As requested by shrike@ to determine whether it is responsible for performance regression
https://code.google.com/p/chromium/codesearch#chromium/src/ui/gfx/mac/io_surface.cc&q=iosurfacelock&sq=package:chromium&type=cs&l=161

"""
I kinda disagree :-)

At this point it appears there are two separate performance problems, but in this thread they have been conflated into the same issue. I say this because when we applied the patch that fixes the performance problem identified by the Yandex folks, that did not fix the performance problem that started this thread. It doesn't seem like working with Yandex further is going to get us closer to locating the cause of the original regression.

I would like us to see if the IOSurface change is the source of the problem. At this point I suspect that it is but we don't know for sure. I'd be happy with wrapping this in an experiment, to affect fewer users. I had also suggested reverting just on M51 beta (I assume that's doable) and looking at the data, and then reapplying the change if it's clearly not the problem. And while it's true that the IOSurface change addresses a security problem, that problem had existed for quite a long time. I am not proposing completely reverting it (yet), just working to see if that change is the source of the regression.
"""