Adding new fuzzer deserialization test code

Adding new fuzzer deserialization test code

BUG=

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=235846

[sugoi1, 2013-11-13 19:34:36]

Here's a 1st version. It's really simple. Is this what you guys had in mind ?

[mbarbella (wrong one), 2013-11-13 20:02:56]

https://codereview.chromium.org/67173009/diff/1/skia/tools/filter_fuzz_stub/f...
File skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc (right):

https://codereview.chromium.org/67173009/diff/1/skia/tools/filter_fuzz_stub/f...
skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc:27:
SkImageFilter::GetFlattenableType());
Once this is done, try to render something as discussed in chat.

[sugoi1, 2013-11-13 20:56:47]

Added rendering code

[mbarbella (wrong one), 2013-11-13 21:08:11]

lgtm

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
File skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc (right):

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc:76:
bitmap.setConfig(SkBitmap::kARGB_8888_Config, BitmapSize, BitmapSize);
Minor thing. Could we only do this if the stream is valid? It probably isn't
much overhead, but it seems like an easy change.

[sugoi1, 2013-11-13 21:17:00]

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
File skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc (right):

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc:76:
bitmap.setConfig(SkBitmap::kARGB_8888_Config, BitmapSize, BitmapSize);
On 2013/11/13 21:08:11, mbarbella1 wrote:
> Minor thing. Could we only do this if the stream is valid? It probably isn't
> much overhead, but it seems like an easy change.

AFAIK, this has to be done before it can be used as a device and, for a large
enough number of test cases, this shouldn't take a significant amount of time,
since it is done only once for all test cases.

[sugoi1, 2013-11-13 21:17:53]

Adding robertphillips@ as a reviewer to get some input from a Skia dev.

[mbarbella (wrong one), 2013-11-13 21:18:33]

On 2013/11/13 21:17:00, sugoi1 wrote:
>
https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
> File skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc (right):
> 
>
https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
> skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc:76:
> bitmap.setConfig(SkBitmap::kARGB_8888_Config, BitmapSize, BitmapSize);
> On 2013/11/13 21:08:11, mbarbella1 wrote:
> > Minor thing. Could we only do this if the stream is valid? It probably isn't
> > much overhead, but it seems like an easy change.
> 
> AFAIK, this has to be done before it can be used as a device and, for a large
> enough number of test cases, this shouldn't take a significant amount of time,
> since it is done only once for all test cases.

Makes sense after looking at it more closely. The overhead won't be an issue.

[robertphillips, 2013-11-14 14:32:04]

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
File skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc (right):

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc:45: // If nothing bad happens
within this call, everything is fine
If I read this correctly, this is drawing the bitmap into itself. It should work
but it might be clearer to make a separate bitmap (e.g., a checkerboard) rather
then reusing the canvas' bitmap.

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc:76:
bitmap.setConfig(SkBitmap::kARGB_8888_Config, BitmapSize, BitmapSize);
You could lazily create the canvas/bitmap but this should be pretty low
overhead.

[sugoi1, 2013-11-14 15:08:46]

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
File skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc (right):

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc:45: // If nothing bad happens
within this call, everything is fine
On 2013/11/14 14:32:04, robertphillips wrote:
> If I read this correctly, this is drawing the bitmap into itself. It should
work
> but it might be clearer to make a separate bitmap (e.g., a checkerboard)
rather
> then reusing the canvas' bitmap.

The only thing I'm worried about in this test is whether or not applying the
image filters would cause a memory issue or crash. Would using a separate bitmap
have an influence on this ?

[robertphillips, 2013-11-14 15:11:53]

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
File skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc (right):

https://codereview.chromium.org/67173009/diff/60001/skia/tools/filter_fuzz_st...
skia/tools/filter_fuzz_stub/filter_fuzz_stub.cc:45: // If nothing bad happens
within this call, everything is fine
It should be okay.

[sugoi1, 2013-11-18 16:46:49]

filter_fuzz_stub is now built when compiling chromium_builder_asan.

[Stephen White, 2013-11-18 18:37:13]

LGTM

[commit-bot: I haz the power, 2013-11-18 20:44:50]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/sugoi@chromium.org/67173009/240001

[commit-bot: I haz the power, 2013-11-18 23:35:24]

Change committed as 235846