Closed Bug 503840 Opened 16 years ago Closed 15 years ago

Crash after update to Weave 0.5pre1 only with History sync enabled [@ SetProperty_tn]

Categories

(Core :: JavaScript Engine, defect, P2)

1.9.1 Branch
x86_64
Windows 7
defect

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: derekharkin, Assigned: dvander)

Details

(Keywords: crash)

Crash Data

User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-GB; rv:1.9.1) Gecko/20090624 Firefox/3.5 (.NET CLR 3.5.30729) Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-GB; rv:1.9.1) Gecko/20090624 Firefox/3.5 (.NET CLR 3.5.30729) ID:20090624025744 After update to Weave version 0.5pre1 Firefox iscontinually crashing during sync only when History sync is enabled. With History sync is disabled no crash occours Crash IDs: aafadd2b-5802-4f00-a477-ecbc92090713 a44daf37-0609-47ca-8917-8cf1b2090713 827996f1-758a-424e-9065-4b11d2090713 Firefox 3.5 Windows 7 x64 Addons: Adblock Plus 1.0.2 ColorfulTabs 3.9.2 Java Console 6.0.14 Microsoft .NET Framework Assistant 1.1 Nightly Tester Tools 2.0.2 NoScript 1.9.5 Personas for Firefox 1.2.1 QuickProxy 2009.03.28 Tweak Network 1.3 Weave 0.5pre1 Versone log at time of crash 2009-07-13 11:12:48 Service.Main INFO Weave 0.5pre1 initializing <-- Restart after crash here. 2009-07-13 11:12:42 Store.HistStore TRACE Creating SQL statement: _visitStm 2009-07-13 11:12:42 Store.HistStore TRACE Creating SQL statement: _pidStm 2009-07-13 11:12:42 Store.HistStore TRACE Creating SQL statement: _urlStm 2009-07-13 11:12:42 Store.HistStore TRACE Creating SQL statement: _findPidByAnnoStm 2009-07-13 11:12:42 Store.HistStore TRACE Creating SQL statement: _annoAttrIdStm 2009-07-13 11:12:42 Engine.History DEBUG Preparing 100 outgoing records Reproducible: Always Steps to Reproduce: 1. Update Weave to 0.5pre1 2. Enable History Sync 3. Perform a Sync 4. Crash
Signature SetProperty_tn UUID aafadd2b-5802-4f00-a477-ecbc92090713 Time 2009-07-13 02:44:34.335968 Uptime 39 Last Crash 115 seconds before submission Product Firefox Version 3.5 Build ID 20090624025744 Branch 1.9.1 OS Windows NT OS Version 6.1.7201 CPU x86 CPU Info GenuineIntel family 6 model 15 stepping 11 Crash Reason EXCEPTION_ACCESS_VIOLATION Crash Address 0x2 User Comments Processor Notes Related Bugs OPEN * 499610 NEW TM: Crash during local mochitest-browser-chrome run [@ SetProperty_tn] Crashing Thread Frame Module Signature [Expand] Source 0 js3250.dll SetProperty_tn js/src/jstracer.cpp:8351 1 @0x4c94e9e 2 @0x19d6fb 3 js3250.dll js_MonitorLoopEdge js/src/jstracer.cpp:4862 4 js3250.dll js_Interpret js/src/jsinterp.cpp:3308
Assignee: nobody → general
Severity: normal → critical
Component: Sync → JavaScript Engine
Keywords: crash
Product: Weave → Core
QA Contact: sync → general
Summary: Crash after update to Weave 0.5pre1 only with History sync enabled → Crash after update to Weave 0.5pre1 only with History sync enabled [@ SetProperty_tn]
Version: unspecified → 1.9.1 Branch
I'm assuming you have jit.chrome set to true? Try toggling that off and turn history sync back on?
Yes I had jit.chrome set to true. I turned it off and reenabled history sync and no crash.
Severity: critical → normal
Flags: blocking1.9.2?
Priority: -- → P2
Aiing: Did you have Weave 0.4 installed with jit.chrome=true and that didn't crash?
Yes. I have had jit.chrome=true for some time and I was on 0.4 before upgrading to 0.5pre1.
Sorry, to clariry - Before upgrading I was not getting constant crashes like I was after upgrading to 0.5pre1.
I'm not sure if this is the same crash, but Fennec has jit.chrome default true, so the latest Weave is causing Fennec to crash. I don't have the stack from a maemo build (all it says is Segmentation Fault), but running Fennec on OS X: 0 TypeMap::matches(TypeMap&) const + 928 1 0 + 350105421 2 0 + 3221208456 3 js_MonitorLoopEdge(JSContext*, unsigned int&) + 832 4 jsInterpret + 44470 5 jsInvoke + 1723 6 js_fun_call + 391 7 js_Interpret + 37353 8 js_Invoke + 1723 9 js_fun_call + 391 Viewing just the weave verbose log, I see a number of these before crashing: 2009-07-15 15:18:12 Store.HistStore DEBUG visit 1247696186652913 Some crashes show 17 of those entries.. other times it's 12 entries. But the weave code that prints those out: let visit; while ((visit = record.visits.pop())) { if (curvisits.filter(function(i) i.date == visit.date).length) continue; this._log.debug(" visit " + visit.date); this._hsvc.addVisit(uri, visit.date, null, visit.type, (visit.type == 5 || visit.type == 6), 0); }
Status: UNCONFIRMED → NEW
Ever confirmed: true
Group: core-security
The crash from comment 7 is different from this bug I believe.. Here's a crash report from os x: http://crash-stats.mozilla.com/report/index/9cde1dfd-df99-430e-af72-7458c2090716?p=1 I triggered it while *uploading* history items. The previous comment was triggering crashes while downloading items on Fennec. Any way to know for sure what js functions were being called before the crash? When I turn on "Trace level" logging for weave, I see 3 "Outgoing:" messages before the crash, so it's probably this loop _uploadOutgoing: function SyncEngine__uploadOutgoing() { .. for (let id in this._tracker.changedIDs) { let out = this._createRecord(id); this._log.trace("Outgoing:\n" + out); // skip getting siblings of already processed and deleted records if (!out.deleted && !(out.id in meta)) this._store.createMetaRecords(out.id, meta); out.encrypt(ID.get("WeaveCryptoID")); up.pushData(JSON.parse(out.serialize())); // FIXME: inefficient Sync.sleep(0); } I'll try reducing this.. tips? debug firefox build?
Oh hey! It's already been fixed on 1.9.1 and m-c. Somewhere here: http://hg.mozilla.org/releases/mozilla-1.9.1/pushloghtml?fromchange=f223409207c0&tochange=b91c9ee69e6e I would guess it's something aborting the trace now.. ? Bug 499772 - TM: TraceRecorder::test_property_cache needs JSClass.getProperty checks when a property isn't found on an object. r=jorendorff, r=brendan Deep abort is not detected in JSOP_IN (500108, r=dvander).
Flags: blocking1.9.2? → blocking1.9.2+
Assigning to dvander.
Assignee: general → dvander
dvander, we need to make some progress here.
looks like this is WFM
Status: NEW → RESOLVED
Closed: 15 years ago
Resolution: --- → WORKSFORME
Crash Signature: [@ SetProperty_tn]
Group: core-security → core-security-release
Group: core-security-release
You need to log in before you can comment on or make changes to this bug.