1011040 - b2gpopulate uses incorrect SD card location when ran against Flame

Status: RESOLVED FIXED

(Testing :: General, defect)

Description

[:Eli Perelman]

When running b2gpopulate against a Flame, the population process fails with an error that it cannot proceed because of a read-only filesystem. In digging into the issue, it appears that the Flame uses a different directory than previous devices using b2gpopulate. In order to get past the issue, I had to modify [1] and replace all references to `sdcard` to `/storage/sdcard0`. The population process worked correctly after that point.

[1] https://github.com/mozilla/b2gpopulate/blob/master/b2gpopulate/b2gpopulate.py

Comment 1

[Dave Hunt [:davehunt] [he/him] ⌚BST]

I don't have a Flame (yet) but I suspect this is related to bug 999506. This is fixed in the latest gaiatest, so should be fairly trivial to get b2gpopulate updated. I have noticed that the cleanup is failing on the Tarako, which may also be related. I'll have a look when I get a chance, but it may not be until next week due to travel commitments.

Comment 2

[Dave Hunt [:davehunt] [he/him] ⌚BST]

Jon: It's possible the reference workload scripts in gaia will also need to be updated for the Flame device, as I see those are also using /sdcard/.

Comment 3

[Dave Hunt [:davehunt] [he/him] ⌚BST]

Attachment: Link to Github pull-request: https://github.com/mozilla/b2gpopulate/pull/24

Eli: Could you try running with this patch against Flame? I still don't have one, but this should work thanks to bug 1000918.

Comment 4

[William Lachance (:wlach)]

Comment on attachment 8425520 [details] [review]
Link to Github pull-request: https://github.com/mozilla/b2gpopulate/pull/24

Sadly I'm not sure if this is going to work, as I think the code in devicemanager is kind of broken:

http://hg.mozilla.org/mozilla-central/file/7618a104e29d/testing/mozbase/mozdevice/mozdevice/devicemanagerADB.py#l470

It looks like we're setting the device root to <sdcard location>/tests when we really should be setting it to <sdcard location> (<sdcard location>/tests should be for actual test information)

I think we're going to need to file a new bug to properly fix mozdevice

Comment 5

[Dave Hunt [:davehunt] [he/him] ⌚BST]

(In reply to William Lachance (:wlach) from comment #4)
> Comment on attachment 8425520 [details] [review]
> Link to Github pull-request: https://github.com/mozilla/b2gpopulate/pull/24
> 
> Sadly I'm not sure if this is going to work, as I think the code in
> devicemanager is kind of broken:
> 
> http://hg.mozilla.org/mozilla-central/file/7618a104e29d/testing/mozbase/
> mozdevice/mozdevice/devicemanagerADB.py#l470
> 
> It looks like we're setting the device root to <sdcard location>/tests when
> we really should be setting it to <sdcard location> (<sdcard location>/tests
> should be for actual test information)
> 
> I think we're going to need to file a new bug to properly fix mozdevice

It works, but I would agree that the deviceRoot should probably not include /tests/. Perhaps we need a testRoot?

Comment 6

[William Lachance (:wlach)]

Comment on attachment 8425520 [details] [review]
Link to Github pull-request: https://github.com/mozilla/b2gpopulate/pull/24

According to davehunt this should actually work (the gaia apps don't care if the data is in subdirectory called /tests/).

We can give this an r+.

We already have a testRoot variable which is confusingly always the same as deviceRoot. It's probably not worth fixing this properly in devicemanager at this point, but we should make sure :bc's new ADB module (see bug 1012711) has a clearer concept of "writable storage area" vs. "temporary directory we can write test data to".

Comment 7

[:Eli Perelman]

Comment on attachment 8425520 [details] [review]
Link to Github pull-request: https://github.com/mozilla/b2gpopulate/pull/24

I can confirm this worked for me.

Comment 8

[Dave Hunt [:davehunt] [he/him] ⌚BST]

Landed in:
https://github.com/mozilla/b2gpopulate/commit/858b225ff00903c3f68bd05b4aefeecbf26ab74f

Released:
https://pypi.python.org/pypi/b2gpopulate/0.23

Comment 9

[Jon Hylands [:jhylands]]

I'll be updating reference workloads shortly with a patch similar to this.