Looks like some precision issues related to compositing scrollbars. Doesn't seem very serious though. I don't have a Windows touch machine to test on at the moment, does this happen on iframes/subscrollers as well?

Taking this to try and reproduce on Windows + touch.

It's a bit concerning that we're showing what looks like uninitialized content.

On Linux using the mobile device emulation mode I notice slightly different issue which is that 1 pixel of the PDF content is visible outside, on the right side of the scrollbar.  This is also briefly noticeable in the "Actual result.mp4" video above at time 0:05.

I'm wasn't able to reproduce locally on a Razer Blade Stealth (2016) with touch screen.

Could you please attach a copy of the output form visiting the "chrome://gpu/" page?

Kindly find below copy of chrome://gpu.

Deleted: chrome gpu.png 889 KB

	chrome gpu.png 889 KB View Download

I was able to reproduce the issue on the Razer Blade Stealth with Crome 56, and latest Chrome Canary 58.0.3000.0.

It looks like the damage rect for the plugin is incorrect.  Only the thin vertical line on the right is being swapped to the screen.  Disabling partial swap with --disable-partial-swap seems to fix this.

It's tricky to reproduce.
 * Pinch-zoom in on the page slightly.
 * Use touch input to scroll the PDF frame via it's scroll bar.
 * If it doesn't break, try different zoom scales

Yeah, looks like the plugin is being double-scaled with pinch zoom. Maybe some zoom issues are putting its damage rects in the wrong place.

cc malaykeshav, as this looks similar to  issue 660744 

It seems like the layer with the plugin on it is being scrolled, but it isn't being treated as damaged. If I modify LayerTreeImpl::DidUpdateScrollOffset to do 

"if (true || node->scroll_offset != scroll_tree.current_scroll_offset(layer_id)) {" then it seems to work,  though I don't know enough about the code to understand what's going wrong otherwise.

Dropping this from my queue.

+pdr, chrishtr who are working on scrolling and scrollbars. 

 Issue 722494  has been merged into this issue.

From #13's bug, if I run a debug build with 1.4 dsf, then I get a cyan strip running next to the scrollbar, which indicates "non painted region". It seems like we don't have recordings there. 

I'm not sure if that's a separate issue though. chrishtr@ could you triage?

#13 might have been related to  Issue 711112 

This needs re-triage to determine severity.

I'm not actively looking at this so unassigning myself.