[Lightning] Re: PATCH - Correct comparison of float branch test on i386

[Paulo César Pereira de Andrade]

Paolo Bonzini wrote:
> On 08/18/2010 09:15 PM, Paulo César Pereira de Andrade wrote:
>> +#define jit_bner_f(d, s1, s2)               jit_fp_btest((UCOMISSrr ((s1), 
>> (s2)),
>> _OO (0x7a03), _OO (0x7405), JMPm (((d))))) /* JP to JMP, JZ past JMP */
>> +#define jit_bner_d(d, s1, s2)               jit_fp_btest((UCOMISDrr ((s1), 
>> (s2)),
>> _OO (0x7a03), _OO (0x7405), JMPm (((d))))) /* JP to JMP, JZ past JMP */
>
> BTW, these should be 0x7a02 (preexisting bug).

  Thanks. I was also wondering about the multiple jumps, but did
not bother much for now, as I plan to write a test suite covering
all jump combinations soon in my testing tool.

  Currently working on an args.tst, to test arguments to functions,
while also working on a patch I should propose soon, for > 6 integer
arguments, > 8 float arguments and varargs for x86_64.

> Paolo

Paulo