[Intel-gfx] [PATCH i-g-t 1/4] lib: Extract helpers for determining submission method

Chris Wilson chris at chris-wilson.co.uk
Fri Oct 13 11:26:59 UTC 2017


Quoting Michał Winiarski (2017-10-13 12:00:17)
> Couple of tests are using either determining submission method, or
> pretty printing. Let's move those to helpers in lib.
> 
> Signed-off-by: Michał Winiarski <michal.winiarski at intel.com>
> Cc: Arkadiusz Hiler <arkadiusz.hiler at intel.com>
> Cc: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Katarzyna Dec <katarzyna.dec at intel.com>
> Cc: Petri Latvala <petri.latvala at intel.com>
> ---
>  lib/igt_aux.c               | 18 +++++++++++++
>  lib/igt_aux.h               |  2 ++
>  lib/igt_gt.c                | 63 +++++++++++++++++++++++++++++++++++++++++++++
>  lib/igt_gt.h                |  7 +++++
>  tests/gem_ctx_thrash.c      | 19 ++------------
>  tests/gem_eio.c             | 45 ++------------------------------
>  tests/gem_exec_await.c      | 39 ++--------------------------
>  tests/gem_exec_fence.c      | 41 +++--------------------------
>  tests/gem_exec_latency.c    | 31 +---------------------
>  tests/gem_exec_nop.c        | 31 +---------------------
>  tests/gem_exec_schedule.c   | 40 ++--------------------------
>  tests/gem_exec_whisper.c    | 31 +---------------------
>  tests/gem_read_read_speed.c | 17 +-----------
>  tests/gem_sync.c            | 31 +---------------------
>  14 files changed, 106 insertions(+), 309 deletions(-)
> 
> diff --git a/lib/igt_aux.c b/lib/igt_aux.c
> index fa6594c3..0c77f85f 100644
> --- a/lib/igt_aux.c
> +++ b/lib/igt_aux.c
> @@ -1455,6 +1455,24 @@ igt_show_stat(proc_t *info, int *state, const char *fn)
>         ++*state;
>  }
>  
> +void igt_show_submission_method(int fd)

Still igt when it appears to be gem specific.

I know I'm just as guilty of overusing the igt prefix, but I really do
think we should try to claw it back so that we know the platform
agnostic test tooling from the rest.

> +unsigned gem_submission_method(int fd)
> +{
> +       unsigned flags = 0;
> +       bool active;
> +       int dir;
> +
> +       dir = igt_sysfs_open_parameters(fd);
> +       if (dir < 0)
> +               return 0;

Ok, we really need to kill of the modparam probing here and get this out
of say PARAM_HAS_EXECBUF2 (by expanding that to this capability mask).

Because in about 4 weeks, we're going to lose some more of these
modparams... And planning for future expansion is a bonus.
-Chris


More information about the Intel-gfx mailing list