Bug 89947 - Pressing repo refresh button with blank entry causes crash
Summary: Pressing repo refresh button with blank entry causes crash
Status: RESOLVED DUPLICATE of bug 87984
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: LibreOffice (show other bugs)
Version:
(earliest affected)
4.0.6.2 release
Hardware: Other All
: medium normal
Assignee: Not Assigned
URL:
Whiteboard:
Keywords: haveBacktrace
Depends on:
Blocks: LO-File-Dialog
  Show dependency treegraph
 
Reported: 2015-03-11 08:21 UTC by Yousuf Philips (jay) (retired)
Modified: 2016-05-27 09:03 UTC (History)
5 users (show)

See Also:
Crash report or crash signature:


Attachments
backtrace (13.37 KB, text/plain)
2015-03-11 08:21 UTC, Yousuf Philips (jay) (retired)
Details
win backtrace (11.04 KB, text/plain)
2015-07-04 16:31 UTC, Gordo
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Yousuf Philips (jay) (retired) 2015-03-11 08:21:39 UTC
Created attachment 114030 [details]
backtrace

Steps:
1) Open LibreOffice
2) Activate LO's custom dialogs if not already active (Tools > Options > General)
3) Click 'Open File' button in start center
4) Click the 'Servers...' button
5) Set type to 'CMIS'
6) Click the refresh button next to the repository field.

Regression that goes back to 4.0, as that is when CMIS was added to the type drop down.

Version: 4.5.0.0.alpha0+
Build ID: 4cf69df2f543b023f4ec3d4279c5ac11a2a819d9
TinderBox: Linux-rpm_deb-x86@45-TDF, Branch:master, Time: 2015-02-21_03:46:34

The commandline had the following error:

GLib:ERROR:/build/buildd/glib2.0-2.32.4/./glib/gmain.c:2542:g_main_dispatch: assertion failed: (current->dispatching_sources == &current_source_link)
Comment 1 A (Andy) 2015-03-11 21:17:03 UTC
Reproducible with LO 4.4.1.2, Win 8.1

After step 6 the LO application is no longer responding to any mouse clicks and can only be closed with the Windows Task Manager.
Comment 2 Julien Nabet 2015-03-15 21:55:08 UTC
Just for the record, on pc Debian x86-64 with master sources updated 2 days ago, I don't reproduce this.
Comment 3 Yousuf Philips (jay) (retired) 2015-03-16 02:40:16 UTC
Crashed on me on the latest master.

Version: 4.5.0.0.alpha0+
Build ID: de145fb467f2a168261915d63b47d3cd3b03d732
TinderBox: Linux-rpm_deb-x86@45-TDF, Branch:master, Time: 2015-03-15_09:51:29
Comment 4 Gordo 2015-07-04 16:31:28 UTC
Created attachment 117047 [details]
win backtrace

It has vclptr in it but this bug predates that.

4.4.4.3

Version: 5.1.0.0.alpha1+
Build ID: 89b5967658392d27fb3147e85abb2b5c1c34b101
TinderBox: Win-x86@39, Branch:master, Time: 2015-06-24_04:10:17
Comment 5 Michael Meeks 2015-07-09 12:42:13 UTC
Works for me -> master as of 5d19f90eb247275420a6774403cb77940ac6131b on Linux, dbgutil build - throws up a dialog saying:

"Nonexistent file."

Can you confirm platform / version for this ?
Comment 6 Gordo 2015-07-09 12:55:58 UTC
The bt is from
Windows Vista 64
Version: 5.1.0.0.alpha1+
Build ID: 89b5967658392d27fb3147e85abb2b5c1c34b101
TinderBox: Win-x86@39, Branch:master, Time: 2015-06-24_04:10:17

Still reproducible in
Windows Vista 64
Version: 5.1.0.0.alpha1+
Build ID: d3b6f3790953bdfeaeebcd3ba9ec370d94ca4ebf
TinderBox: Win-x86@39, Branch:master, Time: 2015-07-09_00:11:56
Comment 7 Michael Meeks 2015-07-09 13:16:45 UTC
Oooh - I get a "solar mutex not locked" error dialog on Windows - what fun; chasing.
Comment 8 Michael Meeks 2015-07-09 13:39:35 UTC
Michael - might be interesting to you ?

 	msvcr120d.dll!_wassert(const wchar_t * expr, const wchar_t * filename, unsigned int lineno) Line 369	C
 	vcllo.dll!ImplDbgTestSolarMutex() Line 817	C++
 	tllo.dll!DbgFunc(unsigned short nAction, void * pParam) Line 303	C++
 	svllo.dll!SfxBroadcaster::RemoveListener(SfxListener & rListener) Line 130	C++

void SfxBroadcaster::RemoveListener( SfxListener& rListener )
{
    DBG_TESTSOLARMUTEX();

        fails here.

 	svllo.dll!SfxListener::~SfxListener() Line 61	C++
 	sfxlo.dll!SfxRequest_Impl::~SfxRequest_Impl() Line 88	C++
 	sfxlo.dll!SfxRequest_Impl::`scalar deleting destructor'(unsigned int)	C++
 	sfxlo.dll!SfxRequest::~SfxRequest() Line 133	C++
 	sfxlo.dll!SfxRequest::`vector deleting destructor'(unsigned int)	C++
 	sfxlo.dll!SfxDispatcher::PostMsgHandler(SfxRequest * pReq) Line 1106	C++

Its rather unclear from the above frame to here who drops the solar mutex.

 	sfxlo.dll!SfxDispatcher::LinkStubPostMsgHandler(void * instance, void * data) Line 1074	C++
 	sfxlo.dll!Link<void *,long>::Call(void * data) Line 127	C++
 	sfxlo.dll!GenLink::Call(void * pCaller) Line 45	C++
 	sfxlo.dll!SfxHintPoster::Event(SfxHint * pPostedHint) Line 60	C++
 	sfxlo.dll!SfxHintPoster::DoEvent_Impl(SfxHint * pPostedHint) Line 53	C++
 	sfxlo.dll!SfxHintPoster::LinkStubDoEvent_Impl(void * instance, void * data) Line 50	C++
 	vcllo.dll!Link<void *,long>::Call(void * data) Line 127	C++
 	vcllo.dll!ImplHandleUserEvent(ImplSVEvent * pSVEvent) Line 2028	C++
 	vcllo.dll!ImplWindowFrameProc(vcl::Window * _pWindow, SalFrame * __formal, unsigned short nEvent, const void * pEvent) Line 2580	C++
 	vcllo.dll!SalFrame::CallCallback(unsigned short nEvent, const void * pEvent) Line 246	C++
 	vcllo.dll!ImplHandleUserEvent(HWND__ * hWnd, long lParam) Line 4128	C++

This guy acquires the SolarMutex.

>	vcllo.dll!SalFrameWndProc(HWND__ * hWnd, unsigned int nMsg, unsigned int wParam, long lParam, int & rDef) Line 5784	C++
 	vcllo.dll!SalFrameWndProcW(HWND__ * hWnd, unsigned int nMsg, unsigned int wParam, long lParam) Line 5943	C++
 	user32.dll!_InternalCallWinProc@20()	Unknown
 	user32.dll!_UserCallWinProcCheckWow@32()	Unknown
 	user32.dll!_DispatchMessageWorker@8()	Unknown
 	user32.dll!_DispatchMessageW@4()	Unknown
...
Comment 9 Julien Nabet 2015-07-09 14:09:02 UTC
Michael Meeks: I suppose you were talking about Michael Stahl in your last comment, so I put him on cc.
Comment 10 David Gerard 2015-07-10 19:47:53 UTC
See also bug 87984 - if this is reproduced by the same steps, these are duplicates.
Comment 11 Michael Stahl (allotropia) 2015-07-16 14:04:39 UTC
(In reply to Michael Meeks from comment #7)
> Oooh - I get a "solar mutex not locked" error dialog on Windows - what fun;
> chasing.

i don't, on current master on Windows or Linux.

at least not with the UI interactions that are possible without actually
authenticating to an actual server.

i don't reproduce a crash either.

if i use the "OpenText ELS 10.2.0" option, which apparently has
a syntactially correct URL, and refresh without entering
user credentials, i get a little hang where apparently
libcmis is blocking on network input for a minute or so,
before it pops up the "invalid device" error.
Comment 12 Gordo 2015-07-20 12:21:35 UTC
As per the CMIS threads at http://nabble.documentfoundation.org/template/NamlServlet.jtp?macro=user_nodes&user=487093 the dialogues have changed.

When choosing the Type "Other CMIS" and clearing the Host then clicking on the refresh button I still get a crash.

Windows Vista 64
Version: 5.1.0.0.alpha1+
Build ID: fcdddbd30a8b5cf6a5cc4d2ff28b7d4a20f8ec6b
TinderBox: Win-x86@39, Branch:master, Time: 2015-07-20_05:21:02
Comment 13 Caolán McNamara 2016-05-27 09:03:24 UTC
I can't reproduce this anymore, maybe was the same thing as 87984 after all

*** This bug has been marked as a duplicate of bug 87984 ***