vmpstr@: Could you take a look at this? Looks like https://chromium-review.googlesource.com/c/chromium/src/+/755759 is relevant.

I get similar errors on linux desktop too:

[1:14:1117/103808.784746:FATAL:software_image_decode_cache.cc(556)] Check failed: result. frame_key[content_id: 1,frame_index: 0,subset_rect: 0,0 0x0] src_rect[48,0 48x48] target_size[12x12] target_color_space{primaries:BT709, transfer:IEC61966_2_1, matrix:RGB, range:FULL} filter_quality[2] can_use_original_size_decode [0] should_use_subrect [0] hash [3605204184479776501]

I'm going to revert the offending patches since I think the problem here is the rework of the sw cache missed a few cases.

For the OP, we're running in RGBA_4444 which doesn't support scaling, so we should never even be in the function that triggers the scale.

For #2, I _suspect_ the problem is the color space that isn't being handled correctly, but tbh I don't see why either peek pixels or extract pixels would fail. I'll make sure to check when relanding.

droger@, was it a particular site that triggered the issue? Are you running with any specific flags? How often could you reproduce the problem?

+droger for the question in #4

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a599a458d66a0fbe729e7cf946be0090af303d59

commit a599a458d66a0fbe729e7cf946be0090af303d59
Author: Vladimir Levin <vmpstr@chromium.org>
Date: Mon Nov 20 20:50:44 2017

Revert software image cache rework patches.

This patch reverts two commits:
Revert "Reland "cc: Rework software image decode cache.""
  This reverts commit 0a6f9bcd69c6f7c3d25675e794603c3c47a965da.
Revert "cc/images: Resuse the closest available decode for scaling sw images."
  This reverts commit 7a28b96c5c6833576e9a40af8b0dc81b73c5cbe5.

The reason is that the rework missed a few cases that are causing DCHECKs
or incorrect behavior in release builds. Since these are not currently
blocking any work, I will reland them with the proper fixes.

TBR=khushalsagar@chromium.org, ericrk@chromium.org

Bug:  786285 
Cq-Include-Trybots: master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel
Change-Id: I7fc81026e162d0854fc828bdd4457a3212b27db9
Reviewed-on: https://chromium-review.googlesource.com/779939
Commit-Queue: vmpstr <vmpstr@chromium.org>
Reviewed-by: Eric Karl <ericrk@chromium.org>
Reviewed-by: vmpstr <vmpstr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#517920}
[modify] https://crrev.com/a599a458d66a0fbe729e7cf946be0090af303d59/cc/paint/paint_image.cc
[modify] https://crrev.com/a599a458d66a0fbe729e7cf946be0090af303d59/cc/paint/paint_image.h
[modify] https://crrev.com/a599a458d66a0fbe729e7cf946be0090af303d59/cc/tiles/software_image_decode_cache.cc
[modify] https://crrev.com/a599a458d66a0fbe729e7cf946be0090af303d59/cc/tiles/software_image_decode_cache.h
[modify] https://crrev.com/a599a458d66a0fbe729e7cf946be0090af303d59/cc/tiles/software_image_decode_cache_unittest.cc

I had this error a lot when I was testing signin on accounts.google.com, like 90% of the time or more. I ended up turning off images in settings to be able to load the page. I was not using renderer-related flags (at first I was using site isolation, but then I disabled it and kept getting the error).

Alright thanks for the info. I reverted this for now, so this is fixed.