On Mon, 2005-01-10 at 11:08 -0800, Mark Wong wrote: > On Mon, Jan 10, 2005 at 10:05:20AM -0800, Dave Hansen wrote: > > Can you figure out *which* call (file/line#) to __pa() was causing the > > BUG? Also, you're not really running on an ARM proc, right? > > > > Oh, right, i386. In any event, if that didn't just show how often I > work with kernel code, is a prink() on every line sufficient to > determine where it dies? Yeah, that should do it. It will be overkill, but it will do it :) Be prepared to dig through a ton of messages, though. > As for finding what is calling __pa() > doesn't look as straightforward to me. Any calls to virt_to_page() or virt_to_phys() call __pa() eventually. Does that help? You can also do addr2line on some addresses that are just before an after the BUG() address to see where it was, if you compiled with -g. Or, just send me your vmlinux, and I'll take a look. -- Dave
