From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([208.118.235.92]:40317)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1TM2XM-0001iu-33
	for qemu-devel@nongnu.org; Wed, 10 Oct 2012 16:05:48 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1TM2XB-0005lb-GJ
	for qemu-devel@nongnu.org; Wed, 10 Oct 2012 16:05:47 -0400
Received: from mail-pa0-f45.google.com ([209.85.220.45]:56693)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1TM2XA-0005ef-6M
	for qemu-devel@nongnu.org; Wed, 10 Oct 2012 16:05:37 -0400
Received: by mail-pa0-f45.google.com with SMTP id fb10so969211pad.4
	for <qemu-devel@nongnu.org>; Wed, 10 Oct 2012 13:05:32 -0700 (PDT)
Sender: Richard Henderson <rth7680@gmail.com>
Message-ID: <5075D50A.9070001@twiddle.net>
Date: Wed, 10 Oct 2012 13:05:30 -0700
From: Richard Henderson <rth@twiddle.net>
MIME-Version: 1.0
References: <1349814458-21739-1-git-send-email-aurelien@aurel32.net>
	<1349814458-21739-4-git-send-email-aurelien@aurel32.net>
In-Reply-To: <1349814458-21739-4-git-send-email-aurelien@aurel32.net>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [PATCH 03/14] target-mips: fix FPU exceptions
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Aurelien Jarno <aurelien@aurel32.net>
Cc: qemu-devel@nongnu.org

On 10/09/2012 01:27 PM, Aurelien Jarno wrote:
> -    return float64_sqrt(fdt0, &env->active_fpu.fp_status);
> +    set_float_exception_flags(0, &env->active_fpu.fp_status);
> +    fdt0 = float64_sqrt(fdt0, &env->active_fpu.fp_status);
> +    update_fcr31(env);
> +    return fdt0;

While accurate, I can't help but think there'd be less memory traffic
if the invariant "fp_status == 0" is maintained between insns.  Thus
you don't need to reset the flags to 0 in each insn, merely change

static inline void update_fcr31(CPUMIPSState *env)
{
    int tmp = ieee_ex_to_mips(get_float_exception_flags(&env->active_fpu.fp_status));
    if (tmp) {
        set_float_exception_flags(0, &env->active_fpu.fp_status);
        SET_FP_CAUSE(env->active_fpu.fcr31, tmp);
        if (GET_FP_ENABLE(env->active_fpu.fcr31) & tmp) {
            helper_raise_exception(env, EXCP_FPE);
        } else {
            UPDATE_FP_FLAGS(env->active_fpu.fcr31, tmp);
        }
    }
}

I'll also note that we don't get the proper PC for the trap there.
We don't save the PC in the translator before the insn, or, more
properly invoke do_restore_state.  That clearly ought to be a
separate change, however.


r~