With stylo-chrome enabled, tsvg_static is regressed by >10% on Linux. This is weird, in sense that tsvg_static is a content page load test, which is not supposed to be regressed by stylo-chrome. From the data points of its subtests, it seems to be a consistent ~5ms regression regardless off the file size, so it may be related to chrome which displays it. And it only shows up in SVG test because they have especially small base value, so they are more sensitive to small changes on performance. Because the test themselves are very short on absolute time, it is pretty hard to reproduce with profiler attached. When profiler is attached, there doesn't seem to be any meaningful difference shown up at all. Given it's Linux-only, and regression is small on absolute number, it's probably not a big concern for shipping stylo-chrome.
You need to log in before you can comment on or make changes to this bug.