GitLab

The move of V8PerFrameMemoryReporterImpl from content to blink
introduced a bug where the PerFrameV8MemoryUsageData::associated_memory
mojo map started to use WTF::HashMap as the backing, which makes key=0
invalid because the hash traits for int64_t reserve that value for the
deleted sentinel.

This CL also fixes another bug of dereferencing a non-existent entry
in the |frames| HashMap.

Additionally this CL adds a test for V8PerFrameMemoryReporterImpl to
make sure that the fixes work.

Bug: 1085129
Change-Id: I04b0b7230fde191571f407b4931d6dba543bd2d8
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2339317
Commit-Queue: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Chris Hamilton <chrisha@chromium.org>
Reviewed-by: Kentaro Hara <haraken@chromium.org>
Reviewed-by: Daniel Cheng <dcheng@chromium.org>
Reviewed-by: Joe Mason <joenotcharles@chromium.org>
Cr-Commit-Position: refs/heads/master@{#795937}