Hi Kamil, On Wed, 16 Sep 2026 09:57:41 +0200 Kamil Konieczny <[email protected]> wrote:
> Hi Deborah, > On 2026-09-15 at 17:13:26 -0700, Deborah Brouwer wrote: > > Sparse VM_BIND support was introduced with Panthor DRM driver version > > 1.9. If a driver isn't advertising support for v1.9, then tests that > > call DRM_PANTHOR_VM_BIND_OP_MAP_SPARSE should not be run. > > Can you just call that ioctl and if EINVALID skip? > Checking version of a driver seems strange. That's how we check feature-support in mesa, so I'd be tempted to do the same in IGT, unless there's a strong reason not to. If we were to check against EINVAL to detect when a feature is not supported, it gets hard to know when it's an actual uAPI regression vs when it's tested against a driver version that's too old and doesn't support the extra arguments/flags. Regards, Boris
