Paul Koster wrote:
Chris Kennedy wrote:
The above does not really explain why this happens with recent drivers
(though not that often) and never with old versions (notably 0.3.2f),
using the same hardware, memory, kernel, etc. Furthermore, I cannot
recall anytime that I ran out of memory in normal cases (uptimes of
weeks/months).
Yes, actually I think the buffer sizes have changed since then and
possibly were 16k buffers before and now 32k. I suspect that since that
changed, and possibly other buffer sizes, the memory allocation is more
stressed and possibly bringing out preemption bugs where we need to do
something in preemption mode (seem ok in SMP and normally I think). So
kind of explains it, although looking at module load messages for each
and checking what is being allocated would show if that's the case.
Is there any relevant information we could provide to help the
developers determine possible bugs in the driver if it happens again?
(In the meantime I'll update to latest svn where dynbuf is turned off
by default, maybe it matters?)
Would be interesting to try that, mainly try and fiddle around between
versions and seeing if anything helps and reporting details about the
tests done and success (detailing what caused it) or failure (detailing
where things break), that's the best information (and module loadup
messages from each version too, often that provides all the logistics of
the versions buffer and DMA setup.
Thanks,
Chris
--
===
Chris Kennedy
[EMAIL PROTECTED]
-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
ivtv-devel mailing list
ivtv-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ivtv-devel
