[PATCH] selftests/configfs: fix the build on 32-bit architectures

[PATCH] selftests/configfs: fix the build on 32-bit architectures

[Sasha Levin]

The fixture setup checks the result of mkdtemp() with ASSERT_NE(), which
hands the pointer to the harness. The harness casts both operands to
intmax_t to print them, and on 32-bit architectures gcc warns about the
cast from a pointer to an integer of a different size. The configfs
test builds with -Werror, so it fails to build there:

  kselftest_harness.h:797:49: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]

Compare the result against NULL inside ASSERT_TRUE(), so that the
harness only sees an int.

Fixes: a7fdcb2edd0f ("selftests/configfs: add tests for the userspace interface")
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
KernelCI's 32-bit arm kselftest builds fail on this. Built
configfs_test.c with gcc 16 -m32, with and without _TIME_BITS=64 (fails
before, builds after), clang 21 -m32, and both compilers for x86_64.

The harness cast warns for every pointer comparison on 32-bit, a few
thousand warnings in one KernelCI arm build, and configfs only breaks
because of its -Werror. Making the harness print pointers through
uintptr_t would be a separate change.

 tools/testing/selftests/filesystems/configfs/configfs_test.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/testing/selftests/filesystems/configfs/configfs_test.c b/tools/testing/selftests/filesystems/configfs/configfs_test.c
index c6a1049e5852e..94ae8f7103092 100644
--- a/tools/testing/selftests/filesystems/configfs/configfs_test.c
+++ b/tools/testing/selftests/filesystems/configfs/configfs_test.c
@@ -122,7 +122,7 @@ FIXTURE_SETUP(configfs)
 	ASSERT_EQ(unshare(CLONE_NEWNS), 0);
 	ASSERT_EQ(mount(NULL, "/", NULL, MS_REC | MS_PRIVATE, NULL), 0);
 
-	ASSERT_NE(mkdtemp(tmpl), NULL);
+	ASSERT_TRUE(mkdtemp(tmpl) != NULL);
 	strcpy(self->mnt, tmpl);
 
 	ASSERT_EQ(mount("configfs", self->mnt, "configfs", 0, NULL), 0);
-- 
2.53.0

Re: [PATCH] selftests/configfs: fix the build on 32-bit architectures

[Breno Leitao]

On Tue, Nov 03, 2026 at 09:16:27AM -0500, Sasha Levin wrote:
> The fixture setup checks the result of mkdtemp() with ASSERT_NE(), which
> hands the pointer to the harness. The harness casts both operands to
> intmax_t to print them, and on 32-bit architectures gcc warns about the
> cast from a pointer to an integer of a different size. The configfs
> test builds with -Werror, so it fails to build there:
> 
>   kselftest_harness.h:797:49: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
> 
> Compare the result against NULL inside ASSERT_TRUE(), so that the
> harness only sees an int.
> 
> Fixes: a7fdcb2edd0f ("selftests/configfs: add tests for the userspace interface")
> Signed-off-by: Sasha Levin <sashal@kernel.org>

Thanks for the fix.

Reviewed-by: Breno Leitao <leitao@debian.org>

> ---
> KernelCI's 32-bit arm kselftest builds fail on this. Built
> configfs_test.c with gcc 16 -m32, with and without _TIME_BITS=64 (fails
> before, builds after), clang 21 -m32, and both compilers for x86_64.
> 
> The harness cast warns for every pointer comparison on 32-bit, a few
> thousand warnings in one KernelCI arm build, and configfs only breaks
> because of its -Werror. Making the harness print pointers through
> uintptr_t would be a separate change.

Christophe posted a harness fix for this in January, but it was never
merged?

    https://lore.kernel.org/all/948937bfe461cf0fb6ade2623e51f673f845fa51.1768298756.git.chleroy@kernel.org/

It still applies to linux-next, and with it configfs_test.c compiles on
32-bit without this patch.

Either way, I will merge your patch tomorrow if no other review pops up.

Thanks,
--breno