[PATCH] mtd: spi-nor: boya: add support for boya by25q128as

Tue Apr 6 06:35:14 BST 2021

On 2/13/21 7:10 PM, George Brooke wrote:
> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
> 
> Hello Tudor,

Hi, George,

Sorry for the long delay :(. I added in To: Chris and Andreas, they
encounter a similar problem with other vendors.

> 
> Tudor.Ambarus at microchip.com writes:
> 
>> Hi, George,
>>
>> On 2/7/21 2:30 PM, George Brooke wrote:
>>> EXTERNAL EMAIL: Do not click links or open attachments unless
>>> you know the content is safe
>>>
>>> Adds support for the Boya Microelectronics BY25Q128AS 128 Mbit
>>> flash.
>>> I tested this on the Creality WB-01 embedded device which uses
>>> this,
>>> although that was with OpenWrt which is still using 5.4 so I
>>> had to
>>> do a bit of porting work. Don't see how that would make much of
>>> a
>>> difference though.
>>>
>>> Signed-off-by: George Brooke <figgyc at figgyc.uk>
>>> ---
>>>  drivers/mtd/spi-nor/Makefile |  1 +
>>>  drivers/mtd/spi-nor/boya.c   | 23 +++++++++++++++++++++++
>>>  drivers/mtd/spi-nor/core.c   |  1 +
>>>  drivers/mtd/spi-nor/core.h   |  1 +
>>>  4 files changed, 26 insertions(+)
>>>  create mode 100644 drivers/mtd/spi-nor/boya.c
>>>
>>> diff --git a/drivers/mtd/spi-nor/Makefile
>>> b/drivers/mtd/spi-nor/Makefile
>>> index 653923896205..7d1551fbfbaa 100644
>>> --- a/drivers/mtd/spi-nor/Makefile
>>> +++ b/drivers/mtd/spi-nor/Makefile
>>> @@ -2,6 +2,7 @@
>>>
>>>  spi-nor-objs                   := core.o sfdp.o
>>>  spi-nor-objs                   += atmel.o
>>> +spi-nor-objs                   += boya.o
>>>  spi-nor-objs                   += catalyst.o
>>>  spi-nor-objs                   += eon.o
>>>  spi-nor-objs                   += esmt.o
>>> diff --git a/drivers/mtd/spi-nor/boya.c
>>> b/drivers/mtd/spi-nor/boya.c
>>> new file mode 100644
>>> index 000000000000..014b0087048a
>>> --- /dev/null
>>> +++ b/drivers/mtd/spi-nor/boya.c
>>> @@ -0,0 +1,23 @@
>>> +// SPDX-License-Identifier: GPL-2.0
>>> +/*
>>> + * Copyright (C) 2005, Intec Automation Inc.
>>> + * Copyright (C) 2014, Freescale Semiconductor, Inc.
>>> + */
>>> +
>>> +#include <linux/mtd/spi-nor.h>
>>> +
>>> +#include "core.h"
>>> +
>>> +static const struct flash_info boya_parts[] = {
>>> +       /* Boya */
>>> +       { "by25q128as", INFO(0x684018, 0, 64 * 1024, 256,
>>
>> The manufacturer’s identification code is defined by one or more
>> eight (8) bit fields each consisting of seven (7) data bits plus
>> one (1)
>> odd parity bit. It is a single field limiting the possible
>> number of
>> vendors to 126. To expand the maximum number of identification
>> codes a
>> continuation scheme has been defined.
>>
>> According to JEP106BA, the manufacturer ID for Boya should be
>> preceded by
>> eight continuation codes. So I would expect the manufacturer ID
>> for this
>> flash to be: 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
>> 0x68.
>>
>> Without the continuation codes, we will have collisions between
>> manufacturer IDs, Convex Computer being an example.
>>
>> I see that the datasheet [1] for this flash doesn't specify
>> anything
>> about the continuation codes, so I suspect that Boya just got it
>> wrong.
> 
> It appears you are right. I thought it would be the best idea to
> actually
> interact with the flash chip and read its responses, so I found a
> tool
> called spincl [2] to send some commands to it with a Raspberry Pi:
> 
> $ spincl -ib -m0 -c0 -s0 -p0 6 0x90
> 0x00 0x00 0x00 0x00 0x68 0x17

I wasn't aware of the 0x90 command, it's not in the
include/linux/mtd/spi-nor.h either. Do you know what 0x17 means?

> $ spincl -ib -m0 -c0 -s0 -p0 4 0x9F
> 0x00 0x68 0x40 0x18
> and indeed there doesn't seem to be any continuation codes.
> 
>> We'll have to check other datasheets from Boya and see if they
>> got
>> their manufacturer ID wrong for all their flashes. We'll have to
>> add
> 
> The BY25Q64AS [3] appears to be described similarly.
> 
>> some fixup mechanism for the manufacturers ID handling, in order
>> to
>> avoid collisions with other manufacturers IDs.
> 
> I looked into this a bit more, and what I'm realising is that I'm
> not sure if
> there even is any mechanism to deliver the continuation codes
> within the
> base SPI-NOR standard? Take esmt.c: the f25l32qa has a device id
> 0x8c4116.
> JEP106BA attribytes 8c to Monolithic Memories in the first bank,
> while
> Elite Flash Storage (presumably an alias of ESMT) should be
> identifying
> as 0x7f, 0x7f, 0x7f, 0x8c. Its datasheet [4] appears to be equally
> sparse on
> detail. To my untrained eye, this seems to be the exact same
> situation we
> find ourselves in here. (You probably know a lot more about this
> then I do -
> if I'm wrong do point it out!)

I don't, but on a quick search I see that Elite Flash Storage may be
an alias for ESMT, so probably you're right. I wasn't aware of the
esmt problem, thanks for pointing it out.

> 
> That said I can't seem to find any formalised definition of what
> the 0x90
> "manufacturer and device ID" command is actually supposed to do in
> the case of
> a manufacturer ID that isn't in the first bank. Likewise with 0x9f
> "JEDEC ID".

I don't know if there's a standardized definition. Maybe we can shuffle
through datasheets and check what manufacturers are saying about this.

> Do you know of any flashes made by companies not in the first bank
> that
> identify themselves correctly?

No, I've worked only with flashes that have their ID in the first bank.

> 
> As far as I can tell the only proper method available is in SFDP,
> where the
> Parameter ID of a vendor specific table would be the bank number
> in the MSB
> and the actual manufacturer code in the LSB. However, this is a
> very
> over-engineered solution, wouldn't work on devices with no vendor
> specific
> tables, and in this case, I couldn't even get the SFDP table to
> read out of
> the by25q128as at all:
> $ spincl -ib -m0 -c0 -s0 -p0 32 0x5A
> 0x00 0x00 0x00 0x00 0x00 0xff 0xff 0xff 0xff 0xff 0xff 0xff 0xff
> 0xff 0xff ...
> (This may very well be a peculiarity of the way I'm accessing it,
> but in any
> case I don't think this fix would be a very good idea.)

yeah, maybe the tool is broken, because I see that the SFDP table
is defined in the datasheet. The manufacturer specific table
identification looks broken too. MSB should have been 0x08 and LSB
0x68, whereas they have for MSB 0xff and LSB 0x0b? We can hack
the sfdp read to dump the tables. Or use Michael's sfdp dump sysfs
patches:

https://patchwork.ozlabs.org/project/linux-mtd/list/?series=235475
> 
> I'm not sure what would actually make a good fixup in this
> scenario. We'd need
> something that could differentiate a Boya chip from any other
> hypothetical
> 0x68 manufacturer flash.
> Maybe even the missing SFDP output would make a good detection
> routine, but
> that would need testing with a better SPI inspection method and
> across multiple
> flash chips because I doubt that's intentional - surely if SFDP
> didn't work it
> would just not be described in the datasheet. Probably better to
> think of
> something other than that though.
> 

The first collision between a flash from the first bank and any other
flashes from the rest of the banks can be avoided if we move all the
manufacturers with continuation codes at the end of the manufacturers
array, so that at the flash detection code they are treated as last.

Collision between flashes with continuation codes is more difficult
to solve and would require differentiation at run-time. Will have to
check jesd216 to see if there's other unique method to identify a flash.
Can you check it?

Cheers,
ta

> Thanks,
> figgyc
> 
> [2] https://github.com/CrosseyeJack/spincl
> [3]
> https://datasheet.lcsc.com/szlcsc/1904091402_BOYAMICRO-BY25Q64ASSIG_C383793.pdf
> [4]
> https://www.esmt.com.tw/upload/pdf/ESMT/datasheets/F25L32QA.pdf
> 
>>
>> Cheers,
>> ta
>>
>> [1]
>> https://datasheet.lcsc.com/szlcsc/1904091402_BOYAMICRO-BY25Q128ASSIG_C383794.pdf
>>
>>> +                       SECT_4K | SPI_NOR_DUAL_READ |
>>> SPI_NOR_QUAD_READ |
>>> +                       SPI_NOR_HAS_LOCK | SPI_NOR_HAS_TB)
>>> +       },
>>> +};
>>> +
>>> +const struct spi_nor_manufacturer spi_nor_boya = {
>>> +       .name = "boya",
>>> +       .parts = boya_parts,
>>> +       .nparts = ARRAY_SIZE(boya_parts),
>>> +};
>>> diff --git a/drivers/mtd/spi-nor/core.c
>>> b/drivers/mtd/spi-nor/core.c
>>> index 20df44b753da..4d0d003e9c3f 100644
>>> --- a/drivers/mtd/spi-nor/core.c
>>> +++ b/drivers/mtd/spi-nor/core.c
>>> @@ -2160,6 +2160,7 @@ int spi_nor_sr2_bit7_quad_enable(struct
>>> spi_nor *nor)
>>>
>>>  static const struct spi_nor_manufacturer *manufacturers[] = {
>>>         &spi_nor_atmel,
>>> +       &spi_nor_boya,
>>>         &spi_nor_catalyst,
>>>         &spi_nor_eon,
>>>         &spi_nor_esmt,
>>> diff --git a/drivers/mtd/spi-nor/core.h
>>> b/drivers/mtd/spi-nor/core.h
>>> index d631ee299de3..d5ed5217228b 100644
>>> --- a/drivers/mtd/spi-nor/core.h
>>> +++ b/drivers/mtd/spi-nor/core.h
>>> @@ -409,6 +409,7 @@ struct spi_nor_manufacturer {
>>>
>>>  /* Manufacturer drivers. */
>>>  extern const struct spi_nor_manufacturer spi_nor_atmel;
>>> +extern const struct spi_nor_manufacturer spi_nor_boya;
>>>  extern const struct spi_nor_manufacturer spi_nor_catalyst;
>>>  extern const struct spi_nor_manufacturer spi_nor_eon;
>>>  extern const struct spi_nor_manufacturer spi_nor_esmt;
>>> -- 
>>> 2.30.0
>>>