Re: SIGSEGV in LuaJIT 2.1 VM
- From: Peter Cawley <corsix@xxxxxxxxxx>
- To: luajit@xxxxxxxxxxxxx
- Date: Fri, 18 Nov 2016 17:51:53 +0000
On Fri, Nov 18, 2016 at 2:52 PM, Tomas Kvasnicka <nzt4567@xxxxxxx> wrote:
Interesting idea indeed - this got me checking some other core dumps
available to see if RAX is always filled with such low values. The answer is
that mostly no - even values like 0x41978ae8 occur in RAX and these would
probably be dereferenced without any problem (sections marked as "ALLOC LOAD
HAS_CONTENTS”).
If you have this corpus of examples, could I ask the following questions of it:
1. Is the faulting instruction always the first instruction of a trace?
2. For each N in 0 through 255, how many examples do you see where
faulting-rip mod 256 == N?
Other related posts: