[U-Boot-Users] [PATCH] cmd_nand: Fix warning: dereferencing type-punned pointer will break strict-aliasing rules

[Jean-Christophe PLAGNIOL-VILLARD]

On 16:39 Wed 02 Apr     , Wolfgang Denk wrote:
> In message <1207136270-19315-1-git-send-email-plagnioj at jcrosoft.com> you wrote:
> > cmd_nand.c:353: warning: dereferencing type-punned pointer will break strict-aliasing rules
> > cmd_nand.c:356: warning: dereferencing type-punned pointer will break strict-aliasing rules
> > 
> > Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj at jcrosoft.com>
> > 
> > diff --git a/common/cmd_nand.c b/common/cmd_nand.c
> > index 7b1f830..2906aed 100644
> > --- a/common/cmd_nand.c
> > +++ b/common/cmd_nand.c
> > @@ -84,7 +84,7 @@ static int nand_dump(nand_info_t *nand, ulong off)
> >  
> >  /* ------------------------------------------------------------------------- */
> >  
> > -static inline int str2long(char *p, ulong *num)
> > +static inline int str2long(char *p, loff_t *num)
> 
> I think this is plain wrong. It may silence the warning, but it is
> wrong.
> 
> The type is the result of calling simple_strtoul(), and this function
> returns, um, and unsigned long. And not an offset type.
I agree
> 
> > -arg_off_size(int argc, char *argv[], nand_info_t *nand, ulong *off, ulong *size)
> > +arg_off_size(int argc, char *argv[], nand_info_t *nand, loff_t *off, size_t *size)
> 
> No. That's all wrong.
I disagree, the true type of the offest is loff_t and size is size_t in
the linux/mtd.

Maybe we need to do the cast to ulong in this function.

Best Regards,
J.