Usage of MTD_UADDR_UNNECESSARY broken?

[Erwin Authried]

Am Fre, den 12.11.2004 schrieb Alexander Hoffmann um 16:15:
> Hi Thayne,
> 
> sorry for the delayed response.
> 
> Thayne Harbaugh wrote:
> 
> >On Mon, 2004-11-08 at 13:55 +0100, Alexander Hoffmann wrote:
> >  
> >
> >>Ben Dooks wrote:
> >>
> >>    
> >>
> >>>On Mon, Nov 08, 2004 at 12:54:16PM +0100, Alexander Hoffmann wrote:
> >>>      
> >>>
> >>>>Hi everyone,
> >>>>
> >>>>can anybody please explain me the exact difference between 
> >>>>MTD_UADDR_DONT_CARE and MTD_UADDR_UNNECESSARY .
> >>>>Because if I use MTD_UADDR_UNNECESSARY an not existing field in the 
> >>>>unlock_addrs array is beeing referenced
> >>>>(/drivers/mtd/chips/jedec_probe.c, function cfi_jedec_setup, line 1740):
> >>>>
> >>>>/* Mask out address bits which are smaller than the device type */
> >>>>mask = ~(p_cfi->device_type-1);
> >>>>p_cfi->addr_unlock1 = unlock_addrs[uaddr].addr1 & mask;
> >>>>p_cfi->addr_unlock2 = unlock_addrs[uaddr].addr2 & mask;
> >>>>        
> >>>>
> >>>hmm, thought this masking had been eliminated in later copies of
> >>>the mtd code?
> >>>      
> >>>
> >
> >Yes, the masking has been eliminated, but someone left the comment in
> >(doh!).
> >
> >  
> >
> >>Ok, you are right. But this doesn't change the fact that
> >>
> >>unlock_addrs[uaddr].addr1
> >>
> >>refers to an nonexisting field in the unlock_addrs array.
> >>    
> >>
> >
> >I don't see how the code you that you described is being reached.  It
> >looks like the start of jedec_probe_chip() checks for UNNECESSARY and
> >returns 0 (although I would expect 1) and so cfi_jedec_setup() should
> >never be called with UNNECESSARY (even for subsequent chips).
> >
> I am working with the cdb89712 development board from cirrus. This board 
> has an "Intel 28F320B3B" chip
> (device_id  =  0x8897). Apparently, jedec_probe() finds 
> MTD_UADDR_0x0555_0x02AA.
> for this chip, while the jedec_table[] specifies it as 
> MTD_UADDR_UNNECESSARY.
> Since the probed unlock type is overidden by the static one, the code 
> _does_ reach
> jedec_setup().
> 
> What I haven't really understood is this: if the code refuses chips of  
> type UNNECESSARY
> (the return code 0 from jedec_probe() is an error), then why are so many 
> chips declared
> as UNECESSARY in jedec_table[]?

MTD_UADDR_UNNECESSARY is used to specify that the chip does not need any
unlock sequence at all. The "return 0" statement near the  top of
jedec_probe_chip is executed if all possible unlock sequences have been
tried without finding a match in jedec table, thus it indicates an
error. It's correct that your chip finds a match with
MTD_UADDR_0x0555_0x02aa, because that's the first try and the chip
doesn't care at all about the unlock sequence. 

Regards,
Erwin