Hi Thiago,
From: Thiago Jung Bauermann thiago.bauermann@linaro.org Sent: 28 October 2025 05:50 AM To: Yury Khrustalev via linaro-toolchain Cc: Christophe Lyon; Yury Khrustalev; nd Subject: Re: [Linaro-TCWG-CI] glibc-2.42.9000-284-g27effb3d50: 9 regressions on aarch64
Thiago Jung Bauermann thiago.bauermann@linaro.org writes:
Sorry for the delay. It took me a bit to reproduce it outside of our CI environment because I'm not used to glibc development, but I finally managed it on a QEMU VM with -cpu max running Ubuntu 24.04:
It turns out I can also reproduce it on an Ampere machine with a Neoverse-N1 processor.
I was using QEMU with -cpu max because I assumed that the regression was related to SME state, but it seems that it isn't.
Thanks for the updated instructions for reproducing this regression. It really helped and I've managed to reproduce it and find the cause. The incorrect CFI directive corrupts state that is used by GDB to workout callstack information.
Patch [1] is on the way.
I'm still not sure why GDB tests were passing when I ran them via `make check`, so I can't confirm if the fix will resolve these regressions in the CI. It would be great if this could be confirmed from your side.
[1]: https://inbox.sourceware.org/libc-alpha/20251028115009.1308287-1-yury.khrust...
Thanks, Yury