From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:57488)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1Vbaa0-0002Oj-8z
	for qemu-devel@nongnu.org; Wed, 30 Oct 2013 14:33:28 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1VbaZr-0006mQ-1Z
	for qemu-devel@nongnu.org; Wed, 30 Oct 2013 14:33:20 -0400
Received: from mail-qe0-x230.google.com ([2607:f8b0:400d:c02::230]:37289)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <rth7680@gmail.com>) id 1VbaZq-0006mM-Us
	for qemu-devel@nongnu.org; Wed, 30 Oct 2013 14:33:11 -0400
Received: by mail-qe0-f48.google.com with SMTP id d4so1075016qej.21
	for <qemu-devel@nongnu.org>; Wed, 30 Oct 2013 11:33:10 -0700 (PDT)
Sender: Richard Henderson <rth7680@gmail.com>
Message-ID: <527150E2.5070200@twiddle.net>
Date: Wed, 30 Oct 2013 11:33:06 -0700
From: Richard Henderson <rth@twiddle.net>
MIME-Version: 1.0
References: <1383073495-5332-1-git-send-email-sebastian@macke.de>
	<1383073495-5332-12-git-send-email-sebastian@macke.de>
In-Reply-To: <1383073495-5332-12-git-send-email-sebastian@macke.de>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [PATCH 11/13] target-openrisc: use jmp_pc as flag
 variable for branches
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: Sebastian Macke <sebastian@macke.de>
Cc: openrisc@openrisc.net, openrisc@lists.opencores.org, qemu-devel@nongnu.org, proljc@gmail.com

On 10/29/2013 12:04 PM, Sebastian Macke wrote:
>          {
>              int lab = gen_new_label();
> -            dc->btaken = tcg_temp_local_new();
> -            tcg_gen_movi_tl(jmp_pc, dc->pc+8);
> -            tcg_gen_movi_tl(dc->btaken, 0);
> +            tcg_gen_movi_tl(jmp_pc, 0);
>              tcg_gen_brcondi_i32(op0 == 0x03 ? TCG_COND_NE : TCG_COND_EQ,
>                                  cpu_srf, 0, lab);
> -            tcg_gen_movi_tl(dc->btaken, 1);
>              tcg_gen_movi_tl(jmp_pc, tmp_pc);
>              gen_set_label(lab);

You can now use movcond here:

  tcg_gen_movi_i32(jmp_pc, tmp_pc);
  tcg_gen_movcond_i32(jmp_pc, cpu_srf, zero, jmp_pc, zero,
                      op0 == 0x03 ? TCG_COND_NE : TCG_COND_EQ);

Although I'd wonder about just using setcond instead, since I think
the value stored in jmp_pc here is also stored in dc->j_target, leading
to the right behaviour...

>      case JUMP_BRANCH:
>          {
>              int l1 = gen_new_label();
> -            tcg_gen_brcondi_tl(TCG_COND_NE, dc->btaken, 0, l1);
> +            tcg_gen_brcondi_tl(TCG_COND_NE, jmp_pc, 0, l1);
>              gen_goto_tb(dc, 1, dc->pc);
>              gen_set_label(l1);
> -            tcg_temp_free(dc->btaken);
>              gen_goto_tb(dc, 0, dc->j_target);
>              break;

... here.

> +    case JUMP_BRANCH_DELAYED:
> +        {
> +            int l1 = gen_new_label();
> +            tcg_gen_brcondi_tl(TCG_COND_NE, jmp_pc, 0, l1);
> +            gen_goto_tb(dc, 1, dc->pc);
> +            gen_set_label(l1);
> +            tcg_gen_mov_tl(cpu_pc, jmp_pc);
> +            tcg_gen_exit_tb(0);
> +            break;
...
> +
> +    dc->delayed_branch = !!(dc->tb_flags & D_FLAG);
> +    if ((dc->delayed_branch) && (dc->tb_flags&B_FLAG)) {
> +        dc->j_state = JUMP_BRANCH_DELAYED;
> +    }
> +


And thus I can't see how these additions are actually useful?

If I've missed something, and the last hunk needs to be retained,
then please fix the coding style:

  if (dc->delayed_branch && (dc->tb_flags & B_FLAG)) {


r~