add blitprocs to shaders

add blitprocs to shaders

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1793883002

Committed: https://skia.googlesource.com/skia/+/830dfd87a7707f687e13da5be645e75c746a2cf5

[reed1, 2016-03-12 22:07:52]

Description was changed from

==========
add blitprocs to shaders

BUG=skia:
==========

to

==========
add blitprocs to shaders

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[reed1, 2016-03-12 22:08:02]

The CQ bit was checked by reed@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-03-12 22:08:05]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1793883002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1793883002/1

[commit-bot: I haz the power, 2016-03-12 22:16:56]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-03-12 22:16:56]

Dry run: This issue passed the CQ dry run.

[reed1, 2016-03-12 23:01:34]

reed@google.com changed reviewers:
+ fmalita@chromium.org, herb@google.com

[reed1, 2016-03-12 23:01:34]

whatcha think for gradients and bitmaps?

I think only supporting srcover and src may be important, since this optional
api is just around for performance (shadeSpan4f will still be in play).

[reed1, 2016-03-13 00:44:32]

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp
File src/core/SkBlitter_PM4f.cpp (right):

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp...
src/core/SkBlitter_PM4f.cpp:164: if (UseBProc) {
Hmm, I wonder if it's always better to use the BProc, even if fConstInY is
true... In that case it might be cheaper to call the shader once into 4f, and
then just call the blits...

[f(malita), 2016-03-13 15:48:36]

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp
File src/core/SkBlitter_PM4f.cpp (right):

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp...
src/core/SkBlitter_PM4f.cpp:192: fBProc(&fBState, x, y, fDevice, 1, nullptr);
fBProc(&fBState, x, y, fDevice, width, nullptr);

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp...
src/core/SkBlitter_PM4f.cpp:192: fBProc(&fBState, x, y, fDevice, 1, nullptr);
In the case of blitV/blitRect we discard some geometry info.  I wonder whether
shaders could take advantage of it if provided...  Linear gradients may try to
be smarter about finding the initial scanline interval when seeing the whole
picture for example.

Would it make sense for the API to use the most general form (blitRect)?

[reed1, 2016-03-13 15:58:53]

Not sure the frequency of fConstInY in practice, not the speed up between the
old code (where we cache the first span) and asking each shader to do it
themselves.

[reed1, 2016-03-13 16:00:54]

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp
File src/core/SkBlitter_PM4f.cpp (right):

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp...
src/core/SkBlitter_PM4f.cpp:192: fBProc(&fBState, x, y, fDevice, 1, nullptr);
On 2016/03/13 15:48:36, f(malita) wrote:
> In the case of blitV/blitRect we discard some geometry info.  I wonder whether
> shaders could take advantage of it if provided...  Linear gradients may try to
> be smarter about finding the initial scanline interval when seeing the whole
> picture for example.
> 
> Would it make sense for the API to use the most general form (blitRect)?

More to pass and more for each proc to do, so perhaps we should investigate
frequence of rects verses spans, and the perf difference.

[f(malita), 2016-03-16 15:00:39]

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp
File src/core/SkBlitter_PM4f.cpp (right):

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp...
src/core/SkBlitter_PM4f.cpp:192: fBProc(&fBState, x, y, fDevice, 1, nullptr);
On 2016/03/13 15:48:36, f(malita) wrote:
> fBProc(&fBState, x, y, fDevice, width, nullptr);

Looks like we need width instead of 1 here.

[reed1, 2016-03-16 16:58:39]

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp
File src/core/SkBlitter_PM4f.cpp (right):

https://codereview.chromium.org/1793883002/diff/1/src/core/SkBlitter_PM4f.cpp...
src/core/SkBlitter_PM4f.cpp:192: fBProc(&fBState, x, y, fDevice, 1, nullptr);
On 2016/03/16 15:00:39, f(malita) wrote:
> On 2016/03/13 15:48:36, f(malita) wrote:
> > fBProc(&fBState, x, y, fDevice, width, nullptr);
> 
> Looks like we need width instead of 1 here.

Doh! copy/paste error from blitV, Done.

[reed1, 2016-03-16 17:00:28]

The CQ bit was checked by reed@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-03-16 17:00:37]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1793883002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1793883002/20001

[commit-bot: I haz the power, 2016-03-16 17:19:46]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-03-16 17:19:47]

Dry run: This issue passed the CQ dry run.

[reed1, 2016-03-16 17:29:33]

ptal

[f(malita), 2016-03-16 17:34:56]

lgtm

[reed1, 2016-03-16 19:27:39]

The CQ bit was checked by reed@google.com

[commit-bot: I haz the power, 2016-03-16 19:27:53]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1793883002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1793883002/20001

[commit-bot: I haz the power, 2016-03-16 19:29:04]

Description was changed from

==========
add blitprocs to shaders

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
add blitprocs to shaders

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...

Committed:
https://skia.googlesource.com/skia/+/830dfd87a7707f687e13da5be645e75c746a2cf5
==========

[commit-bot: I haz the power, 2016-03-16 19:29:05]

Committed patchset #2 (id:20001) as
https://skia.googlesource.com/skia/+/830dfd87a7707f687e13da5be645e75c746a2cf5