Re: [PATCH] x86/mm/mem_encrypt: Disable all instrumentation for SME early boot code

[Gary R Hook]

On 4/29/19 3:51 PM, Borislav Petkov wrote:
> [CAUTION: External Email]
> 
> On Mon, Apr 29, 2019 at 08:16:07PM +0000, Gary R Hook wrote:
>> Yes, option 4 would be a combination of using a local copy of strncmp()
> 
> Why the local copy?

Seemed suitable, since it's tiny. But I'm not married to the idea.

> 
>> and disabling instrumentation (KASAN, KCOV, whatever) for
>> arch/x86/lib/cmdline.c when SME is enabled.
> 
> I think this should suffice. You only disable instrumentation when
> CONFIG_AMD_MEM_ENCRYPT=y and not do any local copies but use the generic
> functions.

Okay, super.  I'll post a v2 that does that.

Do we want to document the subordinate functions in their respective 
source files so that a future editor would (hopefully) be made aware of 
this use?

grh