Re: [RFC] Host1x/TegraDRM UAPI

[Dmitry Osipenko <digetx-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

23.06.2020 15:09, Mikko Perttunen пишет:
> struct drm_tegra_channel_submit {
>         __u32 channel_id;
>         __u32 flags;
> 
>         /**
>          * [in] Timeout in microseconds after which the kernel may
>          *   consider the job to have hung and may reap it and
>          *   fast-forward its syncpoint increments.
>          *
>          *   The value may be capped by the kernel.
>          */
>         __u32 timeout;
> 
>         __u32 num_syncpt_incrs;
>         __u32 num_relocations;
>         __u32 num_commands;
> 
>         __u64 syncpt_incrs;
>         __u64 relocations;
>         __u64 commands;

Do we really need to retain the multi-gather support? The code-bloat
(both userspace and kernel driver) is very significant just for
preparing and patching of the multi-buffer cmdstreams.

If userspace runs out of a free space within the pushbuffer, then it
should simply reallocate a larger pushbuffer.

I'm suggesting that we should have a single gather per-job, any objections?