Re: [PATCH v5 6/7] selftests: prctl: introduce tests for disabling THPs completely

[Mark Brown <broonie@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 1984 bytes --]

On Fri, Aug 15, 2025 at 02:54:58PM +0100, Usama Arif wrote:
> The test will set the global system THP setting to never, madvise
> or always depending on the fixture variant and the 2M setting to
> inherit before it starts (and reset to original at teardown).
> The fixture setup will also test if PR_SET_THP_DISABLE prctl call can
> be made to disable all THPs and skip if it fails.

I don't think this is an issue in this patch but with it we're seeing
build failures in -next on arm64 with:

  make KBUILD_BUILD_USER=KernelCI FORMAT=.xz ARCH=arm64 HOSTCC=gcc CROSS_COMPILE=aarch64-linux-gnu- CROSS_COMPILE_COMPAT=arm-linux-gnueabihf- CC="ccache aarch64-linux-gnu-gcc" O=/tmp/kci/linux/build -C/tmp/kci/linux -j98 kselftest-gen_tar

  ...

    CC       prctl_thp_disable
  prctl_thp_disable.c: In function ‘test_mmap_thp’:
  prctl_thp_disable.c:64:39: error: ‘MADV_COLLAPSE’ undeclared (first use in this function); did you mean ‘MADV_COLD’?
     64 |                 madvise(mem, pmdsize, MADV_COLLAPSE);
        |                                       ^~~~~~~~~~~~~
        |                                       MADV_COLD

since the headers_install copy of asm-generic/mman-common.h doesn't
appear to being picked up with the above build invocation (most others
are fine).  I'm not clear why, it looks like an appropriate -isystem
ends up getting passed to the compiler:

  aarch64-linux-gnu-gcc -Wall -O2 -I /linux/tools/testing/selftests/../../..  -isystem /tmp/kci/linux/build/usr/include -isystem /linux/tools/testing/selftests/../../../tools/include/uapi -U_FORTIFY_SOURCE -D_GNU_SOURCE=     prctl_thp_disable.c vm_util.c thp_settings.c -lrt -lpthread -lm -o /tmp/kci/linux/build/kselftest/mm/prctl_thp_disable

but the header there is getting ignored AFAICT.  Probably the problem is
fairly obvious and I'm just being slow - I'm not quite 100% at the
minute.

Thanks to Aishwarya for confirming which patch triggered the issue.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]