Re: building netbsd-9 2 'sync' processes stuck in 'tstile'

["John D. Baker" <jdbaker%consolidated.net@localhost>]

On Sun, 9 May 2021, Robert Elz wrote:

> Then from your later message, the reply
> to Greg Woods...
> 
>   | trace: pid 2923 lid 1 at 0xffff810060d26e80
>   | sleepq_block() at sleepq_block+0x19a
>   | turnstile_block() at turnstile_block+0x2de
>   | mutex_vector_enter() at mutex_vector_enter+0x2b4
>   | do_sys_sync() at do_sys_sync+0x45
>   | sys_sync() at sys_sync+0x9
>   | syscall() at syscall+0x157
>   | --- syscall (number 36) ---
> 
> That's the mountpoint lock (mnt_updating).
> sync iterates over all mounted filesystems
> performing a fikesystem dependant sync on
> each.  It would have been nice to look and
> see which one was locked and never being
> released.   
> 
> I don't suppose you know exactly what was
> mounted before the reboot?

It would have been the same things that are always mounted:

$ mount
/dev/wd4a on / type ffs (log, local)
/dev/wd4e on /var type ffs (log, nosuid, local)
tmpfs on /tmp type tmpfs (nodev, nosuid, local)
kernfs on /kern type kernfs (local)
procfs on /proc type procfs (local)
ptyfs on /dev/pts type ptyfs (local)
/dev/dk0 on /r0 type ffs (log, nodev, nosuid, NFS exported, local)
/r0/nbsd on /x type null (nodev, nosuid, local)
tmpfs on /var/shm type tmpfs (local)
pid582@yggdrasil:/net on /net type nfs (hidden)
pid582@yggdrasil:/m on /m type nfs (hidden)
pid582@yggdrasil:/home on /home type nfs (hidden)

"/dev/dk0" is the RAID.

The 'amd'-managed mountpoints were quiescent and thus not actually mounted.

At the time, 'ps' reported the stuck 'sync' processes with
"D" (one with "D+") in the "STAT" column.

>   | At this point I rebooted again and booted
>   | a -current kernel which
> 
> Are you still running that, or back to -9 ?

At the time, I tried to go multi-user, but a lot of stuff failed with
"bad syscall" as I booted my custom config rather than my tweaked GENERIC.

As I write this, the machine is now running my tweaked 9.99.82 GENERIC
kernel (i.e., w/"options RF_INCLUDE_RAID5_RS=1") w/netbsd-9 userland.


As I see the pull-ups have been done, I'll be updating my netbsd-9
tree and rebuilding.

-- 
|/"\ John D. Baker, KN5UKS               NetBSD     Darwin/MacOS X
|\ / jdbaker[snail]consolidated[flyspeck]net  OpenBSD            FreeBSD
| X  No HTML/proprietary data in email.   BSD just sits there and works!
|/ \ GPGkeyID:  D703 4A7E 479F 63F8 D3F4  BD99 9572 8F23 E4AD 1645