> Can Documentation/oops-tracing.txt really be applied though? It gives > instruction when klogd is available to symbolize the addresses, or > when ksymoops can be run right after the Ooops. > > Because I dynamically load modules, is it not likely that if any > modules are in the trace, by the time I reboot and can ksymoops the > crash data (captured on the serial console for instance) the address > pointers will have changed due to the dynamic module loading? For modules loaded at boot time it works well because they get the same address on every boot. That is good enought for me as I can take precautions if I suspect a module being buggy ... > > This simply can't happen if you don't use btaudio. > > Sorry, again bad wording on my part. I meant that I use the btaudio.o > driver that comes with my kernel distribution. I don't "supplant" it > with a newer/different version as I do with v4l2 and bttv. Ok. Can you try the latest version then (either the patch from http://bytesex.org/patches/ or the 2.4.19-pre2 kernel)? Gerd -- #define ENOCLUE 125 /* userland programmer induced race condition */
