Add plot-based purging to GrLayerCache

Add plot-based purging to GrLayerCache

This CL allows a GrPlot full of atlased layer to be purged from the atlas to make room for new layers.

Committed: https://skia.googlesource.com/skia/+/320c92380fe6a43dffbcd3f9e7c99897da44298d

[jvanverth1, 2014-07-28 19:33:57]

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.cpp
File src/gpu/GrLayerCache.cpp (right):

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.cpp...
src/gpu/GrLayerCache.cpp:173: for (int i = 0; i < 2; ++i) {
Why 2?

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.cpp...
src/gpu/GrLayerCache.cpp:223: // by delaying we may be able to reuse some of the
atlased layers later.
aggressively

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.h
File src/gpu/GrLayerCache.h (right):

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.h#n...
src/gpu/GrLayerCache.h:209: int fPlotLocks[kNumPlotsX * kNumPlotsY];
It's not clear to me why you couldn't use the drawToken mechanism in GrPlot. Can
a layer be locked for longer than a single render?

[robertphillips, 2014-07-28 20:27:47]

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.cpp
File src/gpu/GrLayerCache.cpp (right):

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.cpp...
src/gpu/GrLayerCache.cpp:173: for (int i = 0; i < 2; ++i) {
I added a comment. Is it clear with that addition?

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.cpp...
src/gpu/GrLayerCache.cpp:223: // by delaying we may be able to reuse some of the
atlased layers later.
On 2014/07/28 19:33:57, jvanverth1 wrote:
> aggressively

Done.

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.h
File src/gpu/GrLayerCache.h (right):

https://codereview.chromium.org/411703003/diff/60001/src/gpu/GrLayerCache.h#n...
src/gpu/GrLayerCache.h:209: int fPlotLocks[kNumPlotsX * kNumPlotsY];
To summarize the off-line discussion for posterity:

I appears that we cannot currently unify the two systems. When the DrawTarget is
immediate mode the drawToken mechanism will report too early that cached layers
are available for purging. The glyph caches do not face this problem since they
render the glyph immediately after it is generated so the immediate mode
DrawTarget is never lying. In the layer cache case, the final use of the created
layers is deferred until after all the layers are generated - so the immediate
mode DrawTarget would be lying about their purgeability.

[jvanverth1, 2014-07-29 12:54:34]

lgtm

[robertphillips, 2014-07-29 12:55:19]

The CQ bit was checked by robertphillips@google.com

[commit-bot: I haz the power, 2014-07-29 12:56:17]

CQ is trying da patch. Follow status at

https://skia-tree-status.appspot.com/cq/robertphillips@google.com/411703003/8...

[commit-bot: I haz the power, 2014-07-29 13:07:23]

Change committed as 320c92380fe6a43dffbcd3f9e7c99897da44298d