Created attachment 6277 [details] ghostscript-bbox-close.patch When bbox_close_device() is called on a compositor wrapper bbox, there is no check before calling gs_closedevice on the target device. Original bug report: https://bugzilla.redhat.com/show_bug.cgi?id=591624
Patch looks "obviously" correct and committed as r11235 . But it is a bit worrying that the target device is NULL there (it shouldn't be). So leave this bug open when we find the regression and re-assign.
Created attachment 6278 [details] www.ti.com/ww/en/analog/led/slyt394a.pdf The test file (just in case it disappears off the internet). Page 1 has a problem: Processing pages 1 through 64. Page 1 jasper (code 0) write component failed jasper (code 0) error: cannot decode code stream unable to decode JPX image data. Luratech decoder can process it. But regardless, page two crashes both jasper and luratech binaries, so is unrelated.
Reassign to author of the problematic patch from git-bisect: r10369 | ray | 2009-11-23 17:35:16 +0000 (Mon, 23 Nov 2009) | 32 lines Fix for segfault reported in 690880 with separation devices. The is_open flag was not being properly maintained when put_params resulted in a target device closing due to DeviceN color state changing. This commit causes a regression with crash with the attached test file.
Tested with 9.22 and the command line: gs Bug691296.pdf Command line not given, but due to reference to the bbox device, I deduce that the default x11 device (on linux) was used. This works for me no problem. Note that we now use openjpeg, which also complains on page 1 with: openjpeg warning: Non conformant codestream TPsot==TNsot. but no crash/segfault occurs. I assume this has been fixed.