Here is the builder where this is breaking: https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Mac%20FYI%20Retina%20Debug%20%28NVIDIA%29

And the first build where the breakage started: https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Mac%20FYI%20Retina%20Debug%20%28NVIDIA%29/2371

and a log of the failure:

[  FAILED  ] ClearTest.ChangeFramebufferAttachmentFromRGBAtoRGB/ES2_OPENGL, where GetParam() = ES2_OPENGL (103 ms)
[ RUN      ] ClearTest.ChangeFramebufferAttachmentFromRGBAtoRGB/ES3_OPENGL
../../third_party/angle/src/tests/gl_tests/ClearTest.cpp:147: Failure
The difference between (0) and pixel[1] is 127, which exceeds 1.0, where
(0) evaluates to 0,
pixel[1] evaluates to 127, and
1.0 evaluates to 1.
Stack trace:
0   libbase.dylib                       0x000000010500d6cc base::debug::StackTrace::StackTrace() + 28
1   angle_end2end_tests                 0x000000010386bd7a StackTraceGetter::CurrentStackTrace(int, int) + 90
2   angle_end2end_tests                 0x0000000103890637 testing::internal::UnitTestImpl::CurrentOsStackTraceExceptTop(int) + 71
3   angle_end2end_tests                 0x000000010388f9ab testing::internal::AssertHelper::operator=(testing::Message const&) const + 139
4   angle_end2end_tests                 0x0000000102e4b49b (anonymous namespace)::ClearTest_ChangeFramebufferAttachmentFromRGBAtoRGB_Test::TestBody() + 3051

Since its angle_end2end_tests, and IIUC these only exercise the code in third_party/angle/, I assumed its something the angle roll. But the only change in the roll in that build is a logging change: https://chromium.googlesource.com/angle/angle.git/+log/a75aa3b2c314..7c37ca183640

+geofflang, could you help triage?