Hello, firefox fails to build due to sandbox violation when using virtualx used for pgo: ^[[31;01m*^[[0m --------------------------- ACCESS VIOLATION SUMMARY --------------------------- ^[[31;01m*^[[0m LOG FILE: "/var/log/sandbox/sandbox-23114.log" ^[[31;01m*^[[0m VERSION 1.0 FORMAT: F - Function called FORMAT: S - Access Status FORMAT: P - Path as passed to function FORMAT: A - Absolute Path (not canonical) FORMAT: R - Canonical Path FORMAT: C - Command Line F: open_wr S: deny P: /dev/nvidiactl A: /dev/nvidiactl R: /dev/nvidiactl C: /var/tmp/portage/www-client/firefox-20.0.1/work/mozilla-release/obj-x86_64-unknown-linux-gnu/dist/firefox/firefox-bin -no-remote -profile /var/tmp/portage/www-client/firefox-20.0.1/work/mozilla-release/obj-x86_64-unknown-linux-gnu/_profile/pgo/pgoprofile/ http://localhost:8888/index.html ^[[31;01m*^[[0m -------------------------------------------------------------------------------- In fact this looks like virtualx.eclass bug for me: access to video acceleration device should be allowed if an appropriate GL interface is used. As a workaround addpredict /dev/nvidiactl should fix this.
Created attachment 349298 [details] emerge --info
Created attachment 349300 [details] environment
Created attachment 349302 [details] build.log.xz
Created attachment 349308 [details, diff] firefox-20.0.1.ebuild.patch This patch enables addpredict for all drivers, works fine for me.
pgo flag is masked we do not have man power to handle all bugs that are related to pgo build failures.
I understand this, but you have a tested patch here! PGO itself may be buggy, but a lot of stuff in portage in buggy. On all three mine boxes (with different hardware and setups) it works quite fine and really helps the browser on not-so-powerful hardware.
I will port the patch to 21.0 in next few days and land it.
Thanks.
Fix has been added to fx-22.0_beta6 which is in the overlay, it will make its way to tree with final release of 22.0 thanks for reporting.