[COMMITTED 2.39-2.41] nptl: clear the whole rseq area before registration

[Michael Jeanson]

On 2025-03-13 01:24, Florian Weimer wrote:
> 
> We encountered this downstream as well, but decided that backporting
> this was not required because for ABI compatibility reasons, the kernel
> can require zero-initialization only for new fields added in the
> extensible part.  Is this conclusion incorrect?

No you're correct and since those versions don't have extensible ABI 
support, this patch is indeed not required.