Re: UFS bad magic number kernel panic on Asus PN43

On Mon, 9 Jun 2025 12:52:31 +0200
Gerrit Kühn <gerrit.kuehn@aei.mpg.de> wrote:

> Hello,
> 
> I am not sure if this is an actual filesystem issue, but right now it
> looks like one to me (maybe related to some unknown hardware issue,
> though), so I will try here first.
> I have a newly installed system (14.2) on a newly bought Asus PN43
> hardware (Intel Alder Lake, N200) using a Samsung SSD 990 EVO 1TB for
> root fs. The settings are pretty much default from the installer, i.e.,
> UFS root, softupdates and soft updates journaling were enabled.
> 
> The system installed and booted just fine, but when trying to bootstrap
> pkg it panicked like this:
> 
> ---
> UFS /dev/nda0p2 (/) cg 809: bad magic number 0x0 should be 0x90255
> panic: softdep_deallocate_dependencies: dangling deps
> cpuid = 0
> time = 1749395925
> KDB: stack backtrace:
> #0 0xffffffff80b8b89d at kdb_backtrace+0x5d
> #1 0xffffffff80b3dc01 at vpanic+0x131
> #2 0xffffffff80b3dac3 at panic+0x43
> #3 0xffffffff80e7018a at softdep_deallocate_dependencies+0x6a
> #4 0xffffffff80bff707 at brelse+0x197
> #5 0xffffffff80e60d8f at ffs_getcg+0x28f
> #6 0xffffffff80e5ed6f at ffs_nodealloccg+0xbf
> #7 0xffffffff80e5e7b3 at ffs_valloc+0x4b3
> #8 0xffffffff80ea5ab7 at ufs_mkdir+0x107
> #9 0xffffffff810ec378 at VOP_MKDIR_APV+0x28
> #10 0xffffffff80c383b6 at kern_mkdirat+0x286
> #11 0xffffffff810262c5 at amd64_syscall+0x115
> #12 0xffffffff80ffccab at fast_syscall_common+0xf8
> Uptime: 2m18s
> ---
> 
> This was reproducable. After a few unsuccessful tries I decided to boot
> into single user mode, disabled soft updates journaling, did a manual
> fsck (reported all issues fixed afterwards), and rebooted once more.
> This brought the system into a basically usable state:
> I could bootstrap pkg now and install other software without much of an
> issue.
> However, I still see (rare) messages like this in dmesg, especially when
> doing a lot of disc writes (I think):
> 
> ---
> /: bad dir ino 53285122 at offset 0: mangled entry
> /: bad dir ino 104647168 at offset 0: mangled entry
> --- 
> 
> I see no other hardware-related problems so far, "smartctl -l selftest
> /dev/nvme0" completed without any error.
> There are a few acpi error messages during boot, though (don't know if
> they are related):
> 
> ---
> ACPI Error: AE_NOT_FOUND, During name lookup/catalog
> (20221020/psobject-372)
> Firmware Error (ACPI): Could not resolve symbol
> [\134_SB.PC00.TXHC.RHUB.SS02], AE_NOT_FOUND (20221020/dswload2-315)
> ACPI Error: AE_NOT_FOUND, During name lookup/catalog
> (20221020/psobject-372)
> ---
> 
> 
> Any idea what is causing the UFS issues (and how to fix them
> properly?).
> 

I have a laptop with an Intel Alder Lake-M and I was also seeing errors
with my UFS file system when lots of files were being installed.

Adding vm.pmap.pcid_enabled=0 to /boot/loader.conf fixed it for me.

So you could try adding it to your loader.conf.

-- 
Gary Jennejohn