Re: Problem with building XS module on AS 1401

On 08/07/2011 11:59 PM, Erland Sommarskog wrote:
>     int
>     executebatch(sqlsrv, rows_affected = NULL)
>      SV * sqlsrv;
>      SV * rows_affected;
>
> In one place in one test script, this XS routine returns undef. This is a
> core routine, so it is called well over thousand times in the test suite.
> And in one specific case it returns undef. I've traced it so far that the
> underlying C++ function does really execute to the end and returns TRUE.
>
> Apparently something goes wrong in the part that comes after? Why? Maybe
> something was clobbered, so maybe I have a bad pointer somewhere. But
> they usually cause a lot more problem than a single miss in 1000 executions.
>
> Anyway, that's why I stick with -O2.

Check the C code generated by xsubpp. Then insert ludicrous amounts of 
printf()'s into the code that comes after the actual C++ function 
invocation.

Since you're passing around SV*'s the bug might be in the code that you 
wrote to handle the Perl datastructures if it's more than just a simple 
SvIV on rows_affected or SvPV_nolen() on sqlsrv. Sometimes, the time of 
destruction of "mortal"s may not be what you expect (generally during 
nextstate ops, though), so if you have a mortal SV that gets destroyed 
early... err. No. Then you typically get a segmentation fault, not an 
SV* that just happens to point at &PL_sv_undef. Nevermind my drivel. :)

Hope this helps a little bit.
Steffen