On 07/13/2012 05:22 PM, Timur Tabi wrote:
> Scott Wood wrote:
>
>> I know the spec wouldn't change, except the version number. But as I
>> said above, there would be no known v2 implementations with the bug.
>> You would only check the bad CRC location if you see v1 data, because
>> there are known buggy v1 implementations.
>
> I already have that:
>
> if ((e.version == NXID_VERSION) && (crc != be32_to_cpup(crcp))) {
>
> NXID_VERSION is equal to 1, so we only do the check for the old CRC if we
> have a v1 EEPROM.
But you continue to generate v1 EEPROMs. If we get the people who
define the format to accept v2, then we can generate v2 after the fix is
applied, and the (very small) risk of a real CRC failure combined with a
spurious CRC success in the old location would only apply on EEPROMs
which haven't been saved since the update.
-Scott
_______________________________________________
U-Boot mailing list
U-Boot@lists.denx.de
http://lists.denx.de/mailman/listinfo/u-boot
