On Jun 11, 10:29am, Manuel Bouyer wrote:
> Subject: Re: Processes Hanging in a Disk Wait
> Robert Black said:
> >
> > Hello,
> >
> > Some time ago I know there was a problem with several ports with processes
> > occassionally hanging in a disk-wait (in vflushbuf). Does anyone know
whether
> > this problem was ever resolved?
> >
>
> I don't think it has been : Yesterday I saw this on my NetBSD/pmax :
> #ps axu | fgrep 2309
> bouyer    2309  0.0  0.0  1216    0 p4  D+    5:54PM    0:00.32 as -g -o
regis.
> #ps axl | fgrep 2309
> 132  2309  2305  59  -5  0  1216    0 vflush D+   p4    0:00.32 as -g -o regi
> #date
> Mon Jun 10 19:50:02 MET DST 1996
>
> 2 hours to assemble a file is too much :)
>
> I was also unable to cd to the directory, the process fall in the same,
> unkillable state. Also, at reboot, the machine hand at 'syncing disks...'

Yup, thats the one. The vnode information when its stuck in that state seems to
be:

num_output = 1;
read = 0;
write = 0;
referenced = 0;
flags = VXLOCK | VBWAIT;

I hope this means something to someone.

I can reliably reproduce this by doing an untar of xc-1.tar from the X11R6.1
distribution onto a Quantum Fireball. Doing the same onto other disks
occasionally produces the problem.

Cheers

Rob Black

--