Page 2 of 2
Re: VMLaunch hang without debug breakpoint
Posted: Mon Jan 30, 2012 4:02 pm
by Cognition
It could be a stack selector problem, but that should show up during any entry into kernel space. If it's not page faulting at any point and the stack pointer value seems sane than it's probably memory corruption elsewhere in the system. Probably the IDT or GDT, the next logical step would be to verify the IDT entries to see if they're sane.
Re: VMLaunch hang without debug breakpoint
Posted: Mon Jan 30, 2012 6:20 pm
by Kamala
> It could be a stack selector problem, but that should show up during any entry into kernel space.
I suspect that was the first entry to kernel space after starting to run as guest.
> If it's not page faulting at any point and the stack pointer value seems sane than it's probably memory corruption elsewhere in the system.
I don't think it is a page fault.
> Probably the IDT or GDT, the next logical step would be to verify the IDT entries to see if they're sane.
I don't remember it being corrupted though I should check again to be sure.
Two questions -
1) When the fault is during stack switch, does the processor provide specific error info as to what in specific could have gone wrong that I am neglecting to look at?
2) What does an initial breakpoint instruction in guest do that obfiscated the problem? I don't seem to encounter the same double fault if the first instruction executed as guest is an int 3. It appear to do something that fixes up the kernel mode stack the right way in the tss structure.
Thanks.
Kamala