I can confirm neither the boot hang with HPET enabled nor the other unexplained boot hang after "NET: Registered protocol family 2" occur on 2.6.24.2 on the same hardware. I still don't have an appropriate way to debug the issue on 2.6.25-rc1, though...
-j On Thu, 2008-02-14 at 00:25 +1300, Jasper Bryant-Greene wrote: > I have been attempting to test 2.6.25-rc1 on my Sun Ultra 40 (two > dual-core Opteron 2222 CPUs). > > There is some issue with the HPET clocksource which causes it to hang > the boot process if enabled, which I will look into in due course (the > HPET works fine on at least 2.6.19). > > More pressing is that the system will not boot at all even with HPET > disabled - kernel gets no further than "NET: Registered protocol family > 2". > > I cannot think of a way to find out what the problem is, because the > system has only USB ports, and the USB keyboard is not working at this > stage, either because the system has locked up hard, or because the USB > support has not yet loaded (I tried both standard USB-HID and boot > protocol keyboard support, both compiled in). Thus I cannot use > magic-sysrq, or even scrollback to look at boot messages. There is also > no serial port to assist with the latter... > > Any thoughts on a procedure for debugging this would be much > appreciated. > > -j
signature.asc
Description: This is a digitally signed message part