[coreboot] Patch set updated for coreboot: b0cd5ca Add AMD Family 10h PH-EO support
xdrudis at tinet.cat
Wed Sep 14 01:04:33 CEST 2011
On Tue, Sep 13, 2011 at 12:18:24PM +0200, Paul Menzel wrote:
> > src/vendorcode/amd/agesa/f12/Proc/CPU/Family/0x10/RevE
> > /F10MicrocodePatch010000bf.c
> 2. I do not understand the commit message. What do you mean with patch
> file? In this patch you are only adding a header file.
It is adding microcode patches to a header file.
The lines added contain binary patches for a binary
"interpreter/program/something" stored in the CPU that is called
microcode. Roughly, there's some kind of EPROM (or ROM+shadow RAM or
some form of memory) in the CPU that gets indexed by x86 opcodes and
other inputs and outputs control signals to drive the CPU circuitry in
a way consistent with the instruction set semantics and CPU
specification. The information in this EPROM is the microcode. The
content in the altered "source" file is not a new copy of the
microcode but just parts of it presumably with some control
information about which version it is or which it should replace and
where to apply the code portions, or whatever. That's why the original
file had MicrocodePatch in its name.
I don't know many more details, since neither the microcode source,
the language it is written in or the CPU design are public at all, but
I think it is enough to understand why QingPei Wang called it "patch
file". You can't say "I add microcode for revision E Fam 10 CPUs"
because the file does not contain the whole microcode of the CPU, just
some (small?) modifications to the microcode contained in the CPU as
shipped from factory. "patch file" is not talking of a patch to the
coreboot code, but of a patch to the microcode, the whole binary patch
will be contained in the coreboot image if you so compile it. Call the
patch set sent "metapatch" if you like.
I agree that initial capital case is clearer, although I don't see it
so important as to raise it.
The only reason I can imagine for not taking the patch set (updated
later) is lack of extensive tests or license compatibility, but both
would apply to microcode already in coreboot so my guess is it'll be
Thanks for reviewing contributions. And thanks to contributors.
More information about the coreboot