The customer reports and I've verified that gs8.54, gs8.61, and gshead (r8434) report an error when reading the attached PDF file. The 1766208.PDF file is the original file, 1766208_simplified.pdf is a much simplified version and should not need to be private (I've redacted all of the private text and cropped the page)/ The command line I used for testing: bin/gs -sDEVICE=bbox ./1766208_simplified.pdf Note that gs8.54 produces the error messages in the bug summary; later versions produce a more generic message: "/undefinedresult in --run--".
Created attachment 3630 [details] 1766208.PDF
Created attachment 3631 [details] 1766208_simplified.pdf
Created attachment 3660 [details] hack91.pdf A really small sample file that sets degenerate text matrix and exercises a composite font.
Created attachment 3662 [details] patch Don't try to obtain the bounding box of a glyph when FontBBox is not valid and CTM is singular. Set an empty box and consider it valid. The bounding box is not used when the CTM is singular. Regression testing shows no differences.
The patch is committed as a rev. 8466. It fixes the simplified sample file (attachment #3 [details]) but the original sample file exposes one more bug - an mix-up between stringwidth rendering and "3 Tr" PDF mode.
Created attachment 3663 [details] hack8l.pdf PDF file that exercises "3 Tr" and singular CTM.
Created attachment 3666 [details] patch 2 Add a flag that marks PDF text rendering mode 3 to avoid confusion with other non-rendering text operations such as stringwidth. This avoids unnecessary calculation of the current point, which fails when the CTM is singular. This patch continues conservative approach to the text rendering modes started in rev. 4006.
The patch 2 is committed as a rev. 8470. Regression testing shows no differences.