576094 - plugin subprocess not terminating after reftest run on win7

Status: RESOLVED INVALID

(Testing :: General, defect)

Description

[Armen [:armenzg]]

ctalbert: armenzg: this looks like a crash that we intentionally caused because we detected that one of the subprocesses for plugin handling did not go away after the test.
armenzg: and is this expected to work like this?
ctalbert: It's expected to crash like this when it gets into this state.  It's supposed to clean up the extra plugin processes as it goes.  So, I'd still file a bug on it - "plugin subprocess not terminating after reftest run on win7" or some such.

This is pretty important to fix so we can run the reftests-d2d tests which is a Q2 goal (it will slip which is fine).

You can also check bug 576072 to see why this crash does not turn the build ORANGE or RED.

Here is few examples of the crash:
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1277927600.1277929130.25657.gz
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1277927600.1277929133.25661.gz
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1277927600.1277929133.25661.gz&fulltext=1

Crash reason:  EXCEPTION_ACCESS_VIOLATION
Crash address: 0x0

Thread 0 (crashed)
 0  nptest.dll!IntentionalCrash [nptest.cpp:accdced3e5b8 : 103 + 0x6]
    eip = 0x7244163a   esp = 0x0033ef94   ebp = 0x0033efb0   ebx = 0x0053d6c0
    esi = 0x005501c0   edi = 0x0033eff8   eax = 0x00000000   ecx = 0x0033ef54
    edx = 0x00470174   efl = 0x00010202
    Found by: given as instruction pointer in context

Comment 1

[Armen [:armenzg]]

BTW should the tests continue running after a controlled crash?

Comment 2

[Jesse Ruderman]

Are you sure this isn't just bug 539823?

Comment 3

[Armen [:armenzg]]

Now that I understand better what controlled crashes are and I am not confused anymore.
I can tell that this bug is invalid since we get test summaries even when there are controlled crashes (I thought we didn't even though I can now see it on the log).

Therefore, this is invalid.

> REFTEST INFO | Result summary:
> REFTEST INFO | Successful: 4445 (4427 pass, 18 load only)
> REFTEST INFO | Unexpected: 0 (0 unexpected fail, 0 unexpected pass, 0 unexpected asserts, 0 unexpected fixed asserts, 0 failed load, 0 exception)
> REFTEST INFO | Known problems: 236 (181 known fail, 0 known asserts, 37 random, 18 skipped)
> REFTEST INFO | Total canvas count = 14
> REFTEST TEST-START | Shutdown