ftrace:test.d--dynevent--add_remove_btfarg.tc in ubuntu_kselftests_ftrace failed on some architectures

[Masahiro Yamada]

In our SRU cycles for Noble kernels, ftrace:test.d--dynevent--add in ubuntu_kselftests_ftrace fails on some architectures.

This test works only when CONFIG_FPROBE=y and CONFIG_FPROBE_EVENTS=y.

In our annotation file, they are enabled only for some architectures.

CONFIG_FPROBE policy<{'amd64': 'y', 'riscv64': '-', 's390x': 'y'}>
CONFIG_FPROBE_EVENTS policy<{'amd64': 'y', 'riscv64': '-', 's390x': 'y'}>

For the other architectures (e.g. arm64), this test always fails.

It should be skipped when CONFIG_FPROBE=n or CONFIG_FPROBE_EVENTS=n.

It was already fixed in upstream by the following commit:

commit 2fd3ef1b9265eda7f53b9506f1ebfb67eb6435a2
Author: Masami Hiramatsu (Google) <email address hidden>
Date: Sat May 4 09:33:10 2024 +0900

    selftests/ftrace: Fix BTFARG testcase to check fprobe is enabled correctly

[Masahiro Yamada (myamada)]

One way to go is to backport 2fd3ef1b926 directly to the noble kernel.

Another is to backport it to 6.8.y stable kernel, and then to the noble kernel as a part of regular cycles.

I am asking Greg-KH to backport it to the stable kernel.
I will update this if there is a progress.