Re: [PATCH v4 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL

[Huang, Kai]

On Sat, 2023-08-26 at 00:50 +0000, Huang, Kai wrote:
> On Sat, 2023-08-26 at 02:35 +0300, kirill.shutemov@xxxxxxxxxxxxxxx wrote:
> > On Tue, Aug 15, 2023 at 11:02:01PM +1200, Kai Huang wrote:
> > > +	/*
> > > +	 * Failure of __tdcall_hypercall() indicates a failure of the TDVMCALL
> > > +	 * mechanism itself and that something has gone horribly wrong with
> > > +	 * the TDX module.  __tdx_hypercall_failed() never returns.
> > > +	 */
> > > +	if (__tdcall_hypercall(TDG_VP_VMCALL, &margs))
> > > +		__tdx_hypercall_failed();
> > 
> > Do we even need __tdx_hypercall_failed() anymore? Just call panic()
> > directly, no?
> > 
> 
> __tdx_hypercall() is currently annotated with 'noinstr' (which is also the
> behaviour of the old code).  We need it otherwise we will get build warning like
> below:
> 
> https://lore.kernel.org/lkml/a23ce8fd289141cea3a1b4f3dace221dca847238.camel@xxxxxxxxx/T/#m205b29ce3e33983ec6be733c2afffbfd5cb653ff

Forgot to say, __tdx_hypercall() is a common code shared by the compressed code,
which also has it's own version of __tdx_hypercall_failed().  We need such
wrapper anyway because compressed code calls error() while the main kernel calls
panic().