Re: [PATCH v3 2/3] mtd: spi-nor: add 4bit block protection support

[Michael Walle <michael@walle.cc>]

Hi Tudor,

Am 2020-02-10 10:47, schrieb Tudor.Ambarus@microchip.com:
> Hi, Michael,
> 
> On Monday, February 10, 2020 10:33:41 AM EET Michael Walle wrote:
> 
> cut
> 
>> > On Monday, February 3, 2020 3:56:58 PM EET Vignesh Raghavendra wrote:
>> >> >>>>>>>>> /*
>> >> >>>>>>>>> * Need smallest pow such that:
>> >> >>>>>>>>> *
>> >> >>>>>>>>> @@ -1908,7 +1972,17 @@ static int stm_lock(struct
>> >> >>>>>>>>> spi_nor
>> >> >>>>>>>>> *nor,
>> >> >>>>>>>>> loff_t ofs, uint64_t len)
>> >> >>>>>>>>> *   pow = ceil(log2(size / len)) = log2(size)
>> >> >>>>>>>>> -
>> >> >>>>>>>>> floor(log2(len))
>> >> >>>>>>>>> */
>> >> >>>>>>>>> pow = ilog2(mtd->size) - ilog2(lock_len);
>> >> >>>>>>>>> -     val = mask - (pow << SR_BP_SHIFT);
>> >> >>>>>>>>> +
>> >> >>>>>>>>> +     if (nor->flags & SNOR_F_HAS_SR_BP3) {
>> >> >>>>>>>>> +             val = ilog2(nor->n_sectors) + 1 - pow;
>> >> >>>>>>>>
>> >> >>>>>>>> Why do you use a new calculation here? As far as I can
>> >> >>>>>>>> see,
>> >> >>>>>>>> the
>> >> >>>>>>>> method is
>> >> >>>>>>>> the same except that is has one bit more. That also
>> >> >>>>>>>> raises
>> >> >>>>>>>> the
>> >> >>>>>>>> question why
>> >> >>>>>>>> n_sectors is now needed?
>> >>
>> >> Flash devices have variable sector size, 64KB, 128KB or 256KB... While
>> >> mapping of number of sectors locked to BP bits is dependent on rules 1
>> >> to 3 you mentioned below, the size or area of flash protected depends
>> >> on
>> >> sector size.
>> >>
>> >> So, the current formula in spi-nor.c (ignoring TB and other
>> >> boilerplate):
>> >>
>> >> pow = ilog2(mtd->size) - ilog2(lock_len);
>> >> val = mask - (pow << shift);
>> >>
>> >> This works only for devices with 64KB sector size as 8MB flash with
>> >> 64KB
>> >> sector size would have 128 sectors (BP0-2 => 0b111 => 2^7).
>> >>
>> >> A more generic formula would be:
>> >>
>> >> Find n where 2^(n - 1) = len/sector-size
>> >> OR 2^ (n - 1) = len * n_sectors / mtd->size
>> >>
>> >> Which solves to:
>> >>
>> >> pow = ilog2(mtd->size) - ilog2(lock_len);
>> >> val = ilog2(nor->n_sectors) + 1 - pow;
>> >
>> > The current mainline locking support is limited. Michael spotted a good
>> > improvement, but I think there are still others that we should
>> > consider.
>> 
>> Sure, as I said my patch was just to show, that there is an underlying
>> problem
>> and that we should not take the 4th BP bit to differentiate between 
>> the
>> two
>> different formulas.
> 
> Right, this is the goal.
> 
> Let me try to extend the description of the proposal.
> 
>> 
>> > We should use a single formula, for all the BP cases. How about the
>> > following:
>> >
>> > bp_slots_available = (bp_mask >> shift) + 1 - 2;
> 
> This formula is derived from Michael's patch.
> 
> A slot (to me) is a horizontal line in the Memory protection table. 
> Maybe we
> can find a better/standardized name for this.
> 
> So for BP0-2, bp_slots_available = 6, and for BP0-3, bp_slots_available 
> = 14.
> Notice that I stripped the two special cases: lock none and lock all.
> 
>> > bp_slots_needed = ilog2(nor->info->n_sectors);
> 
> With bp_slots_needed I tried to describe how many slots are needed if 
> the
> protected density for the first slot is at minimum (sector size).
> 
>> >
>> > if (bp_slots_needed > bp_slots_available) {
>> >
>> >       bp_slot_count = bp_slots_available;
>> >       bp_min_slot_size = nor->info->n_sectors <<
>> >
>> >               (bp_slots_needed - bp_slots_available);
>> 
>> mhh, what is the unit of bp_min_slot_size? bytes or sectors? I guess 
>> it
>> should
> 
> It's bytes. Take a look at W25Q128JV. The sector size for this flash is
> 64KByte. The flash has 256 sectors. For this specific case:
> 	bp_slots_available = 6;
> 	bp_slots_needed = 8;
> 
> The if condition is true, so
> 	bp_slot_count = 6;
> 	bp_min_slot_size = 64k << (8 - 6); //256k

But nor->info->n_sectors is not 64k, its 256. Do you mean sector_size 
(like in
my example below? Then we are on the same page ;)


> which is exactly the protected density for the first slot. The 
> protected
> densities of the other slots can be computed by multiplying with powers 
> of 2.
> 
>> be bytes, eg for a 8MiB flash it would be 128kiB and for a 16MiB flash
>> it would
>> be 256kiB (if there are 3 BP bits).
>> 
>> > } else {
>> >
>> >       bp_slot_count = bp_slots_needed;
>> >       bp_min_slot_size = mtd->size >> bp_block_count;
> 
> typo: s/bp_block_count/bp_slot_count
>> 
>> this is a complicated way of saying its the size of one sector, isn't
>> it?
>> can't we use nor->info->sector_size here? Eg.
>> 
>> if (bp_slots_needed > bp_slots_available) {
>>         bp_slot_count = bp_slots_available;
>>         bp_min_slot_size = nor->info->sector_size <<
>>                 (bp_slots_needed - bp_slots_available);
>> } else {
>>         bp_slot_count = bp_slots_needed;
>>         bp_min_slot_size = nor->info->sector_size;
>> }
> 
> you're right, we're in the else case, where the assumption that the 
> minimum
> protected density is sector size is true, we can use directly 
> nor->info-
>> sector_size.
> 
>> 
>> > }
>> >
>> > When both can_be_bottom and can_be_top are true, we prefer the top
>> > protection,
>> > which is incorrect/buggy/sub-optimal. If the received offset is not
>> > aligned to
>> > one of the start addresses of the bp slots, then we should up/down
>> > align the
>> > offset to the closest bp slot, depending on TB and which (top or
>> > bottom) fits
>> > better. Based on the updated offset and length we can compute the lock
>> > range,
>> > and after that:
>> >
>> > n = ilog2(bp_lock_range/bp_min_slot_size) + 1;
>> > val = mask - (n << shift);
>> 
>> btw. we should catch the two special cases:
>>   - lock none -> 0 (that was already the case)
>>   - lock all -> all BP bits
>> 
>> The latter is important if "bp_slots_needed < bp_slots_available"
>> because there
>> are multiple settings for protect all. Most flashes will define any
>> remaining
>> setting for "protect all", but I've also seen flashes where the
>> in-between ones
>> were undefined (not mentioned) and only the "all bit set" was protect
>> all.
> 
> This case is addressed by using bp_slot_count and bp_slots_available. 
> We're in
> the else case from above. From bp_slot_count up to the 
> bp_slots_available,
> those slots are "protect all".

-michael

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/