We should build glibc with -moutline-atomics on aarch64, to avoid a critical performance regression in the mutex implementation on certain CPUs. This depends on a GCC change. After that, it is just a three-line glibc.spec change.
*** Bug 1818100 has been marked as a duplicate of this bug. ***
*** Bug 1828473 has been marked as a duplicate of this bug. ***
Created attachment 1701786 [details] moutline-atomics.patch Marek provided me with a test build of GCC, and I was able to build glibc with it, using the attached patch. The resulting glibc builds on a system with LSE atomics and selects the expected code path, including in the dynamic loader (so its ELF constructor was executed). Once we know the GCC version with the patch, we should probably tighten the gcc build dependency as well.
I found this right before opening a defect requesting much the same thing so +one... :)
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Moderate: glibc security, bug fix, and enhancement update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2021:1585