1257309 - crash in js::GCMarker::processMarkStackTop

Status: RESOLVED DUPLICATE of bug 719114

(Core :: JavaScript: GC, defect)

Description

[(No longer employed by Mozilla) Aaron Klotz]

This bug was filed from the Socorro interface and is 
report bp-672cb09b-c938-4bae-bfb4-24b142160311.
=============================================================
A user on reddit is having a lot of stability issues.
Of the reports posted here:
https://www.reddit.com/r/firefox/comments/4ahp34/frequent_crashes/d11o544

Two of them happen in this function.

Comment 1

[Andrew McCreight [:mccr8]]

There are decent odds this is their hardware failing. Are they overclocking? They could also run some kind of memory checker.

Comment 2

[Terrence Cole [:terrence]]

Something like 8%[1] of DRAM sticks have bit flip errors out of the box, without overclocking. We've had at least 3 people through with these symptoms and *all* of them have been a failing stick of RAM. This user needs to run memtest86[2].

1- http://www.zdnet.com/article/dram-error-rates-nightmare-on-dimm-street/
2- http://www.memtest86.com/

Comment 3

[BugBot [:suhaib / :marco/ :calixte]]

Crash volume for signature 'js::GCMarker::processMarkStackTop':
 - nightly (version 50): 147 crashes from 2016-06-06.
 - aurora  (version 49): 548 crashes from 2016-06-07.
 - release (version 47): 9205 crashes from 2016-05-31.
 - esr     (version 45): 375 crashes from 2016-04-07.

Crash volume on the last weeks:
             Week N-1   Week N-2   Week N-3   Week N-4   Week N-5   Week N-6   Week N-7
 - nightly         20         28         19         22         17         20         11
 - aurora         112         94         58         85         89         60         14
 - beta          1541        411        475       2622        698        281         90
 - release       1450       1410       1282       1409       1471       1313        481
 - esr             42         27         40         42         56         37         33

Affected platforms: Windows, Mac OS X, Linux

Comment 4

[Liz Henry (:lizzard)]

This is the #4 topcrash in 49 beta 5 and 6. And just on a quick glance it looks like crashes with this signature are around 2K/week on release. 
Terrence, can you take another look at the recent beta crashes?  Does it still seem like hardware failure? Thanks.

Comment 5

[BugBot [:suhaib / :marco/ :calixte]]

Crash volume for signature 'js::GCMarker::processMarkStackTop':
 - nightly (version 51): 85 crashes from 2016-08-01.
 - aurora  (version 50): 207 crashes from 2016-08-01.
 - beta    (version 49): 6341 crashes from 2016-08-02.
 - release (version 48): 6930 crashes from 2016-07-25.
 - esr     (version 45): 539 crashes from 2016-05-02.

Crash volume on the last weeks (Week N is from 08-22 to 08-28):
            W. N-1  W. N-2  W. N-3
 - nightly      25      26      13
 - aurora       65      68      38
 - beta       1710    2229     878
 - release    2239    1905    1106
 - esr          44      38      48

Affected platforms: Windows, Mac OS X, Linux

Crash rank on the last 7 days:
           Browser   Content     Plugin
 - nightly #40       #35
 - aurora  #30       #18
 - beta    #4        #4
 - release #5        #5
 - esr     #177

Comment 6

[Terrence Cole [:terrence]]

(In reply to Liz Henry (:lizzard) (needinfo? me) from comment #4)
> This is the #4 topcrash in 49 beta 5 and 6. And just on a quick glance it
> looks like crashes with this signature are around 2K/week on release. 
> Terrence, can you take another look at the recent beta crashes?  Does it
> still seem like hardware failure? Thanks.

Regardless of the cause, processMarkStack top failures are not directly actionable as the heap corruption is occurring elsewhere. Instead, the GC team is taking a broad-based approach addressing code quality and defense in depth. These bugs are generally tracked as dependencies of bug 1008341.

Comment 7

[Liz Henry (:lizzard)]

Thank you!

Comment 8

[BugBot [:suhaib / :marco/ :calixte]]

Crash volume for signature 'js::GCMarker::processMarkStackTop':
 - nightly (version 52): 22 crashes from 2016-09-19.
 - aurora  (version 51): 11 crashes from 2016-09-19.
 - beta    (version 50): 1570 crashes from 2016-09-20.
 - release (version 49): 699 crashes from 2016-09-05.
 - esr     (version 45): 732 crashes from 2016-06-01.

Crash volume on the last weeks (Week N is from 10-03 to 10-09):
            W. N-1  W. N-2
 - nightly       9      13
 - aurora       11       0
 - beta       1132     438
 - release     571     128
 - esr          56      62

Affected platforms: Windows, Mac OS X, Linux

Crash rank on the last 7 days:
           Browser   Content     Plugin
 - nightly #291      #138
 - aurora  #172      #310
 - beta    #10       #9
 - release #109      #82
 - esr     #187

Comment 9

[Gerry Chang [:gchang]]

This has become #4 topcrash in 51.0b1. 

Hi Kanru,
Can you help take a look at this one?

Comment 10

[Gerry Chang [:gchang]]

Hi Milan,
Can you help find an owner for this crash?

Comment 11

[Andrew McCreight [:mccr8]]

This is an old signature that is not really actionable.