1052224 - Disable leak checking in content processes

Status: RESOLVED FIXED

(Testing :: Mochitest, defect)

Description

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

On desktop, the content process pretty much always crashes before it gets a chance to report leaks. If we fix the crash, then leaks appear. It seems most reasonable to disable content process leak checking, fix the crashes, and eventually fix the leaks and re-enable.

Kyle, what effect will that have on your b2g work?

Comment 1

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

None, yet.  Go ahead and do this.

Comment 2

[Andrew McCreight [:mccr8]]

Attachment: Disable leak checking in content processes.

try: https://tbpl.mozilla.org/?tree=Try&rev=0ff144b4e0d8

Comment 3

[Andrew McCreight [:mccr8]]

Comment on attachment 8477089 [details] [diff] [review]
Disable leak checking in content processes.

try run: https://tbpl.mozilla.org/?tree=Try&rev=0ff144b4e0d8

I'm pretty sure the M3 orange is from the other patch in that stack, as it is a shutdown assertion.

Comment 4

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

Comment on attachment 8477089 [details] [diff] [review]
Disable leak checking in content processes.

Review of attachment 8477089 [details] [diff] [review]:
-----------------------------------------------------------------

::: build/automationutils.py
@@ +353,5 @@
>      return
>  
>    # totalBytesLeaked was seen and is non-zero.
> +  # For now, ignore tab process leaks. See bug 1051230.
> +  if (totalBytesLeaked > leakThreshold) and (not processType or processType != "tab"):

so we will still get a warning printed.  Can we somehow indicate in the log file in the else condition below that we are ignoring this as it is a tab process?

Comment 5

[Andrew McCreight [:mccr8]]

Attachment: Disable leak checking in content processes.

Yeah, that's probably a good idea.

Comment 6

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

Comment on attachment 8477518 [details] [diff] [review]
Disable leak checking in content processes.

Review of attachment 8477518 [details] [diff] [review]:
-----------------------------------------------------------------

thanks, that looks much better!

Comment 7

[Andrew McCreight [:mccr8]]

Thanks for the fast reviews.

https://hg.mozilla.org/integration/mozilla-inbound/rev/f8ac66a28f2f

Comment 8

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/f8ac66a28f2f