On 6 February 2018 at 21:19, Peter Seiderer <[email protected]> wrote: > From: Thomas Petazzoni <[email protected]> > > The configure script currently tests the availability of libatomic_ops > by checking the presence of atomic_ops.h. While this is good enough as > an availability test, it is not sufficient as on some platforms, > libatomic_ops provides an actual shared library against which we > should be linked to access libatomic_ops functionality. > > Therefore, we instead use PKG_CHECK_MODULES() to test the availability > of libatomic_ops. Besides testing its availability, this also fills in > the ATOMIC_OPS_LIBS variable with the list of libraries we need to > link with to use libatomic_ops. > > All Mesa drivers that include xf86atomic.h have been updated to link > against ATOMIC_OPS_LIBS. > > Of course, if we're not using libatomic_ops, ATOMIC_OPS_LIBS is empty, > and we don't link against it. > > Signed-off-by: Thomas Petazzoni <[email protected]> > Signed-off-by: Peter Seiderer <[email protected]> > [Bernd: PKG_CHECK_MODULES should not fail when libatomic_ops is missing] > Signed-off-by: Bernd Kuhls <[email protected]> > --- > amdgpu/Makefile.am | 2 +- > configure.ac | 2 +- > etnaviv/Makefile.am | 3 ++- > freedreno/Makefile.am | 3 ++- > intel/Makefile.am | 3 ++- > nouveau/Makefile.am | 2 +- > omap/Makefile.am | 2 +- > radeon/Makefile.am | 2 +- > tegra/Makefile.am | 2 +- You're adding ATOMIC_OPS_LIBS to all the correct places, although ATOMIC_OPS_CFLAGS seems to be missing. Can you re-spin with that one added?
Thanks Emil _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
