Re: igc, macOS avoiding signals

[Gerd Möllmann]

Helmut Eller <eller.helmut@gmail.com> writes:

> On Tue, Dec 31 2024, Gerd Möllmann wrote:
> [...]
>> We have established that calling get_backtrace is safe since it doesn't
>> access memory in our AMC pool, which might have a barrier. Counter
>> argument was that one would have to know too much about what is safe to
>> access and what cannot, and that would be unmaintainable.
>
> I thought, the problem with this was, that writing GC roots in SIGPROF
> is not safe: if we have interrupted MPS it may have partly scanned roots
> or somthing like that.  Is that not a problem?

I think you meant reading roots. We're reading the binding stack which
is a root.

>From my POV, that has the same problem that we currently have already.
The binding stack may be inconsistent because SIGPROF hits at the wrong
place, with a small probability.

With MPS, it might be a problem that the root is currently exact IIRC,
but it could be made ambiguous, if MPS is fixing addresses, in which
case I think we'd be safe.

I'm too lazy to look at the code, ATM, because I find the
process_pending_singals more attractive.