On Thu, Sep 18, 2025 at 06:56:02PM +0100, Mark Brown wrote:

On Thu, Sep 18, 2025 at 05:04:06PM +0100, Will Deacon wrote:

quoted

On Thu, Sep 18, 2025 at 01:38:53PM +0100, Will Deacon wrote:

quoted

quoted

It would be great if Christian could give this the thumbs up, given that
it changes clone3(). I think the architecture parts are all ready at this
point.

quoted

ah, I may have spoken too soon :/

Well, there's also the fact that this is based on the vfs tree (or would
have conflicts with it).

quoted

Catalin pointed me at this glibc thread:

quoted

https://marc.info/?l=glibc-alpha&m=175811917427562

quoted

which sounds like they're not entirely on board with the new ABI.

I think we're getting there on that one, and the main thing they're
asking for is the ability to reuse the GCS after the thread has exited
which would be orthogonal to this stuff.  I see Catalin replied on the
glibc side so I'll direct most of my reply there.

It would be really helpful to get a clear idea of where we're going with
this series, it's been almost landed for an incredibly long time and
having it in that state is getting disruptive to doing cleanup to try to
factor code out of the arches especially with the RISC-V stuff also up
in the air.  I do think the issues glibc have with this are orthogonal
to the changes here so hopefully this can go as is.

It's not great to get that API feedback is coming in just now... and
that it's not reachable via lore. So I've picked this up to have it soak
in -next.

Please keep this thread informed whether you're reaching consensus on
the glibc side...