[igt-dev] [PATCH v2 1/8] tests/panfrost: Make sure we open a DUMB capable node for prime tests

[Petri Latvala]

On Thu, Jun 24, 2021 at 04:29:43PM +0200, Boris Brezillon wrote:
> Hi Petri,
> 
> On Tue, 22 Jun 2021 12:42:01 +0300
> Petri Latvala <petri.latvala at intel.com> wrote:
> 
> > On Tue, Jun 22, 2021 at 11:02:14AM +0200, Boris Brezillon wrote:
> > > v2:
> > > * Use __drm_open_driver_another()
> > > 
> > > Signed-off-by: Boris Brezillon <boris.brezillon at collabora.com>
> > > ---
> > >  tests/panfrost_prime.c | 25 ++++++++++++++++++++++++-
> > >  1 file changed, 24 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/tests/panfrost_prime.c b/tests/panfrost_prime.c
> > > index 351d46f2f7e6..dbc741a9eacb 100644
> > > --- a/tests/panfrost_prime.c
> > > +++ b/tests/panfrost_prime.c
> > > @@ -35,13 +35,33 @@
> > >  #include <sys/ioctl.h>
> > >  #include "panfrost_drm.h"
> > >  
> > > +static bool igt_has_dumb(int fd)
> > > +{
> > > +	uint64_t value = 0;
> > > +	int ret;
> > > +
> > > +	ret = drmGetCap(fd, DRM_CAP_DUMB_BUFFER, &value);
> > > +	igt_assert(ret == 0 || errno == EINVAL || errno == EOPNOTSUPP);
> > > +	return value == 1;
> > > +}
> > > +
> > > +static bool igt_has_prime(int fd, uint64_t flags)
> > > +{
> > > +	uint64_t value = 0;
> > > +	int ret;
> > > +
> > > +	ret = drmGetCap(fd, DRM_CAP_PRIME, &value);
> > > +	igt_assert(ret == 0 || errno == EINVAL || errno == EOPNOTSUPP);
> > > +	return (value & flags) == flags;
> > > +}
> > > +
> > >  igt_main
> > >  {
> > >  	int fd, kms_fd;
> > >  
> > >  	igt_fixture {
> > > -		kms_fd = drm_open_driver_master(DRIVER_ANY);
> > >  		fd = drm_open_driver(DRIVER_PANFROST);
> > > +		kms_fd = __drm_open_driver_another(1, DRIVER_ANY);
> > >  	}
> > >  
> > >  	igt_subtest("gem-prime-import") {
> > > @@ -50,6 +70,9 @@ igt_main
> > >  	        struct drm_panfrost_get_bo_offset get_bo_offset = {0,};
> > >  		int dmabuf_fd;
> > >  
> > > +		igt_skip_on(!igt_has_dumb(kms_fd) ||
> > > +			    !igt_has_prime(kms_fd, DRM_PRIME_CAP_EXPORT));  
> > 
> > For better logs in case any of these trigger, use separate clauses:
> > 
> > igt_require(igt_has_dumb(kms_fd));
> > igt_require(igt_has_prime(kms_fd, DRM_PRIME_CAP_EXPORT));
> > 
> > 
> > Either way, LGTM
> > Acked-by: Petri Latvala <petri.latvala at intel.com>
> 
> Thanks for this ack. I haven't contributed to igt in a while, and I
> don't remember the process to push patches once they've received
> acks/reviews. Should I push directly to the igt/master on gitlab (in
> which case I'd need to be added to the project), or should I let
> someone else push things for me.

After acks/reviews if the CI doesn't give a fuzz you just push. You
have enough activity now to qualify for direct access so... well you
already have the access. Granted 2 days ago by Daniel Stone.


-- 
Petri Latvala