base::SysInfo::AmountOfPhysicalMemory broken in GPU process on Linux |
||
Issue descriptionThe call appears to be broken due to some aspect of the GPU sandbox (needs faccessat). From email thread: One possible explanation would maybe be the lazy caching via g_lazy_physical_memory https://cs.chromium.org/chromium/src/base/sys_info_linux.cc?type=cs&q=AmountOfPhysicalMemory&sq=package:chromium&l=46 ? It looks like that is warmed up before initializing the sandbox: https://cs.chromium.org/chromium/src/content/zygote/zygote_main_linux.cc?q=AmountOfPhysicalMemory&sq=package:chromium&l=338&dr=C.
,
Sep 7 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/e35050e679a4b5f8e9cef9de6458466c0a69fae4 commit e35050e679a4b5f8e9cef9de6458466c0a69fae4 Author: Eric Karl <ericrk@chromium.org> Date: Thu Sep 07 01:44:35 2017 Remove Linux GPU Proc AmountOfPhysicalMemory workaround base::SysInfo::AmountOfPhysicalMemory() doesn't work through the GPU sandbox on linux. Pre-populate the cached value before sandbox init. Bug: 743271 Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel Change-Id: I21505e321beb84af0b03b3ddca2b5c683e403a30 Reviewed-on: https://chromium-review.googlesource.com/653848 Reviewed-by: Antoine Labour <piman@chromium.org> Commit-Queue: Eric Karl <ericrk@chromium.org> Cr-Commit-Position: refs/heads/master@{#500171} [modify] https://crrev.com/e35050e679a4b5f8e9cef9de6458466c0a69fae4/content/gpu/gpu_main.cc [modify] https://crrev.com/e35050e679a4b5f8e9cef9de6458466c0a69fae4/gpu/command_buffer/service/service_discardable_manager.cc
,
Jan 17 2018
|
||
►
Sign in to add a comment |
||
Comment 1 by piman@chromium.org
, Aug 23 2017