qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Paolo Bonzini <pbonzini@redhat.com>
To: Richard Henderson <rth@twiddle.net>, qemu-devel@nongnu.org
Cc: aurelien@aurel32.net, HPENNER@de.ibm.com
Subject: Re: [Qemu-devel] [PATCH 2/2] tcg/optimize: Remember garbage high bits for 32-bit ops
Date: Fri, 23 May 2014 21:46:31 +0200	[thread overview]
Message-ID: <537FA597.3050703@redhat.com> (raw)
In-Reply-To: <1400871431-12655-3-git-send-email-rth@twiddle.net>

Il 23/05/2014 20:57, Richard Henderson ha scritto:
> For a 64-bit host, the high bits of a register after a 32-bit operation
> are undefined.  Adjust the temps mask for all 32-bit ops to reflect that.
>
> Signed-off-by: Richard Henderson <rth@twiddle.net>
> ---
>  tcg/optimize.c | 33 ++++++++++++++++++++++++++-------
>  1 file changed, 26 insertions(+), 7 deletions(-)
>
> diff --git a/tcg/optimize.c b/tcg/optimize.c
> index 83e1387..19e4831 100644
> --- a/tcg/optimize.c
> +++ b/tcg/optimize.c
> @@ -166,11 +166,18 @@ static void tcg_opt_gen_mov(TCGContext *s, int op_index, TCGArg *gen_args,
>                              TCGOpcode old_op, TCGArg dst, TCGArg src)
>  {
>      TCGOpcode new_op = op_to_mov(old_op);
> +    tcg_target_ulong mask;
>
>      s->gen_opc_buf[op_index] = new_op;
>
>      reset_temp(dst);
> -    temps[dst].mask = temps[src].mask;
> +    mask = temps[src].mask;
> +    if (TCG_TARGET_REG_BITS > 32 && new_op == INDEX_op_mov_i32) {
> +        /* High bits of the destination are now garbage.  */

Or they are zero on x86_64... perhaps this could be some kind of TCG 
target hook.

> +        mask |= ~0xffffffffull;
> +    }
> +    temps[dst].mask = mask;
> +
>      assert(temps[src].state != TCG_TEMP_CONST);
>
>      if (s->temps[src].type == s->temps[dst].type) {
> @@ -194,13 +201,20 @@ static void tcg_opt_gen_movi(TCGContext *s, int op_index, TCGArg *gen_args,
>                               TCGOpcode old_op, TCGArg dst, TCGArg val)
>  {
>      TCGOpcode new_op = op_to_movi(old_op);
> +    tcg_target_ulong mask;
>
>      s->gen_opc_buf[op_index] = new_op;
>
>      reset_temp(dst);
>      temps[dst].state = TCG_TEMP_CONST;
>      temps[dst].val = val;
> -    temps[dst].mask = val;
> +    mask = val;
> +    if (TCG_TARGET_REG_BITS > 32 && new_op == INDEX_op_mov_i32) {
> +        /* High bits of the destination are now garbage.  */
> +        mask |= ~0xffffffffull;
> +    }
> +    temps[dst].mask = mask;
> +
>      gen_args[0] = dst;
>      gen_args[1] = val;
>  }
> @@ -539,7 +553,7 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr,
>      for (op_index = 0; op_index < nb_ops; op_index++) {
>          TCGOpcode op = s->gen_opc_buf[op_index];
>          const TCGOpDef *def = &tcg_op_defs[op];
> -        tcg_target_ulong mask, affected;
> +        tcg_target_ulong mask, partmask, affected;
>          int nb_oargs, nb_iargs, nb_args, i;
>          TCGArg tmp;
>
> @@ -901,13 +915,18 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr,
>              break;
>          }
>
> -        /* 32-bit ops (non 64-bit ops and non load/store ops) generate 32-bit
> -           results */
> +        /* 32-bit ops (non 64-bit ops and non load/store ops) generate
> +           32-bit results.  For the result is zero test below, we can
> +           ignore high bits, but for further optimizations we need to
> +           record that the high bits contain garbage.  */
> +        partmask = mask;
>          if (!(def->flags & (TCG_OPF_CALL_CLOBBER | TCG_OPF_64BIT))) {
> -            mask &= 0xffffffffu;
> +            mask |= ~(tcg_target_ulong)0xffffffffu;
> +            partmask &= 0xffffffffu;
> +            affected &= 0xffffffffu;
>          }
>
> -        if (mask == 0) {
> +        if (partmask == 0) {
>              assert(nb_oargs == 1);
>              tcg_opt_gen_movi(s, op_index, gen_args, op, args[0], 0);
>              args += nb_args;
>

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

  reply	other threads:[~2014-05-23 19:46 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-05-23 18:57 [Qemu-devel] [PATCH 0/2] tcg/optimize fix for known-zero bits Richard Henderson
2014-05-23 18:57 ` [Qemu-devel] [PATCH 1/2] tcg/optimize: Move updating of gen_opc_buf into tcg_opt_gen_mov* Richard Henderson
2014-05-23 18:57 ` [Qemu-devel] [PATCH 2/2] tcg/optimize: Remember garbage high bits for 32-bit ops Richard Henderson
2014-05-23 19:46   ` Paolo Bonzini [this message]
2014-05-23 20:01     ` Richard Henderson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=537FA597.3050703@redhat.com \
    --to=pbonzini@redhat.com \
    --cc=HPENNER@de.ibm.com \
    --cc=aurelien@aurel32.net \
    --cc=qemu-devel@nongnu.org \
    --cc=rth@twiddle.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).