gcc: bump to releases/gcc-15.

[Arielfoever]

No description provided.

[TommyMurphyTM1234]

Same problem as described here:

• gdb: Bump to next major version. #1710 (comment)
• gdb: Bump to next major version. #1710 (comment)

[cmuellner]

Thanks for the PR!

This PR references the untagged commit 9483020ea576516448f6e820fba4b14257d6268e from the remote branch releases/gcc-15. We prefer referencing tagged commits unless there are good reasons to update to something more recent. So please set the commit to 1b306039ac49f8ad91ca71d3de3150a3c9fa792a, which is tagged with releases/gcc-15.1.0.

We also have an allowlist for failing test cases in this repository (see directory test/allowlist/gcc).
It would be great to clean this list up for the new GCC release (i.e. there should be zero failing tests after the allowlist has been applied):

• add failing tests to the allow list with a reference to a GCC bugzilla ticket (to ensure the bug is tracked upstream)
• remove items in the allowlist that are not needed anymore

[Arielfoever]

Thanks for the PR!

This PR references the untagged commit 9483020ea576516448f6e820fba4b14257d6268e from the remote branch releases/gcc-15. We prefer referencing tagged commits unless there are good reasons to update to something more recent. So please set the commit to 1b306039ac49f8ad91ca71d3de3150a3c9fa792a, which is tagged with releases/gcc-15.1.0.

We also have an allowlist for failing test cases in this repository (see directory test/allowlist/gcc). It would be great to clean this list up for the new GCC release (i.e. there should be zero failing tests after the allowlist has been applied):

* add failing tests to the allow list with a reference to a GCC bugzilla ticket (to ensure the bug is tracked upstream)

* remove items in the allowlist that are not needed anymore

I removed all white list. I will check the result and make a new white list.

[Arielfoever]

Ok I make a script and use it to gen the whitelists by math methods/

[Arielfoever]

@cmuellner @TommyMurphyTM1234 ready to merge

[cmuellner]

Thanks for providing the allow list!

I see tons of RVV issues, a few xtheadvector issues, a few gfortran issues, a few rv32 issues and the two test cases slt-1.c and pr112398.c. @kito-cheng, you might be interested in the updated list.

Please drop the first commit (disabling the fail early) from the PR. I think it would just lead to wasted CPU cycles in most situations.

[Arielfoever]

Thanks for providing the allow list!

I see tons of RVV issues, a few xtheadvector issues, a few gfortran issues, a few rv32 issues and the two test cases slt-1.c and pr112398.c. @kito-cheng, you might be interested in the updated list.

Please drop the first commit (disabling the fail early) from the PR. I think it would just lead to wasted CPU cycles in most situations.

The first commit is now dropped.

[cmuellner]

I like the idea of the script, as updating the allow list was the number one blocker for updating GCC in the past.
Is it right that the four text files are generated from the output of the testsuite-filter script?
Did you create the text files locally, or did you get the build log from the CI/CD artefacts?

[cmuellner]

LGTM

[Arielfoever]

I like the idea of the script, as updating the allow list was the number one blocker for updating GCC in the past. Is it right that the four text files are generated from the output of the testsuite-filter script? Did you create the text files locally, or did you get the build log from the CI/CD artefacts?

from CI log.

[Arielfoever]

I like the idea of the script, as updating the allow list was the number one blocker for updating GCC in the past. Is it right that the four text files are generated from the output of the testsuite-filter script? Did you create the text files locally, or did you get the build log from the CI/CD artefacts?

from CI log.

• Step 1: run ci without fast fail and all removed whitelist.
• Step 2: download the raw log.
• Step 3: edit input file in gen.py
• Step 4: run gen.py
• Step 5: copy the result files to whitelist.

All white list are generate by steps above except that I add one testcase since somehow it's failed sometimes but not always.

[pz9115]

LGTM, thanks for this work.