section is for "fixed issues until previous release".

in reply to TomAoki

@TomAoki

Announcement

– the phrases "known problems" and "please see"

– third link: errata

– near the head of the page.

#FreeBSD

in reply to Graham Perrin

@grahamperrin
Yes. I've picked the errata URI from there.
But even though it's there, many screams on forums.freebsd.org.
This would mean the link alone is insufficient.
And this would be because, in many releases, errata were just skeletons to add information "later", and didn't read at the early phase, then, forgotton.
in reply to TomAoki

@TomAoki re: "many screams on forums.freebsd.org"

Thanks, I avoid the place, since I quit, now I see <forums.freebsd.org/posts/68198…> on page 1 of 4.

The word "petrol" comes to mind.

in reply to Graham Perrin

@TomAoki at <forums.freebsd.org/threads/pos…> I see some things that could be corrected or improved.

I'll summarise in Reddit in due course. In the meantime, people may take a hint from part of <mail-archive.freebsd.org/cgi/m…>.

With all that's past, and ongoing, readers might begin to understand why I object so strongly to things such as:

― describing developers as jerks

― describing the FreeBSD Project as not wanting to solve a problem, that was (still is) a quite despicable lie.

Examples of the truth:

― <old.reddit.com/r/freebsd/comme…> (3rd December) Colin attending to release discussions on the day of announcement whilst at an airport, and then <mastodon.bsd.cafe/@grahamperri…> his 4th December report from attendance at the multi-day AWS re:Invent 2024 event, which was relevant to FreeBSD and coincided with release of 14.2

― <bsd.network/@dvl/1136478726041…>

And so on.

Cc and respect to @joel

#FreeBSD #respect #developers #thanks

in reply to Graham Perrin

@grahamperrin
And me too posting DRM related things on most threads without the info I've found. They should be a "single" thread, though, regarding its technical details.

And complaining users should know it is impossible for small project like FreeBSD project to test all up-to-date, latest hardwares. To do so, ALL responsible key persons of ALL hardware manufacturers SHALL pop into FreeBSD project as developers/testers. Just a dream for now, unfortunately.

in reply to TomAoki

I can not downplay end users' frustrations with the kernel module (kmod) package build scenario that preceded the recent flurry/hurricane of activity through which spoon-feeding, largely by the FreeBSD Project, has begun.

What irks me is, the popular assumption that the Project should have provided everything. I mean, everything, whilst the Foundation pleads politely – repeatedly – for donations. The year during which attraction of sizeable investments has astounded me, but still, we're short of target figures (and I can not assume that a colossal lump sum to plug a gap will fall daintily from the air, as it has in some past years, for the 2024 budget).

The popular assumption of absolute entitlement, when it should have been ENTIRELY possible for a few imaginative members of communities to collaborate – to provide a complementary service.

Imagine: a complementary, community-driven kmods repo for just one architecture/platform, for just one version of the OS. This would have set a good example. Excite people, positively, to do more. To do good.

Essentially: be nice.

Gain colossal #thanks (not necessarily on a Tuesday in BSD Cafe).

Imagine: a place, an extraordinary Mastodon instance perhaps, where communities overlap.

Rewind. History. Did we have hordes screaming about the Project not providing torrents? No. Simple. We have respectable torrents that are not Project-provided – and the vast majority of people are entirely happy with this somewhat unofficial arrangement.

Fast-forward. Some awkwardness.

Now, people can know why my "no particular order" at <mastodon.bsd.cafe/@grahamperri…> was a lie with regard to the second thing there: torrents. Big #thanks to those people.

Me being #uppity here, today, does not dilute the Tuesday #thanks there.

Peace

cc @lproven @emaste @stefano @FreeBSDFoundation @TomAoki #FreeBSD #community


In no particular order …

Eric Turgeon – for package repositories, and more.

John-Mark Gurney, Vincent Milum Jr, and others – for FreeBSD-related torrents, seeds, and so on.

Thomas Hurst – for FreshBSD.

Dan Langille – for FreshPorts.

Lucas Holt – for MidnightBSD.

Dmitry Salychev, Grégory Reinbold, Lars Engels, Marcel Kaiser, and Mehmet Mert Gunduz – for NomadBSD.

Shawn Webb – for co-founding HardenedBSD.

JT Pennington, Ken Moore, Kris Moore, and others – for PC-BSD, Lumina Desktop, TrueOS, TrueNAS, and more.

Bob Bruce, David Greenman, Jack Velte, Jim Mock, Jordan Hubbard, Murray Stokely, Pat Rietz, Rod Grimes, Theresa Elam, and others – for FreeBSD Mall, formerly known as Walnut Creek CDROM, and more.

Thom Holwerda – for OSnews.

Jesse Smith – for DistroWatch.com.

Stefano Marinelli – for BSD Cafe, and more.

Liam Proven – for no bullshit, by him on, you know.

pfSense of humor – for an excellent sense of humour, and more. I'm tempted to say "nothing more", but that'll be stretching his sense of humour. Too far.

The shortlist above should be much, much longer. Eye-wateringly long. The full list would take so long to compile that Wednesday will arrive in all time zones on planet Earth before I finish. Fact: if I don't click 'Post' ― now ― I'll make myself late for work. Again.

E&OE

#ThankYouTuesday


in reply to Graham Perrin

@grahamperrin @lproven @emaste @stefano @FreeBSDFoundation
Just my thought, what should be prioritized would be chosen by:
*How many users (including non-newbies) are affected?
*Does it meet the philosophy of the project?
*Does it feasible (in budgets, human resources, ...)
Not everything.😉
in reply to TomAoki

@TomAoki @grahamperrin @emaste @stefano @FreeBSDFoundation

I do not understand what "does it feasible" means. :-(

I upgraded FreeBSD 13.3 to 14.0. It broke X.org on 2 testbed laptops.

As a direct result I skipped testing or writing about FreeBSD 14.0 or 14.1. Some months after the event I managed to manually get X11 working on my machines but my display manager was gone. I had to log in from the command line like it was 1989 or something.

When 14.2 came out I tried again.

I upgraded to 14.1. It was fine. I upgraded that to 14.2.

This time *it broke the console*. Not merely X but the text console.

I have only been experimenting with FreeBSD for about 3 years now but I am already becoming _deeply_ disenchanted with it.

I write for a living. I have had many angry FreeBSD users writing to me, on email and in social media, denying that what happened to me is possible. I also get that in my articles' comments.

So this year I went to the EuroBSDCon and talked in person to some of the core team and I learned that I was in fact right, that there are undocumented system requirements, and that the things I've found are mostly real but are not considered important.

It is *not good.*

in reply to lproven

@lproven @TomAoki @grahamperrin @emaste @FreeBSDFoundation I use FreeBSD (and other BSDs) mainly on servers. I also have some working workstations, and on these, in some cases, I too had to roll back and revert to a previous boot environment because the machine could no longer reach the login page. A couple of weeks ago, I updated an old laptop with Mint Linux, and when upgrading to version 22, everything seemed to go smoothly, except that it wouldn’t boot anymore. It stops somewhere, I haven't had time to investigate, but it's in the drawer, ready for testing.
This is why I’ve always believed in what Liam experienced personally. As long as you stick to standard configurations, the chances of everything going smoothly are definitely higher, but on laptops or with specific devices (which doesn’t mean rare ones), it’s likely that things will break, at least temporarily. On servers, however, I rarely have problems (unlike with various Linux distributions).
In my opinion, the problem exists but it’s not unique to FreeBSD. The Foundation has announced that there will be a focus on this, and let's hope it bears good results.
in reply to lproven

@lproven assuming that you used freebsd-update (the norm) for upgrades to 14.0: IMHO what's close to worst, for your case and countless others, is that we can never truly know WHY.

Because: no log. That's undeniably not good. Istically.

Optimistically: <bugs.freebsd.org/bugzilla/show…>.

Realistically: it'll not be fixed. Axe candidate, abandoned by author five years ago, et cetera.

Futuristically: pkgbase. Logs, et cetera.

That sort of covers the least of your points. Lazily, but truthfully.

I might touch upon other points, later. Hoping to get some sleep, and headspace, before 18:00 today …

mastodon.bsd.cafe/@grahamperri…

Thanks

Cc @stefano

This entry was edited (1 month ago)
in reply to Graham Perrin

@grahamperrin @stefano

I once asked some Fedora devs why they didn't explain their to-me obscure tool with a 3-or-4-word comparison to Ubuntu, instead of 2 or 3 sentences of technobabble.

They were as shocked as if I asked a Roman Catholic to explain the kosher laws.

I don't know & TBH I don't hugely care why the upgrade nuked a thing I never knowingly installed. Some internal technobabble about source code versions in build systems not being upgraded until something else goes EOL doesn't really cut it in 2024 IMHO.

As for the pkgbase thing: at risk of asking a Jew about Mass or something: can you explain it in easy terms like Debian/APT vs RH or something like that?

AIUI -- not well -- OS updates & upgrades are handled by a totally different tool from packaging? *WHY* FFS?

Now there's a move to a new packaging tool which can do updates too?

It sounds like FreeBSD is struggling to make the transition Debian made in 1999 when "Slink" came out.

Which is entirely on-brand for FreeBSD: staggering along 25 years behind.

And yes, I know, the last 25Y of Linux development has seen just as much bad as good. That's sort of my point here.

in reply to lproven

@lproven @grahamperrin the FreeBSD (and other BSDs) have always shared a concept of keeping the base os and the external packages quite separate. That's the reason why we always had two different programs to manage os and packages upgrades.
The problem is when you have to use the package manager to install some kernel modules...
in reply to lproven

@lproven

「… alternatively, the FreeBSD project gets a subproject going which brings in the WiFi drivers from OpenBSD twice a year or something.」

Smart thinking. +100 to reuse of code, collaboration, and the like.

You're not the first person to ask about OpenBSD. The brief answer, from <wiki.freebsd.org/WiFi/Iwlwifi#…>:

「There were other people looking into this.

When I started there was limited support for various chipsets already supported by iwlwifi, … mangled so that comparing to the original code was no longer possible in an automated way … goals listed above … another driver to change and support (in addition to iwm).」

<old.reddit.com/r/freebsd/comme…>

Cc @TomAoki @emaste @stefano @FreeBSDFoundation

#wireless #wifi #FreeBSD #BSD #OpenBSD #iwlwifi #iwm #iwx

Unknown parent

Graham Perrin

@nuintari

「… 13.2 to 14.0 (IIRC) completely broke shit, and required a patch to 13.2 so the upgrade to 14.0 wouldn't shit itself. …」

<freebsd.org/releases/14.0R/> installation information – in November 2023 – gave us the orderly steps that should have avoided such problems.

A May 2022 report <bugs.freebsd.org/bugzilla/show…> requested a relevant improvement to the FreeBSD Handbook. It took around twenty-two months for the improvement, with modifications, to land. Whilst any number of people might describe the wait – and consequences – as shitty, I do not lose respect for the people who ultimately get things done.

Thanks to committers and all others concerned; R.I.P. reviewer Mike Karels.

<github.com/freebsd/freebsd-src…> (February 2024) might be viewed as overlapping; you may blame me for part of the delay.

Cc @stratacast @pinkopanterata @joel @pauamma @emaste @meena @bsdimp @lme

Unknown parent

stratacast
@nuintari @pinkopanterata @vermaden I'm worried about the 13.x to 14.x jump because I have one backup server on Hetzner and I'm afraid of what will happen if I can't boot it :D
Unknown parent

ClaudioM
@mms As I understand it, the issue is because the drm-kmod package is built against the 14.1 kernel. Since the one built against the 14.2 kernel isn't available as a binary package yet, that's where things go sideways, and you have to build the package from ports so it's compiled for the 14.2 kernel. @emaste @gumnos @vermaden @joel
Unknown parent

Ed Maste
@mms @claudiom @gumnos @vermaden because you'll be using the drm-kmod package from the -latest set which builds ~weekly against the up-to-date FreeBSD src main branch, or more likely you're building drm-kmod yourself when you update and build a new kernel
Unknown parent

Ed Maste
@mms @claudiom @gumnos @vermaden The drm-kmod module needs to match the kernel that's running (needs to be built against the kernel headers for the running kernel). Building from ports forces that to be true, for both releases and branches.
Unknown parent

Joel Carnat ♑ 🤪
@dch I was wondering how servers could be impacted too. As I understand it, any kmod could be concerned. Which means some network cards, kvm clock or even some raid controller (according to a search in packages). But it seems only kmod-drm is concerned; or maybe that’s the only thing I use and then noticed.
@dch