Re[2]: [M-Dev] infinit recursive call to FillThreadTables() on [cancel]

[Volker A. <vol...@ya...>]

On Wed, 21 Nov 2001 18:03:42 +0100 (CET) Vadim Zeitlin <Vad...@dp...> wrote:

> On Wed, 21 Nov 2001 12:25:15 +0100 (CET) Volker Apelt
> <vol...@ya...> wrote:
> 
> VA> Find a huge folder (mine has >6300 articles). 
> VA> click once on it. A progress dialog is shown.
> 
>  The one about "retrieving headers" or the one about "threading"?

About retrieving
 
> VA> click on [cancel] while M is still reading the folder.
> VA> 
> VA> crash and core.
> 
>  Weird, why does it dump core? Is it really the stack overflow?

Ah, good guess Vadim, it is:
"Program terminated with signal 11, Segmentation fault."

I have prepared a 'fresh' core. So, tell me what printout you need,
if you have questions on it.
 
>  FillThreadTables() is a recursive function (I think it could be rewritten
> more efficiently as a non recursive one using a stack of thread nodes but
> I am not going to do it) although it's probably not normal that it is that
> much recursive - I haven't ever seen threads of such (5000) depth...
> Xavier, any idea about this one?

The threading is not really 5000 Messages deep, it should't nee more than
usual (maimum 10 ... 20)

## mail/HeaderInfoImpl.cpp:1279

1277       if (node->branch != 0)
1278          nbChildren += FillThreadTables(node->branch, thrData, threadedIndex,
1279                                         indent, indentIfDummyNode);


Is node->banch guaranteed to be initialized, if [cancel] is pressed ?
How is [cancel] supposed to terminate threading? 

All calls start in this line. This time the Number of stac frames goes from 
start at 5049. 

(gdb) up 5049
#5049 0x81408cb in FillThreadTables (node=0x875fb78, thrData=0x84c51f8, threadedIndex=@0xbfffea38, indent=0, indentIfDummyNode=false)
    at mail/HeaderInfoImpl.cpp:1279
1279                                         indent, indentIfDummyNode);
(gdb) print node 
$1 = (thread_node *) 0x875fb78
(gdb) print *node 
$2 = {num = 212, sc = 0x875fbe8, branch = 0x8768f20, next = 0x0}
(gdb) up          
#5050 0x81409bf in HeaderInfoListImpl::CombineSortAndThread (this=0x85de788) at mail/HeaderInfoImpl.cpp:1311
1311                              0, m_thrParams.indentIfDummyNode);
(gdb) print *this
$3 = (HeaderInfoListImpl) {<HeaderInfoList> = {<MObjectRC> = {<MObject> = {m_magic = 1234567890, _vptr. = 0x8342e80}, 
      m_nRef = 2}, <No data fields>}, m_mf = 0x85bbf48, m_headers = {<wxBaseArray> = {m_nSize = 6357, m_nCount = 6357, 
      m_pItems = 0x863da20, _vptr. = 0x8342f00}, <No data fields>}, m_count = 6357, m_tableSort = 0x8643d78, m_reverseOrder = false, 
  m_reversedTables = true, m_thrData = 0x84c51f8, m_tableMsgno = 0x0, m_tablePos = 0x0, m_sortParams = {sortOrder = 5, 
    detectOwnAddresses = true, ownAddresses = {m_nSize = 16, m_nCount = 3, m_pItems = 0x861fdd8, m_autoSort = false}}, m_thrParams = {
    useThreading = true, useServer = false, useServerByRefOnly = false, simplifyingRegex = {
      m_pchData = 0x85effc4 "^ *(R[Ee](\\[[0-9]+\\])?: +)*(\\[[^][]+\\] +)?(R[Ee](\\[[0-9]+\\])?: +)*"}, replacementString = {
      m_pchData = 0x84b3684 "\\3"}, gatherSubjects = true, breakThread = true, indentIfDummyNode = false}, m_dontFreeMsgnos = false, 
  m_firstSort = true, m_lastMod = 3}


Bye

Volker
--
Volker Apelt   volker_apelt  .@.  yahoo.de (remove the dots, please)   
Dipl. Chem.    +49 6172 31126