Re: [PATCH v8 06/12] x86/HVM: make hvmemul_blk() capable of handling r/o operations

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 05/05/2020 09:15, Jan Beulich wrote:
> In preparation for handling e.g. FLDENV or {F,FX,X}RSTOR here as well.
>
> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
> ---
> v8: New (could be folded into "x86emul: support MOVDIR{I,64B} insns",
>     but would invalidate Paul's R-b there).
>
> --- a/xen/arch/x86/hvm/emulate.c
> +++ b/xen/arch/x86/hvm/emulate.c
> @@ -1453,7 +1453,7 @@ static int hvmemul_blk(
>      struct hvm_emulate_ctxt *hvmemul_ctxt =
>          container_of(ctxt, struct hvm_emulate_ctxt, ctxt);
>      unsigned long addr;
> -    uint32_t pfec = PFEC_page_present | PFEC_write_access;
> +    uint32_t pfec = PFEC_page_present;
>      int rc;
>      void *mapping = NULL;
>  
> @@ -1462,6 +1462,9 @@ static int hvmemul_blk(
>      if ( rc != X86EMUL_OKAY || !bytes )
>          return rc;
>  
> +    if ( x86_insn_is_mem_write(state, ctxt) )
> +        pfec |= PFEC_write_access;
> +

For the instructions with two memory operands, it conflates the
read-only side with the read-write side.

~Andrew