[aarch64-port-dev ] [Roland Westrelin] Re: Aarch64 port for ZGC, so far

[Andrew Dinn]

On 15/04/2019 15:57, Stuart Monteith wrote:
>    If I understand this correctly, siginfo_t would contain the
> faulting address with the tag wiped. The ucontext would still contain
> the full 64-bit registers, unmolested. I've asked internally, but that
> is how I've interpreted it.
Well, that makes sense (why would the kernel mask the values pushed by
it's interrupt handler at point of push or later?). I guess you have a
de facto proof of that in that ZGC manges to continue beyond the first
SEGV on AArch64.

We have already seen evidence of the fault address being masked -- we
had to make allowance for it when Shenandoah was trying to dereference
the Brooks pointer for a null reference and ended up loading from
0xfffffffffffffff8 (i.e. -8). We had to special case for a segfault
address reported as 0x00fffffffffffff8 (i.e. 1 gazillion_base_2 - 8).

regards,


Andrew Dinn
-----------
Senior Principal Software Engineer
Red Hat UK Ltd
Registered in England and Wales under Company Registration No. 03798903
Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander