Are existing configuration up to date? Consistent with mozharness? Should new ones be added?

I've tested all of these configurations locally and checked that I can run mochitest/reftest/xpcshell against each. More changes to come, probably, but this is a good start.

Comment on attachment 8945629 [details] [diff] [review] update emulator command lines for 'mach android-emulator' Review of attachment 8945629 [details] [diff] [review]: ----------------------------------------------------------------- rubber stamp- although if this is changed, I can only imagine what comes next :)

Pushed by gbrown@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/dc65d5d9ad9d Update 'mach android-emulator' configurations; r=jmaher

Pushed by gbrown@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/54a14f3a2284 Follow-up to fix lint error; r=me,a=test-only

https://hg.mozilla.org/mozilla-central/rev/dc65d5d9ad9d https://hg.mozilla.org/mozilla-central/rev/54a14f3a2284

The current 'x86-7.0' avd is up to date: It is based on rev 8 of the android-24/x86_64 system-image (most recent revision available today). system-images;android-24;default;x86_64 | 8 | Intel x86 Atom_64 System Image | system-images/android-24/default/x86_64/ [ro.bootimage.build.date]: [Wed Jul 12 19:46:52 UTC 2017] [ro.bootimage.build.fingerprint]: [Android/sdk_phone_x86_64/generic_x86_64:7.0/NYC/4174735:userdebug/test-keys]

Let's consider using swiftshader_indirect instead of swiftshader, for recent emulator versions.

Let's consider changing the default to something modern, and prefer x86 over arm: The Fennec 4.3 test environment's importance is waning, and x86 provides a much better experience.

Pushed by gbrown@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/18845b2e2a11 Remove Android 6.0/x86-6.0 options for mach android-emulator; r=me,a=test-only https://hg.mozilla.org/integration/mozilla-inbound/rev/1eee411e9a96 Change mach android-emulator default to x86-7.0; r=me,a=test-only

https://hg.mozilla.org/mozilla-central/rev/18845b2e2a11 https://hg.mozilla.org/mozilla-central/rev/1eee411e9a96

Re-triaging per https://bugzilla.mozilla.org/show_bug.cgi?id=1473195 Needinfo :susheel if you think this bug should be re-triaged.

(In reply to Geoff Brown [:gbrown] from comment #6) > The current 'x86-7.0' avd is up to date: It is based on rev 8 of the > android-24/x86_64 system-image (most recent revision available today). That is still the most recent revision today.

I tried adding an x86-9.0 variant, based on revision 4 of system-images;android-28;default;x86_64, for those living on the edge. I created an avd with: avdmanager --verbose create avd -d pixel_xl -n mozemu28c -k 'system-images;android-28;default;x86_64' --sdcard 600M and customized it similar to x86-7.0. That worked but was notably slower than x86-7.0. I then tried to package it for tooltool export and ran into a problem: I couldn't get the emulator (27.3.10) to find vendor.img from a local system-img. It was okay if vendor.img was in its default location in android-sdk-linux, but not when I moved all the .img files to the avd directory and updated config.ini appropriately. I note that vendor.img is not part of x86-7.0 or earlier avds. Perhaps this issue could be overcome with some ingenuity or special-case code for 9.0, but combined with the poor performance, I decided to put this avd aside for now. Let's revisit in a future bug when someone needs a newer test avd.

Pushed by gbrown@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/ed8ea654ccde Switch android emulator gpu from swiftshader to swiftshader_indirect; r=me,a=test-only