633270 - Run another Map/Reduce to better categorize MethodJIT crashes from beta 11

Status: RESOLVED FIXED

(Socorro :: General, task)

Description

[(not currently active) Ted Mielczarek]

+++ This bug was initially created as a clone of Bug #627111 +++

I'd like another run of the M/R job from bug 627111 now that beta 11 has been released. Munging your commandline, it should probably be:

hadoop jar categorizeCrash.jar -Dstart.date=20110209 -Dend.date=20110210
-Dbrowser.version=4.0b11 -Dsignature.match=EnterMethodJIT output.2/

I wiped out the data in khan:/home/minidumps, so you can put the results there.

Comment 1

[Anurag Phadke[:[e-mail]]]

Ted - I uploaded the data @ khan:/home/minidumps for the date range: 20110208 to 20110209, data for 20110210 isn't available yet.

Comment 2

[(not currently active) Ted Mielczarek]

Thanks! This is a fairly small sample, only 426 dumps. The breakdown is:

    199 CORRUPT_CODE
     91 UNKNOWN
     60 NO_JIT_MEMORY
     58 EIP_IN_BETWEEN
     10 BAD_EIP_INSTRUCTION
      8 INSTRUCTION_POINTER_IN_INACCESSIBLE_MEM

Of the NO_JIT_MEMORY dumps, 57 of them are Windows XP. I hacked up jit-crash-categorize to refine NO_JIT_MEMORY for Windows 7 dumps since we have the memory map info there. It seems pretty likely that most of those 57 remaining dumps are cases where the instruction pointer is off in unmapped memory, we just don't have the memory map info to back that up on WinXP.

Comment 3

[(not currently active) Ted Mielczarek]

Attachment: Output from running jit-crash-categorize on a sample of b11 minidumps

Here's the output from jit-crash-categorize.