On 3/2/26 1:21 PM, Warner Losh wrote:
On Mon, Mar 2, 2026 at 10:50 AM Pierrick Bouvier
<[email protected] <mailto:[email protected]>> wrote:
On 3/2/26 7:15 AM, Warner Losh wrote:
>
>
> On Mon, Mar 2, 2026 at 7:03 AM Florian Hofhammer
> <[email protected] <mailto:[email protected]>
<mailto:[email protected]
<mailto:[email protected]>>> wrote:
>
> The syscall emulation code previously wasn't interruptible via
> cpu_loop_exit(), as this construct relies on a longjmp target
that
> is not
> live anymore in the syscall handling code. Consequently,
longjmp() would
> operate on a (potentially overwritten) stale jump buffer.
This patch
> adds an additional
> setjmp and the necessary handling around it to make longjmp()
(and by
> proxy cpu_loop_exit() safe to call even within a syscall context.
>
> Signed-off-by: Florian Hofhammer <[email protected]
<mailto:[email protected]>
> <mailto:[email protected]
<mailto:[email protected]>>>
> ---
>
>
> So bsd-user would need something similar, I'd think. But it still
uses
> the older-style
> TARGET_EJUSTRETURN. With half of the archs still out of tree, I'm
not
> entirely
> sure the best way to approach this.
>
> The code does what you say, which is good, but there may be some
subtle
> reason to do something else.
>
> Reviewed-by: Warner Losh <[email protected] <mailto:[email protected]>
<mailto:[email protected] <mailto:[email protected]>>>
>
At the moment, syscall filtering is not available for bsd-user, it's
(sadly) only implemented for linux-user. Which implies that current
changes are not needed for bsd-user neither, since they can't be tested
anyway.
OK. I'll add it to the backlog list then...
Warner
Good news is that it should not be too hard to enable.
More generally, plugins were not enabled for bsd-user until recently (I
added it in 9b23bf5024b5c7d91cfe9233f71859208f4ffa94 around a year ago).
Regards,
Pierrick
