Framework Laptop 13 Ryzen 7040 BIOS 3.05 Release and Driver Bundle

[mikeymop]

After about a day’s worth of use, the only alarming issue I see is that dmesg is spammed with

xhci_hcd 0000:c1:00.3: Refused to change power state from D0 to D3hot

Is this the IRQ1 failure mentioned in the known issues?

[Rei]

So far don’t encounter any issues with Game mode turned off.
I never set sg_display=0

[TRACKING] Graphical corruption in Fedora 39 (AMD 3.03 BIOS)

I disabled UMA_GAME_OPTIMIZED today. Btw its called “Gaming” after 3.05. I tried to trigger the corruption, like I did last time… open many tabs running YouTube, Totem running 1080p video, and plugging and unplugging my usb-c dock with monitor connected. I opened 30 tabs in the background and plug/unplug dock, close/open lid, on/off fractional scaling, open/close lid - didn’t trigger white graphical corruption. I used mine whole day without any issue. worth noting, I’m on kernel 6.8, bios 3.…

[Rijnder_Wever]

@mikeymop

Is this the IRQ1 failure mentioned in the known issues?

No, the error describes an usb device/port that does not want to go in or out of a particulair sleep state.

[Rijnder_Wever]

I’m not entirely sure if this relates to the new 3.05 BIOS, but hibernation is broken on my end (it worked fine a while ago). Specifically, dmesg is flooded with errors, and my laptop wakes up not long after running systemctl hibernation. Could be merely coincidental, however.

Any else has having hibernation problems on linux after installing the new BIOS?

[sanch3z]

I’m having the same issue on kernel 6.7.9.

Apr 07 12:57:25 kernel: tsc: Fast TSC calibration using PIT
Apr 07 12:57:25 kernel: tsc: Detected 3293.768 MHz processor
Apr 07 12:57:25 kernel: clocksource: refined-jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 7645519600211568 ns
Apr 07 12:57:25 kernel: clocksource: hpet: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 133484873504 ns
Apr 07 12:57:25 kernel: clocksource: tsc-early: mask: 0xffffffffffffffff max_cycles: 0x2f7a4d5e84a, max_idle_ns: 440795314252 ns
Apr 07 12:57:25 kernel: clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 7645041785100000 ns
Apr 07 12:57:25 kernel: clocksource: Switched to clocksource tsc-early
Apr 07 12:57:25 kernel: clocksource: acpi_pm: mask: 0xffffff max_cycles: 0xffffff, max_idle_ns: 2085701024 ns
Apr 07 12:57:25 kernel: tsc: Refined TSC clocksource calibration: 3293.769 MHz
Apr 07 12:57:25 kernel: clocksource: tsc: mask: 0xffffffffffffffff max_cycles: 0x2f7a4deb66d, max_idle_ns: 440795294980 ns
Apr 07 12:57:25 kernel: clocksource: Switched to clocksource tsc
Apr 07 12:57:25 kernel: kvm_amd: TSC scaling supported
Apr 07 12:58:36 kernel: clocksource: timekeeping watchdog on CPU6: Marking clocksource 'tsc' as unstable because the skew is too large:
Apr 07 12:58:36 kernel: clocksource:                       'hpet' wd_nsec: 511992864 wd_now: 409f1946 wd_last: 402f3d50 mask: ffffffff
Apr 07 12:58:36 kernel: clocksource:                       'tsc' cs_nsec: 512540105 cs_now: 49d42e9653 cs_last: 496f8edfc6 mask: ffffffffffffffff
Apr 07 12:58:36 kernel: clocksource:                       Clocksource 'tsc' skewed 547241 ns (0 ms) over watchdog 'hpet' interval of 511992864 ns (511 ms)
Apr 07 12:58:36 kernel: clocksource:                       'tsc' is current clocksource.
Apr 07 12:58:36 kernel: tsc: Marking TSC unstable due to clocksource watchdog
Apr 07 12:58:36 kernel: TSC found unstable after boot, most likely due to broken BIOS. Use 'tsc=unstable'.
Apr 07 12:58:36 kernel: clocksource: Checking clocksource tsc synchronization from CPU 8 to CPUs 0,6-7,10,12-14.
Apr 07 12:58:36 kernel: clocksource: Switched to clocksource hpet

[Muggle_Twist]

The skew is quite small it seems. It’s probably safe to run with the watchdog disabled.

tsc=nowatchdog

Add this to your kernel commandline to disable the watchdog

[Matthew_Shapiro]

Yes, in kernels > 6.6 < 6.8.5 (unreleased). Hibernation works fine with Linux kernel 6.6.25 and bios 3.05.

[Matthew_Shapiro]

Graphics work fine with Manjaro KDE 6.6.25 and BIOS 3.05 with sg_display removed from kernel command line.

Game mode is still enabled.

[Hussain_Arif]

No issues so far! I normally get a white screen on 2 hours of uptime. Currently on 2 hrs 49 mins uptime. Thanks for your work!

[Ulmondil]

I have been using the beta BIOS for 3 days with Ubuntu 23.10, kernel 6.5, ryzen 5. Thanks to FW for their work.

@tbd

Can someone confirm if this message still shows up in dmesg?

Same message.

@Mario_Limonciello

People who have needed game mode or sg_display=0 for the white screen issue can you please back both of these out?

I tried removing first sg_display and then game memory mode (btw I might go back to game mode in the future for other reasons). The first time I bootep up, the system hung on the login screen (gdm) and became unusable: no keyboard/mouse response, no tty reachable, I had to reboot via ssh. I tried again and everything worked fine. And again without problem. I now blame gnome/mutter/gdm for the crash and I won’t bother finding the relevant logs, but I will if it happens again. After this first episode now everything looks fine. I’ll report any related problems.

@dimitris

All seems good so far (just minutes into the new BIOS) but it also brought back a wake-back-up-on-lid-close issue that’s almost (not 100%) identical to the original report in this old thread:

I tried and I confirm I have these issues:

• laptop wakes up from s2idle, when the lid is closed, if it was suspended with open lid
• and also, laptop wakes up from s2idle when the power cable is plugged/unplugged during sleep

Tomorrow maybe I’ll try to add the udev rules. And edit here if that workaround solves the issue.

Was this behaviors present also with 3.03? I never noticed/encountered it.

Are the logs from amd_s2idle.py useful for anyone? I have them in case.

[EDIT] test done: with the udev rule everything is ok, thanks @dimitris @Mario_Limonciello

[Mario_Limonciello]

(edit; sorry reread your post more thoroughly).

Yes the udev rules will help your s2idle issues.

[Ferdinand_Bachmann]

Just updated to BIOS 3.05 on my Framework 13 with AMD 7840U, 32GB RAM:

• good: suspend works
• bad: resume on lid-close bug regressed (linux kernel workaround only triggers on BIOS 3.03), so I had to re-add the udev rule
• good: hibernate works
• good: graphical glitches / white screen issue did not regress (was fixed with BIOS 3.03 and kernel 6.8 for me already; Game Mode is OFF in BIOS, and no sg_display=0 kernel cmdline option)
• good: no clocksource “TSC unstable” errors in boot log

[abittner]

@Kieran_Levin

Linux/LVFS

We are waiting for metadata to rebuild in 4 hours

Updating via LVFS is available in the testing channel during the beta period.

I dont quite understand this part about lvfs, why is this 3.05 beta? when the beta? was 3.03? in that other thread ( Framework 13 AMD Ryzen 7040 BIOS 3.03b BETA - #124 by Wade ) ? or is only the lvfs delivery still considered beta for this 3.05 for an initial time span?

thanks.

[Azure]

@Rei

why is it called BETA? is it not a final release?

@abittner

why is this 3.05 beta? when the beta? was 3.03? in that other thread ( Framework 13 AMD Ryzen 7040 BIOS 3.03b BETA - #124 by Wade ) ? or is only the lvfs delivery still considered beta for this 3.05 for an initial time span?

For anyone who is new to Framework’s BIOS updates, all updates are released in a Beta for two weeks to see if there are any initial issues that the team were not able to be catch in internal testing. It is not recommended to download at this time unless you are okay with issues that could come up, but it allows the people who are willing to test it to do so. If there are no major issues in two weeks, it will be made an official release that will be available on the BIOS update support page here.

[Hussain_Arif]

Update: I currently have an issue where the laptop doesn’t detect the external monitor if the monitor was plugged in before the laptop was turned on.
I have to plug the monitor after the laptop has booted. Other than that, everything is fine

[abittner]

this might possibly also be highlevel software (OS) related I guess? what OS etc are you enjoying? here be details. might help them devs and debuggers.

[Hussain_Arif]

Fedora 39, latest kernel(6.8.4). This bug didn’t occur on the previous bios. Since there might be a software issue, I’ll do a liveboot of Fedora with the monitor already plugged in.

Update: Seems to have fixed itself.

[Kieran_Levin]

@Meddlemaker

However, when I went back to turn off booting from USB in the bios options, password I used before now no longer works. Used to show when typing up to 10 characters, now there is 64 character p/w indicator. I can’t edit bios anymore. I’m pretty sure I’m not mistyping or leaving all caps on, or something like that. Even walked and took a few hours break before trying again.

Do you know what your previous bios was? We would like you investigate your report more.

You can try a full bios defaults reset. You have to open the input cover, find the small switch in the center top of the mainboard. And press it for 1 second, release for 1 second, 10 times, then power on to perform a full bios reset.

[Wade]

Unfortunately I got an error during install and the laptop no longer POSTS etc:

Please advise, I used fwudmgr, ArchLinux, LTS kernel. Updating from the previous beta from a few days ago.

Update: Support has handled this on an existing ticket and may be related to other issues.

[rage_311]

My BIOS upgrade from Windows 11 and new driver installation went flawlessly. Thanks!

Is there any advantage to installing the Framework provided GPU/Adrenalin driver that’s on the 23.X major version vs the driver available from AMD on the 24.X versioning? Guaranteed compatibility, performance, something else?