Maybe I'm not familiar enough with this part of the stack, but what does PCI access have to do with Intel atomic instructions?

@MathStuff It looks like I misread the configure.ac. Even so, the motivation for my change remains the same: The build fails for me if I don't depend on libpciaccess.

Reviewing configure.ac again this morning, it appears to set the default value of INTEL to 'auto'. It then runs a bunch of checks and leaves INTEL turned on if 'drm_cv_atomic_primitives = none'. Then it aborts on this check (unless libpciaccess is provided):

'
if test "x$INTEL" != "xno"; then
PKG_CHECK_MODULES(PCIACCESS, [pciaccess >= 0.10])
fi
'

If my change breaks the package for some folks, we turn the new dependency into variant.

libdrm does depend on libpciaccess, so the change looks good to me.

OK, that's all good. The commit message is just confusing :) .

indeed 😄