Re: [Xen-devel] [PATCH 2/3] xen/arm: vsmc: Don't implement function ID that doesn't exist

[Julien Grall <julien.grall@xxxxxxxxxx>]

Hi,

On 26/01/18 18:12, Volodymyr Babchuk wrote:
> On 26.01.18 20:07, Julien Grall wrote:
> > On 26/01/18 18:03, Volodymyr Babchuk wrote:
> > > > diff --git a/xen/include/asm-arm/smccc.h b/xen/include/asm-arm/smccc.h
> > > > index f543dea0bb..303517459f 100644
> > > > --- a/xen/include/asm-arm/smccc.h
> > > > +++ b/xen/include/asm-arm/smccc.h
> > > > @@ -82,9 +82,23 @@ static inline uint32_t smccc_get_owner(register_t 
> > > > funcid)
> > > >   #define ARM_SMCCC_OWNER_TRUSTED_OS_END  63
> > > >   /* List of generic function numbers */
> > > > -#define ARM_SMCCC_FUNC_CALL_COUNT       0xFF00
> > > > -#define ARM_SMCCC_FUNC_CALL_UID         0xFF01
> > > > -#define ARM_SMCCC_FUNC_CALL_REVISION    0xFF03
> > > So, I thing it is bette to leave
> > > #define ARM_SMCCC_FUNC_CALL_*
> > > and introduce
> > I am ok to use FID, but I don't see any reason to keep 
> > ARM_SMCCC_FUNC_* one around. No-one can use them directly in Xen and 
> > this would add confusion.
> Right. So you can nuke them.
> Another idea is to leave only one definition and use it like this:
>
>   ARM_SMCCC_FID_CALL(COUNT, HYPERVISOR)
>
> This will reduce number of similar macro definitions. What do you think?
Well technically the function names are "Call Count", "Call UID", 
"Revision". Therefore the name is already wrong for revision.
I also don't see a good way to use this macro more than 3 times. Indeed 
fastcall, 32-conv is pretty specific.
So here the benefits seems really limited.

> I'm okay with any variant.
I will stick with the one suggested in the patch I sent.

Cheers,


--
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel