408518 - Topcrash [@ sqlite3_enable_shared_cache]

Status: RESOLVED WORKSFORME

(Firefox :: Bookmarks & History, defect, P2)

Description

[Mats Palmgren (inactive)]

It's currently #12 on the top crash list for 3.0b3pre builds:
http://crash-stats.mozilla.com/topcrasher/byversion/Firefox/3.0b3pre

Comment 1

[Mats Palmgren (inactive)]

There are a few different stacks leading up to the crash:

from nsFaviconService::SetFaviconData:
http://crash-stats.mozilla.com/report/index/ca6e74b0-ab5c-11dc-8d5b-001a4bd43ed6#frames

from nsNavBookmarks::RemoveItem:
http://crash-stats.mozilla.com/report/index/275590ca-ab5d-11dc-aec6-001a4bd43ef6#frames
also:
bp-2729d8cf-ab5e-11dc-b878-001a4bd46e84
bp-b7c3cf91-ab5c-11dc-99cb-001a4bd46e84
bp-a86152d1-ab5b-11dc-ada3-001a4bd43e5c

from nsNavHistoryExpire::ExpireHistoryParanoid:
http://crash-stats.mozilla.com/report/index/27400491-ab5f-11dc-8c2e-001a4bd43e5c#frames
also:
bp-45051d6e-ab5f-11dc-960a-001a4bd43e5c

Comment 2

[Mats Palmgren (inactive)]

There are also a bunch of crash reports with address 0x1:
http://crash-stats.mozilla.com/report/index/c8ea72ab-ab12-11dc-9fcb-001a4bd43e5c#frames
are these evidence of stack corruption?

Comment 3

[Shawn Wilsher :sdwilsh]

They seem to all end up calling sqlite_enable_shared_cache three times in a row, which doesn't make much sense...

Comment 4

[Mike Schroepfer]

Moving to blocking list to make sure we look into this.  If it turns out to not be a big issue we can remove later.

Comment 5

[Mike Beltzner [:beltzner, not reading bugmail]]

Dietrich, Mano: the entry points seem to be from Places. Any thoughts?

Comment 6

[Shawn Wilsher :sdwilsh]

Might be bug 420505 issue 2?

But then, looking at a recent report (0724518f-e7cd-11dc-9889-001a4bd43ed6), it looks more like a sqlite bug...

Comment 7

[Shawn Wilsher :sdwilsh]

I contacted the sqlite folks.  E-mail thread here:
http://www.mail-archive.com/sqlite-users%40sqlite.org/msg31932.html

Comment 8

[Shawn Wilsher :sdwilsh]

Note:  we saw a huge spike of these crashes with build 2008020500.  Did anything big land then?

Comment 9

[Marco Bonardo [:mak]]

(In reply to comment #8)
> Note:  we saw a huge spike of these crashes with build 2008020500.  Did
> anything big land then?

i don't see anything places backend related... jemalloc landed in that period if i'm not wrong

Comment 10

[Shawn Wilsher :sdwilsh]

upstream ticket - http://www.sqlite.org/cvstrac/tktview?tn=2970

Although I want to investigate the jemalloc issue.  However - wasn't that landed after beta three, where we've seen a lot of crashes still?

Comment 11

[Shawn Wilsher :sdwilsh]

I need to identify what landed with build 200820500 that may have caused the massive spike in crashes.  From there, I'm not sure where to go.

Comment 12

[Brendan Eich [:brendan]]

If it was jemalloc, it just moved the victims around and different (and more numerous) clumps of them were run over by whatever impurity is biting here. Has anyone done a detailed crash report analysis, a la dbaron's detailed forensic lessons? I'm not volunteering him, but the skills need to be spread.

/be 

Comment 13

[Shawn Wilsher :sdwilsh]

I won't have cycles for this for at least a month.  If it's still a concern, someone else should pick it up.

Comment 14

[Mike Schroepfer]

Ondrej can you pick this up?

Comment 15

[Ondrej Brablc]

Bug 420164 could be completely different problem, but last time we found with Michal a valgrind a problem (bug 419654), it was crashing on completely different places quite randomly.

I will check the available information (the links above are Error 404) but I'm afraid I will have to ask Michal for help again (I do not have Linux nor valgrind knowledge).

Comment 16

[Shawn Wilsher :sdwilsh]

This seems to have disappeared it would seem:
http://crash-stats.mozilla.com/report/list?range_unit=weeks&query_search=signature&query_type=contains&signature=sqlite3_enable_shared_cache&query=sqlite&range_value=1

Comment 17

[Mike Connor [:mconnor]]

Indeed.  Its no longer in the top100, not going to continue to block on this, but we should be sure to follow up, since it seems oddly spiky...

Comment 18

[Mike Shaver (:shaver emeritus)]

Should chase the valgrind error, for sure.

Comment 19

[Shawn Wilsher :sdwilsh]

valgrind error may have been fixed by bug 423964

Comment 20

[Shawn Wilsher :sdwilsh]

It looks like this has completely died off.

Comment 21

[Gervase Markham [:gerv]]

Bug 451915 - move Firefox/Places bugs to Firefox/Bookmarks and History. Remove all bugspam from this move by filtering for the string "places-to-b-and-h".

In Thunderbird 3.0b, you do that as follows:
Tools | Message Filters
Make sure the correct account is selected. Click "New"
Conditions: Body   contains   places-to-b-and-h
Change the action to "Delete Message".
Select "Manually Run" from the dropdown at the top.
Click OK.

Select the filter in the list, make sure "Inbox" is selected at the bottom, and click "Run Now". This should delete all the bugspam. You can then delete the filter.

Gerv