On 5/23/23 01:18, Daniel P. Berrangé wrote:
On Tue, May 23, 2023 at 10:02:51AM +0200, Paolo Bonzini wrote:
On Tue, May 23, 2023 at 10:00 AM Daniel P. Berrangé <berra...@redhat.com> wrote:
I'm curious why we need to do anything ? I would have thought that meson
should handle 'b_pie' itself, passing the right args to $CC that it feels
are appropriate. I don't recall seeing other apps using meson trying to
handle b_pie logic - what's special about QEMU ? IOW, is it possible to
delete this entire b_pie condition and thus avoid worrying about this
problem ?
The issue is that Meson only has "enable PIE" or "leave PIE to the
compiler default", while QEMU also has "disable PIE"---which is the
messy one.
Does QEMU actually need "disable PIE" ? It existed in configure of
course, is there a reason we need to continue to support it in meson ?
We need it for aarch64 runner, because the OS built glibc static libraries with -fpie
instead of -fPIE, and we get a relocation overflow at link time.
Upstream glibc has been fixed, but there has been no update to ubuntu packages.
r~
