Re: [RFC v12 60/65] target/arm: cpu-pauth: new module for ARMv8.3 Pointer Authentication

[Claudio Fontana]

On 3/28/21 9:05 PM, Richard Henderson wrote:
> On 3/26/21 1:36 PM, Claudio Fontana wrote:
>> Pointer Authentication is an AARCH64-only ARMv8.3 optional
>> extension, whose cpu properties can be separated out in its own module.
>>
>> Signed-off-by: Claudio Fontana<cfontana@suse.de>
>> ---
>>   target/arm/cpu.h           |  3 --
>>   target/arm/tcg/cpu-pauth.h | 34 ++++++++++++++++++++
>>   target/arm/cpu.c           |  4 +--
>>   target/arm/cpu64.c         | 35 ++------------------
>>   target/arm/tcg/cpu-pauth.c | 66 ++++++++++++++++++++++++++++++++++++++
>>   target/arm/tcg/meson.build |  1 +
>>   6 files changed, 105 insertions(+), 38 deletions(-)
>>   create mode 100644 target/arm/tcg/cpu-pauth.h
>>   create mode 100644 target/arm/tcg/cpu-pauth.c
> 
> No move + rename at once.
> 
> Also, you've started using tcg_sve_* and I think that might as well apply to 
> these, in that second step.
> 
> r~
> 

The idea for tcg_sve_* was in contrast to cpu_sve_*, which contains the common 
parts of cpu_sve.

So the idea for SVE is:

cpu-sve.c     : CPU SVE module, contains the common functions.
tcg/tcg-sve.c : TCG-specific parts of cpu-sve
kvm/kvm-sve.c : KVM-specific parts of cpu-sve

Now for PAuth we only have a TCG implementation,
so that is the reason that this patch uses the more general name.

If still tcg/tcg-pauth.c seems better we can go for it, but it just seems not 
very logical without a corresponding general cpu-pauth.c

Ciao,

Claudio