Re: [Qemu-devel] [RISU RFC PATCH v1 2/7] risugen_x86_asm: add module

[Jan Bobek <jan.bobek@gmail.com>]

[-- Attachment #1.1: Type: text/plain, Size: 2166 bytes --]

On 6/27/19 5:05 AM, Richard Henderson wrote:
> On 6/19/19 7:04 AM, Jan Bobek wrote:
>> +sub rex_encode(%)
>> +{
>> +    my (%args) = @_;
>> +
>> +    $args{w} = 0 unless defined $args{w};
>> +    $args{r} = 0 unless defined $args{w};
>> +    $args{x} = 0 unless defined $args{w};
>> +    $args{b} = 0 unless defined $args{w};
> 
> What makes you believe that REX.[RXB] are dependent on REX.W?
> Or are these merely cut-and-paste errors?

Oops, that's a typo.

>> +sub modrm_encode(%)
>> +{
>> +    my (%args) = @_;
>> +
>> +    die "MOD field out-of-range: $args{mod}"
>> +        unless 0 <= $args{mod} && $args{mod} <= 3;
>> +    die "REG field out-of-range: $args{reg}"
>> +        unless 0 <= $args{reg} && $args{reg} <= 7;
>> +    die "RM field out-of-range: $args{rm}"
>> +        unless 0 <= $args{rm} && $args{rm} <= 7;
>> +
>> +    return (value =>
>> +            ($args{mod} << 6)
>> +            | ($args{reg} << 3)
>> +            | $args{rm},
>> +            len => 1);
>> +}
>> +
>> +sub sib_encode(%)
>> +{
>> +    my (%args) = @_;
>> +
>> +    die "SS field out-of-range: $args{ss}"
>> +        unless 0 <= $args{ss} && $args{ss} <= 3;
>> +    die "INDEX field out-of-range: $args{index}"
>> +        unless 0 <= $args{index} && $args{index} <= 7;
>> +    die "BASE field out-of-range: $args{base}"
>> +        unless 0 <= $args{base} && $args{base} <= 7;
>> +
>> +    return (value =>
>> +            ($args{ss} << 6)
>> +            | ($args{index} << 3)
>> +            | $args{base},
>> +            len => 1);
>> +}
> 
> These are interdependent, in that SIB requires MODRM.RM == 4.  But I don't see
> anything that either enforces that or validates it, either here or within
> write_insn below.

It also requires MODRM.MOD != 3. You are right, this is not validated
by the risugen_x86_asm module at all; the caller is supposed to know
what they're doing. I also don't validate if displacement is present
when required; the exact conditions are somewhat complicated, already
present in one form or another in risugen_x86 and it generally seemed
like not worth the effort.

-Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]