qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Laurent Desnogues <laurent.desnogues@gmail.com>
To: Juha.Riihimaki@nokia.com
Cc: qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH 07/12] target-arm: fix neon vsri, vshl and vsli ops
Date: Wed, 21 Oct 2009 12:46:38 +0200	[thread overview]
Message-ID: <761ea48b0910210346t3b868d32x529b0893b79f9710@mail.gmail.com> (raw)
In-Reply-To: <FE83A2F3-69F5-4647-A3FE-CD18F4892693@nokia.com>

On Wed, Oct 21, 2009 at 12:17 PM,  <Juha.Riihimaki@nokia.com> wrote:
> Shift immediate value is incorrectly overwritten by a temporary
> variable in the processing of NEON vsri, vshl and vsli instructions.
>
> Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
> ---
> diff --git a/target-arm/translate.c b/target-arm/translate.c
> index 59bf7bc..c92ecc6 100644
> --- a/target-arm/translate.c
> +++ b/target-arm/translate.c
> @@ -4094,7 +4094,7 @@ static int disas_neon_data_insn(CPUState * env,
> DisasContext *s, uint32_t insn)
>      int pairwise;
>      int u;
>      int n;
> -    uint32_t imm;
> +    uint32_t imm, imm2;

I would prefer mask to imm2, but that's personal taste :-)

>      TCGv tmp;
>      TCGv tmp2;
>      TCGv tmp3;
> @@ -4624,31 +4624,31 @@ static int disas_neon_data_insn(CPUState *
> env, DisasContext *s, uint32_t insn)
>                              switch (size) {
>                              case 0:
>                                  if (op == 4)
> -                                    imm = 0xff >> -shift;
> +                                    imm2 = 0xff >> -shift;
>                                  else
> -                                    imm = (uint8_t)(0xff << shift);
> -                                imm |= imm << 8;
> -                                imm |= imm << 16;
> +                                    imm2 = (uint8_t)(0xff << shift);
> +                                imm2 |= imm2 << 8;
> +                                imm2 |= imm2 << 16;
>                                  break;
>                              case 1:
>                                  if (op == 4)
> -                                    imm = 0xffff >> -shift;
> +                                    imm2 = 0xffff >> -shift;
>                                  else
> -                                    imm = (uint16_t)(0xffff << shift);
> -                                imm |= imm << 16;
> +                                    imm2 = (uint16_t)(0xffff << shift);
> +                                imm2 |= imm2 << 16;
>                                  break;
>                              case 2:
>                                  if (op == 4)
> -                                    imm = 0xffffffffu >> -shift;
> +                                    imm2 = 0xffffffffu >> -shift;
>                                  else
> -                                    imm = 0xffffffffu << shift;
> +                                    imm2 = 0xffffffffu << shift;
>                                  break;
>                              default:
>                                  abort();
>                              }
>                              tmp2 = neon_load_reg(rd, pass);
> -                            tcg_gen_andi_i32(tmp, tmp, imm);
> -                            tcg_gen_andi_i32(tmp2, tmp2, ~imm);
> +                            tcg_gen_andi_i32(tmp, tmp, imm2);
> +                            tcg_gen_andi_i32(tmp2, tmp2, ~imm2);
>                              tcg_gen_or_i32(tmp, tmp, tmp2);
>                              dead_tmp(tmp2);
>                          }

I mostly agree with this, except there's a mistake already
present in the original code:  for a size of 2 the shift amount
can be 32 (look at VSRI in the ARM Architecture Reference
Manual).  Note in this case, shift will be -32.


Laurent

  reply	other threads:[~2009-10-21 10:46 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-10-21 10:17 [Qemu-devel] [PATCH 07/12] target-arm: fix neon vsri, vshl and vsli ops Juha.Riihimaki
2009-10-21 10:46 ` Laurent Desnogues [this message]
2009-10-22  6:49   ` Juha.Riihimaki
2009-10-22  7:18     ` Laurent Desnogues
2009-10-22  7:33       ` Juha.Riihimaki
2009-10-22  7:40         ` Laurent Desnogues

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=761ea48b0910210346t3b868d32x529b0893b79f9710@mail.gmail.com \
    --to=laurent.desnogues@gmail.com \
    --cc=Juha.Riihimaki@nokia.com \
    --cc=qemu-devel@nongnu.org \
    /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).