i know that the kernels(except voodoo) wont work without cpuid opcode patching in amd cpus,and it was removed using "marwins amd utility" before...
http://en.wikipedia.org/wiki/CPUID states that the registers r loaded with cpu specific info....
so is it possible (in future?) to create a kext(or mergin this feature with decryptor.kext) that changes the register values to intel specific one so there will not be any requirement of patched kernel in future for amd?...

Voodoo kernel does that. It has "on the fly" patching built in. No need for AMD patching, no need for any "patcher kext".

i know that
read the question properly before answering ...i have clearly stated "except voodoo"
what i want to know is is it possible to run sys using "vanilla kernel" in future if whatever i said is implemented.. ...

I read the question properly before answering

There is no need for any "patcher kext". And cos of that no one will do one.

Or is it that you wan't some new feature? Then explain what it shall be. Word a propper question...

i didnt write "patched kext" anywhere in my post...
mods please answer to the question

There would be a lot of factors involved in properly answering this question..

1. How soon after the kernel is loaded would the kernel extension responsible be loaded?

If it's not fast enough, the system would panic, and the point would be moot.

2. Is there justification enough for it?

IMHO not really, because as Valentine already stated, Voodoo is acceptable enough for the vast majority of AMD users, and if not, they would need excellent reasoning behind it to get a talented enough developer to *make* the kernel extension, again provided that the first factor I mentioned above were satisfied.

3. Is there enough basic support for an AMD processor to use the native kernel to execute the libsa ( "stand-alone library" ) before the kextd (kext daemon, loads kexts into the kernel) becomes available to actually LOAD the new theoretical extension?

I don't think so. To quote from Amit Singh's book "Mac OS X Internals":



Long story short: I don't think it's practical when Voodoo already works, and I seriously doubt the native kernel's ability to not only use libsa properly with an AMD proc, but also to load the kextd and then the opcode patching provided by the "kext" in question.

Even so, the kernel has to be running for these libraries to load and run anyway, so if the kernel can't boot these libraries wont load anyway (from how it sounds). In kernel patching is perfect.

You not really asked, if an kext will ever emulate an Intel CPUID (MSR)?

Even if we emulate Intel CPUID by kernel or by kext we are not sure that processor specific features are the same.
I investigated codes of ACPI_SMC_PlatformPlugin. It asks kernel for CPUID and then analyses it.
Preliminary what I know:
cpu_family must be 6.
cpu_model = {13..18} - I am not clear here.

Hrm... you need to adjust _all_ CPUID values i think, so cpu vendor, model, family, extmodel, extfamily.
Vendor has to be 0x756E6547 for Intel CPUs.

thanks for the ans people..
one more thing is segmentation fault(SEGSEGV signal) related to this...bcos toh modbin etc kernels SEGSEGV signal generated for some process if CPUID not patched...?
is it like patching removes the cpuid instruction and changes the address of preceding instructions and segments addresses?...........