Fix integer overflow when stretching bitmap for printing

Fix integer overflow when stretch bitmap

When an image object is zoomed in by a big factor, the scaling factor in the transformation matrix is big as well, resulting in a large |dest_width| and |dest_height| value(they can be think of as the equivalent pixel size of the entire image, although most of it is outside the device).

BUG=395636
R=vitalybuka@chromium.org

Committed: https://pdfium.googlesource.com/pdfium/+/2fd400c

[Vitaly Buka (NO REVIEWS), 2014-07-30 21:24:43]

https://codereview.chromium.org/432543002/diff/1/core/src/fxge/win32/fx_win32...
File core/src/fxge/win32/fx_win32_device.cpp (right):

https://codereview.chromium.org/432543002/diff/1/core/src/fxge/win32/fx_win32...
core/src/fxge/win32/fx_win32_device.cpp:534: if (m_DeviceClass == FXDC_PRINTER
&& (pBitmap->GetWidth()/abs(dest_width) >
abs(dest_height)/pBitmap->GetHeight())) {
I guess better solution is just to cast to int64

((long long)pBitmap->GetWidth() * pBitmap->GetHeight() > (long
long)abs(dest_width) * abs(dest_height)

[Vitaly Buka (NO REVIEWS), 2014-07-30 21:27:27]

On 2014/07/30 21:24:43, Vitaly Buka wrote:
>
https://codereview.chromium.org/432543002/diff/1/core/src/fxge/win32/fx_win32...
> File core/src/fxge/win32/fx_win32_device.cpp (right):
> 
>
https://codereview.chromium.org/432543002/diff/1/core/src/fxge/win32/fx_win32...
> core/src/fxge/win32/fx_win32_device.cpp:534: if (m_DeviceClass == FXDC_PRINTER
> && (pBitmap->GetWidth()/abs(dest_width) >
> abs(dest_height)/pBitmap->GetHeight())) {
> I guess better solution is just to cast to int64
> 
> ((long long)pBitmap->GetWidth() * pBitmap->GetHeight() > (long
> long)abs(dest_width) * abs(dest_height)

Bo, how that was possible that area of image in pixels was bigger than 2^31?

[Bo Xu, 2014-07-30 23:35:51]

On 2014/07/30 21:27:27, Vitaly Buka wrote:
> On 2014/07/30 21:24:43, Vitaly Buka wrote:
> >
>
https://codereview.chromium.org/432543002/diff/1/core/src/fxge/win32/fx_win32...
> > File core/src/fxge/win32/fx_win32_device.cpp (right):
> > 
> >
>
https://codereview.chromium.org/432543002/diff/1/core/src/fxge/win32/fx_win32...
> > core/src/fxge/win32/fx_win32_device.cpp:534: if (m_DeviceClass ==
FXDC_PRINTER
> > && (pBitmap->GetWidth()/abs(dest_width) >
> > abs(dest_height)/pBitmap->GetHeight())) {
> > I guess better solution is just to cast to int64
> > 
> > ((long long)pBitmap->GetWidth() * pBitmap->GetHeight() > (long
> > long)abs(dest_width) * abs(dest_height)
> 
> Bo, how that was possible that area of image in pixels was bigger than 2^31?

The dest_width and dest_height are the pixels of the entire enlarged (zoomed in)
image. In this case, when zoomed to 1mile, dest_width and dest_height are 63517,
multiply them would overflow

[Vitaly Buka (NO REVIEWS), 2014-07-31 00:00:07]

lgtm

https://codereview.chromium.org/432543002/diff/1/core/src/fxge/win32/fx_win32...
File core/src/fxge/win32/fx_win32_device.cpp (right):

https://codereview.chromium.org/432543002/diff/1/core/src/fxge/win32/fx_win32...
core/src/fxge/win32/fx_win32_device.cpp:526: if (abs(dest_width) *
abs(dest_height) < pBitmap1->GetWidth() * pBitmap1->GetHeight() * 4 ||
same here?

https://codereview.chromium.org/432543002/diff/20001/core/src/fxge/win32/fx_w...
File core/src/fxge/win32/fx_win32_device.cpp (right):

https://codereview.chromium.org/432543002/diff/20001/core/src/fxge/win32/fx_w...
core/src/fxge/win32/fx_win32_device.cpp:525: CFX_ByteString info =
CFX_WindowsDIB::GetBitmapInfo(pBitmap);
don't you expect the same here?

https://codereview.chromium.org/432543002/diff/20001/core/src/fxge/win32/fx_w...
core/src/fxge/win32/fx_win32_device.cpp:534: if (m_DeviceClass == FXDC_PRINTER
&& ((long long)pBitmap->GetWidth() * pBitmap->GetHeight() > (long
long)abs(dest_width) * abs(dest_height))) {
looks like FX_INT64 is more consistent here

[Bo Xu, 2014-07-31 16:51:11]

https://codereview.chromium.org/432543002/diff/20001/core/src/fxge/win32/fx_w...
File core/src/fxge/win32/fx_win32_device.cpp (right):

https://codereview.chromium.org/432543002/diff/20001/core/src/fxge/win32/fx_w...
core/src/fxge/win32/fx_win32_device.cpp:525: CFX_ByteString info =
CFX_WindowsDIB::GetBitmapInfo(pBitmap);
On 2014/07/31 00:00:06, Vitaly Buka wrote:
> don't you expect the same here?

Yes.

https://codereview.chromium.org/432543002/diff/20001/core/src/fxge/win32/fx_w...
core/src/fxge/win32/fx_win32_device.cpp:534: if (m_DeviceClass == FXDC_PRINTER
&& ((long long)pBitmap->GetWidth() * pBitmap->GetHeight() > (long
long)abs(dest_width) * abs(dest_height))) {
On 2014/07/31 00:00:06, Vitaly Buka wrote:
> looks like FX_INT64 is more consistent here

That's right.

[Bo Xu, 2014-07-31 16:56:18]

Committed patchset #3 manually as r2fd400c (presubmit successful).

[Nico, 2014-07-31 17:02:03]

(nit: Please wrap your CL descriptions to about 80 columns, as `git log` doesn't
do any wrapping on its own)

[Bo Xu, 2014-07-31 17:03:26]

On 2014/07/31 17:02:03, Nico (away) wrote:
> (nit: Please wrap your CL descriptions to about 80 columns, as `git log`
doesn't
> do any wrapping on its own)

Thanks, will do.