Re: Re: [COMMITTERS] pgsql: Initialise perl library as documented in perl API.

Andrew Dunstan wrote:
>
>
> Tom Lane wrote:
>> Andrew Dunstan <andrew@dunslane.net> writes:
>>  
>>> That's what we have a buildfarm for ;-) There's  a failure on FBSD 
>>> too by the look of it. I'll dig some more to see what I can find.
>>>     
>>
>> I see this when building HEAD on Fedora 10:
>>
>> plperl.c: In function 'plperl_init_interp':
>> plperl.c:450: warning: null argument where non-null required 
>> (argument 3)
>>
>> Seems like a good hint ...
>>
>>            
>>   
> Yeah. I didn't get that. But the odd thing is that on 5.8 especially 
> it shouldn't matter.
>
> perl 5.8.8's perl.h has:
>
>    #ifndef PERL_SYS_INIT3
>    #  define PERL_SYS_INIT3(argvp,argcp,envp) PERL_SYS_INIT(argvp,argcp)
>    #endif
>
>
> and the only place it's defined elsewhere that I can see is for OS2 
> (for anyone still running it!). These two Unixware machines have 
> 5.8.8. and the FBSD machine has 5.8.7. So surely it can't be that, 
> unless I'm missing something badly.
>
> The unixish.h file has this on 5.8:
>
>    #  define PERL_SYS_INIT(c,v) MALLOC_CHECK_TAINT2(*c,*v)
>    PERL_FPU_INIT MALLOC_INIT
>
> I'm guessing the problem is actually somewhere in there.
>
>

[hours of digging later]

On FBSD at least, this is failing in the function Perl_do_taint(). I 
can't see anything too terrible in the source for this, but to dig 
further I'd have to build another perl with debugging turned on. Anyway, 
it turns out that this actually isn't called at all if Perl is 
configured to use its own malloc() routines instead of those supplied by 
the system, as is the case on Fedora, for example, but not FBSD.

So we have a bit of a dilemma. We know that this initialization is 
required by the Perl API, and we know that some platforms fail without 
it, and we also know that this fails on some platforms which use the 
system's malloc() for perl.

I think we need to float a bug upstream to the perl guys on this, but as 
a holding position I suggest that we alter the #ifdef test to avoid 
calling PERL_SYS_INIT3() where MYMALLOC is defined. It's ugly, but I 
can't think of another simple way around it (and we've done worse things 
to accommodate platform weirdness ;-) )

BTW, it's not caused by passing NULL as the third argument of 
PERL_SYS_INIT3() - changing that was the first thing I tried.


cheers

andrew