1011569 - Remove PreferMemoryOverShmem

Status: RESOLVED FIXED

(Core :: Graphics: Layers, defect)

Description

[Chris Lord [:cwiiis]]

PreferMemoryOverShmem has been superseded by IsSameProcess, we can remove the former and reduce code complexity a little.

Comment 1

[Chris Lord [:cwiiis]]

Attachment: Replace PreferMemoryOverShmem with !IsSameProcess

Comment 2

[Andreas Gal :gal]

Actually I would like to remove the non-Shmem case altogether. What do you think?

Comment 3

[Chris Lord [:cwiiis]]

(In reply to Andreas Gal :gal from comment #2)
> Actually I would like to remove the non-Shmem case altogether. What do you
> think?

TL;DR: Broadly in favour.


I'd like to, but last I checked it was an appreciable performance hit on Android at least.

I think we'd need to profile any significant regression and understand where it's coming from before doing that. It may not actually be a regression now we have the tile cache, but if it is we definitely need to investigate why (which may lead to some nice low-hanging fruit :))

That shouldn't stop this landing though (if it passes review).

Comment 4

[Andreas Gal :gal]

This should only be a hit if we don't cache content clients properly. Especially with tiling we should have a 100% cache hit rate for that.

Comment 5

[Nicolas Silva [:nical]]

There are still other places where we do a poor job of caching textures (pretty much everywhere except tiled layers) I don't think we can get rid of MemoryTextureClient/Host just yet (although I am all for removing it as soon as it doesn't cause a performance hit).

Comment 6

[Chris Lord [:cwiiis]]

Pushed to inbound: https://hg.mozilla.org/integration/mozilla-inbound/rev/a6f85ff386c1

Comment 7

[Carsten Book [:Tomcat]]

sorry had to backout for test failures like https://tbpl.mozilla.org/php/getParsedLog.php?id=39938148&tree=Mozilla-Inbound

Comment 8

[Chris Lord [:cwiiis]]

The assertions were likely caused by shmem being used where previously it wasn't because IsSameProcess is broken (bug 1012639). I'll re-land once that bug is resolved.

Comment 9

[Chris Lord [:cwiiis]]

Pushed to inbound: https://hg.mozilla.org/integration/mozilla-inbound/rev/e7537b4f7a6c

Comment 10

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/e7537b4f7a6c