Fix gaussian blur for small sigma.

Fix gaussian blur for small sigma.

I broke this in https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719, 569883
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1529313003

Committed: https://skia.googlesource.com/skia/+/c834ab178e56009875dee8d2265dba35cf6e1e55

[Stephen White, 2015-12-17 01:45:47]

Description was changed from

==========
Fix gaussian blur for small sigma.

I broke this in
https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719,569883
==========

to

==========
Fix gaussian blur for small sigma.

I broke this in
https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719,569883
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Stephen White, 2015-12-17 01:47:16]

The CQ bit was checked by senorblanco@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-12-17 01:47:19]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1529313003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1529313003/1

[Stephen White, 2015-12-17 01:47:28]

Description was changed from

==========
Fix gaussian blur for small sigma.

I broke this in
https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719,569883
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Fix gaussian blur for small sigma.

I broke this in
https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719,569883
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Stephen White, 2015-12-17 01:47:28]

senorblanco@chromium.org changed reviewers:
+ bsalomon@google.com

[Stephen White, 2015-12-17 01:47:37]

Brian: PTAL. Thanks!

[Stephen White, 2015-12-17 01:48:38]

Description was changed from

==========
Fix gaussian blur for small sigma.

I broke this in
https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719,569883
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Fix gaussian blur for small sigma.

I broke this in
https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719,569883
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[commit-bot: I haz the power, 2015-12-17 02:01:43]

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

[commit-bot: I haz the power, 2015-12-17 02:01:44]

Dry run: This issue passed the CQ dry run.

[bsalomon, 2015-12-17 16:00:18]

lgtm

[Stephen White, 2015-12-17 16:09:29]

The CQ bit was checked by senorblanco@chromium.org

[commit-bot: I haz the power, 2015-12-17 16:09:40]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1529313003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1529313003/1

[commit-bot: I haz the power, 2015-12-17 16:10:19]

Description was changed from

==========
Fix gaussian blur for small sigma.

I broke this in
https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719,569883
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Fix gaussian blur for small sigma.

I broke this in
https://skia.googlesource.com/skia/+/c57e0ded7d535523cfc6bf07c78e5f3479bb8c42.
That change added support for cropping during blur, but
did not do the correct thing for the small-sigma 2D matrix convolution
optimization when blurring from small-to-large textures.

The fix is pass the correct dstRect and srcOffset to convolve_gaussian_2d.
I also changed convolve_gaussian_1d() and convolve_gaussian_2d() to take
the original (non-negated) srcOffset, and to negate them when constructing
the local matrix.

Test: added a new (blur 0.3 0.3) column to imagefilterscropexpand, so that
GM will have to be rebaselined.

BUG=skia:4719,569883
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...

Committed:
https://skia.googlesource.com/skia/+/c834ab178e56009875dee8d2265dba35cf6e1e55
==========

[commit-bot: I haz the power, 2015-12-17 16:10:20]

Committed patchset #1 (id:1) as
https://skia.googlesource.com/skia/+/c834ab178e56009875dee8d2265dba35cf6e1e55