hw/drivers/flash/spiflash: Add option to ignore JEDEC ID

[StarGate01]

This PR introduces three new optional configuration parameters: SPIFLASH_IGNORE_MANUFACTURER, SPIFLASH_IGNORE_MEMORY_TYPE, and SPIFLASH_IGNORE_MEMORY_CAPACITY. (Default: 0).

These parameters selectively disable the whitelist checks of the SPI Flash JEDEC IDs.

Rationale

I am currently porting the InfiniTime OS (written for the PineTime smartwatch) to a popular Chinese smartwatch series called "P8". mynewt-core is used for the bootloader.

These P8 watches come in a whole zoo of variants, each being compatible with the firmware but containing sometimes a bit different chips due to the silicon shortage and general chip availability on the global market. This is usually no problem to handle.

However, the SPI driver of mynewt-core enforces a strict whitelist on the reported SPI Flash JEDEC ID. I see how this is useful in normal applications where the hardware configuration is known in detail and compatibility has to be ensured, however this leads to my bootloader refusing to work on prior unknown / new smartwatch variants, bricking them.

In the past, others and I have added definitions for SPI IDs (e.g. #2798 , #2582 ), but I can't keep up with all the models and I don't want to brick any more watches due to a purely whitelisting (not technical) issue.

I am open to suggestions, I think these optional configuration parameters would be quite useful to have.

[StarGate01]

@kasjer Is this something you would consider merging? The feature would be opt-in.

[kasjer]

Hi, I am out of office, idea is good so definitely I will support it, I have not seen the code, I will take a look next week. Thanks for your submittion. Jerzy czw., 21 lip 2022, 10:17 użytkownik Christoph Honal < ***@***.***> napisał:

…

@kasjer <https://github.com/kasjer> Is this something you would consider merging? The feature would be opt-in. — Reply to this email directly, view it on GitHub <#2858 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFP6YYGIWOUIS7EBDXHZV7DVVEBQNANCNFSM52T7XHKA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[kasjer]

@StarGate01 maybe you could change your commit a little bit to be more consistent with mynewt style that requires opening brace { to be on the same line as if.
Since new syscfg values have default value 0 it can be safely used in plain if (not #if)
Instead of having:

#if !MYNEWT_VAL(SPIFLASH_IGNORE_MANUFACTURER)
        assert(manufacturer == supported_chips[0].fc_jedec_id.ji_manufacturer);
        if (manufacturer != supported_chips[0].fc_jedec_id.ji_manufacturer) {
            rc = -1;
            goto err;
        }
#endif

we would have:

        assert(!MYNEWT_VAL(SPIFLASH_IGNORE_MANUFACTURER) &&
               manufacturer == supported_chips[0].fc_jedec_id.ji_manufacturer);
        if (!MYNEWT_VAL(SPIFLASH_IGNORE_MANUFACTURER) &&
            manufacturer != supported_chips[0].fc_jedec_id.ji_manufacturer) {
            rc = -1;
            goto err;
        }

generate code would be the same and compiler would check syntax that otherwise would not be checked in some build depending on ignore flags and mynewt style would be preserved without additional pre-processor conditions added for style sake only.

You can amend your commit and force push.

[StarGate01]

@kasjer Thank you for the style comments; I have applied your suggestions.

[apache-mynewt-bot]

Style check summary

No suggestions at this time!

[kasjer]

@StarGate01 thanks for contributions