726833 - SUTAgent can get "stuck" if prompt occurs on 1024 byte boundary

Status: RESOLVED FIXED

(Testing :: General, defect)

Description

[William Lachance (:wlach)]

Ok, this is pretty obscure, but I managed to reproduce it locally. Basically, if you send a command and the prompt occurs on a 1024 byte boundary, the agent will hang. I noticed this with a test that launched an application and checked for its pid inside the ps list. e.g.

send cmd: ps

recv'ing...
response: 10078	31831	org.mozilla.fennec_aurora
10076	27867	com.mozilla.SUTAgentAndroid
10057	1846	android.process.acore
10002	1270	com.android.launcher
10032	1325	com.google.process.gapps
10023	1263	com.android.inputmethod.latin
1001	1266	com.android.phone
1000	1099	system
10044	1800	com.google.android.apps.maps:LocationFriendService
10015	27630	com.google.android.googlequicksearchbox
10044	1677	com.google.android.apps.maps:FriendService
10044	27821	com.google.android.apps.maps:NetworkLocationService
10049	27788	com.lge.omadmclient:remote
10003	27703	com.android.cardock
10006	27644	com.android.music
10007	27638	com.android.protips
1000	1739	com.android.settings
10062	1751	com.google.android.apps.genie.geniewidget
10010	1806	com.tmobile.selfhelp
10044	2080	com.google.android.apps.maps
10066	1666	com.google.android.gm
10050	1787	com.google.android.apps.uploader
10008	1401	android.process.media
1000	2142	com.lge.providers.flex
10024	1267	com.android.kineto
10049	1268	com.lge.omadmclient
10045	1269	com.nuance.nmc.sihome
$
recv'ing...
response: >

The solution is simple: we should accumulate data that we get in a response and look for the prompt in that. Don't just look for the prompt in the latest data we rec'vd.

Comment 1

[William Lachance (:wlach)]

Attachment: Fix hang if prompt occurs on 1024-byte boundary

Comment 2

[William Lachance (:wlach)]

Doing a try run to make sure this doesn't accidentally break something.

Comment 3

[Mozilla RelEng Bot]

Try run for 094fe79a6c25 is complete.
Detailed breakdown of the results available here:
    https://tbpl.mozilla.org/?tree=Try&rev=094fe79a6c25
Results (out of 29 total builds):
    success: 24
    warnings: 5
Builds (or logs if builds failed) available at:
http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/[e-mail]-094fe79a6c25

Comment 4

[William Lachance (:wlach)]

(In reply to Mozilla RelEng Bot from comment #3)
> Try run for 094fe79a6c25 is complete.
> Detailed breakdown of the results available here:
>     https://tbpl.mozilla.org/?tree=Try&rev=094fe79a6c25
> Results (out of 29 total builds):
>     success: 24
>     warnings: 5
> Builds (or logs if builds failed) available at:
> http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/wlachance@mozilla.
> com-094fe79a6c25

The few failures here appear to be the usual noise (timeouts, etc.)

Comment 5

[Joel Maher ( :jmaher ) (UTC -8)]

inbound:
https://hg.mozilla.org/integration/mozilla-inbound/rev/e16d9a4fe053

Comment 6

[Marco Bonardo [:mak]]

https://hg.mozilla.org/mozilla-central/rev/e16d9a4fe053