[XeTeX] unicode-math testing

[George N. White III]

On Fri, May 28, 2010 at 7:20 PM, Nikos Platis <nplatis at freemail.gr> wrote:

> By now this is mostly off-topic, but I checked the test case on
> another computer with TeXLive 2009 x86_64 and get segmentation fault
> as well.
>
> I also just installed TL 2010/pretest (Linux x86_64) and again get the
> segmentation fault.
>
> Has anyone else tried the file on this architecture?
>
> Nikos Platis
>
>
> On Thu, May 27, 2010 at 11:45, Nikos Platis <nplatis at freemail.gr> wrote:
>> Interesting...
>>
>> On a fully updated TeXLive 2009 on Linux i386, with XeTeX
>> 3.1415926-2.2-0.9995.2, the sample runs fine.
>>
>> On a fully updated TeXLive 2009 on Linux x86_64, with XeTeX
>> 3.1415926-2.2-0.9995.2, I get a segmentation fault!
>> This seems to happen when processing unicode-math-table.tex, the log stops at
>>
>> Defining \um_config_mathbfsfup_Latin:n on line 1658
>> Defining \um_config
>>
>> If I can provide more debug information, I can do so today. I will try
>> the sample on another machine with more or less the same configuration
>> tomorrow and report my results.
>>
>> Nikos Platis

TL2010pretest on Ubuntu  Linux x86_64 segfualts for me as well.  strace
ends with:

read(5, "UnicodeMathSymbol{\"1D69D}{\\mttt "..., 4096) = 4096
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
+++ killed by SIGSEGV +++
Segmentation fault

valgrind says:

==4835== Conditional jump or move depends on uninitialised value(s)
[...]

 (./unicode-math-table.tex==4835== Invalid read of size 2
==4835==    at 0x425628: ??? (in /usr/local/texlive/2010/bin/x86_64-linux/xetex)
[...]
==4835==    by 0x5303C4C: (below main) (libc-start.c:226)
==4835==  Address 0x18eea82 is not stack'd, malloc'd or (recently) free'd
==4835==
==4835==
==4835== Process terminating with default action of signal 11 (SIGSEGV)
==4835==  Access not within mapped region at address 0x18EEA82

When run under gdb the job runs successfully.


-- 
George N. White III <aa056 at chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia