Framework 16 lockups and flakiness

Hello! So I got a Framework 16, I love it, it is wonderful. It has a bunch of apparent hardware flakiness though. I’ve been able to work around everything I’ve run into so far, but I wanted to offer it as feedback and I’m happy to help track it down if anyone from Framework is interested in hearing more details or wants me to try anything.

This seems to be necessary in order to prevent hard system crashes (with even alt-SysRq not working):

  boot.kernelParams = [
    "amdgpu.dc=1"           # Enable display core
    "amdgpu.gpu_recovery=1" # Enable GPU recovery on hangs
    "amd_pstate=passive"    # Sometimes active mode causes issues
  ];

One very easy way to reproduce the crash is going to claude.ai and interacting with it, using Librewolf. Something that Claude does on Librewolf causes the crash almost immediately. Setting layers.acceleration.disabled in the browser settings seems to avoid the crash.

The network adapter seems to shut down and require reconnect very frequently, or have trouble connecting. Setting:

sudo iwconfig wlp1s0 power off

… to disable power management seems to resolve the network adapter issues, and it just works.

It also seems to have trouble with suspend and resume. Booting always from a total fresh startup, and enabling all the above workarounds, seems to make it stable. But shutting the lid and then opening up the lid still sometimes causes a hard crash within 10-20 minutes.

Like I say, I’ve done workarounds so that things seem to be working, just wanted to pass the word along. If it’s helpful for me to create tickets or anything for any of the above issues I’m happy to do so. I’m up to date on all firmware updates.

Which Linux distro are you using?

NixOS

Which release version?

24.11.718829.ad5c70bcc5cc (Vicuna)

Which kernel are you using?

6.6.93

Which BIOS version are you using?

03.05

Which Framework Laptop 16 model are you using?

Framework Laptop 16 DIY Edition (AMD Ryzen™ 7040 Series), Ryzen™ 7 7840HS

I’ve had an FW16 for over a year now, and in fact, now have two. The original has been running Ubuntu 24.04 the whole time; the new one is running Windows 11 (not my choice, it belongs to my employer).

Just wanted to say that I haven’t seen any of the flakiness that you’re reporting here. Maybe it’s related to NixOS, or some difference in the hardware (I don’t have the discrete graphics module, maybe you do?), or maybe you’ve received a slightly flaky motherboard?

Yeah, could be drivers, too. I don’t really know; I’m just reporting what is happening with my laptop.

Update: I tried with kernel 6.15.3 after seeing that recommended in another thread, and it seems like that may have resolved the GPU issues at least. I’ll keep an eye on it but that might be the solution.

2 Likes

Update: Okay, it’s not resolved after all. I just got a GPU crash (lockup of the system with graphical corruption, can’t switch to console, alt-sysrq still working though) with:

  • Kernel 6.15.3
  • Mesa 25.1.4
  • All firmware up to date

That’s after a fresh boot, no suspend or resume involved.

Kernel 6.15.3 also seems to still have the network flakiness (frequent disconnections). The workarounds I talked about above still seem to work and produce a stable system, on either kernel version, but without them even new kernels show the bad behavior on both. (edit: see below)

I haven’t investigated the suspend/resume issues yet, since I’m still trying to deal with the other brokenness.

Edit: Wait, I think I am wrong, I think the crash I just saw was with the amdgpu parameters set that previously had been working to prevent the crash (and with kernel 6.15.3). And, I just saw Claude lock up in the way that it was doing when the workarounds were not enabled, with everything at most recent versions and with the workaround kernel parameters.

I think the only config I have tried that hasn’t crashed yet is kernel 6.6.93 with the amdgpu workaround parameters. I’ll go back to that configuration and see if it stays stable…

I’d recommend also possibly making a ticket to our Support team if you continue to experience crashes.

1 Like

Done, I made two separate tickets for the issues with the GPU and the wireless.

1 Like

Update: I’ve given up and ordered an Intel wireless card. Apparently these Mediatek cards are known to be very poorly supported on Linux in general. Honestly, I’m not real happy with this laptop anymore. I paid $1700 for a laptop which can’t manage to cope with maintaining a network connection or running normal stuff without crashing. I tried to go through support, and they sort of did everything “right” I guess; they won’t help me with the network issues because I can’t get it to happen under Ubuntu, which I guess is fair, and for the GPU lockups they want me to take apart my computer, muck around with the mainboard, take all the expansion cards out and run it that way for a few days and try to get the crash to happen again, and that’s after a bunch of other general debugging steps and running for days at a time in configurations which make it a little awkward for me to get stuff done, trying to reproduce the crash. They also won’t tell me when I asked what is the full scope of how much I might need to do in terms of making a part-time job out of debugging this thing on my end before they give me some kind of substantive help with it.

I’m now trying to just deal with upstream about the GPU crashes instead, under the maybe wrong assumption that it’s a driver bug (since resetting the GPU brings it back to working, and presumably this all works under Windows). I thought I had it fixed running a kernel from the AMD people but this morning it crashed and I lost some more work. I may at some point cave and just do the whole process frame.work support is trying to ask me to do, and just see what happens as a result of that. But in general, my laptop’s still not working and I’m not real happy about how productive it’s been trying to get help with it.

Hi @Andrew_Moise,

I am sorry to hear about your experiences with this issue.

I am looking at the ticket now (graphics crashing). Absolutely agreed, this is frustrating. Looks like it did not make it to my team and was still with the other agents.

Right off the bat, I am seeing why the issue likely occurred. Your ticket is closed, but I would be happy to help you understand why it happened, what can be done and see if we can get you setup with a path forward.

Let me know.

Matt Hartley
Linux Support Lead

Sure, sounds good. What should I do, open another ticket?

No, I have your GPU ticket in front of me, I will reopen and reply.

Sure, sounds good. What do you want me to do?

Email sent. :slight_smile: