Re: [Valgrind-users] question about the seriousness of an error

[Edward M. <Edw...@ce...>]

On Friday 30 April 2004 19:26, David Hawkins wrote:
> > His response was that it looked like there were two char[3]
> > arrays, but the last element of the array was unitialised,
> > causing the valgrind error. He believed that compilers
> > should normally initialize a character array to
> > binary zero and so it shouldn't be a problem.
>
> In general if the array is 'on the stack', i.e. a local
> variable within the scope of a function call, then the
> variables are NOT initialized.
>
> Go ahead and take a look at the assembly output, if you
> make sure you haven't omitted frame pointers, then you'll
> see the compiler setup the frame pointer, and then
> move the frame pointer by an amount equivalent to the
> number of bytes of local variables. Variables in
> the source are referenced then with respect to the
> frame pointer. There is usually NO initialization
> of these variables.

Wow. Thanks for the prompt response. On monday I will indeed have a go at 
reading the assembly output.

Can I just ask though whether if valgrind reporting this means that it is 
definitely uninitialized? I mean, if this error crops up again and again 
(which it probably will) if I *know* that valgrind reports it correctly then 
I can just tell people that they have a problem they must fix, and I don't 
need to bother with proving it by assembly.

Cheers (and thanks again for replying so fast and so comprehensively)

Ed