[PVFS-developers] pvfs-lib cleanup/sanity checking

Rob Ross rross at mcs.anl.gov
Fri Jul 16 13:37:52 EDT 2004


Hi Don,

This seems like a good thing to me.  Has using this patch uncovered 
anything?

Thanks,

Rob

On Fri, 16 Jul 2004, Porter Don wrote:

> 
>  D'oh.  Here's the actual patch.
> 
> -----Original Message-----
> From: pvfs-developers-bounces at beowulf-underground.org
> To: 'pvfs-developers at www.beowulf-underground.org'
> Sent: 7/16/04 12:16 PM
> Subject: [PVFS-developers] pvfs-lib cleanup/sanity checking
> 
> Hey guys,
> 
> I came across some cases where the fdesc structure has pointers to
> memory
> that are not being explicitly freed in iodcomm.c and mgrcomm.c.  The
> fn_p
> pointer was also not being initialized to NULL in all cases.  
> 
> It turns out that there is an overlying semantic that reserved entries
> in
> pfds[] should not have those members.  That being said, this patch is
> really
> more code clean up and sanity checking than anything else.  It ensures,
> unless there is a case I missed, that fdesc's are always memset to zero
> and
> that part_p and fn_p are always freed if they are non-null.  Also, if a
> reserved fd has either of these fields, an error message is printed out.
> It
> seems that this would be helpful in catching bugs in this code if there
> were
> future changes that broke this semantic.
> 
> Thoughts?
> 
> Don
> _______________________________________________
> PVFS-developers mailing list
> PVFS-developers at www.beowulf-underground.org
> http://www.beowulf-underground.org/mailman/listinfo/pvfs-developers
> 
> 


More information about the PVFS-developers mailing list