Fix fuzzing bug due to float representation limitations.

The gel bbox was being stored internally as floats (despite
only holding ints). This means that as numbers get large the
bbox can become approximate, rather than exact. If the bbox
becomes smaller than it should, this causes crashes in the
scanline filling code.

This is seen with:

  tests_private/fuzzing/mupdf2/17f8aee51ac776994af0b36195cdadd7_signal_sigsegv_5607be_7308_5912.pdf

The solution is simply to use ints rather than floats.

Thanks to Mateusz Jurczyk and Gynvael Coldwind of the Google Security
Team for providing the example files.

1 files changed, 1 insertions, 1 deletions

diff --git a/source/fitz/draw-edge.c b/source/fitz/draw-edge.c
index 7d57f3b6..769116ac 100644
--- a/source/fitz/draw-edge.c
+++ b/source/fitz/draw-edge.c
@@ -160,7 +160,7 @@ struct fz_edge_s
 struct fz_gel_s
 {
 	fz_rect clip;
-	fz_rect bbox;
+	fz_irect bbox;
 	int cap, len;
 	fz_edge *edges;
 	int acap, alen;