Re: [PATCH v3 1/2] mtd: spi-nor: add support to non-uniform SFDP SPI NOR flash memories

[Tudor Ambarus <tudor.ambarus@microchip.com>]

Hi, Yogesh,

On 10/17/2018 10:46 AM, Yogesh Narayan Gaur wrote:
> Hi Boris,
> 
>> -----Original Message-----
>> From: Boris Brezillon [mailto:boris.brezillon@bootlin.com]
>> Sent: Wednesday, October 17, 2018 1:00 PM
>> To: Yogesh Narayan Gaur <yogeshnarayan.gaur@nxp.com>
>> Cc: Cyrille Pitchen <cyrille.pitchen@wedev4u.fr>; Tudor Ambarus
>> <tudor.ambarus@microchip.com>; marek.vasut@gmail.com;
>> dwmw2@infradead.org; computersforpeace@gmail.com; richard@nod.at;
>> linux-kernel@vger.kernel.org; nicolas.ferre@microchip.com;
>> cyrille.pitchen@microchip.com; linux-mtd@lists.infradead.org; linux-arm-
>> kernel@lists.infradead.org; Cristian.Birsan@microchip.com
>> Subject: Re: [PATCH v3 1/2] mtd: spi-nor: add support to non-uniform SFDP SPI
>> NOR flash memories
>>
>> On Wed, 17 Oct 2018 09:10:45 +0200
>> Boris Brezillon <boris.brezillon@bootlin.com> wrote:
>>
>>> On Wed, 17 Oct 2018 09:07:24 +0200
>>> Boris Brezillon <boris.brezillon@bootlin.com> wrote:
>>>
>>>> On Wed, 17 Oct 2018 02:07:43 +0000
>>>> Yogesh Narayan Gaur <yogeshnarayan.gaur@nxp.com> wrote:
>>>>
>>>>>>
>>>>> Actually there is no entry of s25fs512s in current spi-nor.c file.
>>>>> For my connected flash part, jedec ID read points to s25fl512s. I
>>>>> have asked my board team to confirm the name of exact connected
>>>>> flash part. When I check the data sheet of s25fs512s, it also
>>>>> points to the same Jedec ID information. { "s25fl512s",
>>>>> INFO(0x010220, 0x4d00, 256
>>>>> * 1024, 256, ....}
>>>>>
>>>>> But as stated earlier, if I skip reading SFDP or read using 1-1-1
>>>>> protocol then read are always correct. For 1-4-4 protocol read are
>>>>> wrong and on further debugging found that Read code of 0x6C is
>>>>> being send as opcode instead of 0xEC.
>>>>>
>>>>> If I revert this patch, reads are working fine.
>>>>
>>>> Can you try with the following patch?
>>>>
>>>
>>> Hm, nevermind. The problem is actually not related to 4B vs non-4B
>>> mode but 1-1-4 vs 1-4-4 modes.
> Yes, that's only I have stated in my first mail that instead of 1-4-4 mode read opcode is being sent for 1-1-4 mode.
>>>
>>
>> Can you try with this patch applied?
>>
> With suggested patch, read for protocol 1-4-4 working correctly.
> 
> 	[    1.625360] m25p80 spi0.0: found s25fl512s, expected m25p80                                                                   
> 	[    1.631094] m25p80 spi0.0: failed to parse SMPT (err = -22)                                                                   
> 	[    1.636661] 261 8c4c780 opcode(read:eb, pp:2, erase:d8)                                                                       
> 	[    1.641878] 266 8c4c780 opcode(read:ec, pp:12, erase:dc)                                                                      
> 	[    1.647200] m25p80 spi0.0: s25fl512s (65536 Kbytes)         
> 
> Without this patch, param_headers are getting freed and restoring previous erase map i.e. opcode related to 1-1-4 protocol.
> 

Can you add some prints in spi_nor_parse_smpt() to isolate what's failing? We
should understand whether it's something wrong in spi_nor_parse_smpt() or the
s25fs512s smpt table does not respect the standard.

Thanks,
ta

> 
>> --->8---
>> diff --git a/drivers/mtd/spi-nor/spi-nor.c b/drivers/mtd/spi-nor/spi-nor.c index
>> 9407ca5f9443..cf33d834698c 100644
>> --- a/drivers/mtd/spi-nor/spi-nor.c
>> +++ b/drivers/mtd/spi-nor/spi-nor.c
>> @@ -3132,6 +3132,17 @@ static int spi_nor_parse_sfdp(struct spi_nor *nor,
>>                 switch (SFDP_PARAM_HEADER_ID(param_header)) {
>>                 case SFDP_SECTOR_MAP_ID:
>>                         err = spi_nor_parse_smpt(nor, param_header);
>> +                       if (err) {
>> +                               dev_warn(dev,
>> +                                        "failed to parse SMPT (err = %d)\n",
>> +                                        err);
>> +                               /*
>> +                                * SMPT parsing is optional, let's not drop
>> +                                * all information we extracted so far just
>> +                                * because it failed.
>> +                                */
>> +                               err = 0;
>> +                       }
>>                         break;
>>
>>                 default:
>