OSLib bug fixes (various)

[Stewart Brodie]

In message <Marcel-1.53-0112123148-72b#m5i at DEllis.cam.pace.co.uk>
          Daniel Ellis <Daniel.Ellis at pace.co.uk> wrote:

> On Fri 12 Jan, Tony van der Hoff wrote:
> > 
> > >+ Fixed null pointer dereferences in IndexHelp, BindHelp and DefMod -cstrong
> > The SCL, at least, treats null pointers to a string argument to *printf
> > the same as a pointer to a null string. However, I accept that this
> > behaviour is not defined, so I guess your proposal makes it more
> > portable.
> 
> I'm not sure what you mean there.  There's no check in the SCL for the
> argument to printf being null.  printf(0) gives oflaofla... and if zero page
> (reading) is protected will cause a data abort.
> 
> > >+ os_int_on() and os_int_off() added as these can be called from user
> > >   and 32 bit modes.
> > Really? With valid APCS - this was the reason they weren't included in
> > the first place.
> 
> These SWIs don't affect any registers, so why should APCS be of any
> significance?

The flags - OSLib still uses APCS-3/26 for desktop RISC OS machines.  They
can be used in APCS-3/32 programs though.

> > >+ DefMod checks SWI numbers with |xos_swi_number_from_string|.
> 
> Surely this will error if the module supplying the SWI isn't resident?

Yes - plus it will fail for everything in a cross-compilation environment.

-- 
Stewart Brodie, Senior Software Engineer    (Views expressed are my own and
Pace Micro Technology PLC                   not those of my employer)
645 Newmarket Road
Cambridge, CB5 8PB, United Kingdom         WWW: http://www.pacemicro.com/