523494 - Crash in [@ nsCSSSelector::Reset() ]

Status: RESOLVED INCOMPLETE

(Core :: CSS Parsing and Computation, defect)

Description

[Marcia Knous [:marcia]]

Seen while exploring Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.2b1) Gecko/20091019 Firefox/3.6b1 GTB6

Here is my crash report: http://crash-stats.mozilla.com/report/index/bp-fb377e17-d5c8-4800-9872-e7bb02091020

STR:
1. Was trying to repro Bug 503418, so I set a Master Password, turned on FIPS, then with a set of existing tabs I entered PB mode. I think the crash happened right when I entered the mode.

Comment 1

[glen beasley]

Marcia if you check your installation you will find the required NSS FIPS signature .chk files are missing. 
basically the installation requires that the following libraries 

libfreebl3.dylib  libnssdbm3.dylib  libsoftokn3.dylib
require the following signature files:
libfreebl3.chk  libnssdbm3.chk  libsoftokn3.chk

I just tested

Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.2b2pre) Gecko/20091020 Namoroka/3.6b2pre

Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.1.5pre) Gecko/20091020 Shiretoko/3.5.5pre

and both of these recent nightly builds are missing the .chk files. 

I am not sure why 

http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/3.6b1-candidates/build1/mac/

has the .chk files and the latest nightly builds don't.

Comment 2

[glen beasley]

I believe this bug depends on 522220

Comment 3

[Nick Thomas [:nthomas] (UTC+12)]

But the 3.6b1 release builds do have the chk files, so this must be something else.

Comment 4

[glen beasley]

(In reply to comment #3)
> But the 3.6b1 release builds do have the chk files, so this must be something
> else.

While I agree 3.6b1 release build has the .chk files the more recent nightly builds do not.

Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.2b2pre)
Gecko/20091020 Namoroka/3.6b2pre

Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.1.5pre)
Gecko/20091020 Shiretoko/3.5.5pre

There are two issues:

1) nightly builds require .chk files I believe bug 522220 should 
address this issue. 
2) PSM should not crash if it is unable to put NSS in FIPS mode. 
I plan to address this issue in bug 511320.

I think this bug should be marked as duplicate of bug 503418

Comment 5

[Nick Thomas [:nthomas] (UTC+12)]

Both of those issues appear to be unrelated to this bug, or at least to the stack reported in comment #0 (which contains no NSS symbols). find_objects_by_template doesn't appear in the stack either, is there some other reason this would be duped to 503418 ?

Comment 6

[Carsten Book [:Tomcat]]

Crash Stats currently report 215 crashes for this signature

http://crash-stats.mozilla.com/report/list?product=Firefox&version=Firefox%3A3.5.3&query_search=signature&query_type=exact&query=nsCSSSelector%3A%3AReset%28%29&date=&range_value=1&range_unit=weeks&do_query=1&signature=nsCSSSelector%3A%3AReset%28%29

Comment 7

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

The line numbers in the crash reports on Mac mostly seem to point to something wrong with mAttrList.

Comment 8

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Marcia, is that crash reproducible for you? If yes, does the same kind happen for 1.9.1 too?

Comment 9

[Marcia Knous [:marcia]]

Despite trying valiantly, have not been able to reproduce this bug yet on any of the other branches.

(In reply to comment #8)
> Marcia, is that crash reproducible for you? If yes, does the same kind happen
> for 1.9.1 too?

Comment 10

[Marcia Knous [:marcia]]

I spied one of these crashes on the Mac trunk from yesterday: https://crash-stats.mozilla.com/report/index/550171d6-e702-4c81-99af-30bbb2100504. Report indicates the user crashed while downloading a file from http://www.echofon.com/twitter/mac

Comment 11

[RC]

Running Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.3a5pre) Gecko/20100504 Minefield/3.7a5pre

The last five crashes have occurred only when downloading a file and without using any download managers of any kind. One of those crashes was with the OS X Echofon desktop client.

http://crash-stats.mozilla.com/report/index/bp-550171d6-e702-4c81-99af-30bbb2100504
http://crash-stats.mozilla.com/report/index/bp-bef648c8-2970-4b70-832f-007b72100503
http://crash-stats.mozilla.com/report/index/bp-fc9034e7-06f2-4740-8047-baad82100427

These are all separate file download crashes. Hoping there's some correlation here that I'm missing and might help.

Comment 12

[Marcia Knous [:marcia]]

RC: Are you able to reproduce the crash in safe mode? If not, then it this crash could be extension related.

(In reply to comment #11)
> Running Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.3a5pre)
> Gecko/20100504 Minefield/3.7a5pre
> 
> The last five crashes have occurred only when downloading a file and without
> using any download managers of any kind. One of those crashes was with the OS X
> Echofon desktop client.
> 
> http://crash-stats.mozilla.com/report/index/bp-550171d6-e702-4c81-99af-30bbb2100504
> http://crash-stats.mozilla.com/report/index/bp-bef648c8-2970-4b70-832f-007b72100503
> http://crash-stats.mozilla.com/report/index/bp-fc9034e7-06f2-4740-8047-baad82100427
> 
> These are all separate file download crashes. Hoping there's some correlation
> here that I'm missing and might help.

Comment 13

[RC]

Marcia: I'll give it a go. It happens so randomly it's difficult to repro.

Comment 14

[Marcia Knous [:marcia]]

Adjusting the summary so this gets picked up in crash stats. It is still happening with 20101028 trunk builds, at least on Mac.

Comment 15

[Robert Kaiser]

This signature rose high on Nightly since 2011-10-15, have we uncovered the underlying problem or put a different new one on top of this one?

Comment 16

[Robert Kaiser]

(In reply to Robert Kaiser (:[e-mail]) from comment #15)
> This signature rose high on Nightly since 2011-10-15, have we uncovered the
> underlying problem or put a different new one on top of this one?

Actually, this was all one build that had some larger problems and crashed all over the place, so disregard this comment. It's still a residual crash across versions, but nothing really high up.

Comment 17

[Mats Palmgren (inactive)]

The volume is still significant in current versions.
For the past 28 days we have:
Product         Version         Percentage      Number Of Crashes
Firefox         37.0.2          36.52 %         566
Firefox         38.0.1          21.81 %         338
Firefox         38.0b99         7.03 %          109
Firefox         38.0.5b3        5.29 %          82
Firefox         38.0.5b99       3.87 %          60
Firefox         38.0b9          3.81 %          59
Firefox         38.0.5b2        2.52 %          39
Firefox         38.0.5b1        1.29 %          20
Firefox         38.0            1.29 %          20
Firefox         40.0a2          0.90 %          14
Firefox         37.0.1          0.90 %          14
Firefox         38.0b6          0.90 %          14
Firefox         38.0b8          0.71 %          11
Firefox         34.0.5          0.58 %          9
Firefox         36.0.1          0.52 %          8
Firefox         39.0a2          0.52 %          8
Firefox         35.0.1          0.52 %          8
Firefox         41.0a1          0.45 %          7
Firefox         38.0b4          0.45 %          7
...

Comment 18

[Wayne Mery (:wsmwk)]

I wonder if any current crashes relate to the original reason for creating this bug report.

Based on spot check, the majority of non-beta stacks look to be fairly worthless.  

30% of crashes in the past month in [1] are 48.0 beta - [2]. Is there anything actionable?

[1] https://crash-stats.mozilla.com/signature/?date=%3E2016-07-08&signature=nsCSSSelector%3A%3AReset&_columns=date&_columns=product&_columns=version&_columns=build_id&_columns=platform&_columns=reason&_columns=address&_sort=-version&_sort=-date&page=2
[2] https://crash-stats.mozilla.com/signature/?date=%3E2016-07-08&release_channel=beta&version=48.0b99&version=48.0b10&version=48.0b9&version=48.0b8&version=48.0b7&version=48.0b5&version=48.0b6&version=48.0b4&version=48.0b3&version=48.0b11&signature=nsCSSSelector%3A%3AReset&_columns=date&_columns=product&_columns=version&_columns=build_id&_columns=platform&_columns=reason&_columns=address&_sort=-date&page=1

Comment 19

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

Crash volume for signature 'nsCSSSelector::Reset':
 - nightly (version 51): 2 crashes from 2016-08-01.
 - aurora  (version 50): 3 crashes from 2016-08-01.
 - beta    (version 49): 26 crashes from 2016-08-02.
 - release (version 48): 118 crashes from 2016-07-25.
 - esr     (version 45): 90 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       1       1       0
 - aurora        2       0       1
 - beta          7      12       1
 - release      38      42      18
 - esr           9       3       3

Affected platforms: Windows, Mac OS X, Linux

Crash rank on the last 7 days:
           Browser   Content     Plugin
 - nightly           #637
 - aurora
 - beta    #2792     #534
 - release #551
 - esr     #929

Comment 20

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

Crash volume for signature 'nsCSSSelector::Reset':
 - nightly (version 54): 2 crashes from 2017-01-23.
 - aurora  (version 53): 1 crash from 2017-01-23.
 - beta    (version 52): 22 crashes from 2017-01-23.
 - release (version 51): 71 crashes from 2017-01-16.
 - esr     (version 45): 336 crashes from 2016-08-10.

Crash volume on the last weeks (Week N is from 02-06 to 02-12):
            W. N-1  W. N-2  W. N-3  W. N-4  W. N-5  W. N-6  W. N-7
 - nightly       2       0
 - aurora        0       0
 - beta         13       2
 - release      45       7       0
 - esr          15      18      18       8      20       8      13

Affected platforms: Windows, Mac OS X, Linux

Crash rank on the last 7 days:
           Browser   Content   Plugin
 - nightly           #507
 - aurora            #864
 - beta    #1639     #574
 - release #1572     #614
 - esr     #675

Comment 21

[Julien Cristau [:jcristau]]

Too late for firefox 52, mass-wontfix.

Comment 22

[Mats Palmgren (inactive)]

It appears nsCSSSelector::Reset no longer exist.  I'm assuming it was
replaced by stylo.  All remaining crash in nsCSSSelector::Reset
are from non-stylo builds.