Narrow Bisect info : 
https://chromium.googlesource.com/chromium/src/+log/d7cf0e8a1e7dfdc71476a2ec4169c4d4d48fc943..85a1bf73c32a1cda0d34fceb7a20f96b26bba035?pretty=fuller&n=10000

Suspecting: r512173 from Narrow bisect

@kojii : Could you please help to reassign if your change is not the cause for this change.

Note : Unable to bisect using per-revision bisect hence providing bisect using old script.

Tagging with blocker label, please undo if not the case.

Not that one, it affects only when LayoutNG runtime flag is enabled, and it's off by default unless the user turns it on by chrome:flags.

This looks like a paint issue, but I can't find anything in the range.

fsamuel@, any ideas? If not, it'd be great to try another bisect. I added the label.

With response to comment #3 :
Re-bisected this issue on different windows machine and getting the same range as mentioned in comment #1.

This is probably related to my change. I will investigate today.

I'm not sure this is actually a bug. It seems like before we didn't update ScreenInfo when you resize using dev tools and so nothing happened now we do, and so layout causes flicker. Reassigning to dsinclair@ for triage.

If this is just flickering when resized in devtools, I don't think it needs to be RBS. Removing label and sending to the PDF triage queue.

Is this something that's easy to fix?

Even in the "before" video, the non-content area is already painted black upon resizing. It's just now that sometimes the content area goes black too. As most users aren't doing this with DevTools open, I'm going to lower the priority.

Is the updating of ScreenInfo something that the pdf plugin controls? or is this done at a higher level?

i.e., is there anything we can do at the pdf plugin level to fix this?

The propagation of ScreenInfo happens in BrowserPlugin. I'm not sure why the PDF is flashing though. It may be doing a full repaint which maybe isn't necessary? I'm honestly not sure. Passing back to dsinclair@ to investigate. If you can verify that it's not a PDF bug then I'll investigate on the compositing side. At the moment, only PDF does this.

Setting PDF bugs assigned to me back to untriaged so they can get re-assigned as needed.

Since this is just in devtools, I don't think it's worth fixing.