From: Stefan Markovic <smarkovic@wavecomp.com>
To: Aleksandar Markovic <aleksandar.markovic@rt-rk.com>,
"qemu-devel@nongnu.org" <qemu-devel@nongnu.org>,
"jancraig@amazon.com" <jancraig@amazon.com>,
Aleksandar Markovic <amarkovic@wavecomp.com>
Subject: Re: [Qemu-devel] [PATCH 6/6] target/mips: MXU: Add handlers for an align instruction
Date: Tue, 18 Dec 2018 15:19:18 +0000 [thread overview]
Message-ID: <47ccfd89-4927-98ff-ad18-3042d3c9e74d@wavecomp.com> (raw)
In-Reply-To: <20181217200444.14812-7-aleksandar.markovic@rt-rk.com>
On 17.12.18. 21:04, Aleksandar Markovic wrote:
> From: Aleksandar Markovic <amarkovic@wavecomp.com>
>
> Add translation handler for S32ALNI MXU instruction.
>
> Signed-off-by: Aleksandar Markovic <amarkovic@wavecomp.com>
> ---
> target/mips/translate.c | 197 +++++++++++++++++++++++++++++++++++++++-
> 1 file changed, 194 insertions(+), 3 deletions(-)
Reviewed-by: Stefan Markovic <smarkovic@wavecomp.com>
> diff --git a/target/mips/translate.c b/target/mips/translate.c
> index 339de8c32b..96905b78ac 100644
> --- a/target/mips/translate.c
> +++ b/target/mips/translate.c
> @@ -25147,6 +25147,199 @@ static void gen_mxu_Q8MAX_Q8MIN(DisasContext *ctx)
> }
>
>
> +/*
> + * MXU instruction category: align
> + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> + *
> + * S32ALN S32ALNI
> + */
> +
> +/*
> + * S32ALNI XRc, XRb, XRa, optn3
> + * Arrange bytes from XRb and XRc according to one of five sets of
> + * rules determined by optn3, and place the result in XRa.
> + *
> + * 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0
> + * +-----------+-----+---+-----+-------+-------+-------+-----------+
> + * | SPECIAL2 |optn3|0 0|x x x| XRc | XRb | XRa |MXU__POOL16|
> + * +-----------+-----+---+-----+-------+-------+-------+-----------+
> + *
> + */
> +static void gen_mxu_S32ALNI(DisasContext *ctx)
> +{
> + uint32_t optn3, pad, XRc, XRb, XRa;
> +
> + optn3 = extract32(ctx->opcode, 23, 3);
> + pad = extract32(ctx->opcode, 21, 2);
> + XRc = extract32(ctx->opcode, 14, 4);
> + XRb = extract32(ctx->opcode, 10, 4);
> + XRa = extract32(ctx->opcode, 6, 4);
> +
> + if (unlikely(pad != 0)) {
> + /* opcode padding incorrect -> do nothing */
> + } else if (unlikely(XRa == 0)) {
> + /* destination is zero register -> do nothing */
> + } else if (unlikely((XRb == 0) && (XRc == 0))) {
> + /* both operands zero registers -> just set destination to all 0s */
> + tcg_gen_movi_i32(mxu_gpr[XRa - 1], 0);
> + } else if (unlikely(XRb == 0)) {
> + /* XRb zero register -> just appropriatelly shift XRc into XRa */
> + switch (optn3) {
> + case MXU_OPTN3_PTN0:
> + tcg_gen_movi_i32(mxu_gpr[XRa - 1], 0);
> + break;
> + case MXU_OPTN3_PTN1:
> + case MXU_OPTN3_PTN2:
> + case MXU_OPTN3_PTN3:
> + tcg_gen_shri_i32(mxu_gpr[XRa - 1], mxu_gpr[XRc - 1],
> + 8 * (4 - optn3));
> + break;
> + case MXU_OPTN3_PTN4:
> + tcg_gen_mov_i32(mxu_gpr[XRa - 1], mxu_gpr[XRc - 1]);
> + break;
> + }
> + } else if (unlikely(XRc == 0)) {
> + /* XRc zero register -> just appropriatelly shift XRb into XRa */
> + switch (optn3) {
> + case MXU_OPTN3_PTN0:
> + tcg_gen_mov_i32(mxu_gpr[XRa - 1], mxu_gpr[XRb - 1]);
> + break;
> + case MXU_OPTN3_PTN1:
> + case MXU_OPTN3_PTN2:
> + case MXU_OPTN3_PTN3:
> + tcg_gen_shri_i32(mxu_gpr[XRa - 1], mxu_gpr[XRb - 1], 8 * optn3);
> + break;
> + case MXU_OPTN3_PTN4:
> + tcg_gen_movi_i32(mxu_gpr[XRa - 1], 0);
> + break;
> + }
> + } else if (unlikely(XRb == XRc)) {
> + /* both operands same -> just rotation or moving from any of them */
> + switch (optn3) {
> + case MXU_OPTN3_PTN0:
> + case MXU_OPTN3_PTN4:
> + tcg_gen_mov_i32(mxu_gpr[XRa - 1], mxu_gpr[XRb - 1]);
> + break;
> + case MXU_OPTN3_PTN1:
> + case MXU_OPTN3_PTN2:
> + case MXU_OPTN3_PTN3:
> + tcg_gen_rotli_i32(mxu_gpr[XRa - 1], mxu_gpr[XRb - 1], 8 * optn3);
> + break;
> + }
> + } else {
> + /* the most general case */
> + switch (optn3) {
> + case MXU_OPTN3_PTN0:
> + {
> + /* */
> + /* XRb XRc */
> + /* +---------------+ */
> + /* | A B C D | E F G H */
> + /* +-------+-------+ */
> + /* | */
> + /* XRa */
> + /* */
> +
> + tcg_gen_mov_i32(mxu_gpr[XRa - 1], mxu_gpr[XRb - 1]);
> + }
> + break;
> + case MXU_OPTN3_PTN1:
> + {
> + /* */
> + /* XRb XRc */
> + /* +-------------------+ */
> + /* A | B C D E | F G H */
> + /* +---------+---------+ */
> + /* | */
> + /* XRa */
> + /* */
> +
> + TCGv_i32 t0 = tcg_temp_new();
> + TCGv_i32 t1 = tcg_temp_new();
> +
> + tcg_gen_andi_i32(t0, mxu_gpr[XRb - 1], 0x00FFFFFF);
> + tcg_gen_shli_i32(t0, t0, 8);
> +
> + tcg_gen_andi_i32(t1, mxu_gpr[XRc - 1], 0xFF000000);
> + tcg_gen_shri_i32(t1, t1, 24);
> +
> + tcg_gen_or_i32(mxu_gpr[XRa - 1], t0, t1);
> +
> + tcg_temp_free(t1);
> + tcg_temp_free(t0);
> + }
> + break;
> + case MXU_OPTN3_PTN2:
> + {
> + /* */
> + /* XRb XRc */
> + /* +-------------------+ */
> + /* A B | C D E F | G H */
> + /* +---------+---------+ */
> + /* | */
> + /* XRa */
> + /* */
> +
> + TCGv_i32 t0 = tcg_temp_new();
> + TCGv_i32 t1 = tcg_temp_new();
> +
> + tcg_gen_andi_i32(t0, mxu_gpr[XRb - 1], 0x0000FFFF);
> + tcg_gen_shli_i32(t0, t0, 16);
> +
> + tcg_gen_andi_i32(t1, mxu_gpr[XRc - 1], 0xFFFF0000);
> + tcg_gen_shri_i32(t1, t1, 16);
> +
> + tcg_gen_or_i32(mxu_gpr[XRa - 1], t0, t1);
> +
> + tcg_temp_free(t1);
> + tcg_temp_free(t0);
> + }
> + break;
> + case MXU_OPTN3_PTN3:
> + {
> + /* */
> + /* XRb XRc */
> + /* +-------------------+ */
> + /* A B C | D E F G | H */
> + /* +---------+---------+ */
> + /* | */
> + /* XRa */
> + /* */
> +
> + TCGv_i32 t0 = tcg_temp_new();
> + TCGv_i32 t1 = tcg_temp_new();
> +
> + tcg_gen_andi_i32(t0, mxu_gpr[XRb - 1], 0x000000FF);
> + tcg_gen_shli_i32(t0, t0, 24);
> +
> + tcg_gen_andi_i32(t1, mxu_gpr[XRc - 1], 0xFFFFFF00);
> + tcg_gen_shri_i32(t1, t1, 8);
> +
> + tcg_gen_or_i32(mxu_gpr[XRa - 1], t0, t1);
> +
> + tcg_temp_free(t1);
> + tcg_temp_free(t0);
> + }
> + break;
> + case MXU_OPTN3_PTN4:
> + {
> + /* */
> + /* XRb XRc */
> + /* +---------------+ */
> + /* A B C D | E F G H | */
> + /* +-------+-------+ */
> + /* | */
> + /* XRa */
> + /* */
> +
> + tcg_gen_mov_i32(mxu_gpr[XRa - 1], mxu_gpr[XRc - 1]);
> + }
> + break;
> + }
> + }
> +}
> +
> +
> /*
> * Decoding engine for MXU
> * =======================
> @@ -25809,9 +26002,7 @@ static void decode_opc_mxu__pool16(CPUMIPSState *env, DisasContext *ctx)
> generate_exception_end(ctx, EXCP_RI);
> break;
> case OPC_MXU_S32ALNI:
> - /* TODO: Implement emulation of S32ALNI instruction. */
> - MIPS_INVAL("OPC_MXU_S32ALNI");
> - generate_exception_end(ctx, EXCP_RI);
> + gen_mxu_S32ALNI(ctx);
> break;
> case OPC_MXU_S32NOR:
> gen_mxu_S32NOR(ctx);
next prev parent reply other threads:[~2018-12-18 15:22 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-12-17 20:04 [Qemu-devel] [PATCH 0/6] target/mips: Amend MXU support Aleksandar Markovic
2018-12-17 20:04 ` [Qemu-devel] [PATCH 1/6] target/mips: MXU: Add missing opcodes/decoding for LX* instructions Aleksandar Markovic
2018-12-18 14:21 ` Stefan Markovic
2018-12-27 17:15 ` Janeczek, Craig
2018-12-27 18:44 ` Aleksandar Markovic
2018-12-27 18:50 ` Janeczek, Craig
2018-12-27 19:23 ` Aleksandar Markovic
2018-12-27 19:37 ` Janeczek, Craig
2018-12-27 20:12 ` Aleksandar Markovic
2018-12-17 20:04 ` [Qemu-devel] [PATCH 2/6] target/mips: MXU: Add generic naming for optn2 constants Aleksandar Markovic
2018-12-17 20:04 ` [Qemu-devel] [PATCH 3/6] target/mips: MXU: Improve textual description Aleksandar Markovic
2018-12-18 14:30 ` Stefan Markovic
2018-12-17 20:04 ` [Qemu-devel] [PATCH 4/6] target/mips: MXU: Add handlers for logic instructions Aleksandar Markovic
2018-12-18 14:39 ` Stefan Markovic
2018-12-17 20:04 ` [Qemu-devel] [PATCH 5/6] target/mips: MXU: Add handlers for max/min instructions Aleksandar Markovic
2018-12-18 15:05 ` Stefan Markovic
2018-12-27 17:18 ` Janeczek, Craig
2018-12-27 20:14 ` Aleksandar Markovic
2018-12-23 17:22 ` Aleksandar Markovic
2018-12-25 19:34 ` Richard Henderson
2021-03-15 21:26 ` Philippe Mathieu-Daudé
2018-12-17 20:04 ` [Qemu-devel] [PATCH 6/6] target/mips: MXU: Add handlers for an align instruction Aleksandar Markovic
2018-12-18 15:19 ` Stefan Markovic [this message]
2018-12-27 21:27 ` [Qemu-devel] [PATCH 0/6] target/mips: Amend MXU support Aleksandar Markovic
2018-12-31 14:40 ` Aleksandar Markovic
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=47ccfd89-4927-98ff-ad18-3042d3c9e74d@wavecomp.com \
--to=smarkovic@wavecomp.com \
--cc=aleksandar.markovic@rt-rk.com \
--cc=amarkovic@wavecomp.com \
--cc=jancraig@amazon.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).