I am seeing the Nuwa process get killed on recent mozilla-central. To reproduce, use a memory heavy app like the poppit game in the marketplace. For example, I get this (Nuwa is pid 1742): bkelly@lenir:/srv/gaia-master$ adb shell b2g-info Expected to read 4 fields from fscanf(/proc/1742//stat), but got 2. | megabytes | NAME PID NICE USS PSS RSS VSIZE OOM_ADJ USER b2g 1702 0 44.6 48.8 57.8 179.6 0 root 1742 0 4.1 7.5 15.7 51.1 10 root Homescreen 1848 18 10.2 14.7 25.4 67.2 8 app_1848 Poppit!" 1910 1 65.4 70.7 82.3 478.4 2 app_1910 Then: bkelly@lenir:/srv/gaia-master$ adb shell b2g-info | megabytes | NAME PID NICE USS PSS RSS VSIZE OOM_ADJ USER b2g 1702 0 44.2 48.6 56.6 179.6 0 root Homescreen 1848 18 10.3 15.8 24.9 67.2 8 app_1848 Poppit!" 1910 1 71.5 77.8 87.8 485.3 2 app_1910 plugin-containe 2619 18 0.3 0.6 1.1 32.8 10 root And b2g-ps confirms its gone: bkelly@lenir:/srv/gaia-master$ adb shell b2g-ps APPLICATION USER PID PPID VSIZE RSS WCHAN PC NAME b2g root 1702 1 183948 55328 ffffffff 400ff604 S /system/b2g/b2g Homescreen app_1848 1848 1 68820 24480 ffffffff 4005b604 S /system/b2g/plugin-container Poppit!" app_1910 1910 1 494652 82720 ffffffff 4005a8b0 R /system/b2g/plugin-container plugin-containe root 2623 1702 41904 6708 00000000 b0003d56 R /system/b2g/plugin-container It appears Nuwa is getting an OOM adjust of 10. Is this normal and expected? I've been told Nuwa should not be killed under low memory conditions.

Bug 952693 is entitled "The Nuwa process should never be OOM killed" so I'm guessing this is not okay.

(In reply to Andrew McCreight [:mccr8] from comment #1) > Bug 952693 is entitled "The Nuwa process should never be OOM killed" so I'm > guessing this is not okay. That pretty implies this is a blocker then. This is a followup to a 1.3+.

We were accidentally adjusting the priority and OOM score of the Nuwa process so that could have been causing this issue. The patch I've attached in bug 968297 should prevent the Nuwa process' priority from being adjusted so hopefully this might be a duplicate of bug 968297.

Ok, so lets retest this once bug 968297 has landed in mozilla-central. I don't have time at the moment to roll a new build with the patch.

This appears to have been fixed by bug 968297.