[PATCH 1/2] mtd: spi-nor: spansion: Consider reserved bits in CFR5 register

[Dhruva Gole]

On 24/01/23 16:13, Pratyush Yadav wrote:
> On Mon, Jan 23 2023, Dhruva Gole wrote:
>
>> Hi Pratyush,
>>
>> On 23/01/23 20:07, Pratyush Yadav wrote:
>>> +Cc Dhruva
>> I had already reviewed this, but now I have locally applied the patches,
>> to linux master and built and tested - seems okay,
>>> Hi Tudor,
>>>
>>> On Tue, Jan 10 2023, Tudor Ambarus wrote:
>>>
>>>> CFR5[6] is reserved bit and must be always 1. Set it to comply with flash
>>>> requirements. While fixing SPINOR_REG_CYPRESS_CFR5V_OCT_DTR_{EN, DS}
>>>> definition, stop using magic numbers and describe the missing bit fields
>>>> in CFR5 register. This is useful for both readability and future possible
>>>> addition of Octal STR mode support.
>>>>
>>>> Fixes: c3266af101f2 ("mtd: spi-nor: spansion: add support for Cypress Semper flash")
>>>> Cc: stable at vger.kernel.org
>>>> Reported-by: Takahiro Kuwano <Takahiro.Kuwano at infineon.com>
>>>> Signed-off-by: Tudor Ambarus <tudor.ambarus at linaro.org>
>>>> ---
>>>>  drivers/mtd/spi-nor/spansion.c | 9 +++++++--
>>>>  1 file changed, 7 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/drivers/mtd/spi-nor/spansion.c b/drivers/mtd/spi-nor/spansion.c
>>>> index b621cdfd506f..07fe0f6fdfe3 100644
>>>> --- a/drivers/mtd/spi-nor/spansion.c
>>>> +++ b/drivers/mtd/spi-nor/spansion.c
>>>> @@ -21,8 +21,13 @@
>>>>  #define SPINOR_REG_CYPRESS_CFR3V               0x00800004
>>>>  #define SPINOR_REG_CYPRESS_CFR3V_PGSZ          BIT(4) /* Page size. */
>>>>  #define SPINOR_REG_CYPRESS_CFR5V               0x00800006
>>>> -#define SPINOR_REG_CYPRESS_CFR5V_OCT_DTR_EN    0x3
>>>> -#define SPINOR_REG_CYPRESS_CFR5V_OCT_DTR_DS    0
>>>> +#define SPINOR_REG_CYPRESS_CFR5_BIT6           BIT(6)
>>> Perhaps comment here that this bit is reserved. Otherwise it is not
>>> obvious what this does and why we are setting it without going through
>>> git-blame. No need for a re-roll, I think it is fine if you add this
>>> when applying.
>>>
>>>> +#define SPINOR_REG_CYPRESS_CFR5_DDR            BIT(1)
>>>> +#define SPINOR_REG_CYPRESS_CFR5_OPI            BIT(0)
>>>> +#define SPINOR_REG_CYPRESS_CFR5V_OCT_DTR_EN                            \
>>>> +       (SPINOR_REG_CYPRESS_CFR5_BIT6 | SPINOR_REG_CYPRESS_CFR5_DDR |   \
>>>> +        SPINOR_REG_CYPRESS_CFR5_OPI)
>>>> +#define SPINOR_REG_CYPRESS_CFR5V_OCT_DTR_DS    SPINOR_REG_CYPRESS_CFR5_BIT6
>>> I would say don't fix what isn't broken. But if you do, test it. Do you
>>> or Takahiro have a Cypress S28* flash to test this change on? If no,
>>> then perhaps Dhruva can help here since TI uses this flash on a bunch of
>>> their boards?
>>>
>>> The change looks fine to me with the above comment added, I just would
>>> like someone to test it once.
>> Tested OSPI_S_FUNC_DD_RW_ERASESIZE_UBIFS from ltp-ddt and test seemed to pass on my
>> AM625 SK EVM having an OSPI NOR S28HS512T Flash.
> Thanks.
>
> BTW, one interesting bit about this in case you didn't know already.
> Since this is playing with Octal DTR enable/disable, you might also want
> to double-check you are actually using Octal DTR mode. This can be done
> by looking at the SPI NOR debugfs entry (usually in
> /sys/kernel/debug/spi-nor). You can "cat params" and see what protocols
> are being used, and make sure 8D-8D-8D is being used.
Yes, I believe 8D-8D-8D is being used.
Here's the output:
...
root at am62xx-evm:~# cat /sys/kernel/debug/spi-nor/spi0.0/params
name            s28hs512t
id              34 5b 1a 0f 03 90
size            64.0 MiB
write size      16
page size       256
address nbytes  4
flags           4B_OPCODES | HAS_4BAIT | HAS_16BIT_SR | IO_MODE_EN_VOLATILE | SOFT_RESET

opcodes
 read           0xee
  dummy cycles  24
 erase          0xd8
 program        0x12
 8D extension   repeat

protocols
 read           8D-8D-8D
 write          8D-8D-8D
 register       8D-8D-8D

erase commands
 21 (4.00 KiB) [2]
 dc (256 KiB) [3]
 c7 (64.0 MiB)

sector map
 region (in hex)   | erase mask | flags
 ------------------+------------+----------
 00000000-0001ffff |     [  2 ] |
 00020000-0003ffff |     [   3] | overlaid
 00040000-03ffffff |     [   3] |
...

Also, the raw read speed is about 40 MBPS
The frequency we set from DT is 25 MHz,
assuming 1MB is transfered every cycle's rising and falling edge,
ie. 25*2 = 50 MBPS is ideal speed.

>
> AM625 SK _should_ be using 8D-8D-8D mode already, but I think it is
> useful if you know how to confirm this assumption :-)
>
>>> Reviewed-by: Pratyush Yadav <ptyadav at amazon.de>
>> For this series,
>>
>> Tested-by: Dhruva Gole <d-gole at ti.com>
>>
>>>>  #define SPINOR_OP_CYPRESS_RD_FAST              0xee
>>>>
>>>>  /* Cypress SPI NOR flash operations. */

-- 
Best regards,
Dhruva Gole
Texas Instruments Incorporated