Random kernel Oops at boot with "acpi=ht", disappearing with "acpi=off"

Hi,
This laptop suffers of random kernel hangs at boot: I tested kernel 4.1.8, 4.2.1 and 4.3-rc3 and they are all affected.
Every time I turn on my laptop I have to boot several times to be able to reach the sddm login because often I get hangs which prevent the system to boot. Once booted I don't have any problem anymore, until the next boot.

With "acpi=off" the system boots flawlessly, while with "acpi=ht" I still get random hangs, so I guess the problem is in the ACPI table parsing code itself, or perhaps the SMP code.

I recompiled kernel 4.3-rc3 with CONFIG_ACPI_DEBUG=y and I attached full log from a serial console, booting with "debug acpi.debug_level=0x2003" (warn, error and tables debug enabled): https://bugzilla.kernel.org/attachment.cgi?id=189041

Distro is Arch Linux. Please let me know if you need further logs.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=105251

Niccolò Belli
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Does it work any better if you blacklist the crashing module "snd_soc_rl6347a" (or set CONFIG_SND_SOC_RL6347A=n)?
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Hi,
I tried to blacklist the crashing module (snd_soc_rl6347a) without any success: it still loads. I booted kernel with modprobe.blacklist=snd_soc_rl6347a and I put "blacklist snd_soc_rl6347a" in /etc/modprobe.d/blacklist.conf. I even tried to recompile the kernel without that module, but there is no option in "make nconfig" to disable it, while adding "# CONFIG_SND_SOC_RL6347A is not set" to the config doesn't help because it keeps showing CONFIG_SND_SOC_RL6347A=m in "make nconfig" if I search for it with F8. I attached another log with modprobe.blacklist=snd_soc_rl6347a plus modprobe.d entries: https://bugzilla.kernel.org/attachment.cgi?id=189361

Thanks,
Niccolò Belli

What if you disable the whole ALSA SoC?

CONFIG_SND_SOC=n
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Il lunedì 5 ottobre 2015 12:46:45 CEST, Mika Westerberg ha scritto:

Now I don't get any Oops but it doesn't reach the login anymore: instead it goes into emergency mode! :(
Log: https://bugzilla.kernel.org/attachment.cgi?id=189471

Thanks,
Niccolò Belli
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

It says:

Mounting /boot...
[FAILED] Failed to mount /boot.
See 'systemctl status boot.mount' for details. [DEPEND] Dependency failed for Local File Systems.

Did you run 'systemctl status boot.mount'?
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Il lunedì 5 ottobre 2015 15:54:49 CEST, Mika Westerberg ha scritto:

Please do not consider my previous e-mail, becuase the failure was due to file loss: in fact sometimes the system hangs while shutting down (I think because of this bug) and so some modules of my just compiled kernel were loss. I recompiled the kernel once again and synced the disks just to be sure: I still get random crashes even with CONFIG_SND_SOC=n. See the new log: https://bugzilla.kernel.org/attachment.cgi?id=189481

Thank you very much for your help, it's much apreciated. Niccolò Belli
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

It looks like a memory corruption somewhere. Unfortunately it is not obvious from the log where :-(

Il giovedì 8 ottobre 2015 12:06:15 CEST, Mika Westerberg ha scritto:

It seems I'm the only one with this issue, both Dell and an Aaron Lu (an Intel employee which uses Fedora) told me they don't experience it: https://bugzilla.kernel.org/show_bug.cgi?id=105251
So maybe it has something to do with my configuration, maybe the combo dm-crypt + btrfs which is pretty unusual, but I really don't know.

If you can tell me how to obtain a more useful log I can provide it.

Thanks,
Niccolò Belli

You may want to try another distro if it reproduces.

Other than dmesg, I have no idea :-/ Maybe full crashdump but then it needs someone who is able to analyze it properly.
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

It seems that only Arch triggers the kernel bug: https://bugzilla.kernel.org/show_bug.cgi?id=105251#c9

This is really, really strange...

Niccolò Belli

I tried to use Ubuntu's kernel with Ubuntu's initramfs on Arch, but somehow I didn't manage to boot the system.

So I kept the Ubuntu kernel and I generated a new initramfs for it: I still get crashes at boot, maybe a bit less frequently.

Then I decided to build a kernel statically without using initramfs at all: I don't get crashes anymore!!!

So even Ubuntu's kernel *IS* bugged, but somehow only Arch's initramfs exposes the bug.

Unfortunately my laptop uses dm-crypt, so I *need* an initramfs to boot it (I was booting a flash drive in my tests).

Niccolò Belli
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html