Coda File System

Re: Coda & 2.1.89

From: Peter J. Braam <braam_at_cs.cmu.edu>
Date: Sat, 7 Mar 1998 16:43:31 -0500 (EST)
Steven,

Did you get my message this morning?  If it's so repeatable: excellent.
When you have started venus, or in fact just after doing insmod coda

coda-debug 4095
coda-trace 1

Then do your build and mail me the log (compressed :))

Sorry for all the hassle, it's great that you keep crashing things for us,
seriously.

- Peter -

On Sat, 7 Mar 1998, Steven N. Hirsch wrote:

> All,
> 
> Ok, after running into the problems I posted to the list w/ 2.1.86 + coda
> patches, I downloaded and built 2.1.89.  This hangs 100% of time during
> kernel compiles, leaving the process involved unkillable and requiring a
> reboot to clear:
> 
>   PID TTY STAT  TIME COMMAND
>   242  p3 S    0:00 (bash)
>   247  p3 S    0:01 (make)
>  1213  p3 S    0:00 make -C arch/i386/boot bzImage 
>  1243  p3 S    0:00 /bin/sh -c tools/build -b bbootsect bsetup compressed/bvmli
>  1244  p3 D    0:00 tools/build -b bbootsect bsetup compressed/bvmlinux.out CUR
> 
> 
>  FLAGS   UID   PID  PPID PRI  NI   SIZE   RSS WCHAN       STA TTY TIME COMMAND
> 100100     0   232   160   0   0  14880  5516 do_select   S    4  2:11 venus -in
>    100   600   242   241   0   0   1176   144             S   p3  0:00 (bash)
> 100000   600   247   242   0   0   1180   384 wait4       S   p3  0:01 (make)
>      0   600  1213   247   0   0   1008   472 wait4       S   p3  0:00 make -C a
> 100000   600  1243  1213   0   0   1180   512             S   p3  0:00 /bin/sh -
>      0   600  1244  1243   0   0     48    12 down_failed D   p3  0:00 tools/bui
> 
> It's clear that 'build' is stuck waiting on "down_failed" (not sure what
> this is).  
> 
> Like I said, 100% repeatable.  It generally happens either during
> generation of the kconfig.tk file right up front or, in this case, when
> trying to write out a longish file.
> 
> Let me know if there are any steps you'd like done to aid debugging.  I
> really can't make heads or tails out of the current filesystem / dcache
> layer, and with all the churn going on it's a losing battle at best.
> 
> Still, I can break things on demand if that helps <g>.
> 
> Steve
> 
> 
Received on 1998-03-07 16:45:38