Add CTM to the cached layers' key and reduce render target pingponging in layer pre-rendering

Add CTM to the cached layers' key and reduce render target pingponging in layer pre-rendering

The CTM is required on the key so we will re-render when necessary.
The SkGpuDevice change ensures big layers in the midst of atlas-able layers don't cause a render target switch.

Committed: https://skia.googlesource.com/skia/+/0c4233285919c10df63c6cb207e580a7d1ff66b4

[bsalomon, 2014-07-30 21:23:09]

https://codereview.chromium.org/433553002/diff/20001/src/gpu/GrLayerCache.h
File src/gpu/GrLayerCache.h (right):

https://codereview.chromium.org/433553002/diff/20001/src/gpu/GrLayerCache.h#n...
src/gpu/GrLayerCache.h:85: return SkChecksum::Murmur3(reinterpret_cast<const
uint32_t*>(&key), sizeof(Key));
assert that key is tightly packed?

what if matrix::getType hasn't been called on one CTM but has on the other?

Stephen just encountered this. Maybe we should have an opaque struct that can be
copied out of an SkMatrix that just holds the matrix values (privately) and is
used for keys. Something like:

struct MatrixKey {
private:
  float m[9];
}

void SkMatrix::asKey(MatrixKey* key) const;

[robertphillips, 2014-07-31 13:23:36]

https://codereview.chromium.org/433553002/diff/20001/src/gpu/GrLayerCache.h
File src/gpu/GrLayerCache.h (right):

https://codereview.chromium.org/433553002/diff/20001/src/gpu/GrLayerCache.h#n...
src/gpu/GrLayerCache.h:85: return SkChecksum::Murmur3(reinterpret_cast<const
uint32_t*>(&key), sizeof(Key));
I have added the assert and forced the computation of the matrix type.

The MatrixKey proves problematic b.c. I want to use the ctm stored in the key as
a matrix. It would be possible to reinflate the matrix key but things stop being
simple.

[bsalomon, 2014-07-31 14:05:35]

lgtm

[robertphillips, 2014-07-31 17:51:33]

The CQ bit was checked by robertphillips@google.com

[commit-bot: I haz the power, 2014-07-31 17:51:56]

CQ is trying da patch. Follow status at

https://skia-tree-status.appspot.com/cq/robertphillips@google.com/433553002/1...

[commit-bot: I haz the power, 2014-07-31 18:02:42]

Change committed as 0c4233285919c10df63c6cb207e580a7d1ff66b4