Arch Linux freezes resuming from sleep

jebez · 2025-08-06 01:10:50

Is the system installed on an F2FS partition?

Yes, see my config under.

/dev/disk/by-id/nvme-WDS250G3X0C-00SJG0_185250422455

Nothing works...

Is it because of F2FS? Will it works on ext4?

Last edited by jebez (2025-08-06 21:28:54)

seth · 2025-08-07 14:51:24

seth wrote:

Bold move… but it's probably rather related to https://wiki.archlinux.org/title/Solid_ … leshooting
"nvme_core.default_ps_max_latency_us=0 pcie_aspm=off iommu=soft" to the https://wiki.archlinux.org/title/Kernel_parameters and if that doesn't cut it, add "pcie_port_pm=off" to the list.

I was mostly asking to make sure we're even looking at the right drive here.
Rule out ASPM and/or APST being the cause.

F2FS isn't exactly as tested and reliable as ext4 or even btrfs, but for now it's not the most obvious contender.
If it's however not the nvme power saving mechanisms, I'd indeed look there next and test the behavior w/ ext4

jebez · 2025-08-11 03:13:08

Well I installed Arch Linux on a USB flash drive, the OS partition is F2FS:

systemctl suspend

works!

Also on archlinux-2025.08.01-x86_64.iso :

fsck /dev/nvme0n1p1

good.

So it's not because of F2FS.

I asked Google Gemini (seems better than ChatGPT...):
This problem often happens with certain NVMe SSDs on Linux and is related to power management. When the system resumes, the filesystem is unable to correctly write its state back to the disk, which can lead to the filesystem being remounted as read-only to prevent data corruption.
Suggested:
add nvme.noacpi=1

sudo fwupdmgr get-updates

not worked...

But I found https://www.ferrybig.nl/linux/arch-linu … ng-suspend but

sudo sh -c 'echo 0>/sys/power/pm_async'

new line

not works: pm_async keeps 1, so how to?

seth · 2025-08-11 06:05:01

gemini must have found https://bbs.archlinux.org/viewtopic.php … 2#p2254892 but I'm not sure how it infers that *disabling* the ACPI quirks stands a high chance to… oh, I forgot: "artificial idiocy".

sudo -v
(sleep 5; sudo systemctl suspend) & dmesg -w

Then wake the system - maybe the ringbuffer has more clues about what's wrong w/ the nvme.

jebez · 2025-08-11 11:34:12

See my edited 1st post.

sudo -v
(sleep 5;systemctl suspend)&sudo dmesg -w

https://drive.google.com/file/d/1stFN9K … vIRQ0rfxjm
https://drive.google.com/file/d/1UGCs5J … 6iFsavibkH
https://drive.google.com/file/d/1XZbkWu … o6mMDQRj1R
https://drive.google.com/file/d/1JxmOS0 … 1pWK1vxZre
https://drive.google.com/file/d/1VUSklM … G8I-mLI0Gv
https://drive.google.com/file/d/1dIi2F9 … 9bRan2_Gc7

seth what about pm_async keeps 1, how to?

Last edited by jebez (2025-08-11 11:55:06)

seth · 2025-08-11 12:33:11

echo 0 | sudo tee /sys/power/pm_async

"0>" w/o any blanks redirects the stdin (which you don't provide)

Edit: you perfectly match https://wiki.archlinux.org/title/Solid_ … nd_support …
Post a journal where you set "iommu=off"

Last edited by seth (2025-08-11 12:35:21)

jebez · 2025-08-11 14:32:27

echo 0 |sudo tee /sys/power/pm_async

works, /sys/power/pm_async is 0 but not works, still the freeze.

iommu=off, iommu=soft, amd_iommu=off & amd_iommu=fullflush not work.

Your link https://wiki.archlinux.org/title/Solid_ … nd_support is about broken firmware or hardware of NVMe, but as my 1st post Sleep works on the golden eggs laying goose 11 of Bill Gate$...
So I try to report on https://bugzilla.kernel.org/ ?

Last edited by jebez (2025-08-11 14:42:27)

seth · 2025-08-11 14:39:10

seth wrote:

Post a journal where you set "iommu=off"

is about broken firmware or hardware of NVMe

The wiki wrote:

Controller failure due to broken suspend support

jebez · 2025-08-11 14:59:11

iommu=off

https://drive.google.com/file/d/1lgh4mt … BGuZAN7oGD
https://drive.google.com/file/d/1AHFRgN … i_-9L6mV8j
https://drive.google.com/file/d/1foNLNl … o4zK3UeJoT

seth · 2025-08-11 15:26:33

sudo journalctl -b | curl -F 'file=@-' 0x0.st

Not some images.

jebez · 2025-08-11 15:30:36

Well I edited my penultimate post, iommu=off not works so photos.

seth · 2025-08-11 16:33:13

I don't want to to fail the suspend w/ it - I want to see whether the parameter is applied or gets ignored for some reason.
Just boot with it and post the journal.

jebez · 2025-08-11 20:53:29

iommu=off

sudo journalctl -b -1|curl -F 'file=@-' 0x0.st

http://0x0.st/8Fnl.txt

seth · 2025-08-11 21:00:35

"iommu=soft", NOT! "off"!
Alternatives would be "amd_iommu=off" or better "amd_iommu=fullflush", but we'll go w/ the software IOMMU first.

jebez · 2025-08-11 21:19:14

seth wrote:

Post a journal where you set "iommu=off"

You want off or soft?

iommu=soft
http://0x0.st/8F5E.txt

Last edited by jebez (2025-08-11 22:18:06)

seth · 2025-08-12 05:53:24

iommu-wise looks exactly like http://0x0.st/8hQG.txt … the suspend at the end resulted in the same kind of wakeup-freeze?

Aug 11 22:13:08 archlinux kernel: perf/amd_iommu: Detected AMD IOMMU #0 (2 banks, 4 counters/bank).

add "iommu=soft amd_iommu=off nvme_core.default_ps_max_latency_us=0 pcie_aspm=off", post a new journal
Then, if it still fails, add "pcie_port_pm=off" to the above list and post a journal for that as well.

jebez · 2025-08-12 06:11:16

Your 2 new mixes not work.

iommu=soft amd_iommu=off nvme_core.default_ps_max_latency_us=0 pcie_aspm=off
http://0x0.st/8Crb.txt

iommu=soft amd_iommu=off nvme_core.default_ps_max_latency_us=0 pcie_aspm=off pcie_port_pm=off
http://0x0.st/8Cra.txt

I think no need to insist, a bug of Linux so I'll try https://bugzilla.kernel.org/ .

seth · 2025-08-12 06:31:34

Do you still get

nvme nvme0: Device not ready; aborting reset, CSTS=0x1
nvme nvme0: Disabling after reset failure: -19

after the wakeup?
And just to be clear:

some idiotic AI wrote:

add nvme.noacpi=1

You don't have anything like this still flying around in some modprobe.conf, do you?

jebez · 2025-08-12 07:33:06

sudo -v
(sleep 5;systemctl suspend)&sudo dmesg -w

For the 2 mixes yes.

Idiotic Gemini solved https://en.wikipedia.org/wiki/Zebra_Puzzle ...

What path for modprobe.conf? As I wrote https://bbs.archlinux.org/viewtopic.php … 8#p2254578 I only edited /etc/kernel/cmdline then

sudo mkinitcpio -P

seth · 2025-08-12 11:51:37

Idk what you did, I just wanna make sure this isn't caused by some bogus module parameter.
In doubt check

systool -vm nvme

--- OT ---
https://en.wikipedia.org/wiki/Zebra_Puzzle isn't nearly as impressive as you might think it is. The crux is to abstract the human language and yeah… that's what LLMs to and are pretty good at.
Running a linear solver over that is the most basic task imaginable and no challenge at all.
It looks tough to you because you're presented a large set of equations and try to juggle them in your head, but once you take a structured approach, solving this kind or "puzzle" becomes a boring chore.

jebez · 2025-08-12 12:27:43

[a@archlinux ~]$ systool -vm nvme
Module = "nvme"

  Attributes:
    coresize            = "65536"
    initsize            = "0"
    initstate           = "live"
    refcnt              = "3"
    srcversion          = "BEFCC1A55641705A9EC0FBD"
    taint               = ""
    uevent              = <store method only>
    version             = "1.0"

  Parameters:
    io_queue_depth      = "1024"
    max_host_mem_size_mb= "128"
    noacpi              = "N"
    poll_queues         = "0"
    sgl_threshold       = "32768"
    use_cmb_sqes        = "Y"
    use_threaded_interrupts= "0"
    write_queues        = "0"

  Sections:

I wanna report this misadventure https://gitlab.archlinux.org/archlinux/ … x/-/issues but https://gitlab.archlinux.org/users/sign_in so I emailed at 8:04 UTC to sign in, still no reply...

I solved myself this puzzle, I created the pieces of the puzzle then I had to test different combinations to finally found the good, so I'm rather impressed a software solved that.
Otherwise I asked Gemini "What is the time in function of the angle of 1 of the orbit of 2 masses?", it not found but me yes...

Last edited by jebez (2025-08-12 12:28:06)

seth · 2025-08-12 14:06:41

This isn't a packaging bug anyway, you'll have to report this upstream and you should before that check out for updates for the nvme firmware and your UEFI

jebez · 2025-08-12 18:31:11

Upstream? You mean https://bugzilla.kernel.org/ ? But:
"If you did not compile your own kernel from scratch, you are probably in the wrong place.
Please use the following links to report a bug to your distribution instead:"
https://gitlab.archlinux.org/groups/arc … g/-/issues

Of course the firmware of the NVMe & the BIOS UEFI are updated:

[a@archlinux ~]$ fwupdmgr get-updates
...
└─WDS250G3X0C-00SJG0:
      Device ID:          71b677ca0f1bc2c5b804fa1d59e52064ce589293
      Summary:            NVM Express solid state drive
      Current version:    102000WD
      Vendor:             Sandisk Corp (PCI:0x15B7)
      Serial Number:      185250422455
      GUIDs:              ff2112dc-038c-596d-90ca-d43c5077c6ec ← NVME\VEN_15B7&DEV_5002
                          c528df4b-7972-5880-8cb1-330415e2dc6a ← NVME\VEN_15B7&DEV_5002&SUBSYS_15B75002
                          bcf288b7-13bb-5c90-9362-3b0fa7917a83 ← WDS250G3X0C-00SJG0
      Device Flags:       • Internal device
                          • Updatable
                          • System requires external power source
                          • Needs a reboot after installation
                          • Device is usable for the duration of the update
                          • Can tag for emulation
    
[a@archlinux ~]$ fwupdmgr get-updates
Devices with no available firmware updates: 
 • DT01ACA100
 • WDS250G3X0C-00SJG0
No updatable devices

https://www.gigabyte.com/Motherboard/B4 … rt-dl-bios

seth · 2025-08-12 19:03:42

You'll in the best of cases be asked to compile a patched kernel, but it's not a packaging bug and iff this is fixable/mitigatable (the usual suggestion seems to be to use S3, but your board doesn't seem to support that) only the developers of the nvme module will be competent to do so.
There isn't even a backtrace - there's some nvme that stops responding after an s2idle and all of the usual mitigations have no impact
There's nothing to work with - this could be a problem w/ the board, could be a problem w/ the nvme, could be a problem w/ the combination.

If you boot and suspend from the usb key, does the nvme disappear as well?
Does it then help to rescan the pci bus after the wakeup?

echo 1 > /sys/bus/pci/rescan

jebez · 2025-08-12 20:53:07

S3 ?

seth wrote:

only the developers of the nvme module will be competent to do so.

Where? What website?

Yes.
No.

https://drive.google.com/file/d/15VWnnW … NGUcdYn0r8
https://drive.google.com/file/d/1rbgeHH … liV3uBkxEE
Ctrl-Alt-Del to reboot
F12 for the BIOS boot menu (appears after unusually more time)
https://drive.google.com/file/d/1xmuu9V … 05kjDJbzxt
omg my Arch Linux KDE on NVMe disappeared!
power button to shut down
power button to turn on
F12
https://drive.google.com/file/d/1plkJ77 … ciOerU-bkl
oof it reappeared!

Last edited by jebez (2025-08-12 20:53:20)

Arch Linux

#26 2025-08-06 01:10:50

Re: Arch Linux freezes resuming from sleep

#27 2025-08-07 14:51:24

Re: Arch Linux freezes resuming from sleep

#28 2025-08-11 03:13:08

Re: Arch Linux freezes resuming from sleep

#29 2025-08-11 06:05:01

Re: Arch Linux freezes resuming from sleep

#30 2025-08-11 11:34:12

Re: Arch Linux freezes resuming from sleep

#31 2025-08-11 12:33:11

Re: Arch Linux freezes resuming from sleep

#32 2025-08-11 14:32:27

Re: Arch Linux freezes resuming from sleep

#33 2025-08-11 14:39:10

Re: Arch Linux freezes resuming from sleep

#34 2025-08-11 14:59:11

Re: Arch Linux freezes resuming from sleep

#35 2025-08-11 15:26:33

Re: Arch Linux freezes resuming from sleep

#36 2025-08-11 15:30:36

Re: Arch Linux freezes resuming from sleep

#37 2025-08-11 16:33:13

Re: Arch Linux freezes resuming from sleep

#38 2025-08-11 20:53:29

Re: Arch Linux freezes resuming from sleep

#39 2025-08-11 21:00:35

Re: Arch Linux freezes resuming from sleep

#40 2025-08-11 21:19:14

Re: Arch Linux freezes resuming from sleep

#41 2025-08-12 05:53:24

Re: Arch Linux freezes resuming from sleep

#42 2025-08-12 06:11:16

Re: Arch Linux freezes resuming from sleep

#43 2025-08-12 06:31:34

Re: Arch Linux freezes resuming from sleep

#44 2025-08-12 07:33:06

Re: Arch Linux freezes resuming from sleep

#45 2025-08-12 11:51:37

Re: Arch Linux freezes resuming from sleep

#46 2025-08-12 12:27:43

Re: Arch Linux freezes resuming from sleep

#47 2025-08-12 14:06:41

Re: Arch Linux freezes resuming from sleep

#48 2025-08-12 18:31:11

Re: Arch Linux freezes resuming from sleep

#49 2025-08-12 19:03:42

Re: Arch Linux freezes resuming from sleep

#50 2025-08-12 20:53:07

Re: Arch Linux freezes resuming from sleep

Board footer