On 08/12/2014 03:45 AM, Tom Musta wrote:
> The rlwnm specification includes the ROTL32 operation, which is defined
> to be a left rotation of two copies of the least significant 32 bits of
> the source GPR.
>
> The current implementation is incorrect on 64-bit implementations in that
> it rotates a single copy of the least significant 32 bits, padding with
> zeroes in the most significant bits.
>
> Fix the code to properly implement this ROTL32 operation.
>
> Example:
>
> R3 = 0000000000000002
> R4 = 7FFFFFFFFFFFFFFF
> rlwnm 3,3,4,31,16
> R3 expected : 0000000100000001
> R3 actual : 0000000000000001 (without this patch)
>
> Signed-off-by: Tom Musta <tommu...@gmail.com>
> ---
> target-ppc/translate.c | 18 +++++++++---------
> 1 files changed, 9 insertions(+), 9 deletions(-)
>
> diff --git a/target-ppc/translate.c b/target-ppc/translate.c
> index a27d063..48f13a9 100644
> --- a/target-ppc/translate.c
> +++ b/target-ppc/translate.c
> @@ -1695,7 +1695,7 @@ static void gen_rlwnm(DisasContext *ctx)
> uint32_t mb, me;
> TCGv t0;
> #if defined(TARGET_PPC64)
> - TCGv_i32 t1, t2;
> + TCGv t1;
> #endif
>
> mb = MB(ctx->opcode);
> @@ -1703,14 +1703,11 @@ static void gen_rlwnm(DisasContext *ctx)
> t0 = tcg_temp_new();
> tcg_gen_andi_tl(t0, cpu_gpr[rB(ctx->opcode)], 0x1f);
> #if defined(TARGET_PPC64)
> - t1 = tcg_temp_new_i32();
> - t2 = tcg_temp_new_i32();
> - tcg_gen_trunc_i64_i32(t1, cpu_gpr[rS(ctx->opcode)]);
> - tcg_gen_trunc_i64_i32(t2, t0);
> - tcg_gen_rotl_i32(t1, t1, t2);
> - tcg_gen_extu_i32_i64(t0, t1);
> - tcg_temp_free_i32(t1);
> - tcg_temp_free_i32(t2);
> + t1 = tcg_temp_new_i64();
> + tcg_gen_deposit_i64(t1, cpu_gpr[rS(ctx->opcode)],
> + cpu_gpr[rS(ctx->opcode)], 32, 32);
> + tcg_gen_rotl_i64(t0, t1, t0);
> + tcg_temp_free_i64(t1);
> #else
> tcg_gen_rotl_i32(t0, cpu_gpr[rS(ctx->opcode)], t0);
> #endif
> @@ -1721,6 +1718,9 @@ static void gen_rlwnm(DisasContext *ctx)
> #endif
> tcg_gen_andi_tl(cpu_gpr[rA(ctx->opcode)], t0, MASK(mb, me));
> } else {
> +#if defined(TARGET_PPC64)
> + tcg_gen_andi_tl(t0, t0, MASK(32, 63));
> +#endif
Similarly, checking early for the MB==0, ME==31 case and generating a straight
32-bit rotate would be good. Note that ANDI will already special case for the
constant -1, so there's no need to check for that as a mov.
r~
