On Wed, 2004-05-19 at 12:01, Andy Green wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On Wednesday 19 May 2004 11:51, Dave Jones wrote: > > > There was nothing particularly magical about that kernel to explain > > why it suddenly started working again, and then failed again > > subsequently, which just adds to my confusion :-/ > > > > I'm beginning to wonder if it could be a compiler issue. > > If there's anything I can do to help nail it down, I will try to help. The > board I have does have a serial port and I can hook a laptop up to that to > capture any dumps. > > It is always a clean reboot though, not a hang or panic, maybe a sign of some > gross CPU exception. One way to close in on it might be to make the very > early kernel init spin in a tight loop, move that around and use the reboot > vs hang behaviour to probe where the evil is. Assuming it even gets started > on the kernel init. Unfortunately I don't have the kernel-fu to do this > easily myself. if you're feeling adventurous, adding a for (;;); to the start_kernel() function in init/main.c and seeing if it just hangs instead of rebooting, and if it hangs, move it down a bit, etc should isolate where the bug is being hit, assuming that we're actually getting that far. If we're blowing up any earlier than that things could get really messy having to play with the assembly startup routines in arch/i386/boot Dave
