134552 - Memory leak in nsCellMap::ExpandWithRows()

Status: RESOLVED INCOMPLETE

(Core :: Layout: Tables, defect, P2)

Description

[Peter A Jonsson]

__builtin_new -:- libstdc++-libc6.1-1.so.2 0x407d81e6 
nsCellMap::AppendCell <...> nsCellMap.cpp:1191 libgklayout.so 0x8075dbf3 
nsCellMap::ExpandWithRows <...> nsCellMap.cpp:1425 libgklayout.so 0x8075e59d 
nsCellMap::InsertRows <...> nsCellMap.cpp:1083 libgklayout.so 0x8075d86e 
nsTableCellMap::InsertRows <...> nsCellMap.cpp:440 libgklayout.so 0x8075bdbb

Not catching a return value from AppendCell at nsCellMap.cpp:1425?

Comment 1

[Bernd]

I went to the URL, it does not show this pattern for 2002-04-09.

Comment 2

[Amarendra Hanumanula]

 WFM on both branch and trunk builds : 2002061408. WIN2K. I am not seeing any 
kind of assertion or memory leak.

Comment 3

[Amarendra Hanumanula]

 WFM on both branch and trunk builds : 2002061408. WIN2K. I am not seeing any 
kind of assertion or memory leak.

Comment 4

[Peter A Jonsson]

This is a memory leak, not an assertion of any kind. 

Fresh from CVS:

CellData *
nsTableCellMap::AppendCell(..) {
 <..>
 CellData *result;
 <..>
 result = ...
 return result;
}

From line 1424:

  if (IS_TABLE_CELL(cFrameType)) {
    AppendCell(aMap, (nsTableCellFrame *)cFrame, rowX, PR_FALSE, DamageArea);
  }

Why does this *not* leak?

Comment 5

[karnaze (gone)]

Peter, why does it leak? This code is just adding an entry to the cell map if it 
is a cell. If it is not a cell, it doesn't get added, but it still has a frame. 

Comment 6

[Bernd]

just for the record: origdata is returned from cellmap::AppendCell if this
pointer is not always added to the cellmap then it would be lost

Comment 7

[karnaze (gone)]

mass reassign to default owner

Comment 8

[Boris Zbarsky [:bzbarsky]]

It looks like AppendCell does in fact put the data in the cellmap (except in
cases that assert and should never be hit....).

So I don't think we have a leak here (though we do have some rather unreadable
code).

Comment 9

[Marco Castelluccio [:marco]]

Can we close this bug as we shouldn't have a leak here?