Bug#327082: tetex-bin: pdflatex segfaults on iclusion of png images
Wolfram Quester <wolfi@mittelerde.physik.uni-konstanz.de> wrote:
> Yes, I know this workaround and I'm using it ATM. Does this mean that you
> can reproduce this bug with the image I sent you in?
I can reproduce it on powerpc on sid, not on sarge (in the etch chroot
on bruckner, no tetex is installed, so I can' test.)
$ diff -u sid.pdftex.ldd.1 sarge.pdftex.ldd.1
--- sid.pdftex.ldd.1 Thu Sep 8 13:40:05 2005
+++ sarge.pdftex.ldd.1 Thu Sep 8 13:40:11 2005
@@ -1,8 +1,8 @@
libpng12.so.0 => /usr/lib/libpng12.so.0
libz.so.1 => /usr/lib/libz.so.1
libkpathsea.so.3 => /usr/lib/libkpathsea.so.3
- libstdc++.so.6 => /usr/lib/libstdc++.so.6
+ libstdc++.so.5 => /usr/lib/libstdc++.so.5
libm.so.6 => /lib/libm.so.6
libgcc_s.so.1 => /lib/libgcc_s.so.1
libc.so.6 => /lib/libc.so.6
- /lib/ld.so.1
+ /lib/ld.so.1 => /lib/ld.so.1
So despite the different output format for ld.so, the only difference is
libstdc++. However, libkpathsea (which contains xfopen.o) is a C
library, not C++.
The problematic call is:
,----
| void
| xfclose P2C(FILE *, f, const_string, filename)
| {
| assert (f);
|
| if (fclose (f) == EOF)
| FATAL_PERROR (filename);
| }
`----
So it seems the message means that the file descriptor is lost? I don't
really know how scalar expressions are handled in C, though - I only
read assert(3).
Regards, Frank
--
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer
Reply to: