[Bug 295511] 16-CURRENT hanging on startup, not completing multi-user boot

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=295511

            Bug ID: 295511
           Summary: 16-CURRENT hanging on startup, not completing
                    multi-user boot
           Product: Base System
           Version: CURRENT
          Hardware: Any
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: misc
          Assignee: bugs@FreeBSD.org
          Reporter: agh@riseup.net

Attempting to maintain a 16-CURRENT NAS/build host, in the recent last week or
two frequent hangs (no panics, no dumps, no response to physical shutdown, just
a lock-up) have made it increasingly challenging to tune in on a source of the
problem. As of netlink: Fix lock leak in nl_find_nhop[1], booting often stalls
during rc, most often at ifconfig:

> Starting file system checks:
> /dev/label/fafnir_encfs: FILE SYSTEM CLEAN; SKIPPING CHECKS
> /dev/label/fafnir_encfs: clean, 13611862 free (29670 frags, 1697774 blocks, 0.1% fragmentation)
> Mounting local filesystems:.
> ELF ldconfig path: /lib /usr/lib /usr/lib/compat /usr/local/lib /usr/local/lib/compat/pkg /usr/local/gnat12/lib /usr/local/lib/compat/pkg /usr/local/lib/gcc12 /usr/local/lib/gcc14 /usr/local/lib/perl5/5.42/mach/CORE
> 32-bit compatibility ldconfig path: /usr/lib32 /usr/local/lib32/gcc12 /usr/local/lib32/gcc14
> Setting hostname: fafnir.home.arpa..
> Setting up harvesting: PURE_RDSEED,RANDOMDEV,[CALLOUT],[UMA],[FS_ATIME],SWI,INTERRUPT,NET_NG,[NET_ETHER],NET_TUN,MOUSE,KEYBOARD,ATTACH,CACHED
> Feeding entropy: .
> igb0: Link is up 1000 Mbps Full Duplex
> load: 52.79  cmd: ifconfig 68738 [sbwait] 27.00r 0.00u 0.00s 0% 2848k
> load: 131.25  cmd: sh 65665 [runnable] 210.56r 0.00u 0.00s 0% 3144k
> load: 131.55  cmd: sh 65665 [runnable] 214.67r 0.00u 0.00s 0% 3144k
> load: 131.55  cmd: sh 65665 [runnable] 215.27r 0.00u 0.00s 0% 3144k
> load: 131.55  cmd: sh 65665 [runnable] 215.97r 0.00u 0.00s 0% 3144k
> load: 134.66  cmd: sh 65665 [runnable] 350.54r 0.00u 0.00s 0% 3144k
> load: 134.96  cmd: sh 65665 [runnable] 484.75r 0.00u 0.00s 0% 3144k
> load: 134.99  cmd: sh 65665 [runnable] 587.68r 0.00u 0.00s 0% 3144k

Forcing a reset will result in an un-clean filesystem, and the subsequent boot
will then stall indefinitely at fsck. Perhaps every 30th boot makes it to
multi-user mode, however, as a result of the current predicament, bug #129645,
and bug #293862, I have had to remove mounting of most of the system from
startup.  If I later mount the UFS gjournal, the system will again freeze with
no panic, and no dump on reboot. An older working kernel from the 23rd of April
no longer restores system usability, as it too struggles booting multi-user
mode with the same lock-ups as d66fec481bfd[1]. If a boot with a debug kernel
is successful, it does not trigger the system lock-up, however, it is so
incredibly slow, and if the system locks-up again, there is still no core dump
or debugger. Running poudriere-bulk will always result in a lock-up, either 1
minute in, or hours in to the process, just not when the system is booted with
a debug kernel.

1:
https://codeberg.org/FreeBSD/freebsd-src/commit/d66fec481bfd65cbabb6c12a410d76843e76083e

-- 
You are receiving this mail because:
You are the assignee for the bug.