Closed Bug 961005 Opened 12 years ago Closed 8 years ago

ff26.0,since update to 26.0 ff crash with kde wallet plugin

Categories

(Firefox :: Extension Compatibility, defect)

Product:
Firefox
Component:
Extension Compatibility
Version:
26 Branch
Platform:
x86_64
Linux
Type:
defect
Priority:
Not set
Severity:
critical

Tracking

()

Status:
RESOLVED WONTFIX

People

(Reporter: epistemepromeneur, Unassigned)

Details

(Keywords: crash)

Attachments

(1 file)

firefox-20140612-001701.kcrash.txt
64.49 KB, text/plain
Details
User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:26.0) Gecko/20100101 Firefox/26.0 (Beta/Release) Build ID: 2013120700 Steps to reproduce: since the update of ff to 26.0 then kde kwallet plugin 2.2 (or 2.1) crashes very often i navigate through play.google.com or plus.google.com or any site after a while or immediately when clicking on a link then firefox crashes gmolina the dev of the plugin detects it's a ion pb see the dialog here and how to contact gmolina http://kde-apps.org/content/show.php?content=116886&forumpage=4 some extract of the dialog It seems that the crashes experienced by the users are related to the new Javascript JIT Compiler Called ION Baseline. Disabling it seems to solve the problem. You can disable it by setting javascript.options.baselinejit.* in about:config to false I think is more a bug in the JIT compiler than a compatibility issue. I've tracked the segfault to that module of firefox using its debug info, in order to see what is causing it to segfault I would need far more knowledge of its internals than the knwoledge I have. Sorry about that. another person rwman says : actually you can just turn off ion by setting javascript.options.ion.content to "false". This fixes crashes for me. unrealted trick i discovered: setting javascript.options.ion.chrome to "true" seems to speedup all the browser interface responsiveness! (and it does not cause any crashes for me)
Severity: normal → major
after several days i confirm by setting javascript.options.baselinejit.* in about:config to false there is no more crash
Reproter, could you please provide a crash id from about:crashes Firefox page?
Flags: needinfo?(epistemepromeneur)
no i can't because there is not crash report with about:crashes when crash occurs ff crash report is not launched . also kde crash report says the report is not sufficient even if i install debug symbols . i assume without kwallet plugin debug symbols we can't generate any crash report . do you want the not sufficient kde crash report ?
Flags: needinfo?(epistemepromeneur)
please it would be more useful to contact the dev of kwallet plugin . he investigated the pb . i supplied to you a link to the site of the app. in this page you will find a way to contact him or you will find the site of the dev and a way to contact him . thanks
ff 27 pb still there
So the summary: 1) This is caused by ion but not baseline. 2) The addon overrides our crash reporter with the KDE one somehow, so we're not getting crash reports. :( 3) The addon has a binary blob that is uses via ctypes. 4) Builds with the addon still have a working slow script dialog, so whatever is happening in #2 is not just seeing out watchdog SEGVs.
Status: UNCONFIRMED → NEW
Component: Untriaged → JavaScript Engine: JIT
Ever confirmed: true
Product: Firefox → Core
Mook_as found http://www.guillermomolina.com.ar/es/archivos/category/1-firefox-kwallet-extension?download=23:firefox-kwallet-extension-sources-for-version-2-2 which appears to be the source for the addon. KDEWallet.cpp in the addon creates a KApplication // KWallet requries a functioning KApplication or it will segfault KAboutData aboutData("FirefoxKwalletPlugin", NULL, ki18n("Firefox KWallet Plugin"), "" ); KCmdLineArgs::init( &aboutData ); app = new KApplication(false); I don't know much about KDE applications, but: http://www.purinchu.net/kdelibs-apidocs/kdeui/html/namespaceKCrash.html Seems to say DrKonqi crash handling is enabled by default for KApplication, and that setCrashHandler(null) would disable this. So perhaps someone could add this to the addon and rebuild to get a more useful debug version.
(In reply to Boris Zbarsky [:bz] from comment #6) > 4) Builds with the addon still have a working slow script dialog, so > whatever is happening in #2 is not just seeing out watchdog SEGVs. But FF 26 added the signal-based interrupt mechanism for Ion code... Reporter, can you try Firefox 28 (or newer) and see if setting the JS_DISABLE_SLOW_SCRIPT_SIGNALS=1 environment variable helps? In a shell you can do: export JS_DISABLE_SLOW_SCRIPT_SIGNALS=1 Then start Firefox from that shell: path/to/firefox
So, asked someone to reproduce with: --- kwallet@guillermo.molina/library/KDEWallet.cpp 2013-12-23 05:04:14.000000000 -0500 +++ kwallet@guillermo.molina.new/library/KDEWallet.cpp 2014-06-11 17:33:35.597038075 -0400 @@ -40,6 +40,7 @@ #include <kcmdlineargs.h> #include <kwallet.h> #include <kdebug.h> +#include <kcrash.h> #include "KDEWallet.h" @@ -151,6 +152,7 @@ int32_t KDEWallet_Init() { kError() << "Could not create KApplication"; return false; } + KCrash::setCrashHandler(NULL); return true; } Against the 2.2 sources mentioned prior, in the hopes of getting a crash that Firefox would handle. Instead, he was unable to get a crash at all! The (64 bit only) XPI is here. http://m8y.org/tmp/firefox-kde-wallet-2.2.xpi He says he'll leave it active to see if he can get a crash to happen, but it used to happen very quickly.
For people who are seeing the crash (with the unpatched extension), do they crash reliably on: data:text/html,<script>while(true);</script> ?
stack trace from crash.
That crash is new as of fire fox 30. Bugzilla apparently doesn't do attachments the way I thought. Setting javascript.options.ion.parallel_compilation and javascript.options.ion to false stops the crash. Before just setting javascript.options.baselinejit.* to false seemed to stop it.
Jorge, do you mind terribly contacting the author of this extension? I'm preeeetty sure that disabling the KDE signal handler fixes this, and it'd be nice if users of it could get the fix. Also he could probably tell them to stop disabling the JIT ☺ https://addons.mozilla.org/en-US/firefox/addon/kde-wallet-password-integratio/ http://kde-apps.org/content/show.php?content=116886&forumpage=1 http://guillermomolina.com.ar/
Flags: needinfo?(jorge)
Sent a message to the dev through AMO.
Flags: needinfo?(jorge)
Ok I am changing it in version 2.3
Cool. Worth trying anyway. At the very least, it makes getting useful debug info back to Firefox devs difficult. I'm setting up KDE in a VM to see if I can replicate.
opensuse 13.1 x86_64 kde 4.13.2 ff 30.0 , kwallet plugin 2.3 since several hours no more crashes due to kwallet plugin but ... crashes due to new gui as with my windows os ! i will confirm by using ff next days
opensuse 13.1 x86_64 kde 4.13.2 ff 30.0 , kwallet plugin 2.3 perhaps the numerous crashes are due to kwallett plugin . most of the time there is no crash report but yesterday i sent 4 crash reports there is no dr konqui report https://crash-stats.mozilla.com/report/index/bp-5decd156-df4c-4fce-9dee-1d3c52140620 https://crash-stats.mozilla.com/report/index/bp-b8e70207-7eef-468e-8ad4-f90982140620 https://crash-stats.mozilla.com/report/index/bp-f51cd8e9-a067-4e29-8c14-6e0382140620 https://crash-stats.mozilla.com/report/index/bp-433e4880-55e6-489e-b2fa-642622140620
i have many crashes link to webpage containing template of any kind , login , comment without loging in or adding a comment . also when clicking in mail on link like this kind https://bugzilla.mozilla.org/show_bug.cgi?id=961005
i can't workaround the pb by setting to false javascript.options.baselinejit.chrome.* i am going to go back to 2.3 beta because i must have a working ff
So, promeneur - the kwallet extension you are testing disables the KDE signal handling, and that's where the 4 crash-stats in comment #18 comes from? Shame, 'cause if so that means disabling the signal handling isn't a fix :-/ or not a complete one. At least there are usable crash reports now...
>> So, promeneur - the kwallet extension you are testing disables the KDE signal handling, and that's >> where the 4 crash-stats in comment #18 comes from? yes
since i updated to ff 30 3 days ago it is worst than ever !
i changed yesterday the crypting method of the wallet from blowfish to gpg method . result : ff still crashes
Confirm this issue reproduced on Firefox 30 on Kubuntu 14.04 AMD64 - problem is still here.
Well, I gotta say, isn't so much FF crashing as the kwallet extension ;)
Can someone seeing these crashes try the following simple testcase with the _modified_ extension: data:text/html,<script>while(true);</script> Does that crash? Does it show a "slow script" dialog?
Flags: needinfo?(epistemepromeneur)
>> Can someone seeing these crashes try the following simple testcase with the _modified_ extension: i don't understand what to experiment .
Flags: needinfo?(epistemepromeneur)
if i put the extension in url field then a window opens : warning unresponsive script A script on this page may be busy, or it may have stopped responding. You can stop the script now, or you can continue to see if the script will complete. Script: data:text/html,<script>while(true);</script>:1 continue stop script
if i click on "stop script" then the dialog window closes and a blank page is displayed . the tab stays responsive : i can delete the content of the url filed and delete the tab
OK. So that's not crashing, alright. That means the issue is really not the SIGSEGV handling in the extension....
bz, I'd like to note that in kwallet extension 2.2, entering that while loop causes an instant crash. So, I think signal handling is part of the problem. I'm going to try to get a little test matrix of kwallet extension and firefox versions. So far, I've been unable to get the extension to work in FF nightly, even though the exact same profile works just fine in stable.
> I'd like to note that in kwallet extension 2.2, entering that while loop causes an > instant crash. Aha! So the unmodified extension is definitely mishandling our internal SEGVs. That definitely needs to be fixed. Jorge, can we get the extension author to fix that? It might be worth spinning out whatever problems exist with the modified extension into a separate bug and use this one to track the obvious fix to the extension.
Component: JavaScript Engine: JIT → Extension Compatibility
Flags: needinfo?(jorge)
Product: Core → Firefox
i assume another hypothesis : some crashes are not due to kwallet extension . i was reading 2 webpages : www.lemonde and : and http://www.lemonde.fr/sante/article/2014/06/24/vincent-lambert-le-conseil-d-etat-autorise-l-arret-des-soins_4444362_1651302.html then a crash occurs see see what was logged x@linux-b4lz:~> firefox (process:25733): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed WARNING: QApplication was not created in the main() thread. SSL Observatory: Cert submission failure 0: SSL Observatory: Planning to retry submission... ... SSL Observatory: Planning to retry submission... SSL Observatory: Cert submission failure 0: SSL Observatory: Planning to retry submission... SSL Observatory: Cert submission failure 0: SSL Observatory: Planning to retry submission... [25864] ###!!! ABORT: Aborting on channel error.: file /home/abuild/rpmbuild/BUILD/mozilla/ipc/glue/MessageChannel.cpp, line 1522 [25864] ###!!! ABORT: Aborting on channel error.: file /home/abuild/rpmbuild/BUILD/mozilla/ipc/glue/MessageChannel.cpp, line 1522 Erreur de segmentation x@linux-b4lz:~>
for the 2 pages there is no login ot comment template then kwallet extension is not called . isn't it ?
well, promeneur, can you reproduce crashes in a clean profile without kwallet, vs a clean profile with? that might help w/ the hypothesis.
Hi I am the author of the extension, can you give me a hint on how to handle the internal SEGVs?
By ignoring them. Basically, do not register a SEGV handler at all, because if you do it will interfere with those and with the crash reporter.
Ok, that's the same as nemo's patch?: KCrash::setCrashHandler(NULL); If that is the case, I've done it in version 2.3 of the extension.
I don't know if it is related, but my firefox crahsed with this message: Failed to open VDPAU backend libvdpau_r600.so: no se puede abrir el archivo del objeto compartido: No existe el archivo o el directorio Violación de segmento (`core' generado) Kubuntu 14.04 x64, Firefox 30
> KCrash::setCrashHandler(NULL); At a minimum, yes.
Flags: needinfo?(jorge)
Oh, Guille, if you want to share the crashes with the devs here, you can find them in about:crashes Also. I didn't have any luck using kwallet extension 2.2 against a firefox nightly build. Does that work for you, or do you know about this? Is there a workaround? I wanted to see if I could get better crash info using a debug build, and figured I might as well use the latest nightlies. But I can fall back a bit if they don't work with the extension.
Hi there, about:crashes says that there is no crashes sent. 2.2 extension does not have the KCrash::setCrashHandler(NULL); set, you should use 2.3 I've been using 2.3 with ff 30 with ion set for a day or so, with no crashes except for the libvdpau_r600 crash, but I am not sure if it is the same kind of crashes we had before. I'll find out if I can get that nightly build to test.
I don't know why, but in nightly build, password manager (neither native, nor kwallet) seems to be disabled. There are no messages in console
(In reply to Guille from comment #44) > I've been using 2.3 with ff 30 with ion set for a day or so, with no crashes i use FF 30 with 2.3 since several days (2014-06-20) . i have many crashes each day . i use the sync feature . yesterday i unactivated kwallet extension then no crash during all the day . KCrash::setCrashHandler(NULL); is not sufficient
(In reply to promeneur from comment #46) > (In reply to Guille from comment #44) > > > I've been using 2.3 with ff 30 with ion set for a day or so, with no crashes > > i use FF 30 with 2.3 since several days (2014-06-20) . > > i have many crashes each day . > > i use the sync feature . > > yesterday i unactivated kwallet extension then no crash during all the day . > > KCrash::setCrashHandler(NULL); is not sufficient I am not saying that the problem is gone, I am pointing out that I can not upload a crash report because I can not reproduce the problem.
WRT comment #45, indeed, password management seems to do nothing at all in nightly, but that's only if kwallet extension is installed. On removing it, password manager worked normally.
hello i made a very important experiment . i was very bored with many many crashes then i decided : - to inactivate kwallet extension during 2 days - to set to "true" javascript.options.baselinejit.* during this 2 days i discovered that in ff password manger there were recorded password then i deleted all recorded password and i never recorded any password during these 2 days no crash at all . then i decided to activate again kwallet extension then after one days i can say ff crashes very rarely , you can again use it with kwallet extension . i assume this hypothesis : when using kwallet extension if you have passwords in ff password manager then ff crashes
Can anyone who experience the crashes upload a crash report?
since i have done what i describe in comment #49 i have crashes but no crash report i can just report this about the last crash : x@linux-b4lz:~> firefox (process:19482): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed WARNING: QApplication was not created in the main() thread. Erreur de segmentation x@linux-b4lz:~> >>(process:19482): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed is displayed when launching firefox >>WARNING: QApplication was not created in the main() thread. >>Erreur de segmentation is displayed when crashing
i made an error . >>WARNING: QApplication was not created in the main() thread. is displayed during using ff >>Erreur de segmentation is displayed when ff crashes
promeneur, I think what Guille is asking for in comment #50 is one of the links from about:crashes - I know you posted a few above, but perhaps when describing a specific crash, you could include the link on the page with the matching time of the crash?
as i said in comment #51 >>since i have done what i describe in comment #49 i have crashes but no crash report
> WARNING: QApplication was not created in the main() thread. That sounds like something kde wallet is doing that might be worth fixing... Threading errors are a fairly likely source of crashes, for sure.
(In reply to Boris Zbarsky [:bz] from comment #55) > > WARNING: QApplication was not created in the main() thread. > > That sounds like something kde wallet is doing that might be worth fixing... > Threading errors are a fairly likely source of crashes, for sure. It's a Qt's message. Any KDE App must create a KApplication object, KApplication extends QApplication, the message says that the object should be created inside the main thread. Aparently firefox executes the plugin in other than the main thread. That could be a problem if the plugin would execute Qt's event loop, which is not the case. And after all, it is a Warning. Does anyone know if the ION compiler can print a stack trace? I remember that when I was able to debug the crash, the thread that crashed was the one compiled by ion. And I wasn't able to see what was the code that crashed. I think that is the key to dig into the problem.
> Does anyone know if the ION compiler can print a stack trace? A JS stack? In a current tip build, calling DumpJSStack() in the debugger should do the trick.
(In reply to Boris Zbarsky [:bz] from comment #57) > > Does anyone know if the ION compiler can print a stack trace? > > A JS stack? > > In a current tip build, calling DumpJSStack() in the debugger should do the > trick. I ment as part of the crash report.
Then no, sadly. Crash reports have a C++ stack, but not a JS stack, not least because JS stacks would expose too much user-private information.
don't know if it is a kwallet pb but today i get this : x@linux-b4lz:~> firefox (process:6597): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed 1404196019397 Sync.BrowserIDManager ERROR Failed to fetch a token for authentication: [Exception... "App. Quitting" nsresult: "0x80004004 (NS_ERROR_ABORT)" location: "JS frame :: resource://services-common/async.js :: checkAppReady/onQuitApplication/Async.checkAppReady :: line 123" data: no] x@linux-b4lz:~>
Ok, I've managed to crash it. But it was a bit tricky to obtain the core (Ubuntu has a core management app that does not work). Firefox does not send the crash either, It seems to be pretty serious. The crash happens at the main thread (I suppose), with this log: Jul 1 09:29:22 pc98001 kernel: [ 6536.599351] firefox[8370]: segfault at 7f980a016ab7 ip 00007f980a016ab7 sp 00007fff985db120 error 15 And the extract of the stack: Hilo 1 (Thread 0x7f984250f780 (LWP 8370)): #0 0x00007f980a016ab7 in ?? () No symbol table info available. #1 0x00007fff985db198 in ?? () No symbol table info available. #2 0x00007fff985db438 in ?? () No symbol table info available. #3 0x00007fff985db5f4 in ?? () No symbol table info available. #4 0x00007f983cff6d99 in nsXULElement::QueryInterface (this=0x7f97fed3cc80, aIID=..., aInstancePtr=0x7fff985db900) at /build/buildd/firefox-30.0+build1/content/xul/content/src/nsXULElement.cpp:348 foundInterface = 0x0 status = <optimized out> table = {{iid = 0x7f983df16470 <nsIDOMNode::COMTypeInfo<int>::kIID>, offset = 120}, { iid = 0x7f983df151c0 <nsIDOMElement::COMTypeInfo<int>::kIID>, offset = 120}, { iid = 0x7f983df42560 <nsIDOMXULElement::COMTypeInfo<int>::kIID>, offset = 120}, {iid = 0x0, offset = 0}} aInstancePtr = 0x7fff985db900 aIID = @0x7fff985db1b0: {m0 = 2556277408, m1 = 32767, m2 = 0, m3 = "˥\203=\230\177\000"} this = 0x7f97fed3cc80 #5 0xfffbff9801d5a4c0 in ?? () No symbol table info available. #6 0x00007fff985db288 in ?? () No symbol table info available. #7 0x0000000000000000 in ?? () No symbol table info available. If the rest of the log is needed, I would have to upload it manually, and it is rather large to put it here. Waiting for instructions
I am experiencing this crash with version 2.3 of the KWallet Plugin in Thunderbird (or, to be more precise, Icedove in Debian). The issue appeared after updating from version 24 to 31. The signal handler patch obviously did not help. However, disabling the JIT fixed the problem for me. Does Thunderbird have something like about:crashes (which however doesn't even seem to exist in my Firefox/Ixeweasel)? I managed to catch a backtrace with gdb: #0 js::jit::Assembler::TraceJumpRelocations (trc=0x7fff9d1e7360, code=0x7fff9e62b520, reader=...) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jit/x64/Assembler-x64.cpp:259 #1 0x00007ffff42d7da3 in js::jit::JitCode::trace (this=0x7fff9d1e7360, this@entry=0x7fff9e62b520, trc=0x7fff9e62b520, trc@entry=0x7fff9d1e7360) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jit/Ion.cpp:733 #2 0x00007ffff42cda63 in MarkChildren (code=0x7fff9e62b520, trc=0x7fff9d1e7360) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/gc/Marking.cpp:1290 #3 js::GCMarker::processMarkStackOther (this=0x7fff9d1e7360, tag=<optimized out>, addr=140735850657056) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/gc/Marking.cpp:1461 #4 0x00007ffff42ce308 in processMarkStackTop (budget=..., this=0x7fff9d1e7360) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/gc/Marking.cpp:1499 #5 js::GCMarker::drainMarkStack (this=this@entry=0x7fff9d1e7360, budget=...) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/gc/Marking.cpp:1608 #6 0x00007ffff441fa96 in DrainMarkStack (phase=js::gcstats::PHASE_MARK, sliceBudget=..., rt=0x7fff9d1e7000) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jsgc.cpp:3841 #7 IncrementalCollectSlice (rt=rt@entry=0x7fff9d1e7000, budget=budget@entry=0, reason=reason@entry=JS::gcreason::DOM_WORKER, gckind=gckind@entry=js::GC_SHRINK) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jsgc.cpp:4404 #8 0x00007ffff44214b5 in GCCycle (rt=rt@entry=0x7fff9d1e7000, incremental=incremental@entry=false, budget=budget@entry=0, gckind=gckind@entry=js::GC_SHRINK, reason=reason@entry=JS::gcreason::DOM_WORKER) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jsgc.cpp:4565 #9 0x00007ffff44218ee in Collect (rt=0x7fff9d1e7000, rt@entry=0x0, incremental=incremental@entry=false, budget=budget@entry=0, gckind=gckind@entry=js::GC_SHRINK, reason=JS::gcreason::DOM_WORKER, reason@entry=JS::gcreason::MAYBEGC) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jsgc.cpp:4698 #10 0x00007ffff4421a90 in Collect (reason=JS::gcreason::MAYBEGC, gckind=js::GC_SHRINK, budget=0, incremental=false, rt=rt@entry=0x0) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jsgc.cpp:4730 #11 js::GC (rt=rt@entry=0x0, gckind=gckind@entry=js::GC_SHRINK, reason=reason@entry=JS::gcreason::DOM_WORKER) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jsgc.cpp:4729 #12 0x00007ffff440ac4c in JS::ShrinkingGC (rt=rt@entry=0x0, reason=reason@entry=JS::gcreason::DOM_WORKER) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/js/src/jsfriendapi.cpp:204 #13 0x00007ffff377083a in mozilla::dom::workers::WorkerPrivate::GarbageCollectInternal (this=0x7fffb384fc00, aCx=0x7fffb384fc00, aShrinking=true, aCollectChildren=<optimized out>) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/dom/workers/WorkerPrivate.cpp:5593 #14 0x00007ffff377089c in (anonymous namespace)::GarbageCollectRunnable::WorkerRun (this=<optimized out>, aCx=<optimized out>, aWorkerPrivate=<optimized out>) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/dom/workers/WorkerPrivate.cpp:1660 #15 0x00007ffff3779934 in mozilla::dom::workers::WorkerRunnable::Run (this=0x7fffb04df670) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/dom/workers/WorkerRunnable.cpp:312 #16 0x00007ffff37760e8 in mozilla::dom::workers::WorkerPrivate::ProcessAllControlRunnablesLocked (this=this@entry=0x7fffb384fc00) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/dom/workers/WorkerPrivate.cpp:4480 #17 0x00007ffff3777a1a in mozilla::dom::workers::WorkerPrivate::DoRunLoop (this=0x7fffb384fc00, aCx=aCx@entry=0x7fff96f69ef0) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/dom/workers/WorkerPrivate.cpp:3977 #18 0x00007ffff3767cc7 in (anonymous namespace)::WorkerThreadPrimaryRunnable::Run (this=0x7fffb04df4c0) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/dom/workers/RuntimeService.cpp:2558 #19 0x00007ffff2dce135 in nsThread::ProcessNextEvent (this=0x7fffdc193d80, mayWait=<optimized out>, result=0x7fff895fde1f) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/xpcom/threads/nsThread.cpp:715 #20 0x00007ffff2d8cf1e in NS_ProcessNextEvent (thread=<optimized out>, mayWait=mayWait@entry=false) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/xpcom/glue/nsThreadUtils.cpp:263 #21 0x00007ffff2f69548 in mozilla::ipc::MessagePumpForNonMainThreads::Run (this=0x7fffaf364580, aDelegate=0x7fff970b4da0) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/ipc/glue/MessagePump.cpp:307 #22 0x00007ffff2f5b711 in RunHandler (this=0x7fff970b4da0) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/ipc/chromium/src/base/message_loop.cc:222 #23 MessageLoop::Run (this=this@entry=0x7fff970b4da0) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/ipc/chromium/src/base/message_loop.cc:196 #24 0x00007ffff2dceac5 in nsThread::ThreadFunc (arg=0x7fffdc193d80) at /build/icedove-HPEiZ0/icedove-31.0/mozilla/xpcom/threads/nsThread.cpp:316 #25 0x00007ffff61a5468 in ?? () from /usr/lib/x86_64-linux-gnu/libnspr4.so #26 0x00007ffff7bc70a4 in start_thread (arg=0x7fff895fe700) at pthread_create.c:309 #27 0x00007ffff6ed604d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:111 Please let me know if there is any other information I can provide.
opensuse 13.1 x86_64 , kde 4.14.0 firefox 32.0 kwallet 2.3 since the update 31.1 to 32.0 firefox does not fill the id and pwd fields of any site
also if you fill the id and pwd fields then firefox does not ask for you if you want to record the password if you browse the firefox list of recorded pwds then this list is empty
i found here https://www.mozilla.org/en-US/firefox/32.0.1/releasenotes/ this >> Changed Performance improvements to Password Manager and Add-on Manager
Severity: major → critical
Keywords: crash
Ok, Submited Version 2.4 to revisioning. Including this changes: Version 2.4: * Fixed LDAP password managaement, thanks to Aliaksandr Stelmachonak * Fixed FF32 crashes due to changes in Add-on Manager, thanks helper987654321 on kde-look.org * Implemented modify instead of remove/add Version 2.3: * Added timestamp support (Fix sync bug?) * Added entries for timestamp in kwallet * Added storage version numbering (for upgrades) * Fixed crash bug, thanks to nemo from Bugzilla@Mozilla It is available now at my web page
So. This bug was apparently about a couple of crashes, that it seems were fixed in 2 successive versions of his addon. Could the reporter or one of the regular users (promeneur ?) confirm that this can be marked FIXED as of version 2.4?
I think that if you enable ION compiler it still crashes
O_o You sure? Could you do that and try the snippet from comment #10 ? That's what the signal handling was supposed to avoid.
Version 2.4 is completely disfunctional here: I installed it in Thunderbird, and now Thunderbird fails to open my IMAP folders on startup without any error message. The KWallet prompt does not even appear. If I try to send a message, I get the error that it failed to obtain the password. Enabling or disabling the baselinejit has no effect. This is with Icedove (Thunderbird) 31.0 from Debian. I'm not back at version 2.2. I used to be on 2.3 but cannot find it anymore...
opensuse 13.1 x86_64 , kde 4.14.2 firefox 33.0 kwallet plugin 2.4 i set back javascript.options.baselinejit.* to "true" (default value) result : after 1 day using intensively FF , no more crash . for me pb is solved thanks to everybody
addendum : javascript.options.ion.* are set to their default value "true"
(In reply to Ralf Jung from comment #70) > Version 2.4 is completely disfunctional here: I installed it in Thunderbird, > and now Thunderbird fails to open my IMAP folders on startup without any > error message. The KWallet prompt does not even appear. If I try to send a > message, I get the error that it failed to obtain the password. > Enabling or disabling the baselinejit has no effect. > This is with Icedove (Thunderbird) 31.0 from Debian. > > I'm not back at version 2.2. I used to be on 2.3 but cannot find it > anymore... http://www.guillermomolina.com.ar/en/files/category/1-firefox-kwallet-extension?start=20
I had to disable automatic updates of Thunderbird addons because the KWallet extension 2.4 is now on addons.mozilla.org, and it's still completely dysfunctional - Thunderbird does not even attempt to log in to any servers anymore. The following appears several times in the error log: Timestamp: 10/11/14 14:41:28 Error: this._storage is null Source File: resource://gre/components/nsLoginManager.js Line: 383 Could somebody please point me to where to report bugs in this addon? Thanks.
@ralf jung you can report a bug also here >> http://kde-apps.org/content/show.php/Firefox+addon+for+kwallet?content=116886
I already added a comment with my problem there long ago. It seems to me that upstream is MIA.
Mass-closing old Extension Compatibility bugs that relate to legacy add-ons or NPAPI plug-ins. If you think this bug is still valid, please reopen or comment. Sorry for the bug spam, and happy Friday!
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: