Quoting Nuutti Kotivuori <naked@iki.fi>:
> Karl Fogel wrote:
> > philip@tigris.org writes:
> >> Fix an unitialized memory read identified by valgrind.
> >
> > Hmmm. That's interesting -- valgrind doesn't know enough to know
> > that that code was correct because of the semantics of the
> > conditional's body.
> >
> > No objection to switching the order, if only for clarity, just
> > pointing out that there wasn't really a bug here :-).
>
> As far as I know, valgrind does not check semantics or analyze code in
> any way. It instruments the x86 assembler so to check against these
> things.
>
> So if it mentioned an uninitialized memory read - the read actually
> happened. That is, unitialized memory was read by the assembler code.
>
> Atleast, this is how I think valgrind works - please correct me if I'm
> wrong so I don't confuse more people :)
>
That makes sense. I beleive the code in question was something like:
I don't snippet with me here, but I think if there was an error, it would be
possible that the field wasn't ever initialized.
Kevin Pilch-Bisson.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Aug 28 23:50:03 2002