adler187 has asked for the wisdom of the Perl Monks concerning the following question:

I'm working on an XS perl module and having troubles with an SV. From perlapi, SvPV should coerce any scalar to a string, but whenever I do this, I get a SegFault. Trying example #2 for Inline::C gives me the same. Here's some basic functions I've been testing: 
void foo(SV * sv)
{
    if(sv)
    {
        printf("%s\n", SvPV_nolen(sv));
    }
}

void bar(SV * sv)
{
    printf ("%i\n", SvPOK(sv));
}

Mytest2::foo('asdf'); # prints asdf

Mytest2::bar(1); # prints 0

Mytest2::foo(1); # Segmentation Fault

[download]
Do I really have to do something like this: 
void foo(SV * sv)
{
    char buf[1024];
    
    switch(SvTYPE(sv))
    {
        case SVt_IV:
            snprintf(buf, sizeof(buf), "%s\n", SvPV(sv));
            break;
            
        case SVt_PV:
            snprintf(buf, sizeof(buf), "%d\n", SvIV(sv));
            break;
            
        case SVt_NV:
            snprintf(buf, sizeof(buf), "%f\n", SvNV(sv));
            break;
            
            ... handle scalar refs here ...
    }
    printf("%s\n", buf);
}

[download]
This is on perl 5.14.2, linux x86-64

Replies are listed 'Best First'.
Re: SvPV Segmentation Fault
by BrowserUk (Patriarch) on Apr 25, 2012 at 20:52 UTC

    Different platform, version and build environment, but it confirms that SvPV_nolen() should do what you think it should do:

    #! perl -slw
use strict;
use Inline C => Config => BUILD_NOISY => 1;
use Inline C => <<'END_C',  NAME => '_967160', CLEAN_AFTER_BUILD => 0;

void foo( SV *sv ) {
    if( sv ) printf( "%s\n", SvPV_nolen( sv ) );
    return;
}


END_C

foo( 'fred' );
foo( 1 );

__END__
C:\test>967160
validate Stage
get_maps Stage
Writing Makefile for _967160

Microsoft (R) Program Maintenance Utility Version 9.00.21022.08
Copyright (C) Microsoft Corporation.  All rights reserved.

        C:\Perl64\bin\perl.exe C:\Perl64\lib\ExtUtils\xsubpp  -typ
        cl -c  -I"C:/test"  -nologo -GF -W3 -MD -Zi -Ox -GL -fp:pr
_967160.c
Running Mkbootstrap for _967160 ()
        C:\Perl64\bin\perl.exe -MExtUtils::Command -e chmod -- 644
        C:\Perl64\bin\perl.exe -MExtUtils::Mksymlists  -e "Mksymli
        link -out:blib\arch\auto\_967160\_967160.dll -dll -nologo
   Creating library blib\arch\auto\_967160\_967160.lib and object
Generating code
Finished generating code
        if exist blib\arch\auto\_967160\_967160.dll.manifest mt -n
        if exist blib\arch\auto\_967160\_967160.dll.manifest del b
        C:\Perl64\bin\perl.exe -MExtUtils::Command -e chmod -- 755
        C:\Perl64\bin\perl.exe -MExtUtils::Command -e cp -- _96716
        C:\Perl64\bin\perl.exe -MExtUtils::Command -e chmod -- 644

Microsoft (R) Program Maintenance Utility Version 9.00.21022.08
Copyright (C) Microsoft Corporation.  All rights reserved.

Files found in blib\arch: installing files in blib\lib into archit
Installing C:\test\_Inline\lib\auto\_967160\_967160.dll
Installing C:\test\_Inline\lib\auto\_967160\_967160.exp
Installing C:\test\_Inline\lib\auto\_967160\_967160.lib
Installing C:\test\_Inline\lib\auto\_967160\_967160.pdb
fred
1

C:\test>967160
fred
1

    [download]

    It also works here on 5.14.2:

    <Reveal this spoiler or all in this thread>

    Which leaves platform and build environment to isolate, with the latter the most likely the problem I think. Perhaps you could show your build process console log?

    With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
    Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
    "Science is about questioning the status quo. Questioning authority".
    In the absence of evidence, opinion is indistinguishable from prejudice.

    The start of some sanity?
[reply]
[d/l]
[select]

      Ok, that worked:

      ./svpv.pl 
Starting Build Preprocess Stage                                       
+                                                                     
+                                                                     
+                             
Finished Build Preprocess Stage                                       
+                                                                     
+                                                                     
+                             
                                                                      
+                                                                     
+                                                                     
+                             
Starting Build Parse Stage
Finished Build Parse Stage

Starting Build Glue 1 Stage
Finished Build Glue 1 Stage

Starting Build Glue 2 Stage
Finished Build Glue 2 Stage

Starting Build Glue 3 Stage
Finished Build Glue 3 Stage

Starting Build Compile Stage
  Starting "perl Makefile.PL" Stage
Writing Makefile for _967160
Writing MYMETA.yml
  Finished "perl Makefile.PL" Stage

  Starting "make" Stage
/usr/bin/perl /usr/lib/perl5/5.14.2/ExtUtils/xsubpp  -typemap /usr/lib
+/perl5/5.14.2/ExtUtils/typemap   _967160.xs > _967160.xsc && mv _9671
+60.xsc _967160.c
cc -c  -I/home/kadler -D_REENTRANT -D_GNU_SOURCE -DPERL_USE_SAFE_PUTEN
+V -fno-strict-aliasing -pipe -fstack-protector -D_LARGEFILE_SOURCE -D
+_FILE_OFFSET_BITS=64 -fmessage-length=0 -O2 -Wall -D_FORTIFY_SOURCE=2
+ -fstack-protector -funwind-tables -fasynchronous-unwind-tables -g -W
+all -pipe   -DVERSION=\"0.00\" -DXS_VERSION=\"0.00\" -fPIC "-I/usr/li
+b/perl5/5.14.2/x86_64-linux-thread-multi/CORE"   _967160.c
Running Mkbootstrap for _967160 ()
chmod 644 _967160.bs
rm -f blib/arch/auto/_967160/_967160.so
cc  -shared -L/usr/local/lib64 -fstack-protector _967160.o  -o blib/ar
+ch/auto/_967160/_967160.so        \
        \
  
chmod 755 blib/arch/auto/_967160/_967160.so
cp _967160.bs blib/arch/auto/_967160/_967160.bs
chmod 644 blib/arch/auto/_967160/_967160.bs
  Finished "make" Stage

  Starting "make install" Stage
Files found in blib/arch: installing files in blib/lib into architectu
+re dependent library tree
Installing /home/kadler/_Inline/lib/auto/_967160/_967160.bs
Installing /home/kadler/_Inline/lib/auto/_967160/_967160.so
  Finished "make install" Stage

  Starting Cleaning Up Stage
  Finished Cleaning Up Stage

Finished Build Compile Stage

fred
1

      [download]

      But, I wasn't actually using Inline::C. Just using that as an example of using SvPV and SvPV_nolen to build my XS module. There must be something wrong with building my XS module, then.

      I've been using perlxstut as a basis, specifically examples #4 and #5.

      Here's mylib.c: 
      #include <stdlib.h>

#include "mylib.h"

PerlInterpreter *my_perl;

void foo(SV * sv)
{
    if(sv)
    {
        printf("%s\n", SvPV_nolen(sv));
    }
}

void bar(SV * sv)
{
    printf ("%i\n", SvPOK(sv));
}

      [download]
      Here's mylib.h: 
      #include "EXTERN.h"
#include "perl.h"

#define TESTVAL 4
void foo(SV * sv);

void bar(SV * sv);

      [download]
      Here's Mytest2.xs: 
      #include "EXTERN.h"
#include "perl.h"
#include "XSUB.h"

#include "ppport.h"

#include "mylib/mylib.h"

#include "const-c.inc"

MODULE = Mytest2        PACKAGE = Mytest2        

INCLUDE: const-xs.inc

void foo(SV * sv)

void bar(SV * sv)

      [download]
      Everything else should be exactly as in perlxstut.
[reply]
[d/l]
[select]
        PerlInterpreter *my_perl
        That's very wrong. I don't know why you've put that there. You need to either pass the my_perl arg to the foo/bar functions using aTHX_/pTHX, or (less efficiently), retrieve it locally using dTHX.

        Dave.

[reply]
[d/l]

        I'd start by looking for differences in the generated C files between the Inline::C and XS versions. Also, a comparison of the respective makefile.pl's might shed some light.

        With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
        Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
        "Science is about questioning the status quo. Questioning authority".
        In the absence of evidence, opinion is indistinguishable from prejudice.

        The start of some sanity?
[reply]
Re: SvPV Segmentation Fault
by Khen1950fx (Canon) on Apr 25, 2012 at 22:35 UTC
    I tried it, keeping it simple---no segmentation fault. 
    #!/usr/bin/perl -l

use strict;
use warnings;
use Inline C => <<'END_C';

void foo( SV *sv ) { 
if( sv ) { 
    printf( "%s\n", SvPV_nolen( sv ) );
    return;
    }
}

void bar( SV *sv ) {
    printf( "%i\n", SvPOK( sv ) );
}

END_C

foo('asdf'); 
foo(1); 
bar(0);

    [download]
[reply]
[d/l]

Back to Seekers of Perl Wisdom