qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Paolo Bonzini <pbonzini@redhat.com>
To: qemu-devel@nongnu.org
Subject: [PATCH 21/37] target/i386: reimplement 0x0f 0xd8-0xdf, 0xe8-0xef, 0xf8-0xff, add AVX
Date: Mon, 12 Sep 2022 01:04:01 +0200	[thread overview]
Message-ID: <20220911230418.340941-22-pbonzini@redhat.com> (raw)
In-Reply-To: <20220911230418.340941-1-pbonzini@redhat.com>

These are more simple integer instructions present in both MMX and SSE/AVX,
with no holes that were later occupied by newer instructions.

Simple, non-saturating operations are implemented using gvec; apart
from this, there is not much to talk about.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 target/i386/tcg/decode-new.c.inc |  28 ++++++++
 target/i386/tcg/emit.c.inc       | 113 +++++++++++++++++++++++++++++++
 target/i386/tcg/translate.c      |   4 +-
 3 files changed, 144 insertions(+), 1 deletion(-)

diff --git a/target/i386/tcg/decode-new.c.inc b/target/i386/tcg/decode-new.c.inc
index f20587c096..59f5637583 100644
--- a/target/i386/tcg/decode-new.c.inc
+++ b/target/i386/tcg/decode-new.c.inc
@@ -264,6 +264,34 @@ static const X86OpEntry opcodes_0F[256] = {
     [0x6d] = X86_OP_ENTRY3(PUNPCKHQDQ, V,x, H,x, W,x,  vex4 p_66 avx2_256),
     [0x6e] = X86_OP_ENTRY3(MOVD_to,    V,x, None,None, E,y, vex5 mmx p_00_66),  /* wrong dest Vy on SDM! */
     [0x6f] = X86_OP_GROUP3(0F6F,       V,x, None,None, W,x, vex5 mmx p_00_66_f3),
+
+    /* Incorrectly missing from 2-17 */
+    [0xd8] = X86_OP_ENTRY3(PSUBUSB,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xd9] = X86_OP_ENTRY3(PSUBUSW,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xda] = X86_OP_ENTRY3(PMINUB,   V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xdb] = X86_OP_ENTRY3(PAND,     V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xdc] = X86_OP_ENTRY3(PADDUSB,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xdd] = X86_OP_ENTRY3(PADDUSW,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xde] = X86_OP_ENTRY3(PMAXUB,   V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xdf] = X86_OP_ENTRY3(PANDN,    V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+
+    [0xe8] = X86_OP_ENTRY3(PSUBSB,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xe9] = X86_OP_ENTRY3(PSUBSW,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xea] = X86_OP_ENTRY3(PMINSW,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xeb] = X86_OP_ENTRY3(POR,     V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xec] = X86_OP_ENTRY3(PADDSB,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xed] = X86_OP_ENTRY3(PADDSW,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xee] = X86_OP_ENTRY3(PMAXSW,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xef] = X86_OP_ENTRY3(PXOR,    V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+
+    [0xf8] = X86_OP_ENTRY3(PSUBB,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xf9] = X86_OP_ENTRY3(PSUBW,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xfa] = X86_OP_ENTRY3(PSUBD,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xfb] = X86_OP_ENTRY3(PSUBQ,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xfc] = X86_OP_ENTRY3(PADDB,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xfd] = X86_OP_ENTRY3(PADDW,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    [0xfe] = X86_OP_ENTRY3(PADDD,  V,x, H,x, W,x,  vex4 mmx avx2_256 p_00_66),
+    /* 0xff = UD0 */
 };
 
 static void do_decode_0F(DisasContext *s, CPUX86State *env, X86OpEntry *entry, uint8_t *b)
diff --git a/target/i386/tcg/emit.c.inc b/target/i386/tcg/emit.c.inc
index 3f89d3cf50..1ba7a45668 100644
--- a/target/i386/tcg/emit.c.inc
+++ b/target/i386/tcg/emit.c.inc
@@ -290,6 +290,20 @@ BINARY_INT_MMX(PUNPCKHWD,  punpckhwd)
 BINARY_INT_MMX(PUNPCKHDQ,  punpckhdq)
 BINARY_INT_MMX(PACKSSDW,   packssdw)
 
+BINARY_INT_MMX(PSUBUSB, psubusb)
+BINARY_INT_MMX(PSUBUSW, psubusw)
+BINARY_INT_MMX(PMINUB,  pminub)
+BINARY_INT_MMX(PADDUSB, paddusb)
+BINARY_INT_MMX(PADDUSW, paddusw)
+BINARY_INT_MMX(PMAXUB,  pmaxub)
+
+BINARY_INT_MMX(PSUBSB, psubsb)
+BINARY_INT_MMX(PSUBSW, psubsw)
+BINARY_INT_MMX(PMINSW, pminsw)
+BINARY_INT_MMX(PADDSB, paddsb)
+BINARY_INT_MMX(PADDSW, paddsw)
+BINARY_INT_MMX(PMAXSW, pmaxsw)
+
 /* Instructions with no MMX equivalent.  */
 #define BINARY_INT_SSE(uname, lname)                                               \
 static void gen_##uname(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode) \
@@ -526,6 +540,51 @@ static void gen_MULX(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
 
 }
 
+static void gen_PADDB(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_add(MO_8,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PADDW(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_add(MO_16,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PADDD(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_add(MO_32,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PAND(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_and(MO_64,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PANDN(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_andc(MO_64,
+                      decode->op[0].offset, decode->op[2].offset,
+                      decode->op[1].offset, vec_len, vec_len);
+}
+
 static void gen_PCMPGTB(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
 {
     int vec_len = sse_vec_len(s, decode);
@@ -571,6 +630,60 @@ static void gen_PEXT(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
     gen_helper_pext(s->T0, s->T0, s->T1);
 }
 
+static void gen_POR(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_or(MO_64,
+                    decode->op[0].offset, decode->op[1].offset,
+                    decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PXOR(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_xor(MO_64,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PSUBB(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_sub(MO_8,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PSUBW(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_sub(MO_16,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PSUBD(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_sub(MO_32,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
+static void gen_PSUBQ(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
+{
+    int vec_len = sse_vec_len(s, decode);
+
+    tcg_gen_gvec_sub(MO_64,
+                     decode->op[0].offset, decode->op[1].offset,
+                     decode->op[2].offset, vec_len, vec_len);
+}
+
 static void gen_RORX(DisasContext *s, CPUX86State *env, X86DecodedInsn *decode)
 {
     MemOp ot = decode->op[0].ot;
diff --git a/target/i386/tcg/translate.c b/target/i386/tcg/translate.c
index cf18e12d38..11c17258eb 100644
--- a/target/i386/tcg/translate.c
+++ b/target/i386/tcg/translate.c
@@ -4666,7 +4666,9 @@ static target_ulong disas_insn(DisasContext *s, CPUState *cpu)
 #ifndef CONFIG_USER_ONLY
         use_new &= b <= limit;
 #endif
-        if (use_new && (b >= 0x160 && b <= 0x16f)) {
+        if (use_new &&
+            ((b >= 0x160 && b <= 0x16f) ||
+             (b >= 0x1d8 && b <= 0x1ff && (b & 8)))) {
             return disas_insn_new(s, cpu, b + 0x100);
         }
         break;
-- 
2.37.2




  parent reply	other threads:[~2022-09-11 23:37 UTC|newest]

Thread overview: 86+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-11 23:03 [RFC PATCH 00/37] target/i386: new decoder + AVX implementation Paolo Bonzini
2022-09-11 23:03 ` [PATCH 01/37] target/i386: Define XMMReg and access macros, align ZMM registers Paolo Bonzini
2022-09-11 23:03 ` [PATCH 02/37] target/i386: make ldo/sto operations consistent with ldq Paolo Bonzini
2022-09-12  8:33   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 03/37] target/i386: REPZ and REPNZ are mutually exclusive Paolo Bonzini
2022-09-12  8:37   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 04/37] target/i386: introduce insn_get_addr Paolo Bonzini
2022-09-12  8:39   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 05/37] target/i386: add core of new i386 decoder Paolo Bonzini
2022-09-12  9:27   ` Richard Henderson
2022-09-12 10:54   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 06/37] target/i386: add ALU load/writeback core Paolo Bonzini
2022-09-12 10:02   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 07/37] target/i386: add CPUID[EAX=7, ECX=0].ECX to DisasContext Paolo Bonzini
2022-09-12 10:02   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 08/37] target/i386: add CPUID feature checks to new decoder Paolo Bonzini
2022-09-12 10:05   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 09/37] target/i386: add AVX_EN hflag Paolo Bonzini
2022-09-12 10:06   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 10/37] target/i386: validate VEX prefixes via the instructions' exception classes Paolo Bonzini
2022-09-12 10:39   ` Richard Henderson
2022-09-12 10:42   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 11/37] target/i386: validate SSE prefixes directly in the decoding table Paolo Bonzini
2022-09-12 10:51   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 12/37] target/i386: add scalar 0F 38 and 0F 3A instruction to new decoder Paolo Bonzini
2022-09-12 11:04   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 13/37] target/i386: remove scalar VEX instructions from old decoder Paolo Bonzini
2022-09-12 11:06   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 14/37] target/i386: Prepare ops_sse_header.h for 256 bit AVX Paolo Bonzini
2022-09-12 11:09   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 15/37] target/i386: extend helpers to support VEX.V 3- and 4- operand encodings Paolo Bonzini
2022-09-12 11:11   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 16/37] target/i386: support operand merging in binary scalar helpers Paolo Bonzini
2022-09-12 11:11   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 17/37] target/i386: provide 3-operand versions of unary " Paolo Bonzini
2022-09-12 11:14   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 18/37] target/i386: implement additional AVX comparison operators Paolo Bonzini
2022-09-12 11:19   ` Richard Henderson
2022-09-11 23:03 ` [PATCH 19/37] target/i386: Introduce 256-bit vector helpers Paolo Bonzini
2022-09-12 11:19   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 20/37] target/i386: reimplement 0x0f 0x60-0x6f, add AVX Paolo Bonzini
2022-09-12 11:41   ` Richard Henderson
2022-09-13 10:56     ` Paolo Bonzini
2022-09-13 11:35       ` Richard Henderson
2022-09-12 13:01   ` Richard Henderson
2022-09-11 23:04 ` Paolo Bonzini [this message]
2022-09-12 13:19   ` [PATCH 21/37] target/i386: reimplement 0x0f 0xd8-0xdf, 0xe8-0xef, 0xf8-0xff, " Richard Henderson
2022-09-11 23:04 ` [PATCH 22/37] target/i386: reimplement 0x0f 0x50-0x5f, " Paolo Bonzini
2022-09-12 13:46   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 23/37] target/i386: reimplement 0x0f 0x78-0x7f, " Paolo Bonzini
2022-09-12 13:56   ` Richard Henderson
2022-09-14 16:17     ` Paolo Bonzini
2022-09-11 23:04 ` [PATCH 24/37] target/i386: reimplement 0x0f 0x70-0x77, " Paolo Bonzini
2022-09-12 14:29   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 25/37] target/i386: reimplement 0x0f 0xd0-0xd7, 0xe0-0xe7, 0xf0-0xf7, " Paolo Bonzini
2022-09-12 15:06   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 26/37] target/i386: reimplement 0x0f 0x3a, " Paolo Bonzini
2022-09-12 15:33   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 27/37] target/i386: Use tcg gvec ops for pmovmskb Paolo Bonzini
2022-09-13  8:16   ` Richard Henderson
2022-09-14 22:59     ` Paolo Bonzini
2022-09-15  6:48       ` Richard Henderson
2022-09-11 23:04 ` [PATCH 28/37] target/i386: reimplement 0x0f 0x38, add AVX Paolo Bonzini
2022-09-13  9:31   ` Richard Henderson
2022-09-14 17:04     ` Paolo Bonzini
2022-09-15  6:50       ` Richard Henderson
2022-09-11 23:04 ` [PATCH 29/37] target/i386: reimplement 0x0f 0xc2, 0xc4-0xc6, " Paolo Bonzini
2022-09-13  9:44   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 30/37] target/i386: reimplement 0x0f 0x10-0x17, " Paolo Bonzini
2022-09-13 10:14   ` Richard Henderson
2022-09-14 22:45     ` Paolo Bonzini
2022-09-15  6:51       ` Richard Henderson
2022-09-13 10:38   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 31/37] target/i386: reimplement 0x0f 0x28-0x2f, " Paolo Bonzini
2022-09-13 10:24   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 32/37] target/i386: implement XSAVE and XRSTOR of AVX registers Paolo Bonzini
2022-09-13 10:27   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 33/37] target/i386: Enable AVX cpuid bits when using TCG Paolo Bonzini
2022-09-13 10:28   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 34/37] target/i386: implement VLDMXCSR/VSTMXCSR Paolo Bonzini
2022-09-13 10:32   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 35/37] tests/tcg: extend SSE tests to AVX Paolo Bonzini
2022-09-13 10:33   ` Richard Henderson
2022-09-11 23:04 ` [PATCH 36/37] target/i386: move 3DNow completely out of gen_sse Paolo Bonzini
2022-09-13 10:34   ` Richard Henderson
2022-09-13 10:39 ` [RFC PATCH 00/37] target/i386: new decoder + AVX implementation 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=20220911230418.340941-22-pbonzini@redhat.com \
    --to=pbonzini@redhat.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).