Add option to ignore already-set bits in otp set command

[will-v-pi]

This adds a way to set bits in an OTP row, without having to specify a field - for example for the page lock rows

Allows you to workaround this (for example)

$ picotool otp set crit1 0x1
ROW 0x0040  OLD_VALUE=0x000000: OTP_DATA_CRIT1
        "Page 1 critical boot flags (RBIT-8)"
$ picotool otp set crit1 0x2
ROW 0x0040  OLD_VALUE=0x000001: OTP_DATA_CRIT1
        "Page 1 critical boot flags (RBIT-8)"
ERROR: Cannot clear bits in OTP row(s): current value 000001, new value 000002

using the -s argument

$ picotool otp set crit1 0x2 -s
ROW 0x0040  OLD_VALUE=0x000001: OTP_DATA_CRIT1
        "Page 1 critical boot flags (RBIT-8)"
$ picotool otp get crit1
ROW 0x0040: OTP_DATA_CRIT1
        "Page 1 critical boot flags (RBIT-8)"

    VALUE 0x000003

    field GLITCH_DETECTOR_SENS (bits 5-6) = 0
        "Increase the sensitivity of the glitch detectors from their default."
    field GLITCH_DETECTOR_ENABLE (bit 4) = 0
        "Arm the glitch detectors to reset the system if an abnormal clock/power event is observed."
    field BOOT_ARCH (bit 3) = 0
        "Set the default boot architecture, 0=ARM 1=RISC-V. Ignored if ARM_DISABLE, RISCV_DISABLE or SECURE_BOOT_ENABLE is set."
    field DEBUG_DISABLE (bit 2) = 0
        "Disable all debug access"
    field SECURE_DEBUG_DISABLE (bit 1) = 1
        "Disable Secure debug access"
    field SECURE_BOOT_ENABLE (bit 0) = 1
        "Enable boot signature enforcement, and permanently disable the RISC-V cores."

Fixes #137

[lurch]

ERROR: Cannot clear bits in OTP row(s): current value 000001, new value 000002

I wonder if this error message should point people at the -s option?

Also, this PR should probably update https://github.com/raspberrypi/picotool/tree/develop?tab=readme-ov-file#otp too?

[kilograham]

's' is OK, but I think ignore-set is confusing... I think set-bits-only (or probably just set-bits)-is better, and we should probably disallow it with ECC

[lurch]

's' is OK, but I think ignore-set is confusing... I think set-bits-only (or probably just set-bits)-is better, and we should probably disallow it with ECC

Or I guess you could use:

option('o', "--or-existing").set(settings.otp.or_with_existing) % "Logically OR with any already-set bits"

?
But I'm not familiar enough with OTP and its usage, to know if that's actually a sensible suggestion or not.

[will-v-pi]

's' is OK, but I think ignore-set is confusing... I think set-bits-only (or probably just set-bits)-is better, and we should probably disallow it with ECC

I've changed it to set-bits - although I've left the internal variable as ignore_set as that's what picotool's doing internally

It already fails on ECC rows in the normal way:

picotool/main.cpp

Lines 7107 to 7109 in fb85aca

	if (old_raw_value && settings.otp.ecc) {
	fail(ERROR_NOT_POSSIBLE, "Cannot modify OTP ECC row(s)\n");
	}

[kilograham]

Approving, but

I've changed it to set-bits - although I've left the internal variable as ignore_set as that's what picotool's doing internally

Seems like a weird hill to die on... it doesn't ignore anything; it ORs the old value into the new value ;-)