The file causes a Seg fault when converting to PDF using: gswin32c -sDEVICE=pdfwrite -o bmw.pdf bmw.eps The segfault happens in pdf_find_same_resource when attempting to compare a cos_dict to a cos_array, both in the resourceFunction chain. A patch that prevents comparing incompatible types is attached. Analysis is needed to determine if the function data (created with alloc_aside as cos_generic, then later changed to cos_array) should be in this chain, or in some other chain.
Created attachment 3012 [details] BMW.eps Adobe Illustrator file that triggers the problem (actually EPSF format)
Created attachment 3013 [details] patch for gdevpdfu.c
Customer bug priority
rev 8045 with the patch applied has an unexpected regression with pdfwrite 311- 03.ps . Raster devs are fine. The effect is pretty strange, now analyzing it.
rev 8045 with no path fails also. So the failure is unrelated to the patch. Rev 8022 doesn't fail.
The patch has been committed to HEAD as http://ghostscript.com/pipermail/gs-cvs/2007-June/007629.html The regression mentioned in Comment #4,5 appears to be unrelated to the patch. I'll open a separate bug report for it when I understand the effect better. Now it is too hard to reproduce.
Original bug fixed by my patch. Unrelated problem mentioned in comment #4,5 will be opened as a separate bug when confirmed.
The problem mentioned in Comment #4,5 is described in bug 689284.