23352 - RDF Assertions thrown when starting up mail or composer

Status: VERIFIED FIXED

(Core Graveyard :: RDF, defect, P1)

Description

[kinmoz]

In today's WinNT debug build, when I bring up a Composer or Mail window a couple
of assertions are thrown, if I continue past them, I see what appears to be
embedded composer and mail windows in their sidebars.


Here's the first
assertion I'm seeing:



NTDLL! 77f76148()
nsDebug::Assertion(const char *
0x011fe6d4, const char * 0x011fe6c0, const char * 0x011fe68c, int 851) line 186
+ 13 bytes
nsDebug::PreCondition(const char * 0x011fe6d4, const char *
0x011fe6c0, const char * 0x011fe68c, int 851) line 275 + 21 bytes
InMemoryDataSource::HasAssertion(InMemoryDataSource * const 0x014136c0,
nsIRDFResource * 0x01b09280, nsIRDFResource * 0x00ccbac0, nsIRDFNode *
0x00000000, int 1, int * 0x0012f7e0) line 851 + 32 bytes
RDFXMLDataSourceImpl::HasAssertion(RDFXMLDataSourceImpl * const 0x01413710,
nsIRDFResource * 0x01b09280, nsIRDFResource * 0x00ccbac0, nsIRDFNode *
0x00000000, int 1, int * 0x0012f7e0) line 278
LocalStoreImpl::HasAssertion(LocalStoreImpl * const 0x01413914, nsIRDFResource *
0x01b09280, nsIRDFResource * 0x00ccbac0, nsIRDFNode * 0x00000000, int 1, int *
0x0012f7e0) line 139
CompositeDataSourceImpl::HasAssertion(CompositeDataSourceImpl * const
0x0280da80, nsIRDFResource * 0x01b09280, nsIRDFResource * 0x00ccbac0, nsIRDFNode
* 0x00000000, int 1, int * 0x0012f7e0) line 1031 + 32 bytes
CompositeDataSourceImpl::OnUnassert(CompositeDataSourceImpl * const 0x0280da84,
nsIRDFResource * 0x01b09280, nsIRDFResource * 0x00ccbac0, nsIRDFNode *
0x00000000) line 1428 + 34 bytes
InMemoryDataSource::Unassert(InMemoryDataSource
* const 0x00ccb840, nsIRDFResource * 0x01b09280, nsIRDFResource * 0x00ccbac0,
nsIRDFNode * 0x00000000) line 1123
nsWindowMediator::UpdateWindowTitle(nsWindowMediator * const 0x00cca5a0,
nsIWebShellWindow * 0x01b09e10, const unsigned short * 0x02ad2070) line 568
nsWebShellWindow::SetTitle(nsWebShellWindow * const 0x01b09e18, const unsigned
short * 0x100782d0 gCommonEmptyBuffer) line 2786
nsWebShell::SetTitle(nsWebShell
* const 0x02a88e2c, const unsigned short * 0x100782d0 gCommonEmptyBuffer) line
3869
nsHTMLDocument::SetTitle(nsHTMLDocument * const 0x02a6a698, const nsString
& {...}) line 777
HTMLContentSink::DidBuildModel(HTMLContentSink * const
0x02a69a50, int 0) line 2166 + 38 bytes
CNavDTD::DidBuildModel(CNavDTD * const
0x02ab6180, unsigned int 0, int 1, nsIParser * 0x02a69b70, nsIContentSink *
0x02a69a50) line 688 + 14 bytes
nsParser::DidBuildModel(unsigned int 0) line 586
+ 55 bytes
nsParser::ResumeParse(nsIDTD * 0x00000000, int 1) line 984
nsParser::OnStopRequest(nsParser * const 0x02a69b74, nsIChannel * 0x02a8bb20,
nsISupports * 0x00000000, unsigned int 0, const unsigned short * 0x00000000)
line 1368 + 19 bytes
nsDocumentOpenInfo::OnStopRequest(nsDocumentOpenInfo *
const 0x02a8a290, nsIChannel * 0x02a8bb20, nsISupports * 0x00000000, unsigned
int 0, const unsigned short * 0x00000000) line 243
nsInputStreamChannel::OnStopRequest(nsInputStreamChannel * const 0x02a8bb24,
nsIChannel * 0x02a8ba70, nsISupports * 0x00000000, unsigned int 0, const
unsigned short * 0x00000000) line 345 + 45 bytes
nsOnStopRequestEvent::HandleEvent(nsOnStopRequestEvent * const 0x02a8b770) line
279
nsStreamListenerEvent::HandlePLEvent(PLEvent * 0x02a8cde0) line 93 + 12
bytes
PL_HandleEvent(PLEvent * 0x02a8cde0) line 522 + 10 bytes
PL_ProcessPendingEvents(PLEventQueue * 0x00caae70) line 483 + 9 bytes
_md_EventReceiverProc(HWND__ * 0x00150410, unsigned int 49400, unsigned int 0,
long 13282928) line 951 + 9 bytes
USER32! 77e71250()
00caae70()

Comment 1

[Chris Waterson]

Attachment: proposed fix

Comment 2

[kinmoz]

I filed the composer/mail window in sidebar bug as bug #23358.

Comment 3

[Chris Waterson]

see attached patch. GetTarget() will "succeed" (i.e., NS_SUCCEEDED() == true)
even if no target node is found. So also check target node before
Unassert()-ing.

Comment 4

[kinmoz]

Whoops mid-air collision during bug modification. Setting Priority back to P1
and Target Milestone back to M13.

Comment 5

[Chris Waterson]

Attachment: additional cleanup & sanity checking that'll catch bugs like this more quickly

Comment 6

[Chris Waterson]

rjc: i added some additional sanity checking code to nsInMemoryDataSource.cpp;
basically just protected all JS-accessable callsites with NS_PRECONDITION()/if
(null) checks. Could you code review the -second- patch for me?

Comment 7

[Robert John Churchill]

Looks good.

Comment 8

[Chris Waterson]

fix checked in, r=davidm for the first patch, r=rjc for the second patch

Comment 9

[Tom Everingham]

marking verified per engineer's comments