input overrun on origen with sqlite commands

[Paul Larson]

https://validation.linaro.org/lava-server/scheduler/job/24065/log_file

Looking down near the bottom, it looks like the android test image booted successfully, but when it gets to running the sqlite commands I see:
root@linaro: getprop init.svc.bootanim
running
root@linaro: [ 64.205000] binder: release 2387:2390 transaction 5092 out, still active
[ 64.285000] binder: 2310:2839 transaction failed 29189, size 8-0
[ 64.285000] binder: send failed reply for transaction 5092, target dead
getprop init.svc.bootanim
stopped
root@linaro: sqlite3 /data/data/com.android.providers.settings/databases/settioviders.settings/databases/setti <ngs.db "delete from system where name='stay_on"delete from system where name='stay_on <_while_plugged_in'; insert into system _while_plugged_in'; insert into system <(name, value) values ('stay_on_while_pl(name, value) values ('stay_on_while_pl <ugged_in','3');"
root@linaro: sqlite3 /data/data/com.android.providers.settings/databases/settioviders.settings/databases/setti <ngs.db "delete from system where name='screen_"delete from system where name='screen_ <off_timeout'; insert into system (name,off_timeout'; insert into system (name, < value) values ('screen_off_timeout','- value) values ('screen_off_timeout','- <1');"
root@linaro: sqlite3 /data/data/com.android.p[ 80.345000] ttySAC2: 1 input overrun(s)
roviders.settings/da
SQLite version 3.7.4
Enter ".help" for instructions
Enter SQL statements terminated with a ";"

sqlite> <LAVA_DISPATCHER>2012-07-02 01:03:46 PM ERROR: boot_linaro_android_image failed: Timeout exceeded in read_nonblocking().
<lava_dispatcher.utils.logging_spawn object at 0x7fb53797f350>

So what looks like happened here is that the last sqlite3 command overran the buffer and it missed the remainder of the line, then timed out sitting at a sqlite prompt. Not sure if this is specific to the board, so I put it in dispatcher. I wonder... do we know for sure that the sqlite modifications are still needed?

[Andy Doan (doanac)]

Two comments:

1) it seems like overriding the pexpect.[write/send/sendline] functions to check the length of the buffer we are sending, and break it up with some small delays if needed?

2) paul - the sqllite calls are still needed as far as I understand.

[Yongqin Liu (liuyq0307)]

I have submitted a branch to use the integrated disablesuspend.sh script.
I think that can fix this problem.
Please review the related branch.