GLIBC 2.41 - Am I the only one seeing issues?

OCD-Insomniac · 2025-04-07 18:33:31

Ever since the GLIBC 2.41 update, I've seen so many issues arise. Is it just me...or has this update created issues far and beyond the "DISCORD" specific note on the landing page?

A note about potential workarounds for those who are fairly new to the Arch world regarding this issue...

Most have pointed at using patchelf or execstack to "fix" the .so file in question. It seems to work a lot of the time, but not all the time.

Here's the terminal commands if you need them: (you should double-check them...but I think they're right)

sudo patchelf --clear-execstack <path to your .so file>
-or-
sudo execstack -c <path to your .so file> <----AUR install

The other suggestion that has me sweating bullets is to downgrade glibc back to 2.40 or an even earlier version.
I'm super nervous about trying that so I'm asking the community..those who know more than me..is that even a safe option or will doing that downgrade break my Arch? I remember years ago when we had glibc problems, and downgrading was something I tried, but I also remember it being one helluva ride to get things back in order...ultimately resulting in a full reinstall. Not a pleasant memory.

I'm not usually one to ask...rather prefer usually to do my research and figure it out myself, but I'm not finding much other than those 3 suggestions and the top two hasn't fixed my IPEX issue in several python env packages.

Thanks to anyone willing to share their experience.

mackin_cheese · 2025-04-07 18:40:40

never downgrade a core package. it's a core package for a reason, you downgrade that and you will break your system. This is why partial upgrades are not supported.

OCD-Insomniac · 2025-04-07 18:47:36

mackin_cheese wrote:

never downgrade a core package. it's a core package for a reason, you downgrade that and you will break your system. This is why partial upgrades are not supported.

I appreciate the confirmation. If this is anything like the last time, it'll take awhile for things to mellow. Guess I'll just have to work on other projects that are still working for me.

Thanks again for the reply.

seth · 2025-04-08 07:55:53

Ever since the GLIBC 2.41 update, I've seen so many issues arise. Is it just me...or has this update created issues far and beyond the "DISCORD" specific note on the landing page?

"Big issues, strong issues, tears pouring down its… "

What "many issues" specifically and why do you think it's related to the glibc update?
The most trouble-prone area would be third-party binary only executables that where compiled and linked against an older glibc (like discord in february)

OCD-Insomniac · 2025-04-08 15:00:11

What "many issues" specifically and why do you think it's related to the glibc update?

Well...as I said, I noticed projects broken as soon as I updated. Since I had written a quick and dirty script that keeps a manifest of every package updated so I took a look at it. That is usually a key hint when something breaks...yanno...something that changed..like an update, wouldn't you say? So..yes, that triggered my initial thought that something in the manifest list was responsible. Then I started looking more closely at the error logs. Specifically, ChromaDB in my python project which led me to check out other parts of my requisites...like IPEX.

In my attempt to narrow down where I need to dig deeper, I found this post: https://github.com/intel/intel-extensio … 2719897630. The explanation makes sense and relates to what I was seeing in my ChromaDB logs. The post also links to the GLIBC v2.41 release notes...which shows that the GNU dev team knew there would be packages that would fail outright (see the Depreciated and removed features section).

Additionally, I did temporarily revert GLIBC and ran my project...and it then ran without error. Sooo...yeah, pretty safe bet the issue lies there. I was too chicken to leave the downgraded GLIBC version installed so I re-updated and continued on.

Since I hadn't really seen discussion in the Arch community about it...I was second guessing my assessment despite my testing and research. I thought I'd ask if anyone else here has noticed effects beyond the very specific DISCORD mention on the landing page. I was hoping someone had already figured it out and would offer another potential solution or workaround to this issue I'm having. After all, this is where the Arch experts are (and those who think they are)...so I asked here.

That's my particular case...but to address your initial comment about "Big issues, strong issues, blah blah bah..." To get your list of "\"many issues\" specifically", you need only do a quick google search for "GLIBC 2.41 library issue" to get a pretty good idea that it isn't localized to me alone. For the record, I never said what you quoted...nor did I present my question as such. I have no idea why you elected to feature that statement as your opener...but whatever. You do you....I guess.

seth · 2025-04-08 15:03:32

Well...as I said, I noticed projects broken as soon as I updated.

Where?

ran my project...

But just to be clear, the problem is limited to some local code?
Did you at least recompile it (make clean; make) afterh the update?

OCD-Insomniac · 2025-04-08 15:24:46

To be clear...my particular projects don't usually require compiling. They are super simple little scripts and mods with very little code. They usually just redirect or change locations to match my framework..nothing that really changes how packages functions.

I don't fancy myself as a coder/programmer. Just know enough to accomplish simple goals. The python packages I mentioned..like ChromaDB, and especially IPEX...well, they might require a little nudging to play nice..like importing IPEX right after pytorch, but I don't go deep into their code. In the example I gave, it was ChromaDB's python script for loading their miniLLM that threw fits about onnxruntime. It seems the library can't be seen which so I'm guessing that their script must be trying to execute the onnxruntime stack but can't..possibly due to the stack not being inherently executable as stated in the glibc v2.41 release notes, maybe. I don't know. Like I said, I'm not a programmer.

My understanding is that the new GLIBC, has removed the execution function from dlopen and dlmopen...so if either ChromaDB or onnxruntime were reliant on one of them to execute the onnxruntime stack, it would fail...like it is. Again, I dunno..just seems to make sense to me with my limited knowledge.

seth · 2025-04-08 20:11:29

"GLIBC 2.40 library issue" will give you a handy amount of problems, that's proof of absolutely nothing.

Is python-chromadb from the AUR or a pip, did you re-install, rebuild /that/ (and other non-repo resources you rely on)?

But if your "issues" fit https://github.com/intel/intel-extensio … 2718218146 then https://github.com/intel/intel-extensio … 2764740976 would also work for you?

Also: https://bbs.archlinux.org/viewtopic.php?id=57855
(It explains why dealing in vague claims about "many issues" is a waste of time - and kinda the MO of the *sigh* inescapable orange-man)

tekstryder · 2025-04-09 12:47:08

OP: you may be interested in a related upstream bug report filed by an Arch maintainer:

• https://sourceware.org/bugzilla/show_bug.cgi?id=32653

And very recent glibc merge to address some execstack backward compatibility:

• https://sourceware.org/git/?p=glibc.git … 22a079771e

Arch Linux

#1 2025-04-07 18:33:31

GLIBC 2.41 - Am I the only one seeing issues?

#2 2025-04-07 18:40:40

Re: GLIBC 2.41 - Am I the only one seeing issues?

#3 2025-04-07 18:47:36

Re: GLIBC 2.41 - Am I the only one seeing issues?

#4 2025-04-08 07:55:53

Re: GLIBC 2.41 - Am I the only one seeing issues?

#5 2025-04-08 15:00:11

Re: GLIBC 2.41 - Am I the only one seeing issues?

#6 2025-04-08 15:03:32

Re: GLIBC 2.41 - Am I the only one seeing issues?

#7 2025-04-08 15:24:46

Re: GLIBC 2.41 - Am I the only one seeing issues?

#8 2025-04-08 20:11:29

Re: GLIBC 2.41 - Am I the only one seeing issues?

#9 2025-04-09 12:47:08

Re: GLIBC 2.41 - Am I the only one seeing issues?

Board footer