Re: [PATCHv2 01/12] x86/tdx: Consolidate TDX error handling

[kirill.shutemov@xxxxxxxxxxxxxxx]

On Wed, Jun 25, 2025 at 02:27:21PM -0700, Sean Christopherson wrote:
> On Wed, Jun 25, 2025, Rick P Edgecombe wrote:
> > On Wed, 2025-06-25 at 10:58 -0700, Dave Hansen wrote:
> > > > --- a/arch/x86/kvm/vmx/tdx.c
> > > > +++ b/arch/x86/kvm/vmx/tdx.c
> > > > @@ -202,12 +202,6 @@ static DEFINE_MUTEX(tdx_lock);
> > > >   
> > > >   static atomic_t nr_configured_hkid;
> > > >   
> > > > -static bool tdx_operand_busy(u64 err)
> > > > -{
> > > > -	return (err & TDX_SEAMCALL_STATUS_MASK) == TDX_OPERAND_BUSY;
> > > > -}
> > > > -
> > > > -
> > > 
> > > Isaku, this one was yours (along with the whitespace damage). What do
> > > you think of this patch?
> > 
> > I think this actually got added by Paolo, suggested by Binbin. I like these
> > added helpers a lot. KVM code is often open coded for bitwise stuff, but since
> > Paolo added tdx_operand_busy(), I like the idea of following the pattern more
> > broadly. I'm on the fence about tdx_status() though.
> 
> Can we turn them into macros that make it super obvious they are checking if the
> error code *is* xyz?  E.g.
> 
> #define IS_TDX_ERR_OPERAND_BUSY
> #define IS_TDX_ERR_OPERAND_INVALID
> #define IS_TDX_ERR_NO_ENTROPY
> #define IS_TDX_ERR_SW_ERROR
> 
> As is, it's not at all clear that things like tdx_success() are simply checks,
> as opposed to commands.

I remember Dave explicitly asked for inline functions over macros where
possible.

Can we keep them as functions, but give the naming scheme you proposing
(but lowercase)?

-- 
  Kiryl Shutsemau / Kirill A. Shutemov