DGGEV failure

[pa33]

Hello, the routine DGGEV (specifically DHGEQZ) fails on the generalized system:

A =
[ 1 1 0 ]
[ 0 0 -1 ]
[ 1 0 0 ]

B =
[ -1 0 -1 ]
[ 0 -1 0 ]
[ 0 0 -1 ]

By failure I mean it finds no eigenvalues, and therefore no eigenvectors either.
I think its an issue with the QZ steps not working and the exceptional shifts failing to find new eigenvalues. I have also found 1 other 3x3 integer system which fails, as well as many many 4x4 integer systems which fail. I can provide more matrices if desired.

I should also note that both maple and matlab solve this system correctly for both eigenvalues and eigenvectors.