develooper Front page | perl.perl5.porters | Postings from March 2001

Re: my purify runs with 5.6.1-to-be

Thread Previous | Thread Next
From:
Gurusamy Sarathy
Date:
March 16, 2001 07:58
Subject:
Re: my purify runs with 5.6.1-to-be
Message ID:
200103161558.f2GFwKm12508@smtp3.ActiveState.com
On Fri, 16 Mar 2001 10:48:06 +0300, "Konovalov, Vadim" wrote:
>> Sarathy's "clear_pmop" patch hasn't been applied yet.  Well, it was,
>> but since it causes the fertilizer hit the fan in Digital UNIX really
>> quick, I took it out.  The maddening thing is that the core dumps
>> galore doesn't happen in platforms where Sarathy has access to :-(
>
>Can it be because Sarathy changed the size of structure PMOP by adding extra
>field into it, thus influenting some memory allocations?

Why should this matter?  AFAIK, nothing in the sources relies on the
sizeof PMOP (except for the allocations themselves).

>Anyway, it seems to me that in "clear_pmop" patch there should be no extra
>field for PMOP, at least for economy reasons, and that list of PMOPs should
>be organized in a slightly different manner. Say, there should be one global
>pointer to head of that list -- it seems to me this will solve that broblem.

"What _precisely_ is the problem?"  That we get a random coredump on a
random platform is only a symptom.  We need to find where the problem
is and fix that.

I'd suggest the following approach:

   * find out what platforms it coredumps on
   * if it is only one platform, check if it may due to compiler
     bug (use a different compiler; try turning off optimizations, &c)
   * if it is more than one platform, run something like purify on
     it and watch the address that it SEGVs on and see who modifies
     it

>What do you think?
>should I try this approach?

Perhaps, but I won't have any confidence that you'd have fixed
the real problem until someone manages to explain why the other
approach doesn't work in some places.

>Another question  - Sarathy, did you applied your "clear_pmop" patch into
>maint version of perl?

Nope, and I probably won't for 5.6.1.  This bug is too low priority
to risk applying without understanding why it crashes only in some
places.


Sarathy
gsar@ActiveState.com

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About