| Summary: | SEGV caused by invalid font | ||
|---|---|---|---|
| Product: | Ghostscript | Reporter: | Alex Cherepanov <alex> |
| Component: | PDF Writer | Assignee: | Ken Sharp <ken.sharp> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | ||
| Priority: | P2 | ||
| Version: | master | ||
| Hardware: | PC | ||
| OS: | Windows NT | ||
| Customer: | Word Size: | --- | |
| Attachments: | Sample file from alextest/pdf.0 | ||
Well, at least its not caused by the FreeType code, it fails in the same way with the internal Ghostscript font handling. Increasing priority for a crash. Fixed in revision 11270: http://ghostscript.com/pipermail/gs-cvs/2010-May/011052.html I don't think the font is invalid, the problem was caused by not properly interpreting a CFF operation (cntrmask with operands on the stack), which led to us not skipping over bytes in the data stream. Eventually this caused us to interpret data as operators, with unpredictable, but unfortunate, results. |
Created attachment 6292 [details] Sample file from alextest/pdf.0 gs fails with SEGV running ps2pdf on the sample file. The error is caused by uninitialized ipsp->cs_data->procs in gxtype1.c:458 . The SEGV is easy to fix by checking the stack, but this causes a PostScript error. Somehow, the file runs without an error on raster devices.