diff options
author | Lyude Paul <lyude@redhat.com> | 2019-04-24 14:48:35 -0400 |
---|---|---|
committer | Lyude Paul <lyude@redhat.com> | 2019-05-07 17:38:03 -0400 |
commit | 3f26aa32f2f9a17dd7fdf1a22321be02a6e63374 (patch) | |
tree | 7cd942a6246efb6beeff55d5ad57e8ea76c4055e /lib/igt_gpu_power.c | |
parent | 800618644a61acb1a00aa0459cd3432d68e1eac7 (diff) |
meson: Don't allow building with NDEBUG, ever
Following some discussion and confusion around whether or not assert()
should be used, it seems the decision has come to "yes, sometimes". To
quote Petri Latvala on the appropriate points to use assert() in lib/:
However, it's the thought that matters, and this is slightly going off
on a tangent. Those uses of assert in lib/ are for places where
1) something is fatally wrong and we need to drop everything and stop
executing
2) cannot use igt_assert for it. That's for places where we can say
"you tried testing your kernel but it has a bug". The lib/ asserts
are for "IGT has a bug", or in a couple of cases, "your IGT setup
has a bug".
While we did come to the conclusion that we should possibly consider
introducing a new API to check for bugs with igt (and prevent further
testing if any are found), until then let's at least make sure that
assert() always works where we expect it.
So, accomplish this by raising an error if b_ndebug isn't set to
'false'. Additionally, run the compile check in lib/check-ndebug.h to
make sure that the user does not have -DNDEBUG set in their CFLAGS,
c_args options, etc.
Reviewed-by: Petri Latvala <petri.latvala@intel.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Diffstat (limited to 'lib/igt_gpu_power.c')
0 files changed, 0 insertions, 0 deletions