503770 - Crash [@ memcpy | fillInCell]

Status: RESOLVED FIXED

(Toolkit :: Safe Browsing, defect)

Description

[Gary Kwong [:gkw] [:nth10sd] (NOT official MoCo now)]

Topcrash for the past week at #23 for "Results within 1 weeks of now, and the product is one of Firefox, and the version is one of Firefox:3.5." Occurs mostly on Win and using 3.5 final release.

http://crash-stats.mozilla.com/report/list?product=Firefox&version=Firefox%3A3.5&query_search=signature&query_type=exact&query=&date=&range_value=1&range_unit=days&do_query=1&signature=memcpy%20|%20fillInCell

It'll be cool if there was a list of URLs that crash at this location.

Comment 1

[Shawn Wilsher :sdwilsh]

This is crashing in SQLite - it may be useful if someone is seeing this to get a copy of their database.

Comment 2

[Wayne Mery (:wsmwk)]

(In reply to comment #1)
> This is crashing in SQLite - it may be useful if someone is seeing this to get
> a copy of their database.

this seems not to be in the cards any time soon.

this is now #20 crasher for 3.5.2. only 6 comments in 2,800 crashes, all windows crashes.

eg  bp-d0760136-8042-419b-ade4-85c612090819 In safe mode, just opened firefox to default google homepage, let it sit there. Then it crashed.

Comment 3

[D. Richard Hipp]

The change at:

   http://www.sqlite.org/src/vdiff/68957cf0c4ba

should prevent the crash and cause SQLite to return SQLITE_CORRUPT.  This change is in SQLite 3.6.11 and all subsequent releases of SQLite.  We need more data (such as the database file itself, and perhaps the SQL statement that was being run) in order to figure out if the corruption is real and why it is occurring.

Comment 4

[Shawn Wilsher :sdwilsh]

We are using 3.6.10 for 3.5.  We could take bug 508104 on branch to fix this.

Comment 5

[Jesse Ruderman]

What will happen with this fix?  Will the phishing database be blown away and replaced with a new one, or will the anti-phishing feature silently not work?

The SQL statement referenced by bp-d0760136-8042-419b-ade4-85c612090819 looks
pretty simple:
http://mxr.mozilla.org/mozilla-central/source/toolkit/components/url-classifier/src/nsUrlClassifierDBService.cpp#938

Comment 6

[Shawn Wilsher :sdwilsh]

The backend will throw an exception.  I don't know how the UI handles that, however.

Comment 7

[[:Cww]]

https://support.mozilla.com/tiki-view_forum_thread.php?comments_parentId=429880&forumId=1 Here's someone with this crash (maybe).  Feel free to comment in that thread, it'll email the user.

Comment 8

[Shawn Wilsher :sdwilsh]

(In reply to comment #5)
> What will happen with this fix?  Will the phishing database be blown away and
> replaced with a new one, or will the anti-phishing feature silently not work?
I think that regardless of the answer, not crashing is better than crashing.  If we need to fix safebrowsing code to handle it better, we should do that, but in a different bug.  If we don't blow away the database and crash, we'll be in no worse position if we don't blow away the database and silently fail.

Comment 9

[Shawn Wilsher :sdwilsh]

Fixed on branch via bug 508104.