Re: [RFC PATCH 03/23] Allow glibc to be compiled without EXEC_PAGESIZE

[Sergey Bugaev]

Hello,

On Mon, Mar 25, 2024 at 2:58 PM Florian Weimer <fweimer@redhat.com> wrote:
> > I think the intent here is to initialize _dl_pagesize with a
> > conservative default, to avoid initialization ordering issues.
> > EXEC_PAGESIZE is supposed to be largest supported page size.
>
> This was committed without addressing the comment above.

yes, I also didn't expect this to get pushed until we come to an agreement here.

On topic: I understand that this must have been done this way because
of potential initialization order issues (and the mail I linked also
makes this guess). The question is, is that (working around
initialization order issues) actually required, or is that a leftover
from something that's no longer relevant (or perhaps never was)?
Things seem to still work with this patch on aarch64-gnu for me, both
in SHARED and !SHARED, though I obviously didn't test every potential
codepath.

I was hoping that some broader testing, such as running the testsuite
on CI on existing ports, could answer that question comprehensively --
though now I realize that since this patch leaves the initialization
as-is when EXEC_PAGESIZE is defined (i.e. everywhere but on
aarch64-gnu), CI wouldn't catch any issues that removing it would
cause.

We could define EXEC_PAGESIZE to some conservative value on
aarch64-gnu too, if it turns out that this little workaround is really
required. But it seems cleaner to make sure we don't need to, as
Roland's email suggests, and introducing a new port that doesn't have
a fixed page size (and doesn't come with an EXEC_PAGESIZE value
already defined in kernel headers) seems to be a good opportunity to
do that. That's my reasoning here.

Sergey