278511 - Displaying some emails results in ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start && end', file nsTDependentSubstring.cpp, line 76

Status: RESOLVED FIXED

(Thunderbird :: Mail Window Front End, defect)

Description

[Martin Mokrejs]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8a6) Gecko/20050115
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8a6) Gecko/20050115

Deleting an email in current Inbox folder using 'delete' button gives me. Once I
get binary with all debug symbols, I'll post a stracktrace. It's in cvs some
weeks at least. Still happens with todays cvs.

Begin mail message delivery.
Abort mail message delivery.
Begin mail message delivery.
Abort mail message delivery.
WARNING: nsMsgProtocol::SetContentCharset() not implemented, file
nsMsgProtocol.cpp, line 596
WARNING: GetDefaultCharsetForLocale: need to add multi locale support, file
nsUNIXCharset.cpp, line 189
mailCharsetLoadListener: ISO-8859-2 
WARNING: GetDefaultCharsetForLocale: need to add multi locale support, file
nsUNIXCharset.cpp, line 189
mailCharsetLoadListener: ISO-8859-1 
WARNING: GetDefaultCharsetForLocale: need to add multi locale support, file
nsUNIXCharset.cpp, line 189
###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76
###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76
###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76
###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76
###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76
###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76
###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76
###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76


Reproducible: Always

Comment 1

[Martin Mokrejs]

No, it happens actually when an email is selected in the list and therefore
displayed in the lower-right window. Will attach one such email.

Comment 2

[Martin Mokrejs]

Attachment: testcase

Comment 3

[David :Bienvenu]

Here's the stack trace - it happens to me on messages that contain quoted
replies. It started with some checkin around the 1.8a6 timeframe (I think just
after, but I'm not sure)

nsDependentSubstring::Rebind(const unsigned short * 0x00000000, const unsigned
short * 0x00000000) line 76 + 34 bytes
nsScannerSubstring::AsString() line 249
CCommentToken::GetStringValue() line 1512
nsCParserNode::GetText() line 160 + 17 bytes
SinkContext::AddComment(const nsIParserNode & {...}) line 1595 + 11 bytes
HTMLContentSink::AddComment(HTMLContentSink * const 0x0464b078, const
nsIParserNode & {...}) line 3195 + 18 bytes
CNavDTD::HandleCommentToken(CToken * 0x12269600) line 2117 + 34 bytes
CNavDTD::HandleToken(CNavDTD * const 0x123a2be0, CToken * 0x12269600, nsIParser
* 0x043abc48) line 906 + 12 bytes
CNavDTD::BuildModel(CNavDTD * const 0x123a2be0, nsIParser * 0x043abc48,
nsITokenizer * 0x04f8a410, nsITokenObserver * 0x00000000, nsIContentSink *
0x0464b078) line 455 + 20 bytes
nsParser::BuildModel(nsParser * const 0x043abc48) line 2023 + 34 bytes
nsParser::ResumeParse(int 0x00000001, int 0x00000000, int 0x00000001) line 1890
+ 12 bytes
nsParser::OnDataAvailable(nsParser * const 0x043abc4c, nsIRequest * 0x12134004,
nsISupports * 0x11fe1624, nsIInputStream * 0x1207a3e0, unsigned int 0x00000000,
unsigned int 0x000006d9) line 2568 + 21 bytes
nsDocumentOpenInfo::OnDataAvailable(nsDocumentOpenInfo * const 0x1209a488,
nsIRequest * 0x12134004, nsISupports * 0x11fe1624, nsIInputStream * 0x1207a3e0,
unsigned int 0x00000000, unsigned int 0x000006d9) line 348 + 46 bytes

Comment 4

[Blake Kaplan (:mrbkap) (inactive)]

Attachment: patch v1

I can't say for certain that this patch fixes this bug (could someone test
it?), but this is the only case where we could return NS_OK with an unbound
mComment. Note: we assert in the case: <!>, does that appear in the generated
message source?

Comment 5

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 171421 [details] [diff] [review]
patch v1

r+sr=bzbarsky

Comment 6

[Martin Mokrejs]

Tha patch doesn't help me. Another testcase will follow.


###!!! ASSERTION: nsTDependentSubstring must wrap a non-NULL buffer: 'start &&
end', file nsTDependentSubstring.cpp, line 76
Break: at file nsTDependentSubstring.cpp, line 76
mailCharsetLoadListener: ISO-8859-2 
WARNING: NS_ENSURE_TRUE(nsDoc) failed, file nsXULElement.cpp, line 2557

Comment 7

[Martin Mokrejs]

Attachment: another testcase

Comment 8

[Blake Kaplan (:mrbkap) (inactive)]

Attachment: patch v2

Whoops, I was blind when I made that first patch. This code was apparently
written with the assumption that aScanner.BindSubstring(string, aPos, aPos) was
a bad thing. We now enforce that mComment has been bound to something, so those
BindSubstring calls need to happen.

Martin, would you try this patch and see if it fixes things?

Comment 9

[Martin Mokrejs]

Yes, patch v2 has helped! Thanks!

Comment 10

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 171434 [details] [diff] [review]
patch v2

Looks good.

Comment 11

[Blake Kaplan (:mrbkap) (inactive)]

Fix checked in.