Re: [RFC PATCH 2/4] KVM: emulate: avoid repeated calls to do_insn_fetch_bytes

[Paolo Bonzini <pbonzini@redhat.com>]

Il 07/05/2014 06:21, Bandan Das ha scritto:
>> > +		if (rc != X86EMUL_CONTINNUE)
>> > +			goto done;
>> > +	}
>> > +
>> >  	while (size--) {
>> > -		if (unlikely(ctxt->_eip == fc->end)) {
>> > -			rc = do_insn_fetch_bytes(ctxt);
>> > -			if (rc != X86EMUL_CONTINUE)
>> > -				return rc;
>> > -		}
>> >  		*dest++ = *src++;
>> >  		ctxt->_eip++;
>> >  		continue;
>> > @@ -4273,7 +4282,7 @@ int x86_decode_insn(struct x86_emulate_ctxt *ctxt, void *insn, int insn_len)
>> >  	if (insn_len > 0)
>> >  		memcpy(ctxt->fetch.data, insn, insn_len);
>> >  	else {
>> > -		rc = do_insn_fetch_bytes(ctxt);
>> > +		rc = do_insn_fetch_bytes(ctxt, 1);
> Is this saying that if the cache is full, then we fetch one more byte ?
>

No, it is saying that if the instruction is being executed for the first 
time (we can execute it multiple times if we reenter a repeated 
instruction after a userspace exit) we try to get at least one byte from 
RIP.  Most of the time, do_insn_fetch_bytes will fetch 15 bytes which 
are the maximum length of an instruction.

Passing op_size == 1 matches this change in do_insn_fetch_bytes:

-	if (unlikely(size == 0))
+	if (unlikely(size < op_size))

Paolo