[Buildroot] [PATCH 01/11] package/libselinux: fix build issues caused by 3.8.1 bump

[Thomas Petazzoni]

Hello Julien,

+Waldemar in Cc, uClibc-ng related issue.

On Mon, 15 Sep 2025 00:28:28 +0200
Julien Olivain <ju.o at free.fr> wrote:

> I applied the full series to master, thanks.

Thanks!

> While reviewing, I found one corner case build failure with libselinux 
> 3.9
> on x86_64 uclibc. See:
> https://gitlab.com/jolivain/buildroot/-/jobs/11350151899
> 
> It seems to be a followup issue related to the one fixed in this patch.

Dammit. I must say I don't quite grasp what they are trying to do in
these matchpathcon.c checks.

It seems like:

#if (defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS == 64) && defined(__INO64_T_TYPE) && !defined(__INO_T_MATCHES_INO64_T)
/* alias defined in the public header but we undefine it here */
#undef matchpathcon_filespec_add

/* ABI backwards-compatible shim for non-LFS 32-bit systems */

static_assert(sizeof(unsigned long) == sizeof(__ino_t), "inode size mismatch");
static_assert(sizeof(unsigned long) == sizeof(uint32_t), "inode size mismatch");
static_assert(sizeof(ino_t) == sizeof(ino64_t), "inode size mismatch");
static_assert(sizeof(ino64_t) == sizeof(uint64_t), "inode size mismatch");

is for 32-bit systems? And here you are building for x86-64, but it
gets detected as a 32-bit system, or something like that. Is it due to
the fact that we always build with _FILE_OFFSET_BITS=64 even on 64-bit
systems, where I guess it doesn't make sense because off_t is anyway
going to always be 64-bit.

And indeed, the check used to be:

-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS == 64 && __BITS_PER_LONG < 64
+#if (defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS == 64) && !defined(__INO_T_MATCHES_INO64_T)

which would be false on 64-bit platforms... but now is true on
uClibc-ng.

That check was "improved" for musl:

-#if (defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS == 64) && !defined(__INO_T_MATCHES_INO64_T)
+#if (defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS == 64) && defined(__INO64_T_TYPE) && !defined(__INO_T_MATCHES_INO64_T)

So I guess some additional tweak is needed for uClibc-ng.

I'm not sure what the correct fix is.

Thomas
-- 
Thomas Petazzoni, co-owner and CEO, Bootlin
Embedded Linux and Kernel engineering and training
https://bootlin.com