[Libreoffice-commits] core.git: vcl/source

[Stephan Bergmann]

 vcl/source/outdev/map.cxx |   12 ------------
 1 file changed, 12 deletions(-)

New commits:
commit d8e7bc042bd0912a66c231b101cecda0f838a80f
Author: Stephan Bergmann <sbergman at redhat.com>
Date:   Wed Mar 4 10:01:57 2015 +0100

    Remove hack of replacing 0 with 1/LONG_MAX
    
    ...that hack was in there since the initial import, but it is unclear to me what
    it shall fix, and it rather leads to problems as reported by
    -fsanitize=signed-integer-overflow when OuptutDevice::PixelToLogic of a 1x1 Size
    calls ImplPixelToLogic with n=1, nMapDenom=LONG_MAX causing the computation of
    2*n64 to overflow on 64-bit Linux.
    
    Change-Id: I2465da46175b8a1d4af1c23fb3df993230eec78d

diff --git a/vcl/source/outdev/map.cxx b/vcl/source/outdev/map.cxx
index 7a9cc0c..5a43c07 100644
--- a/vcl/source/outdev/map.cxx
+++ b/vcl/source/outdev/map.cxx
@@ -329,18 +329,6 @@ static void ImplCalcMapResolution( const MapMode& rMapMode,
     rMapRes.mnMapScDenomX = aTempX.GetDenominator();
     rMapRes.mnMapScNumY   = aTempY.GetNumerator();
     rMapRes.mnMapScDenomY = aTempY.GetDenominator();
-
-    // hack: 0/n approximately 1/max
-    if ( !rMapRes.mnMapScNumX )
-    {
-        rMapRes.mnMapScNumX = 1;
-        rMapRes.mnMapScDenomX = LONG_MAX;
-    }
-    if ( !rMapRes.mnMapScNumY )
-    {
-        rMapRes.mnMapScNumY = 1;
-        rMapRes.mnMapScDenomY = LONG_MAX;
-    }
 }
 
 inline void ImplCalcMapResolution( const MapMode& rMapMode,