[RFC PATCH] board: ti: common: board_detect: Fix EEPROM read quirk for 2-byte

[Matwey V. Kornilov]

вт, 29 нояб. 2022 г. в 09:50, Neha Malcom Francis <n-francis at ti.com>:
>
> EEPROM detection logic in ti_i2c_eeprom_get() involves figuring out
> whether addressing is 1-byte or 2-byte. There are currently different
> behaviours seen across boards as documented in commit bf6376642fe8
> ("board: ti: common: board_detect: Fix EEPROM read quirk"). Adding to
> the list, we see that there are 2-byte EEPROMs that read properly
> with 1-byte addressing with no offset.
>
> For ti_i2c_eeprom_am6_get where eeprom parse operation is dynamic, the
> earlier commit d2ab2a2bafd5 ("board: ti: common: board_detect: Fix
> EEPROM read quirk for AM6 style data") tried to resolve this by running
> ti_i2c_eeprom_get() twice. However this commit along with its former
> commit fails on J7 platforms where EEPROM successfully return back the
> header on 1-byte addressing and continues to do so until an offset is
> introduced. So the second read incorrectly determines the EEPROM as
> 1-byte addressing.
>
> A more generic solution is introduced here to solve
> this issue: 1-byte read without offset and 1-byte read with offset. If
> both passes, it follows 1-byte addressing else we proceed with 2-byte
> addressing check.
>
> Tested on J721E, J7200, DRA7xx, AM64x

I'll try to test this on the AM335x boards I have as soon as possible.

>
> Signed-off-by: Neha Malcom Francis <n-francis at ti.com>
> Fixes: d2ab2a2bafd5 (board: ti: common: board_detect: Fix EEPROM read
> quirk for AM6 style data) and bf6376642fe8 (board: ti: common: board_detect:
> Fix EEPROM read quirk)
> ---
>  board/ti/common/board_detect.c | 29 ++++++++++++++++++-----------
>  1 file changed, 18 insertions(+), 11 deletions(-)
>
> diff --git a/board/ti/common/board_detect.c b/board/ti/common/board_detect.c
> index c37629fe8a..b9f2ebf2a0 100644
> --- a/board/ti/common/board_detect.c
> +++ b/board/ti/common/board_detect.c
> @@ -91,6 +91,8 @@ static int __maybe_unused ti_i2c_eeprom_get(int bus_addr, int dev_addr,
>  #if CONFIG_IS_ENABLED(DM_I2C)

Should #else branch also be modified according to the new algo?

>         struct udevice *dev;
>         struct udevice *bus;
> +       uint8_t offset_test;
> +       bool one_byte_addressing = true;
>
>         rc = uclass_get_device_by_seq(UCLASS_I2C, bus_addr, &bus);
>         if (rc)
> @@ -114,8 +116,23 @@ static int __maybe_unused ti_i2c_eeprom_get(int bus_addr, int dev_addr,
>          */
>         (void)dm_i2c_read(dev, 0, ep, size);
>
> +       if (*((u32 *)ep) != header)
> +               one_byte_addressing = false;
> +
> +       /*
> +        * Handle case of bad 2 byte eeproms that responds to 1 byte addressing
> +        * but gets stuck in const addressing when read requests are performed
> +        * on offsets. We perform an offset test to make sure it is not a 2 byte
> +        * eeprom that works with 1 byte addressing but just without an offset
> +        */
> +
> +       rc = dm_i2c_read(dev, 0x1, &offset_test, sizeof(offset_test));
> +
> +       if (*((u32 *)ep) != (header & 0xFF))
> +               one_byte_addressing = false;
> +
>         /* Corrupted data??? */
> -       if (*((u32 *)ep) != header) {
> +       if (!one_byte_addressing) {
>                 /*
>                  * read the eeprom header using i2c again, but use only a
>                  * 2 byte address (some newer boards need this..)
> @@ -444,16 +461,6 @@ int __maybe_unused ti_i2c_eeprom_am6_get(int bus_addr, int dev_addr,
>         if (rc)
>                 return rc;
>
> -       /*
> -        * Handle case of bad 2 byte eeproms that responds to 1 byte addressing
> -        * but gets stuck in const addressing when read requests are performed
> -        * on offsets. We re-read the board ID to ensure we have sane data back
> -        */
> -       rc = ti_i2c_eeprom_get(bus_addr, dev_addr, TI_EEPROM_HEADER_MAGIC,
> -                              sizeof(board_id), (uint8_t *)&board_id);
> -       if (rc)
> -               return rc;
> -
>         if (board_id.header.id != TI_AM6_EEPROM_RECORD_BOARD_ID) {
>                 pr_err("%s: Invalid board ID record!\n", __func__);
>                 return -EINVAL;
> --
> 2.34.1
>


-- 
With best regards,
Matwey V. Kornilov