On 7/10/26 2:01 AM, Torbjorn SVENSSON wrote:
Hi all,
I just stumbled on this issue for arm-none-eabi in releases/gcc-15.
If I build newlib with -Os, I do not see it for any of my test runs
(different multilib flags).
If I, on the other hand, build newlib with -O2, then it, reliably,
triggers for 2 out of the 14 test runs I do.
After picking this patch on top of Arm GNU Toolchain 15.3.rel1 sources
(based on r15-11320-g7909a9f85f30b6), I no longer get the failure.
Ok to pick r16-8451-g7d7cbf984e4a26 for releases/gcc-15 to avoid the use
of uninitialized cleanup guard value?
OK.
Kind regards,
Torbjörn
On 2026-04-02 16:31, Jason Merrill wrote:
On 4/2/26 2:34 AM, Iain Sandoe wrote:
On 2 Apr 2026, at 07:35, Alexandre Oliva <[email protected]> wrote:
If the promise ctor throws, the EH cleanup will access refcount before
it is initialized. Initialize it to zero so that cleanups can take
place.
This seems very hard to hit on systems that ensure the stack starts
out zero-initialized, but on VxWorks in kernel mode, this hits more
often than not on various architectures.
Regstrapping on x86_64-linux-gnu. Already tested on x86_64-vx7r2,
where
pr95615-01-promise-ctor-throws.C went down from a 77% failure rate
to no
failure whatsoever in 512 runs. Ok to install?
looks good from my PoV, but I cannot OK it,
Iain
OK, thanks. Alas, this doesn't also fix 121961.
Jason
