Tue Oct 28 2025
Saurabh Jha <[email protected]> wrote:
> For master, it fails on step 4 because mingw-crt needs long double to be
> 8 bytes. So I couldn't run regression tests for aarch64-w64-mingw32 on
> master. After these commits are applied, the test summary looks like
> this.
>
> # of expected passes            312151
> # of unexpected failures        55879
> # of unexpected successes       6
> # of expected failures          1667
> # of unresolved testcases       1413
> # of unsupported tests          7384
>
> We have some way to go to achieve parity with aarch64-unknown-linux-gnu
> and the future work should make these numbers better.

It looks like it has been tested only for C. Most likely, most of the
tests that cause unexpected failures come from SVE/SME.

https://github.com/eukarpov/gnu-toolchain-windows-arm64
contains recent changes, daily rebase and improvements, such as SEH
implementation
(upstreaming in progress,
https://sourceware.org/pipermail/binutils/2025-August/143401.html),
aarh64-pc-cygwin (the patch series is being prepared for upstreaming)
and the relocation fixes for DLL auto-import.

As of 2024/08/09 the toolchain has reached the following level of quality with
the GCC testing targeting armv8-a for C/C++ without optional extensions such
as SVE (some changes are not upstreamed yet):

Expected passes 573324
Unexpected failures     12118
Unexpected successes    172
Expected failures       4528
Unresolved testcases    7363
Unsupported tests       10774
DejaGnu errors  0
Total   608279
Reliability rate        96%

Regards,
Evgeny

Reply via email to