bcm63xx kernel 5.10

[Florian Fainelli]

On 2/4/2022 2:50 PM, Florian Fainelli wrote:
> 
> 
> On 2/4/2022 2:45 PM, Florian Fainelli wrote:
>>
>>
>> On 2/4/2022 2:34 PM, Daniel González Cabanelas wrote:
>>> El vie, 4 feb 2022 a las 23:02, Florian Fainelli
>>> (<f.fainelli at gmail.com>) escribió:
>>>>
>>>>
>>>>
>>>> On 2/4/2022 11:21 AM, Álvaro Fernández Rojas wrote:
>>>>> So the problem is that SET_FEATURES and GET_FEATURES isn’t 
>>>>> supported by versions 2.1, 2.2 and 4.0 of the nand controller, 
>>>>> which are the ones present on bcm63xx, right?
>>>>
>>>> Yes, I suspect this is the problem since I do not see CMD_LOW_LEVEL_OP
>>>> being defined in the register database for the controllers v2.1 and
>>>> v2.2, v3.3. Staring with v4.0, the controllers do have the
>>>> CMD_LOW_LEVEL_OP. This is the version/feature table that I could
>>>> programmatically compile:
>>>>
>>>> version: 0.1, ll: no
>>>> version: 1.0, ll: no
>>>> version: 2.0, ll: no
>>>> version: 2.1, ll: no
>>>> version: 2.2, ll: no
>>>> version: 3.0, ll: no
>>>> version: 3.2, ll: no
>>>> version: 3.3, ll: no
>>>> version: 3.4, ll: no
>>>> version: 4.0, ll: yes
>>>> version: 5.0, ll: yes
>>>> version: 6.0, ll: yes
>>>> version: 6.2, ll: yes
>>>> version: 7.0, ll: yes
>>>> version: 7.1, ll: yes
>>>> version: 7.2, ll: yes
>>>> version: 7.3, ll: yes
>>>>
>>>> How about this patch, does it work better?
>>>>
>>>> diff --git a/drivers/mtd/nand/raw/brcmnand/brcmnand.c
>>>> b/drivers/mtd/nand/raw/brcmnand/brcmnand.c
>>>> index f75929783b94..06ac593beec0 100644
>>>> --- a/drivers/mtd/nand/raw/brcmnand/brcmnand.c
>>>> +++ b/drivers/mtd/nand/raw/brcmnand/brcmnand.c
>>>> @@ -1727,6 +1727,8 @@ static void brcmnand_cmdfunc(struct nand_chip
>>>> *chip, unsigned command,
>>>>                   break;
>>>>           case NAND_CMD_SET_FEATURES:
>>>>           case NAND_CMD_GET_FEATURES:
>>>> +               if (ctrl->nand_version < 0x0400)
>>>> +                       break;
>>>>                   brcmnand_low_level_op(host, LL_OP_CMD, command, 
>>>> false);
>>>>                   brcmnand_low_level_op(host, LL_OP_ADDR, column, 
>>>> false);
>>>>                   break;
>>>>
>>>> -- 
>>>> Florian
>>>
>>> No, it didn't help:
>>
>> OK, last tentative and then I think you should report this to 
>> linux-mtd such that the MTD maintainers may suggest whether we are 
>> missing something obvious here:
> 
> scratch that, can you try this instead:

And also try this patch because I don't believe there is sufficient 
protection in macronix_nand_block_protection_support to ensure that the 
NAND chip is ONFI capable:

diff --git a/drivers/mtd/nand/raw/nand_macronix.c 
b/drivers/mtd/nand/raw/nand_macronix.c
index 1472f925f386..78f893add636 100644
--- a/drivers/mtd/nand/raw/nand_macronix.c
+++ b/drivers/mtd/nand/raw/nand_macronix.c
@@ -219,9 +219,13 @@ static int mxic_nand_unlock(struct nand_chip *chip, 
loff_t ofs, uint64_t len)

  static void macronix_nand_block_protection_support(struct nand_chip *chip)
  {
+       struct nand_parameters *p = &chip->parameters;
         u8 feature[ONFI_SUBFEATURE_PARAM_LEN];
         int ret;

+       if (!p->onfi || !chip->parameters.supports_set_get_features)
+               return;
+
         bitmap_set(chip->parameters.get_feature_list,
                    ONFI_FEATURE_ADDR_MXIC_PROTECTION, 1);

Thanks!
-- 
Florian