Whoops, didn't mean to close this one.

Looks like this bot has the same issue but a smaller regression range: https://build.chromium.org/p/chromium.gpu.fyi/builders/Mac%20Pro%20Debug%20%28AMD%29/builds/1072

Nothing looks obvious yet though.

I think Mac Pro Debug AMD issue was caused by "memory-infra: Finish moving to Mojo (2nd attempt)" https://codereview.chromium.org/2734193002, which was reverted in https://codereview.chromium.org/2738093003, making https://build.chromium.org/p/chromium.gpu.fyi/builders/Mac%20Pro%20Debug%20%28AMD%29/builds/1087 green.
Doesn't seem related to Linux Debug (Intel HD 530) problem.

Linux Debug (Intel HD 530) was green for 3 build recently.
Looks like https://codereview.chromium.org/2746493002 "Statically link libprotobuf_lite on Linux" made it green in https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Debug%20%28Intel%20HD%20530%29/builds/825, and then it became red again in https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Debug%20%28Intel%20HD%20530%29/builds/828, where this CL was reverted in https://codereview.chromium.org/2739203004.

Searching for protobuf related CLs in https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Debug%20%28Intel%20HD%20530%29/builds/743, I wonder if https://codereview.chromium.org/2724903004/ "Add ELEMENT_TYPE_WEBSOCKET to subresource filter" maybe related.

Not that Linux Debug (Intel HD 530) OS is "linux yakkety", which https://codereview.chromium.org/2746493002 warns about.

"Note", not "Not"

This sounds like a dupe of  bug 700120 

Indeed https://codereview.chromium.org/2746493002 was meant to fix this issue, but it caused some test failures which we're debugging now.

Note that this was ultimately caused by this CL https://codereview.chromium.org/2670623002/

Ah, indeed https://codereview.chromium.org/2670623002 "Linux UI: Switch to the Gtk3 theme" is in https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Debug%20%28Intel%20HD%20530%29/builds/743 blamelist.