First observed on 
https://build.chromium.org/p/chromium.memory.fyi/builders/Linux%20MSan%20Tests/builds/14464

however, there is no culprit in the regression range (381316 - 381331)

Could be that this has regressed in previous build but it passed due to flakyness. (buidl 14479 passed for example)

jkarlin@, you made changes to sync code (https://codereview.chromium.org/1763123002). Can you look into this?

Adding following folks who also made changes to sync related code.

jdonnelly@chromium.org (https://codereview.chromium.org/1785923010)
maxbogue@chromium.org (https://codereview.chromium.org/1789433002)