On Sat, 2004-11-27 at 20:05 +0100, Sascha Herrmann wrote: > Hy, > > > If you compile CONFIG_XMON in the kernel, once it's locked, does > > command-power work to break into the debugger ? Also, does it lockup if > > you remove CONFIG_CPUFREQ from the kernel ? > comman-dpower dosn't work after the lockup. One time i was able to > switch to xmon short before the lockup, so xmon basicly works. Without > CONFIG_CPU_FREQ the effects are the same. Vanilla 2.6.9 boots without > problems. With the sleep patch the kernel hangs on about 3 of 4 boots... > > Is there maybe a chance to put some call into the kernel so that xmon > gets called short before one suspects the hangup? (and would it be of > any use?)
Strange... I did experience the crash at ADB probe time here, and fixing that bit in the cpufreq code fixed it for me ... from what I dug out, it was apparently the interrupts not working any more, though it would "resume" when kicked by a direct call to do_IRQ, and I suspected we were basically losing them during the freq. transition. Now you seem to claim something else is going on... At this point I don't have a clue. I'll go through the patch again see if I can find anything ... Ben.
