Re: [PATCH v13 05/22] x86/virt/tdx: Handle SEAMCALL no entropy error in common code

[Huang, Kai]

On Fri, 2023-09-08 at 09:21 -0700, Dave Hansen wrote:
> On 8/25/23 05:14, Kai Huang wrote:
> > Some SEAMCALLs use the RDRAND hardware and can fail for the same reasons
> > as RDRAND.  Use the kernel RDRAND retry logic for them.
> > 
> > There are three __seamcall*() variants.  Add a macro to do the SEAMCALL
> > retry in the common code and define a wrapper for each __seamcall*()
> > variant.
> > 
> > Signed-off-by: Kai Huang <kai.huang@xxxxxxxxx>
> > ---
> > 
> > v12 -> v13:
> >  - New implementation due to TDCALL assembly series.
> > 
> > ---
> >  arch/x86/include/asm/tdx.h | 27 +++++++++++++++++++++++++++
> >  1 file changed, 27 insertions(+)
> > 
> > diff --git a/arch/x86/include/asm/tdx.h b/arch/x86/include/asm/tdx.h
> > index a252328734c7..cfae8b31a2e9 100644
> > --- a/arch/x86/include/asm/tdx.h
> > +++ b/arch/x86/include/asm/tdx.h
> > @@ -24,6 +24,11 @@
> >  #define TDX_SEAMCALL_GP			(TDX_SW_ERROR | X86_TRAP_GP)
> >  #define TDX_SEAMCALL_UD			(TDX_SW_ERROR | X86_TRAP_UD)
> >  
> > +/*
> > + * TDX module SEAMCALL leaf function error codes
> > + */
> > +#define TDX_RND_NO_ENTROPY	0x8000020300000000ULL
> > +
> >  #ifndef __ASSEMBLY__
> >  
> >  /*
> > @@ -82,6 +87,28 @@ u64 __seamcall(u64 fn, struct tdx_module_args *args);
> >  u64 __seamcall_ret(u64 fn, struct tdx_module_args *args);
> >  u64 __seamcall_saved_ret(u64 fn, struct tdx_module_args *args);
> >  
> > +#include <asm/archrandom.h>
> > +
> > +#define SEAMCALL_NO_ENTROPY_RETRY(__seamcall_func, __fn, __args)	\
> > +({									\
> > +	int ___retry = RDRAND_RETRY_LOOPS;				\
> > +	u64 ___sret;							\
> > +									\
> > +	do {								\
> > +		___sret = __seamcall_func((__fn), (__args));		\
> > +	} while (___sret == TDX_RND_NO_ENTROPY && --___retry);		\
> > +	___sret;							\
> > +})
> 
> This is a *LOT* less eye-bleedy if you do it without macros:
> 
> 
> typedef u64 (*sc_func_t)(u64 fn, struct tdx_module_args *args);
> 
> static inline
> u64 sc_retry(sc_func_t func, u64 fn, struct tdx_module_args *args)
> {
>         int retry = RDRAND_RETRY_LOOPS;
>         u64 ret;
> 
>         do {
>                 ret = func(fn, args);
>         } while (ret == TDX_RND_NO_ENTROPY && --retry);
> 
>         return ret;
> }
> 
> #define seamcall(_fn, _args)           sc_retry(_seamcall,
> (_fn), (_args))
> #define seamcall_ret(_fn, _args)       sc_retry(_seamcall_ret,
> (_fn), (_args))
> #define seamcall_saved_ret(_fn, _args) sc_retry(_seamcall_saved_ret,
> (_fn), (_args))
> 
> The compiler can figure it out and avoid making func() an indirect call
> since it knows the call location at compile time.

Indirect call was a concern when I was implementing those.  I didn't know for 
sure that the compiler can avoid it.  I'll change to use above.  Thanks!

> 
> You can also do the seamcall() #define as a static inline, but it does
> take up more screen real estate.  Oh, and going a wee bit over 80
> columns is OK for those #defines.

Yes I verified the checkpatch.pl wouldn't complain if the #define exceeded 80
characters in one line.  I'll use #define.  Thanks!