Closed Bug 1205987 Opened 9 years ago Closed 3 years ago

startup crash in PR_Read | GetStatusFileContents<T> with freecorder app

Categories

(Core :: General, defect)

40 Branch
x86
Windows NT
defect
Not set
critical

Tracking

()

RESOLVED WORKSFORME
mozilla44
Tracking Status
firefox41 --- wontfix
firefox42 --- affected
firefox43 --- affected
firefox44 --- affected

People

(Reporter: philipp, Unassigned)

References

Details

(Keywords: crash, Whiteboard: qa-not-actionable)

Crash Data

Attachments

(1 file)

This bug was filed from the Socorro interface and is report bp-5117bd90-9b8f-4d27-8547-5e3b52150917. ============================================================= Crashing Thread Frame Module Signature Source 0 nss3.dll PR_Read nsprpub/pr/src/io/priometh.c 1 xul.dll GetStatusFileContents<32> toolkit/xre/nsUpdateDriver.cpp 2 xul.dll GetUpdateStatus toolkit/xre/nsUpdateDriver.cpp 3 xul.dll ProcessUpdates(nsIFile*, nsIFile*, nsIFile*, int, char**, char const*, bool, bool, nsIFile*, void**) toolkit/xre/nsUpdateDriver.cpp 4 xul.dll XREMain::XRE_mainStartup(bool*) toolkit/xre/nsAppRunner.cpp 5 xul.dll XREMain::XRE_main(int, char** const, nsXREAppData const*) toolkit/xre/nsAppRunner.cpp 6 xul.dll XRE_main toolkit/xre/nsAppRunner.cpp 7 firefox.exe do_main browser/app/nsBrowserApp.cpp 8 firefox.exe NS_internal_main(int, char**) browser/app/nsBrowserApp.cpp this low level startup crash has been around for a while, but it's increasing during recent releases. correlations point to the FLVSrvLib.dll module, which seems part of "Freecorder" by "Applian Technologies". 100% (40/40) vs. 0% (73/84474) FLVSrvLib.dll
cc'ing the "usual suspects" on this bug report... although the overall crash numbers for this signature are quite low, this is triggering an exorbitant amount of questions at sumo recently which is a bit worrying: https://support.mozilla.org/en-US/questions/firefox?tagged=bug1205987&show=all
It's strange that this is only now spiking, when v40 has been out for a while and this freecorder library dates back to 2010.
maybe a two-way interference by third-party programs. other correlations that stick out are: 45% (5/11) vs. 1% (695/89857) product_info.dll (part of kaspersky) 0% (0/11) vs. 0% (1/89857) 0% (0/11) vs. 0% (28/89857) 15.0.0.463 0% (0/11) vs. 0% (16/89857) 15.0.1.415 0% (0/11) vs. 0% (1/89857) 15.0.2.337 0% (0/11) vs. 0% (79/89857) 15.0.2.361 0% (0/11) vs. 0% (3/89857) 16.0.0.207 0% (0/11) vs. 0% (3/89857) 16.0.0.360 0% (0/11) vs. 0% (6/89857) 16.0.0.424 0% (0/11) vs. 0% (2/89857) 16.0.0.540 45% (5/11) vs. 1% (554/89857) 16.0.0.614 0% (0/11) vs. 0% (2/89857) 16.0.0.667 45% (5/11) vs. 6% (5658/89857) IPSEng32.dll (part of norton) 0% (0/11) vs. 0% (2/89857) 0% (0/11) vs. 0% (5/89857) 14.0.0.431 0% (0/11) vs. 0% (21/89857) 14.1.0.4 0% (0/11) vs. 0% (19/89857) 14.1.1.2 0% (0/11) vs. 0% (34/89857) 14.1.2.8 0% (0/11) vs. 0% (9/89857) 14.2.0.84 0% (0/11) vs. 0% (35/89857) 14.2.1.9 0% (0/11) vs. 2% (1591/89857) 14.3.0.8 0% (0/11) vs. 0% (287/89857) 14.4.0.24 45% (5/11) vs. 4% (3655/89857) 15.0.0.480 the norton IPSEng32.dll 15.0.0.480 in particular seems to have been shipped with a signature update on 2015-09-14 which fits well into the picture/timeframe of the spike.
The Freecorder DLL FLVSrvLib.dll also is showing up in recent startup crashes with the signature CreateFileW on Windows Vista. It's hard to generalize but a couple use Norton; in non-startup crashes with that signature, there's often an unfamiliar looking LSP. https://crash-stats.mozilla.com/report/list?product=Firefox&signature=CreateFileW
at the moment & because many of the other top crashers have been fixed in the dot-release, this is now the #1 crash in the score board for firefox 41.0.1 (https://crash-analysis.mozilla.com/rkaiser/crash-report-tools/score/?version=41.0.1&limit=30) and still generating quite an unusually high amount of feedback on sumo and many submitted user comments. due of its nature as a persistent startup crash its volume might decrease again as affected users simply move away to other alternatives... needinfoing dmajor, if something could be done about it on our part?
Flags: needinfo?(dmajor)
adding the FLVSrvLib.dll 1.0.0.0 file to the windows dll blocklist
Flags: needinfo?(dmajor)
Attachment #8670913 - Flags: review?(dmajor)
Attachment #8670913 - Flags: review?(dmajor) → review?(benjamin)
Comment on attachment 8670913 [details] [diff] [review] bug1205987v1.patch I'm skeptical that we want this kind of injection at all, but yes let's block this version for now. Kairo have you reached out to applian about this, per our blocklisting guidelines? You should do that before this is released.
Flags: needinfo?(kairo)
Attachment #8670913 - Flags: review?(benjamin) → review+
(In reply to Benjamin Smedberg [:bsmedberg] from comment #7) > Kairo have you reached out to applian about > this, per our blocklisting guidelines? You should do that before this is > released. I'm not usually doing any outreach, AFAIK Release Managers are doing that usually. Or is Jorge doing that when it come to add-on blocks? I hope Ritu (41 release manager) or Sylvestre know.
Flags: needinfo?(sledru)
Flags: needinfo?(rkothari)
Flags: needinfo?(kairo)
Is this from an add-on? If so, what's the ID?
Based on the patch it has to do with flvsrvlib.dll by Applian Technologies. It's described as "FLV Service Library for Ask and Record Toolbar". I don't think it's an addon but I could be wrong.
Flags: needinfo?(rkothari)
afaik, we don't have a contact to reach Applian tech :/
Flags: needinfo?(sledru)
Philipp, is there a reason this should not land in Nightly now?
Flags: needinfo?(madperson)
hi ritu - since this is my first patch & dmajor helped me get going there, i need the help of somebody else to get that checked in... in addition since blocking this .dll is rather a speculative fix and the crash volume for this is quite low on pre-release channels, it would probably need to go into beta builds before we can judge by crash stats that this has the desired effect.
Flags: needinfo?(madperson)
Thanks Philipp. Given that BSmedsberg has r+ this, let's get it checked into Nighlty and see whether it is reproducible or not. You can make Aurora/Beta uplift requests after that.
Keywords: checkin-needed
Crash Signature: [@ PR_Read | GetStatusFileContents<T>] → [@ PR_Read | GetStatusFileContents<T>] [@ nsZipItem::RealSize() ]
i'm able to reproduce the crash in a vm with freecorder 4 & norton trialware. the patch from comment #6 couldn't fix the crash with the inbound try build. this is a crash triggered: https://crash-stats.mozilla.com/report/index/265a05e8-35aa-434c-8ce4-cf50f2151013 - the FLVSrvLib.dll 1.0.0.0 is still showing up in the list of modules despite being on the blocklist.
(In reply to philipp from comment #17) > i'm able to reproduce the crash in a vm with freecorder 4 & norton trialware. > > the patch from comment #6 couldn't fix the crash with the inbound try build. > this is a crash triggered: > https://crash-stats.mozilla.com/report/index/265a05e8-35aa-434c-8ce4- > cf50f2151013 - the FLVSrvLib.dll 1.0.0.0 is still showing up in the list of > modules despite being on the blocklist. Thanks for the follow up Philipp. Given that the fix did not work as expected, I will skip including this in the next 41 dot release.
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla44
Crash Signature: [@ PR_Read | GetStatusFileContents<T>] [@ nsZipItem::RealSize() ] → [@ PR_Read | GetStatusFileContents<T>] [@ nsZipItem::RealSize() ] [@ nsZipItem::RealSize ]
Status: RESOLVED → REOPENED
Flags: needinfo?(madperson)
Resolution: FIXED → ---
Flags: needinfo?(madperson)
since this isn't solvable with a simple blocklist entry, i'm probably the wrong assignee to get this fixed as well :-)
Assignee: madperson → nobody
A French user reported that uninstalling "Freecorder" fixed the start-up crashes. https://forums.mozfr.org/viewtopic.php?f=5&t=126683
Crash Signature: [@ PR_Read | GetStatusFileContents<T>] [@ nsZipItem::RealSize() ] [@ nsZipItem::RealSize ] → [@ PR_Read | GetStatusFileContents<T>] [@ nsZipItem::RealSize() ] [@ nsZipItem::RealSize ] [@ PR_Read | ProcessUpdates ]
Whiteboard: qa-not-actionable

Closing because no crashes reported for 12 weeks.

Status: REOPENED → RESOLVED
Closed: 9 years ago3 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: