On Wed, 2014-10-08 at 15:46 +0300, Dan Carpenter wrote:
> On Wed, Oct 08, 2014 at 05:26:11AM -0700, Joe Perches wrote:
> > On Wed, 2014-10-08 at 13:40 +0300, Dan Carpenter wrote:
> > > The return from myid() isn't aligned correctly for ether_addr_copy().
> >
> > Hey Dan.
> >
> > Actual evidence showing ether_addr_copy conversions
> > may not always be wise.
> >
> > How did you find them?
>
> I was just trying to see how common these kinds of bugs are. It didn't
> take long to find, but my impression is that they are rare and I got
> lucky. These kinds of bugs are tricky to find and we don't have any
> tools for it.
As far as I know, that's true too.
Jes, was the mac_addr field in this struct
ever __aligned(2)?
struct eeprom_priv {
u8 bautoload_fail_flag;
u8 bloadfile_fail_flag;
u8 bloadmac_fail_flag;
/* u8 bempty; */
/* u8 sys_config; */
u8 mac_addr[6]; /* PermanentAddress */
...
}
As far as I can tell from git history, it was
that way at the first check-in.
Dan, did you also look for any other alignment
defects in uses of any is_<foo>_ether_addr calls?
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
