Re: [patch V4 23/36] rseq: Provide and use rseq_set_ids()

[Mathieu Desnoyers]

On 2025-09-11 12:02, Thomas Gleixner wrote:
> On Thu, Sep 11 2025 at 09:40, Mathieu Desnoyers wrote:
> > On 2025-09-08 17:32, Thomas Gleixner wrote:

[...]

> > Also why does rseq_fork() set it to ~0ULL rather than keep the value
> > from the parent when forking a new process ?
> >
> > Whatever was present in the parent process in the rseq area should
> > still be in the child, including the rseq registration.
> >
> > Or am missing something ?
>
> Not really. I just made all those cases (install, fork, exec...) behave
> the same way. It's harmless enough, no?
It does prevent validation of the rseq area content across fork in the
child process. I do not expect this to have a significant impact in
practice though.

Thanks,

Mathieu


--
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com