[SCM] Samba Shared Repository - branch master updated - release-4-0-0alpha8-351-g7be1d72

Jeremy Allison jra at samba.org
Wed Jul 15 09:47:19 MDT 2009


On Wed, Jul 15, 2009 at 07:48:56AM +0200, Volker Lendecke wrote:
> On Tue, Jul 14, 2009 at 06:45:02PM -0500, Jeremy Allison wrote:
> > The branch, master has been updated
> >        via  7be1d727a31b34debbcf8faa1e0bea911112d145 (commit)
> >       from  e84d5bb02d113ae54172755ea284e7f22b677a7b (commit)
> > 
> > http://gitweb.samba.org/?p=samba.git;a=shortlog;h=master
> > 
> > 
> > - Log -----------------------------------------------------------------
> > commit 7be1d727a31b34debbcf8faa1e0bea911112d145
> > Author: Jeremy Allison <jra at samba.org>
> > Date:   Tue Jul 14 16:42:21 2009 -0700
> > 
> >     When tallocing a memory block for the state in a tevent_req struct,
> >     ensure it's zeroed out. Vl & Metze please check.
> >     Jeremy.
> 
> Why that change? This certainly does not do harm, but we've
> deliberately not made this, the caller should decide it. For
> a large or very large state that contains for example
> transfer buffers, this would be a waste.

Ok, that's why I wanted you to look at this. In
the usage I've made of this in libsmb/clifile.c
the allocations here are always something containing
vwv[x] and data[y]. It's easy to miss initializing
these which means uninitialized data sent over the
wire. Having this come back initialized to zero is
very convenient here.

If this is being used to large transfer buffers
in other places then I'll revert and re-add the
memset() calls in clifile.c. Let me know.

Jeremy.


More information about the samba-technical mailing list