[PATCH] gdi32: fix PatBlt() drawing with negative width/height

[Damjan Jovanovic]

In a (x, y, w, h) rectangle passed to PatBlt(), a negative w results in
the rectangle drawn being (x+w, y, -w, h), and negative h results in
(x, y+h, w, -h). Wine instead does (x+w+1, y, -w, h) and
(x, y+h+1, w, -h), so the rectangle drawn is shifted
1 pixel row too far down and/or 1 pixel column too far right.
This patch recalculates the rectangle dimensions correctly.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=27584
Signed-off-by: Damjan Jovanovic <damjan.jov(a)gmail.com>
---
 dlls/gdi32/bitblt.c       | 11 +++++++++++
 dlls/gdi32/tests/bitmap.c | 36 ++++++++++++++++++++++++++++++++++++
 2 files changed, 47 insertions(+)

[Huw Davies]

On Sat, Nov 14, 2020 at 11:56:57AM +0200, Damjan Jovanovic wrote:
> diff --git a/dlls/gdi32/tests/bitmap.c b/dlls/gdi32/tests/bitmap.c
> index fe3482671b2..a3969cf25d6 100644
> --- a/dlls/gdi32/tests/bitmap.c
> +++ b/dlls/gdi32/tests/bitmap.c
> @@ -3074,6 +3074,41 @@ static void test_BitBlt(void)
>      DeleteDC(hdcScreen);
>  }

Hi Damjan,

The tests would be better in tests/dib.c, see under
    /* solid brush PatBlt */
in draw_graphics().  You could add additional PatBlt()s
to the right of the existing one (the dib is 512 pixels
wide).

This also begs the question what happens if the brush
is patterned - is it flipped?  You could add a test
for that under
    /* 8888 DIB pattern brush */
(no need to explore every brush format, just one would
be enough).

Huw.