libobs-opengl: Fix compile error on GCC 14+ and i686

[oreo639]

Description

Under the public gl_egl_query_dmabuf_modifiers_for_format, n_modifiers is defined as size_t, however in the static query_dmabuf_modifiers function, it is defined as EGLuint64KHR.

Since query_dmabuf_modifiers is only called by gl_egl_query_dmabuf_modifiers_for_format and n_modifiers is assigned from an EGLint, make it size_t for both.

This fixes building on i686 with gcc14.

Motivation and Context

On i686 with gcc14, the following compile error is outputted:

 [309/422] Building C object libobs-opengl/CMakeFiles/libobs-opengl.dir/gl-egl-common.c.o
FAILED: libobs-opengl/CMakeFiles/libobs-opengl.dir/gl-egl-common.c.o 
/usr/bin/cc -DENABLE_HEVC -DHAVE_OBSCONFIG_H -DSIMDE_ENABLE_OPENMP -Dlibobs_opengl_EXPORTS -I/builddir/obs-31.0.2/build/libobs-opengl -I/builddir/obs-31.0.2/libobs-opengl -I/builddir/obs-31.0.2/build/config -I/builddir/obs-31.0.2/libobs -I/builddir/obs-31.0.2/deps/glad/include -DNDEBUG -fstack-clash-protection -D_FORTIFY_SOURCE=2 -mtune=i686 -O2    -ffile-prefix-map=/builddir/obs-31.0.2/build=. -std=gnu17 -fPIC -fvisibility=hidden -fopenmp-simd -fno-strict-aliasing -Wdeprecated-declarations -Wempty-body -Wenum-conversion -Werror=return-type -Wextra -Wformat -Wformat-security -Wno-conversion -Wno-float-conversion -Wno-implicit-fallthrough -Wno-missing-braces -Wno-missing-field-initializers -Wno-shadow -Wno-sign-conversion -Wno-trigraphs -Wno-unknown-pragmas -Wno-unused-function -Wno-unused-label -Wparentheses -Wswitch -Wuninitialized -Wunreachable-code -Wunused-parameter -Wunused-value -Wunused-variable -Wvla -Wint-conversion -Wno-missing-prototypes -Wno-strict-prototypes -Wpointer-sign -Wno-error=maybe-uninitialized -Winfinite-recursion -Wno-stringop-overflow -MD -MT libobs-opengl/CMakeFiles/libobs-opengl.dir/gl-egl-common.c.o -MF libobs-opengl/CMakeFiles/libobs-opengl.dir/gl-egl-common.c.o.d -o libobs-opengl/CMakeFiles/libobs-opengl.dir/gl-egl-common.c.o -c /builddir/obs-31.0.2/libobs-opengl/gl-egl-common.c
/builddir/obs-31.0.2/libobs-opengl/gl-egl-common.c: In function 'gl_egl_query_dmabuf_modifiers_for_format':
/builddir/obs-31.0.2/libobs-opengl/gl-egl-common.c:387:73: error: passing argument 4 of 'query_dmabuf_modifiers' from incompatible pointer type [-Wincompatible-pointer-types]
  387 |         if (!query_dmabuf_modifiers(egl_display, drm_format, modifiers, n_modifiers)) {
      |                                                                         ^~~~~~~~~~~
      |                                                                         |
      |                                                                         size_t * {aka unsigned int *}
/builddir/obs-31.0.2/libobs-opengl/gl-egl-common.c:349:57: note: expected 'EGLuint64KHR *' {aka 'long long unsigned int *'} but argument is of type 'size_t *' {aka 'unsigned int *'}
  349 |                                           EGLuint64KHR *n_modifiers)
      |                                           ~~~~~~~~~~~~~~^~~~~~~~~~~

How Has This Been Tested?

This was compiled with gcc 14 and glibc 2.41 on Void Linux.

Types of changes

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• My code has been run through clang-format.
• I have read the contributing document.
• My code is not on the master branch.
• The code has been tested.
• All commit messages are properly formatted and commits squashed where appropriate.
• I have included updates to all appropriate documentation.

[kkartaltepe]

nack on this solution.

If you want to maintain 32bit support consider converting to fixed size types at the api borders instead of introducing architecture dependent types here. Though I think someone else submitted this a while back and we didnt have any active contributors who cared about 32bit and preferred to leave dealing it to active 32bit communities like legacy distros instead.

[oreo639]

If you want to maintain 32bit support consider converting to fixed size types at the api borders

Makes sense.

[norihiro]

Though I think someone else submitted this a while back

It was #11349, but #11349 has a buffer overrun in the 32-bit size_t type system.