In Xpdf 4.05 (and earlier), invalid header info in a DCT (JPEG) stream can lead to an uninitialized variable in the DCT decoder. The proof-of-concept PDF file causes a segfault attempting to read from an invalid address.

In Xpdf 4.05 (and earlier), a PDF object loop in a pattern resource leads to infinite recursion and a stack overflow.

In Xpdf 4.05 (and earlier), very large coordinates in a page box can cause an integer overflow and divide-by-zero.

Out-of-bounds array write in Xpdf 4.05 and earlier, due to incorrect integer overflow checking in the PostScript function interpreter code.

Out-of-bounds array write in Xpdf 4.05 and earlier, triggered by long Unicode sequence in ActualText.

Out-of-bounds array write in Xpdf 4.05 and earlier, due to missing object type check in AcroForm field reference.

Out-of-bounds array write in Xpdf 4.05 and earlier, triggered by negative object number in indirect reference in the input PDF file.

Out-of-bounds array write in Xpdf 4.05 and earlier, triggered by an invalid character code in a Type 1 font. The root problem was a bounds check that was being optimized away by modern compilers.