By: Dresdenboy (M.Waldhauer.delete@this.gmx.de), July 22, 2004 2:55 pm
Room: Moderated Discussions
Anonymous (not@amd.com) on 7/22/04 wrote:
---------------------------
>Dresdenboy (M.Waldhauer@gmx.de) on 7/22/04 wrote:
>---------------------------
>>To me, your variables unknown5 and unknown6 look like they are the
>>microcode ROM line numbers to be patched and unused[6] would be the other
>>6, unused line numbers. I think that still fits to what is described in the patent.
>
>I'm not sure about this. This means only two lines can be patched, which would
>be very limiting. The patents suggest that updates can patch any 64 lines out of
>the 3072 in the microcode array. This is probably not the exact number in K8 cores,
>but it may be close. Probably the addresses to patch are within the data area itself.
>
IMO these up to 8 line addresses just tell the starting points for the patch code with additional information in the patch data (like the number of patched lines starting from that point).
---------------------------
>Dresdenboy (M.Waldhauer@gmx.de) on 7/22/04 wrote:
>---------------------------
>>To me, your variables unknown5 and unknown6 look like they are the
>>microcode ROM line numbers to be patched and unused[6] would be the other
>>6, unused line numbers. I think that still fits to what is described in the patent.
>
>I'm not sure about this. This means only two lines can be patched, which would
>be very limiting. The patents suggest that updates can patch any 64 lines out of
>the 3072 in the microcode array. This is probably not the exact number in K8 cores,
>but it may be close. Probably the addresses to patch are within the data area itself.
>
IMO these up to 8 line addresses just tell the starting points for the patch code with additional information in the patch data (like the number of patched lines starting from that point).