1675159 - Implement Draw / partial present compositor in terms of native compositor interface

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, task)

Comment 1

[Glenn Watson [:gw]]

This will simplify much of the picture cache code, and is also a way to expose partial present dirty rects to the native compositor interface (e.g. for swgl)

Comment 2

[Glenn Watson [:gw]]

This is the ordering of relevant parts of the draw / native compositor code paths in the current implementation (reference for porting draw -> native):

new:
    get_capabilities

render_impl:
    (if compositor changed)
        destroy_surface
        enable_native_compositor

    update_native_surfaces
        create_external_surface
        attach_external_image
        create_surface
        destroy_surface
        create_tile
        destroy_tile

    begin_frame

    draw_frame
        invalidate_tile

        composite_native
            add_surface[]
            start_compositing

        passes[]
            main
                composite_simple
                    get_buffer_age
                    set_buffer_damage_region
            offscreen
                bind
                unbind

    bind_debug_overlay
    unbind_debug_overlay

    end_frame

deinit:
    destroy_surface
    deinit

Comment 3

[Glenn Watson [:gw]]

Attachment: Bug 1675159 - Pt 1 - Pass `buffer_age` via `render` rather than callback.

Implementing the Draw compositor via the native compositor interface
is simpler if the buffer age is passed into the top level render method.

Comment 4

[Glenn Watson [:gw]]

Attachment: Bug 1675159 - Pt 2 - Expose dirty rects to start_compositing.

Move the calculation of the dirty rects array earlier in frame
drawing, and supply that to the start_compositing method of
the compositor trait.

For now, it's assumed that the native compositor wants a single
dirty rect, and doesn't use buffer-age functionality. These
params will be configurable as part of the compositor capabilities
struct in follow up patches.

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/644505d8afbb
Pt 1 - Pass `buffer_age` via `render` rather than callback. r=jnicol

Comment 6

[Dorel Luca [:dluca]]

Backed out changeset 644505d8afbb (bug 1675159) for Browser-chrome failures in browser/base/content/test/general/browser_save_video_frame.js. CLOSED TREE

Log:
https://treeherder.mozilla.org/logviewer?job_id=320752832&repo=autoland&lineNumber=2334

Push with failures:
https://treeherder.mozilla.org/jobs?repo=autoland&group_state=expanded&revision=644505d8afbb7e297270d186a2cc5e37488be3f5

Backout:
https://hg.mozilla.org/integration/autoland/rev/a86eb22de44a700456acd86b74e539ec0c4230df

Comment 7

[Glenn Watson [:gw]]

Discussed in the sheriffs channel that I think these are unrelated, so will try to reland. Looks similar to a preexisting https://bugzilla.mozilla.org/show_bug.cgi?id=1673742, perhaps random timing changes make it more likely with this patch.

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/33fbbff5e725
Pt 1 - Pass `buffer_age` via `render` rather than callback. r=jnicol

Comment 9

[Dorel Luca [:dluca]]

Dave: Can you please take a look at the failures in comment 6 and 7?

Comment 10

[Cristian Brindusan [:cbrindusan]]

https://hg.mozilla.org/mozilla-central/rev/33fbbff5e725

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/93e746226b8f
Pt 2 - Expose dirty rects to start_compositing. r=sotaro

Comment 12

[Andreea Pavel [:apavel]]

https://hg.mozilla.org/mozilla-central/rev/93e746226b8f

Comment 13

[Dave Townsend [:mossop]]

Looks like this is no longer needed?