From: Allan Schaffer (allan++at++sgi.com)
Date: 06/06/2005 14:17:39
Rick van Haasen wrote:
> my performer application runs fine on an Onyx3400 for some 2 days, after
> that the application aborts
> as the result of an assertion failure in lib/libpf/pfProcess.C at line
> 9030,
> function _pff_getFrameCount():
> Assertion failed frameCount == _pff_getFramecount() in pfProcess.C
> This beheviour was reported earlier in the mailing list, however there was
> no solution.
> does anyone know what causes this abortion, and even better, what can be
> done about it?
(I'm assuming pf3.0.2)
That particular line is checking that the frame number in the DRAW
process matches what the pipe thinks it should be. This is actually
different than the previous list postings which had to do the same
general idea but in the CULL process (from what I could tell).
There's nothing obvious that I can see; But given the duration of the
run perhaps there is a counter eventually overflowing. Next chance you
get could you try reproducing it with e.g. town.perfly? That's a test
we leave running for days at a time. We can see if it's something
system-specific by having you run the same app.
A stab-in-the-dark debug option -
setenv LD_LIBRARYN32_PATH /usr/lib32/Performer/Debug
setenv _PF_SEMA_DEBUG 1
Will let you see the frame counter incrementing, look for the numbers in
format [xxxx->yyyy]. The left column is the frame number in APP, the
right column is the frame number in DRAW..
Allan
-- Allan Schaffer allan++at++sgi.com Engr. Dept. Manager, Visual Systems Group 1-650-933-2160 Silicon Graphics http://www.sgi.com
This archive was generated by hypermail 2b29 : Mon Jun 06 2005 - 14:17:39 PDT