Re: problem in usage of grub_errno...

[Joel Buckley]

There is a second bug though...

A possible stack memory-leak is possible.  Replace the
first line with the following to avoid the memory leak:

   if (grub_errno != GRUB_ERR_NONE) {
      grub_error_push();
   }

Otherwise, a useless push is done.

leoJ.


Vesa Jääskeläinen wrote:

> Tomáš Ebenlendr wrote:
>  
>
> > I think there is a 'bug' in the example. Or maybe I'm missing something.
> >    
>
> Nope, it was intentional. :)
>
>  
>
> > On 17 Prosinec 2005, 18:40, Vesa Jääskeläinen napsal(a):
> > ....
> >    
> >
> > > ---
> > > /* Save possible old error message.  */
> > > grub_error_push ();
> > >
> > > /* Do your stuff here.  */
> > > call_possibly_failing_function ();
> > >
> > > if (grub_errno != GRUB_ERR_NONE)
> > >  {
> > >    /* Inform rest of the code that there is error (grub_errno
> > >       is set). */
> > >    return;
> > >  }
> > >
> > > /* Restore old error state by popping previous item from stack. */
> > > grub_error_pop ();
> > > ---
> > >      
> > ....
> >
> > When grub_errno != GRUB_ERR_NONE (branching into the 'if'), then
> > there is no grub_error_pop() in the example.
> > If the whole thing will be called in a loop, then unwanted
> > stack overflow can simply occur.
> >    
>
> Error stack is statically allocated and it is protected so that it
> cannot overflow. Error stack is defined as being static to file
> kern/err.c so there is no direct access possible to error stack.
>
> Error stack is emptied in grub_print_error function. (And this is called
> before prompt is printed to screen)
>
> Idea of the error stack is to record all error messages before they are
> printed.
>
> Or did I miss your idea with the "unwanted stack overflow" ?
>
> Thanks,
> Vesa Jääskeläinen
>
>
> _______________________________________________
> Grub-devel mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/grub-devel
>
>