https://bugzilla.kernel.org/show_bug.cgi?id=198123
--- Comment #48 from shibe@tuta.io --- Turns out, I missed some interesting bits. Here, least significant bit first:
69E4 - - 0 0 - - - 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 69E8 0 0 0 0 0 0 - - 0 0 - - - 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 69EC 0 0 0 0 0 0 - - - - - - - - 0 0 0 0 0 0 0 0 1 - 0 0 - - - 0 0 0 69F0 0 0 + - 0 0 - - - 0 0 0 + - 0 0 - - - 0 0 0 + - 0 0 - - - 0 0 0 69FC 0 + 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 6E68 - 0 0 - 0 - 0 0 0 - - 0 0 - 0 - 0 0 0 - - 0 0 - 0 - 0 0 0 - 0 0
Legend: 0 always 0 in black dumps 1 always 1 in black dumps + always 1 in black dumps, usually 0 in grey dumps - always 0 in black dumps, usually 1 in grey dumps
I've found that 0x69?? addresses are defined as constants with the prefix "EVERGREEN_DC_LUT_". Why could these registers be messed up?