749637 - Tegra board allocation for Security Team (ARM Fuzzing)

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: CIDuty, task)

Description

[Christian Holler (:decoder)]

As previously discussed with ctalbert, part of our Q2 goals in security is to deploy an automated fuzzing framework to test Fennec on Android. For this purpose, we will ultimately want to use our own pool of panda boards. However, since these boards won't be ready until Q3, we decided that a smaller set of Tegra boards should be used as a temporary drop-in so we can start testing and iron out further bugs in our automation and deployment.

As discussed on IRC with jmaher and bear, 5 tegra boards should be allocated for this purpose such that we can access them from a server where we can run the framework to interact with the boards. Jmaher mentioned that the boards would be accessible from the tegravm server (10.250.7.80), where I do have an account already, so this wouldn't require any changes on that side.

Let me know if there's anything else required and thanks in advance for helping with setting this up :)

Comment 1

[Mike Taylor [:bear]]

I'm working thru inventory now to identify the boards.

Comment 2

[Mike Taylor [:bear]]

I have identified the following tegras:

tegra-202
tegra-203
tegra-204
tegra-205
tegra-275

they are now out of releng circulation and ready for your use

Comment 3

[Mike Taylor [:bear]]

info delivered along with notes on how to reboot and manage tegras

enjoy your fuzzing ;)

Comment 4

[Justin Wood (:Callek)]

Also as an FYI --

"How to remote reformat the sdcard"

* Make sure nothing else is actively using tegra at the time
* Connect to tegra via command port (20701)
* execute: |exec mount| (skim/take note of output, this is just a sanity check)
* execute: |exec newfs_msdos -F 32 /dev/block/vold/179:9| (this does the actual clearing)
* execute: |rebt| (this should seem to "hang" at prompt for quite a while)
* -- To dive out of the hung telnet prompt use the escape char (ctrl+]) then enter, then type |quit| and enter to return to your normal prompt.

This method of formatting is *not* a secure format.

Comment 5

[Christian Holler (:decoder)]

Reopening per our discussion on IRC. I've been monitoring the Tegras for two days now and it seems that tegra-202 and tegra-205 keep rebooting themselves shortly after testing has started. The rest seems to run more reliably.

Comment 6

[Mike Taylor [:bear]]

k, let's see how these two perform:

tegra-198
tegra-287

Comment 7

[[e-mail] (:bhearsum)]

This doesn't seem buildduty related, removing the whiteboard tag.

Comment 8

[Justin Wood (:Callek)]

(In reply to Mike Taylor [:bear] from comment #6)
> k, let's see how these two perform:

No complaints on these, so..

enjoy your fuzzing ;)

Comment 9

[Christian Holler (:decoder)]

These boards can be returned now, we are no longer using them for fuzzing. Thanks a lot :)

Comment 10

[Armen [:armenzg]]

Thanks decoder.
We will be using dep bugs to track returning each one of them.

12:20 decoder: Callek: armenzg_brb: I thought the tegras were already given back
12:20 decoder: if not, you can take them back now
12:20 Callek: decoder: ooo, heh not afaik
12:20 Callek: decoder: note that "Mozilla" will be supporting tegras until at least Q4 2014
12:21 Callek: decoder: since thats the only platform we have that supports Android 2.x atm (2.2 on tegras)
12:21 decoder: yea, they havent been very reliable during fuzzing, i dont think it makes much sense to keep them right now
12:21 decoder: (for us)
12:21 Callek: decoder: can you enumerate that explicitly in, lets go for, Bug 749637, then, just so we have a record we can point to
12:22 decoder: Callek: shall I reopen that bug and state that they can be taken back?
12:22 Callek: that works for me!

Comment 11

[John Hopkins (:jhopkins)]

coop: should tegra bugs that are still blocked here be blocked on bug 928122 instead?

Comment 12

[Chris Cooper [:coop] (he/him)]

(In reply to John Hopkins (:jhopkins) from comment #11)
> coop: should tegra bugs that are still blocked here be blocked on bug 928122
> instead?

Yes.