Freezes / Hangs / Frozen screen from a recent update - no input, 2 second freezes to 5 minutes

[masakk1]

I have been getting some VERY annoying freezes. Stopping my work way too frequenctly, it's hard to work like this without loosing my mind.

When did this start?

It's been a problem for the last week or so. But both yesterday and today felt like the worst. I get varied errors, and in this post I only focus on the ones that seem to be due to amdgpu.

My hardware and configuration

with ujust bios-info

Manufacturer: ASUSTeK COMPUTER INC.
Product Name: PRIME B550M-K
Version: 3302
Release Date: 08/01/2023

fastfetch

bluefin-dx:stable
Bluefin (Version: 43.20260303)
Linux 6.18.7-200.fc43.x86_64
3 hours, 53 mins
Forged on Jan 07 2026

AMD Ryzen 5 5600 (12) @ 4.47 GHz
AMD Radeon RX 9070 XT [Discrete] -> AMDGPU
5.38 GiB / 23.37 GiB (23%) -> RAM seems more than enough
901.36 GiB / 928.93 GiB (97%) - btrfs -> I guess it's kinda full? This happened before I installed a game, though!
29.31 GiB / 114.65 GiB (26%) - exfat -> a USB drive
1920x1080 in 24", 60 Hz [External]
2560x1440 in 27", 165 Hz [External] *
Keychron  Keychron Link

GNOME 49.4
Mutter (Wayland)
bash 5.3.0
Ptyxis 49.3
118 (flatpak), 110 (brew), 10 (brew-cask)

What I have tried

1. Asking dosu on the "Ask bluefin" page on the docs. Without much success, other than showing the errors attached, I guess.
2. Searching discord. Including bazzite server. No one seems to report the exact same errors.
3. Disabling extensions
4. Removing the experimental fractional-scaling in gsettings
5. Match both monitor's framerate to 60Hz
6. Turning on and off
7. Doing Ctrl-Alt-3 and Ctrl-Alt-2 didn't work. It doesn't bring me to the TTY
8. Forcing an update on the system

Errors

dmesg shows many different errors after each freeze. But they are mostly this:

• the ringy

[ 9024.113659] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring gfx_0.0.0
[ 9180.457692] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring gfx_0.0.0
[ 9257.552702] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring gfx_0.0.0

• Ignore frame

[ 9259.325514] xhci_hcd 0000:0f:00.3: Frame ID 1491 (reg 11946, index 3) beyond range (1494, 340)
[ 9259.325519] xhci_hcd 0000:0f:00.3: Ignore frame ID field, use SIA bit instead
[ 9259.329521] xhci_hcd 0000:0f:00.3: Frame ID 1491 (reg 11978, index 4) beyond range (1498, 344)
[ 9259.329524] xhci_hcd 0000:0f:00.3: Ignore frame ID field, use SIA bit instead
[ 9259.333527] xhci_hcd 0000:0f:00.3: Frame ID 1491 (reg 12010, index 5) beyond range (1502, 348)
...
[ 9259.349666] xhci_hcd 0000:0f:00.3: Ignore frame ID field, use SIA bit instead
[ 9259.349668] xhci_hcd 0000:0f:00.3: Frame ID 1496 (reg 12139, index 41) beyond range (1518, 364)
[ 9259.349669] xhci_hcd 0000:0f:00.3: Ignore frame ID field, use SIA bit instead
[ 9259.349671] xhci_hcd 0000:0f:00.3: Frame ID 1496 (reg 12139, index 42) beyond range (1518, 364)
[ 9259.349673] xhci_hcd 0000:0f:00.3: Ignore frame ID field, use SIA bit instead

• Even more fence thingy (as you can see, this happened in different timeframes, though not too far from each other) — I'm attaching the a file with the whole error, because it's a lot. I can't tell how much of it is useful

[ 9260.278627] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring gfx_0.0.0
[ 9693.384639] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring gfx_0.0.0
[10069.408390] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring sdma1
[10071.241388] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring sdma0
[10302.897297] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring gfx_0.0.0
[10305.768295] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring gfx_0.0.0
[10443.480257] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring sdma1
[10444.189064] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring gfx_0.0.0

• apparently it even transfer to the CPU? This happens PER CORE, so 8 times in total. I'm only showing the last

IT STARTS WITH SOMETHING LIKE THIS
[ 8170.023315] amdgpu 0000:0d:00.0: amdgpu: Fence fallback timer expired on ring sdma1
[ 8185.419310] watchdog: CPU4: Watchdog detected hard LOCKUP on cpu 4
[ 8185.419313] Modules linked in: exfat uas usb_storage nf_conntrack_netlink xt_nat veth xt_conntrack xt_MASQUERADE bridge stp llc xt_set ip_set xt_addrtype nft_compat uinput snd_seq_dummy snd_hrtimer nf_conntrack_netbios_ns nf_conntrack_broadcast nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_tables qrtr sunrpc binfmt_misc vfat fat intel_rapl_msr amd_atl intel_rapl_common snd_hda_codec_alc882 edac_mce_amd snd_hda_codec_realtek_lib snd_hda_codec_atihdmi snd_hda_codec_generic snd_hda_codec_hdmi ee1004 kvm_amd snd_hda_intel snd_usb_audio snd_hda_codec kvm snd_usbmidi_lib snd_hda_core snd_ump snd_intel_dspcfg eeepc_wmi snd_rawmidi snd_intel_sdw_acpi asus_wmi snd_hwdep snd_seq sparse_keymap irqbypass snd_seq_device platform_profile rapl rfkill snd_pcm r8169 wmi_bmof i2c_piix4 pcspkr snd_timer snd k10temp soundcore i2c_smbus realtek joydev gpio_amdpt gpio_generic scsi_dh_rdac scsi_dh_emc tun scsi_dh_alua iptable_nat ip_tables nf_nat nf_conntrack
[ 8185.419364]  nf_defrag_ipv6 nf_defrag_ipv4 loop dm_multipath nfnetlink zram lz4hc_compress lz4_compress overlay erofs netfs amdgpu amdxcp drm_panel_backlight_quirks gpu_sched drm_buddy drm_ttm_helper ttm video drm_exec i2c_algo_bit drm_suballoc_helper drm_display_helper nvme polyval_clmulni ghash_clmulni_intel nvme_core cec sp5100_tco nvme_keyring nvme_auth hkdf wmi zfs(POE) spl(OE) v4l2loopback(OE) videodev mc pkcs8_key_parser fuse i2c_dev
[ 8185.419391] CPU: 4 UID: 0 PID: 0 Comm: swapper/4 Tainted: P           OE       6.18.7-200.fc43.x86_64 #1 PREEMPT(lazy) 

... IT GOES ON FOR A WHILE ...

[ 8363.978694] watchdog: BUG: soft lockup - CPU#8 stuck for 26s! [firefox-bin:59635]
[ 8363.978695] CPU#8 Utilization every 4000ms during lockup:
[ 8363.978696] 	#1: 100% system,	  0% softirq,	  0% hardirq,	  0% idle
[ 8363.978697] 	#2: 100% system,	  0% softirq,	  1% hardirq,	  0% idle
[ 8363.978699] 	#3: 100% system,	  0% softirq,	  0% hardirq,	  0% idle
[ 8363.978700] 	#4: 100% system,	  0% softirq,	  1% hardirq,	  0% idle
[ 8363.978701] 	#5: 100% system,	  0% softirq,	  0% hardirq,	  0% idle
[ 8363.978702] Modules linked in: exfat uas usb_storage nf_conntrack_netlink xt_nat veth xt_conntrack xt_MASQUERADE bridge stp llc xt_set ip_set xt_addrtype nft_compat uinput snd_seq_dummy snd_hrtimer nf_conntrack_netbios_ns nf_conntrack_broadcast nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_tables qrtr sunrpc binfmt_misc vfat fat intel_rapl_msr amd_atl intel_rapl_common snd_hda_codec_alc882 edac_mce_amd snd_hda_codec_realtek_lib snd_hda_codec_atihdmi snd_hda_codec_generic snd_hda_codec_hdmi ee1004 kvm_amd snd_hda_intel snd_usb_audio snd_hda_codec kvm snd_usbmidi_lib snd_hda_core snd_ump snd_intel_dspcfg eeepc_wmi snd_rawmidi snd_intel_sdw_acpi asus_wmi snd_hwdep snd_seq sparse_keymap irqbypass snd_seq_device platform_profile rapl rfkill snd_pcm r8169 wmi_bmof i2c_piix4 pcspkr snd_timer snd k10temp soundcore i2c_smbus realtek joydev gpio_amdpt gpio_generic scsi_dh_rdac scsi_dh_emc tun scsi_dh_alua iptable_nat ip_tables nf_nat nf_conntrack
[ 8363.978743]  nf_defrag_ipv6 nf_defrag_ipv4 loop dm_multipath nfnetlink zram lz4hc_compress lz4_compress overlay erofs netfs amdgpu amdxcp drm_panel_backlight_quirks gpu_sched drm_buddy drm_ttm_helper ttm video drm_exec i2c_algo_bit drm_suballoc_helper drm_display_helper nvme polyval_clmulni ghash_clmulni_intel nvme_core cec sp5100_tco nvme_keyring nvme_auth hkdf wmi zfs(POE) spl(OE) v4l2loopback(OE) videodev mc pkcs8_key_parser fuse i2c_dev
[ 8363.978764] CPU: 8 UID: 1000 PID: 59635 Comm: firefox-bin Tainted: P           OEL      6.18.7-200.fc43.x86_64 #1 PREEMPT(lazy) 
[ 8363.978767] Tainted: [P]=PROPRIETARY_MODULE, [O]=OOT_MODULE, [E]=UNSIGNED_MODULE, [L]=SOFTLOCKUP
[ 8363.978768] Hardware name: ASUS System Product Name/PRIME B550M-K, BIOS 3302 08/01/2023
[ 8363.978768] RIP: 0010:smp_call_function_many_cond+0x117/0x560
[ 8363.978772] Code: c3 73 30 48 63 eb 49 8b 0e 48 81 fd 00 20 00 00 0f 83 fe 03 00 00 48 03 0c ed c0 10 d9 a4 8b 41 08 a8 01 74 09 f3 90 8b 41 08 <a8> 01 75 f7 83 c3 01 eb b3 48 83 c4 40 5b 5d 41 5c 41 5d 41 5e 41
[ 8363.978773] RSP: 0018:ffffce82587439c0 EFLAGS: 00000202
[ 8363.978774] RAX: 0000000000000011 RBX: 0000000000000004 RCX: ffff8e3416e3cc40
[ 8363.978775] RDX: 0000000000000004 RSI: 0000000000000000 RDI: 0000000000000004
[ 8363.978776] RBP: 0000000000000004 R08: ffff8e2f00c3af80 R09: ffff8e2f00c3adc8
[ 8363.978777] R10: ffff8e3238c77dc0 R11: 0000000000000000 R12: 0000000000000001
[ 8363.978778] R13: 0000000000000008 R14: ffff8e34170339c0 R15: ffffffffa6613c40
[ 8363.978779] FS:  00007f9c1e2d9780(0000) GS:ffff8e3470a29000(0000) knlGS:00007f9200000000
[ 8363.978780] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 8363.978781] CR2: 00007f6799898f40 CR3: 000000022205f000 CR4: 0000000000f50ef0
[ 8363.978782] PKRU: 55555554
[ 8363.978783] Call Trace:
[ 8363.978784]  <TASK>
[ 8363.978785]  ? __pfx_flush_tlb_func+0x10/0x10
[ 8363.978788]  on_each_cpu_cond_mask+0x24/0x40
[ 8363.978791]  flush_tlb_mm_range+0x153/0x1f0
[ 8363.978793]  tlb_finish_mmu+0x79/0x1e0
[ 8363.978796]  do_madvise+0xff/0x150
[ 8363.978801]  __x64_sys_madvise+0x2b/0x40
[ 8363.978803]  do_syscall_64+0x7e/0x7f0
[ 8363.978816]  ? srso_alias_return_thunk+0x5/0xfbef5
[ 8363.978818]  ? __x64_sys_poll+0xc4/0x190
[ 8363.978822]  ? srso_alias_return_thunk+0x5/0xfbef5
[ 8363.978824]  ? do_syscall_64+0xb6/0x7f0
[ 8363.978826]  ? srso_alias_return_thunk+0x5/0xfbef5
[ 8363.978828]  ? do_syscall_64+0xb6/0x7f0
[ 8363.978830]  ? srso_alias_return_thunk+0x5/0xfbef5
[ 8363.978831]  ? do_syscall_64+0xb6/0x7f0
[ 8363.978834]  ? srso_alias_return_thunk+0x5/0xfbef5
[ 8363.978835]  ? irqentry_exit_to_user_mode+0x2c/0x1c0
[ 8363.978837]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
[ 8363.978839] RIP: 0033:0x7f9c1df19b6b
[ 8363.978843] Code: 9c ff ff ff e8 c6 77 ff ff 48 8b 55 f8 64 48 2b 14 25 28 00 00 00 75 02 c9 c3 e8 20 19 01 00 f3 0f 1e fa b8 1c 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 85 72 0d 00 f7 d8 64 89 01 48
[ 8363.978844] RSP: 002b:00007ffcdfcd2908 EFLAGS: 00000216 ORIG_RAX: 000000000000001c
[ 8363.978845] RAX: ffffffffffffffda RBX: 000000000000004f RCX: 00007f9c1df19b6b
[ 8363.978846] RDX: 0000000000000004 RSI: 0000000000002000 RDI: 00007f9bd044f000
[ 8363.978847] RBP: fffffffffffff000 R08: 00007f9bd0400a41 R09: 0000000000002000
[ 8363.978848] R10: 0000000000002000 R11: 0000000000000216 R12: 0000000000000002
[ 8363.978849] R13: 0000000000000002 R14: 0000000000000002 R15: 00007f9c1dd00001
[ 8363.978853]  </TASK>
[ 8366.442139] perf: interrupt took too long (208154 > 37855), lowering kernel.perf_event_max_sample_rate to 1000

What matters most

Honestly, this feels like I should rollback to a previous release. I can hold off some security updates until a new kernel version fixes this, or someone knows how to fix it. It is paramount that I get this machine working to finish my studies.

[masakk1]

Here is the full log:
logs20260309_very_long_CPU_hang.txt

[jfmongrain]

Rebasing to latest may be a temporary workaround (kernel 6.18.16) until the kernel version catches up.
It helped on my system, which is similar to yours.

[masakk1]

I was trying to look at bluefin's releases, to check the last kernel update. But it seems like they didn't update this in 3 weeks?
https://github.com/ublue-os/bluefin/releases

Do they not offer their changelogs as releases anymore?

[jfmongrain]

I think there are currently some problems with the automatic system that produce those, that they are working on fixing.

[masakk1]

I think there are currently some problems with the automatic system that produce those, that they are working on fixing.

Is there anywhere I can find the logs, then? I just need to roll back to the version previous to the kernel change.

Or better yet, what version are you using atm? Since you said you had to roll back, as well.

[castrojo]

https://docs.projectbluefin.io/driver-versions may help with this.

[masakk1]

https://docs.projectbluefin.io/driver-versions may help with this.

Seems to be outdated, too. Unfortunately.

[castrojo]

Also it's not a bad thing to stay on a pinned release, that's its primary purpose!

Working > Not working.

I've used a pinned release for an entire kernel in the past, it's just life dealing with regressions.

[castrojo]

#3595 for future improvements.