Alexander,
On Fri, Apr 09, 2021 at 13:07:37 CEST, Alexander Tormasov via users wrote:
On Thu, Apr 08, 2021 at 18:18:01 CEST, Alexander Tormasov via users wrote:
page fault, pd='init' thread='ep' cpu=0 ip=0xb1d73 address=0x401eeffc stack pointer=0x401eeff0 qualifiers=0x6 irUWp reason=1
The fault address is in the stack area and looks suspicious: 0x401eeff0 is 4100 bytes below 64K of stack. Could it be your stack size is 64K which does not suffice your use case?
it is really hard to say something about, but in the past 64k was definitely enough. I suspect nested "panic in panic" during processing of c++ exceptions…
while it is hard to say something because of strange debugger work (request 4 more caps from init and hang), I try to find any methods to understand/debug situation…
Sorry but I completely ignored the fact that the failing component is the root init instance. So, it might help to just remove ld_verbose="yes" from its config while debugging with Josef's instrumentation. The change seems to trigger unexpected stack usage in init which you don't seem to debug currently.
Regards