Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(477)

Issues Search
    My Issues | Starred     Open | Closed | All

Issue 2526343002: WinHeap test (Closed)

Created:
4 years ago by kraynov
Modified:
4 years ago
Reviewers:
CC:
chromium-reviews, tracing+reviews_chromium.org, wfh+watch_chromium.org, vmpstr+watch_chromium.org
Target Ref:
refs/pending/heads/master
Project:
chromium
Visibility:
Public.

Description

WinHeap test BUG=665516

Patch Set 1 #

Unified diffs Side-by-side diffs Delta from patch set Stats (+108 lines, -20 lines) Patch
M base/trace_event/malloc_dump_provider.cc View 5 chunks +108 lines, -20 lines 0 comments Download

Messages

Total messages: 3 (1 generated)
kraynov
4 years ago (2016-11-24 17:46:57 UTC) #1
kraynov
4 years ago (2016-11-24 17:54:39 UTC) #2 
As part of investigation https://crbug.com/665516 I wrote the "test" which
checks how WinHeap behave in weird situations and how much memory get
non-reported.
Eric said (see bug) that only GPU process seems to use significant amount or
memory outside of main heap. And seems this heap is not LFH (low-fragmentation
heap), so it's indistinguishable from dengarous heaps. Moreover,
https://msdn.microsoft.com/en-us/library/windows/desktop/ee175820 says "Note
that an application should call heap functions only on its default process heap
or on private heaps that the application has created; calling heap functions on
a private heap owned by another component may cause undefined behavior."

This CL https://crrev.com/2519313005 seems to deal with that okay, but anyway I
can't tell it's safe enough to do and unsure this sacrifice worth 0.5-1 MiB of
reported memory.

Powered by Google App Engine
This is Rietveld 408576698