[PATCH 1/3] [MTD] Flex-OneNAND support

Adrian Hunter adrian.hunter at nokia.com
Wed Mar 4 06:41:59 EST 2009


Andrew Morton wrote:
>> ...
>>
>> +static loff_t flexonenand_get_addr(struct onenand_chip *this, int block)
>> +{
>> +	loff_t ofs = 0;
>> +	int die = 0, boundary;
>> +
>> +	if (ONENAND_IS_DDP(this) && block >= this->density_mask) {
>> +		block -= this->density_mask;
>> +		die = 1;
>> +		ofs = this->diesize[0];
>> +	}
>> +
>> +	boundary = this->boundary[die];
>> +	ofs += block << (this->erase_shift - 1);
>> +	if (block > (boundary + 1))
>> +		ofs += (block - boundary - 1) << (this->erase_shift - 1);
> 
> Both `block' and `boundary' have 32-bit types.  Are you sure that the
> left-shift cannot overflow?

Only very recently has MTD supported sizes greater than 32 bits internally
for any type of flash.  The external APIs (ioctls) are still 32-bit based.

For this driver, supporting sizes over 32-bits is a separate issue - and
may never be needed.

>> +	return mtd->ecc_stats.corrected - stats.corrected ? -EUCLEAN : 0;
>> +}
> 
> I wonder what the heck EUCLEAN was invented for and whether MTD's
> extensive use of it is appropriate.

UBI uses it to detect bit-flips so that data can be moved before it
can no longer be read.  So it is pretty much essential for flash
memories.




More information about the linux-mtd mailing list