All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=808621

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=4d704d83d0b0a53534700a73779224dddb76ea0096a91a89b6deef6ada9525c3


Bot(s) for this bug's original alert(s):

win-high-dpi

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12f9c01e840000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/12f9c01e840000

Enable the Oilpan metadata canary in production builds. by palmer@chromium.org
chromium @ 129fac28f2fc2e9003222a0be60997f299841a9a

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

I don't think this CL is the cause of a space regression. It doesn't cause the HeapObjectHeader object to grow in size on 64-bit platforms; the HOH is 64 bits both before and after the patch.

The HOH canary *would* cause a space regression on 32-bit... which is why we don't have it on 32-bit.

haraken: Do you agree?

The analysis, https://00e9e64bac84cbf6616258995e59092803badbfd58d6dc14d1-apidata.googleusercontent.com/download/storage/v1/b/results2-public/o/12f9c01e840000.html?r=chromium%40129fac2&s=%25%CE%94avg&g=name&c=0&n=0, doesn't make a ton of sense to me. But I admit I am a novice at this.

gpumemorybuffer:effective_size is up 2,300%, from 5.4 bytes. (No, I don't know how you can have .4 of a byte.)

v8:allocated_by_malloc:effective_size, up 20% from 83 KiB.

Both seem unrelated to my CL.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12a89a65440000

This metric is quite noisy but there still seems to be a trend of higher values after the alert. Re-running bisect, CC-ing metric owner.

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/12a89a65440000

Enable the Oilpan metadata canary in production builds. by palmer@chromium.org
https://chromium.googlesource.com/chromium/src/+/129fac28f2fc2e9003222a0be60997f299841a9a

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

The Pinpoint results (still) don't make much sense to me, and in any case I don't see how the CL can cause a memory regression (#4).