I confirm this behaviour with build 82 on mako. To illustrate it I made the following test: Start and stop dialer-app every 6s (reasonable delay to give time to the application to start and stop) The application is started with: start application APP_ID=dialer-app and stop with a kill.
I captured RSS measured by ps after each stop and made 2 runs with Mir and 1 with SF. The result is attached. I shows that RSS usage of init with Mir is much higher the SF and growth is exponential while it is linear with Mir. As a side note, the number of samples with Mir is limited to less than 50 because the phone stops responding at this moment while it continues to run fine with SF. This is reproduceable. ** Attachment added: "RSS-usage-Mir_vs_SF.png" https://bugs.launchpad.net/ubuntu/+source/upstart/+bug/1235649/+attachment/3862180/+files/RSS-usage-Mir_vs_SF.png ** Also affects: upstart (Ubuntu Saucy) Importance: Critical Status: Confirmed ** Tags added: qa-manual-testing qa-touch rls-s-incoming -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1235649 Title: session upstart leaks massive amounts of memory on Ubuntu Touch To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/upstart/+bug/1235649/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
