]> pilppa.com Git - linux-2.6-omap-h63xx.git/commitdiff
[POWERPC] Import spu disassembly code into xmon
authorMichael Ellerman <michael@ellerman.id.au>
Wed, 22 Nov 2006 23:46:43 +0000 (00:46 +0100)
committerPaul Mackerras <paulus@samba.org>
Mon, 4 Dec 2006 09:40:29 +0000 (20:40 +1100)
This patch imports and munges the spu disassembly code from binutils.

All files originated from version 1.1 in binutils cvs.
 * spu.h, spu-insns.h and spu-opc.c are unchanged except for pathnames.
 * spu-dis.c has been edited heavily:
   * use printf instead of info->fprintf_func and similar.
   * pass the instruction in rather than reading it.
   * we have no equivalent to symbol_at_address_func, so we just assume
     there is never a symbol at the address given.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
arch/powerpc/xmon/spu-dis.c [new file with mode: 0644]
arch/powerpc/xmon/spu-insns.h [new file with mode: 0644]
arch/powerpc/xmon/spu-opc.c [new file with mode: 0644]
arch/powerpc/xmon/spu.h [new file with mode: 0644]

diff --git a/arch/powerpc/xmon/spu-dis.c b/arch/powerpc/xmon/spu-dis.c
new file mode 100644 (file)
index 0000000..75ac081
--- /dev/null
@@ -0,0 +1,249 @@
+/* Disassemble SPU instructions
+
+   Copyright 2006 Free Software Foundation, Inc.
+
+   This file is part of GDB, GAS, and the GNU binutils.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License along
+   with this program; if not, write to the Free Software Foundation, Inc.,
+   51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.  */
+
+#include <linux/string.h>
+#include "nonstdio.h"
+#include "ansidecl.h"
+#include "spu.h"
+
+extern void print_address (unsigned long memaddr);
+
+/* This file provides a disassembler function which uses
+   the disassembler interface defined in dis-asm.h.   */
+
+extern const struct spu_opcode spu_opcodes[];
+extern const int spu_num_opcodes;
+
+#define SPU_DISASM_TBL_SIZE (1 << 11)
+static const struct spu_opcode *spu_disassemble_table[SPU_DISASM_TBL_SIZE];
+
+static void
+init_spu_disassemble (void)
+{
+  int i;
+
+  /* If two instructions have the same opcode then we prefer the first
+   * one.  In most cases it is just an alternate mnemonic. */
+  for (i = 0; i < spu_num_opcodes; i++)
+    {
+      int o = spu_opcodes[i].opcode;
+      if (o >= SPU_DISASM_TBL_SIZE)
+       continue; /* abort (); */
+      if (spu_disassemble_table[o] == 0)
+       spu_disassemble_table[o] = &spu_opcodes[i];
+    }
+}
+
+/* Determine the instruction from the 10 least significant bits. */
+static const struct spu_opcode *
+get_index_for_opcode (unsigned int insn)
+{
+  const struct spu_opcode *index;
+  unsigned int opcode = insn >> (32-11);
+
+  /* Init the table.  This assumes that element 0/opcode 0 (currently
+   * NOP) is always used */
+  if (spu_disassemble_table[0] == 0)
+    init_spu_disassemble ();
+
+  if ((index = spu_disassemble_table[opcode & 0x780]) != 0
+      && index->insn_type == RRR)
+    return index;
+
+  if ((index = spu_disassemble_table[opcode & 0x7f0]) != 0
+      && (index->insn_type == RI18 || index->insn_type == LBT))
+    return index;
+
+  if ((index = spu_disassemble_table[opcode & 0x7f8]) != 0
+      && index->insn_type == RI10)
+    return index;
+
+  if ((index = spu_disassemble_table[opcode & 0x7fc]) != 0
+      && (index->insn_type == RI16))
+    return index;
+
+  if ((index = spu_disassemble_table[opcode & 0x7fe]) != 0
+      && (index->insn_type == RI8))
+    return index;
+
+  if ((index = spu_disassemble_table[opcode & 0x7ff]) != 0)
+    return index;
+
+  return 0;
+}
+
+/* Print a Spu instruction.  */
+
+int
+print_insn_spu (unsigned long insn, unsigned long memaddr)
+{
+  int value;
+  int hex_value;
+  const struct spu_opcode *index;
+  enum spu_insns tag;
+
+  index = get_index_for_opcode (insn);
+
+  if (index == 0)
+    {
+      printf(".long 0x%x", insn);
+    }
+  else
+    {
+      int i;
+      int paren = 0;
+      tag = (enum spu_insns)(index - spu_opcodes);
+      printf("%s", index->mnemonic);
+      if (tag == M_BI || tag == M_BISL || tag == M_IRET || tag == M_BISLED
+         || tag == M_BIHNZ || tag == M_BIHZ || tag == M_BINZ || tag == M_BIZ
+          || tag == M_SYNC || tag == M_HBR)
+       {
+         int fb = (insn >> (32-18)) & 0x7f;
+         if (fb & 0x40)
+           printf(tag == M_SYNC ? "c" : "p");
+         if (fb & 0x20)
+           printf("d");
+         if (fb & 0x10)
+           printf("e");
+       }
+      if (index->arg[0] != 0)
+       printf("\t");
+      hex_value = 0;
+      for (i = 1;  i <= index->arg[0]; i++)
+       {
+         int arg = index->arg[i];
+         if (arg != A_P && !paren && i > 1)
+           printf(",");
+
+         switch (arg)
+           {
+           case A_T:
+             printf("$%d",
+                                    DECODE_INSN_RT (insn));
+             break;
+           case A_A:
+             printf("$%d",
+                                    DECODE_INSN_RA (insn));
+             break;
+           case A_B:
+             printf("$%d",
+                                    DECODE_INSN_RB (insn));
+             break;
+           case A_C:
+             printf("$%d",
+                                    DECODE_INSN_RC (insn));
+             break;
+           case A_S:
+             printf("$sp%d",
+                                    DECODE_INSN_RA (insn));
+             break;
+           case A_H:
+             printf("$ch%d",
+                                    DECODE_INSN_RA (insn));
+             break;
+           case A_P:
+             paren++;
+             printf("(");
+             break;
+           case A_U7A:
+             printf("%d",
+                                    173 - DECODE_INSN_U8 (insn));
+             break;
+           case A_U7B:
+             printf("%d",
+                                    155 - DECODE_INSN_U8 (insn));
+             break;
+           case A_S3:
+           case A_S6:
+           case A_S7:
+           case A_S7N:
+           case A_U3:
+           case A_U5:
+           case A_U6:
+           case A_U7:
+             hex_value = DECODE_INSN_I7 (insn);
+             printf("%d", hex_value);
+             break;
+           case A_S11:
+             print_address(memaddr + DECODE_INSN_I9a (insn) * 4);
+             break;
+           case A_S11I:
+             print_address(memaddr + DECODE_INSN_I9b (insn) * 4);
+             break;
+           case A_S10:
+           case A_S10B:
+             hex_value = DECODE_INSN_I10 (insn);
+             printf("%d", hex_value);
+             break;
+           case A_S14:
+             hex_value = DECODE_INSN_I10 (insn) * 16;
+             printf("%d", hex_value);
+             break;
+           case A_S16:
+             hex_value = DECODE_INSN_I16 (insn);
+             printf("%d", hex_value);
+             break;
+           case A_X16:
+             hex_value = DECODE_INSN_U16 (insn);
+             printf("%u", hex_value);
+             break;
+           case A_R18:
+             value = DECODE_INSN_I16 (insn) * 4;
+             if (value == 0)
+               printf("%d", value);
+             else
+               {
+                 hex_value = memaddr + value;
+                 print_address(hex_value & 0x3ffff);
+               }
+             break;
+           case A_S18:
+             value = DECODE_INSN_U16 (insn) * 4;
+             if (value == 0)
+               printf("%d", value);
+             else
+               print_address(value);
+             break;
+           case A_U18:
+             value = DECODE_INSN_U18 (insn);
+             if (value == 0 || 1)
+               {
+                 hex_value = value;
+                 printf("%u", value);
+               }
+             else
+               print_address(value);
+             break;
+           case A_U14:
+             hex_value = DECODE_INSN_U14 (insn);
+             printf("%u", hex_value);
+             break;
+           }
+         if (arg != A_P && paren)
+           {
+             printf(")");
+             paren--;
+           }
+       }
+      if (hex_value > 16)
+       printf("\t# %x", hex_value);
+    }
+  return 4;
+}
diff --git a/arch/powerpc/xmon/spu-insns.h b/arch/powerpc/xmon/spu-insns.h
new file mode 100644 (file)
index 0000000..99dc452
--- /dev/null
@@ -0,0 +1,410 @@
+/* SPU ELF support for BFD.
+
+   Copyright 2006 Free Software Foundation, Inc.
+
+   This file is part of BFD, the Binary File Descriptor library.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program; if not, write to the Free Software Foundation,
+   Inc., 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.  */
+
+/* SPU Opcode Table
+
+-=-=-= FORMAT =-=-=-
+                                                                                                    
+       +----+-------+-------+-------+-------+                         +------------+-------+-------+-------+
+RRR    | op |  RC   |  RB   |  RA   |  RT   |          RI7    | op         |  I7   |  RA   |  RT   |
+       +----+-------+-------+-------+-------+                 +------------+-------+-------+-------+
+        0  3       1       1       2       3                   0          1       1       2       3 
+                   0       7       4       1                              0       7       4       1 
+
+       +-----------+--------+-------+-------+                 +---------+----------+-------+-------+
+RI8    | op        |   I8   |  RA   |  RT   |          RI10   | op      |   I10    |  RA   |  RT   |
+       +-----------+--------+-------+-------+                 +---------+----------+-------+-------+
+        0         9        1       2       3                   0       7          1       2       3 
+                           7       4       1                                      7       4       1 
+
+       +----------+-----------------+-------+                 +--------+-------------------+-------+
+RI16   | op       |       I16       |  RT   |          RI18   | op     |       I18         |  RT   |
+       +----------+-----------------+-------+                 +--------+-------------------+-------+
+        0        8                 2       3                   0      6                   2       3 
+                                   4       1                                              4       1 
+
+       +------------+-------+-------+-------+                 +-------+--+-----------------+-------+
+RR     | op         |  RB   |  RA   |  RT   |          LBT    | op    |RO|       I16       |  RO   |
+       +------------+-------+-------+-------+                 +-------+--+-----------------+-------+
+        0          1       1       2       3                   0     6  8                 2       3 
+                   0       7       4       1                                              4       1 
+
+                                                              +------------+----+--+-------+-------+
+                                                       LBTI   | op         | // |RO|  RA   |  RO   |
+                                                              +------------+----+--+-------+-------+
+                                                               0          1    1  1       2       3
+                                                                          0    5  7       4       1
+
+-=-=-= OPCODE =-=-=-
+
+OPCODE field specifies the most significant 11bit of the instruction. Some formats don't have 11bits for opcode field, and in this
+case, bit field other than op are defined as 0s. For example, opcode of fma instruction which is RRR format is defined as 0x700,
+since 0x700 -> 11'b11100000000, this means opcode is 4'b1110, and other 7bits are defined as 7'b0000000.
+
+-=-=-= ASM_FORMAT =-=-=-
+
+RRR category                                           RI7 category                               
+       ASM_RRR         mnemonic RC, RA, RB, RT                 ASM_RI4         mnemonic RT, RA, I4
+                                                               ASM_RI7         mnemonic RT, RA, I7
+
+RI8 category                                           RI10 category                               
+       ASM_RUI8        mnemonic RT, RA, UI8                    ASM_AI10        mnemonic RA, I10    
+                                                               ASM_RI10        mnemonic RT, RA, R10
+                                                               ASM_RI10IDX     mnemonic RT, I10(RA)
+
+RI16 category                                          RI18 category                           
+       ASM_I16W        mnemonic I16W                           ASM_RI18        mnemonic RT, I18
+       ASM_RI16        mnemonic RT, I16
+       ASM_RI16W       mnemonic RT, I16W
+
+RR category                                            LBT category                                    
+       ASM_MFSPR       mnemonic RT, SA                         ASM_LBT         mnemonic brinst, brtarg 
+       ASM_MTSPR       mnemonic SA, RT                                                                 
+       ASM_NOOP        mnemonic                        LBTI category                                   
+       ASM_RA          mnemonic RA                             ASM_LBTI        mnemonic brinst, RA     
+       ASM_RAB         mnemonic RA, RB
+       ASM_RDCH        mnemonic RT, CA
+       ASM_RR          mnemonic RT, RA, RB
+       ASM_RT          mnemonic RT
+       ASM_RTA         mnemonic RT, RA
+       ASM_WRCH        mnemonic CA, RT
+
+Note that RRR instructions have the names for RC and RT reversed from
+what's in the ISA, in order to put RT in the same position it appears
+for other formats.
+
+-=-=-= DEPENDENCY =-=-=-
+
+DEPENDENCY filed consists of 5 digits. This represents which register is used as source and which register is used as target.
+The first(most significant) digit is always 0. Then it is followd by RC, RB, RA and RT digits.
+If the digit is 0, this means the corresponding register is not used in the instruction.
+If the digit is 1, this means the corresponding register is used as a source in the instruction.
+If the digit is 2, this means the corresponding register is used as a target in the instruction.
+If the digit is 3, this means the corresponding register is used as both source and target in the instruction.
+For example, fms instruction has 00113 as the DEPENDENCY field. This means RC is not used in this operation, RB and RA are
+used as sources and RT is the target.
+
+-=-=-= PIPE =-=-=-
+
+This field shows which execution pipe is used for the instruction
+
+pipe0 execution pipelines:
+       FP6     SP floating pipeline
+       FP7     integer operations executed in SP floating pipeline
+       FPD     DP floating pipeline
+       FX2     FXU pipeline
+       FX3     Rotate/Shift pipeline
+       FXB     Byte pipeline
+       NOP     No pipeline
+
+pipe1 execution pipelines:
+       BR      Branch pipeline
+       LNOP    No pipeline
+       LS      Load/Store pipeline
+       SHUF    Shuffle pipeline
+       SPR     SPR/CH pipeline
+
+*/
+
+#define _A0() {0}
+#define _A1(a) {1,a}
+#define _A2(a,b) {2,a,b}
+#define _A3(a,b,c) {3,a,b,c}
+#define _A4(a,b,c,d) {4,a,b,c,d}
+
+/*    TAG              FORMAT  OPCODE  MNEMONIC        ASM_FORMAT      DEPENDENCY      PIPE    COMMENT                         */
+/*                                                                     0[RC][RB][RA][RT]                                       */
+/*                                                                     1:src, 2:target                                         */
+
+APUOP(M_BR,            RI16,   0x190,  "br",           _A1(A_R18),     00000,  BR)     /* BRel          IP<-IP+I16 */
+APUOP(M_BRSL,          RI16,   0x198,  "brsl",         _A2(A_T,A_R18), 00002,  BR)     /* BRelSetLink   RT,IP<-IP,IP+I16 */
+APUOP(M_BRA,           RI16,   0x180,  "bra",          _A1(A_S18),     00000,  BR)     /* BRAbs         IP<-I16 */
+APUOP(M_BRASL,         RI16,   0x188,  "brasl",        _A2(A_T,A_S18), 00002,  BR)     /* BRAbsSetLink  RT,IP<-IP,I16 */
+APUOP(M_FSMBI,         RI16,   0x194,  "fsmbi",        _A2(A_T,A_X16), 00002,  SHUF)   /* FormSelMask%I RT<-fsm(I16) */
+APUOP(M_LQA,           RI16,   0x184,  "lqa",          _A2(A_T,A_S18), 00002,  LS)     /* LoadQAbs      RT<-M[I16] */
+APUOP(M_LQR,           RI16,   0x19C,  "lqr",          _A2(A_T,A_R18), 00002,  LS)     /* LoadQRel      RT<-M[IP+I16] */
+APUOP(M_STOP,          RR,     0x000,  "stop",         _A0(),          00000,  BR)     /* STOP          stop */
+APUOP(M_STOP2,         RR,     0x000,  "stop",         _A1(A_U14),     00000,  BR)     /* STOP          stop */
+APUOP(M_STOPD,         RR,     0x140,  "stopd",        _A3(A_T,A_A,A_B),         00111,        BR)     /* STOPD         stop (with register dependencies) */
+APUOP(M_LNOP,          RR,     0x001,  "lnop",         _A0(),          00000,  LNOP)   /* LNOP          no_operation */
+APUOP(M_SYNC,          RR,     0x002,  "sync",         _A0(),          00000,  BR)     /* SYNC          flush_pipe */
+APUOP(M_DSYNC,         RR,     0x003,  "dsync",        _A0(),          00000,  BR)     /* DSYNC         flush_store_queue */
+APUOP(M_MFSPR,         RR,     0x00c,  "mfspr",        _A2(A_T,A_S),   00002,  SPR)    /* MFSPR         RT<-SA */
+APUOP(M_RDCH,          RR,     0x00d,  "rdch",         _A2(A_T,A_H),   00002,  SPR)    /* ReaDCHannel   RT<-CA:data */
+APUOP(M_RCHCNT,                RR,     0x00f,  "rchcnt",       _A2(A_T,A_H),   00002,  SPR)    /* ReaDCHanCouNT RT<-CA:count */
+APUOP(M_HBRA,          LBT,    0x080,  "hbra",         _A2(A_S11,A_S18),       00000,  LS)     /* HBRA          BTB[B9]<-M[I16] */
+APUOP(M_HBRR,          LBT,    0x090,  "hbrr",         _A2(A_S11,A_R18),       00000,  LS)     /* HBRR          BTB[B9]<-M[IP+I16] */
+APUOP(M_BRZ,           RI16,   0x100,  "brz",          _A2(A_T,A_R18), 00001,  BR)     /* BRZ           IP<-IP+I16_if(RT) */
+APUOP(M_BRNZ,          RI16,   0x108,  "brnz",         _A2(A_T,A_R18), 00001,  BR)     /* BRNZ          IP<-IP+I16_if(RT) */
+APUOP(M_BRHZ,          RI16,   0x110,  "brhz",         _A2(A_T,A_R18), 00001,  BR)     /* BRHZ          IP<-IP+I16_if(RT) */
+APUOP(M_BRHNZ,         RI16,   0x118,  "brhnz",        _A2(A_T,A_R18), 00001,  BR)     /* BRHNZ         IP<-IP+I16_if(RT) */
+APUOP(M_STQA,          RI16,   0x104,  "stqa",         _A2(A_T,A_S18), 00001,  LS)     /* SToreQAbs     M[I16]<-RT */
+APUOP(M_STQR,          RI16,   0x11C,  "stqr",         _A2(A_T,A_R18), 00001,  LS)     /* SToreQRel     M[IP+I16]<-RT */
+APUOP(M_MTSPR,         RR,     0x10c,  "mtspr",        _A2(A_S,A_T),   00001,  SPR)    /* MTSPR         SA<-RT */
+APUOP(M_WRCH,          RR,     0x10d,  "wrch",         _A2(A_H,A_T),   00001,  SPR)    /* ChanWRite     CA<-RT */
+APUOP(M_LQD,           RI10,   0x1a0,  "lqd",          _A4(A_T,A_S14,A_P,A_A), 00012,  LS)     /* LoadQDisp     RT<-M[Ra+I10] */
+APUOP(M_BI,            RR,     0x1a8,  "bi",           _A1(A_A),               00010,  BR)     /* BI            IP<-RA */
+APUOP(M_BISL,          RR,     0x1a9,  "bisl",         _A2(A_T,A_A),   00012,  BR)     /* BISL          RT,IP<-IP,RA */
+APUOP(M_IRET,                  RR,     0x1aa,  "iret",         _A1(A_A),       00010,  BR)     /* IRET          IP<-SRR0 */
+APUOP(M_IRET2,                 RR,     0x1aa,  "iret",         _A0(),          00010,  BR)     /* IRET          IP<-SRR0 */
+APUOP(M_BISLED,                RR,     0x1ab,  "bisled",       _A2(A_T,A_A),   00012,  BR)     /* BISLED        RT,IP<-IP,RA_if(ext) */
+APUOP(M_HBR,           LBTI,   0x1ac,  "hbr",          _A2(A_S11I,A_A),        00010,  LS)     /* HBR           BTB[B9]<-M[Ra] */
+APUOP(M_FREST,         RR,     0x1b8,  "frest",        _A2(A_T,A_A),   00012,  SHUF)   /* FREST         RT<-recip(RA) */
+APUOP(M_FRSQEST,       RR,     0x1b9,  "frsqest",      _A2(A_T,A_A),   00012,  SHUF)   /* FRSQEST       RT<-rsqrt(RA) */
+APUOP(M_FSM,           RR,     0x1b4,  "fsm",          _A2(A_T,A_A),   00012,  SHUF)   /* FormSelMask%  RT<-expand(Ra) */
+APUOP(M_FSMH,          RR,     0x1b5,  "fsmh",         _A2(A_T,A_A),   00012,  SHUF)   /* FormSelMask%  RT<-expand(Ra) */
+APUOP(M_FSMB,          RR,     0x1b6,  "fsmb",         _A2(A_T,A_A),   00012,  SHUF)   /* FormSelMask%  RT<-expand(Ra) */
+APUOP(M_GB,            RR,     0x1b0,  "gb",           _A2(A_T,A_A),   00012,  SHUF)   /* GatherBits%   RT<-gather(RA) */
+APUOP(M_GBH,           RR,     0x1b1,  "gbh",          _A2(A_T,A_A),   00012,  SHUF)   /* GatherBits%   RT<-gather(RA) */
+APUOP(M_GBB,           RR,     0x1b2,  "gbb",          _A2(A_T,A_A),   00012,  SHUF)   /* GatherBits%   RT<-gather(RA) */
+APUOP(M_CBD,           RI7,    0x1f4,  "cbd",          _A4(A_T,A_U7,A_P,A_A),  00012,  SHUF)   /* genCtl%%insD  RT<-sta(Ra+I4,siz) */
+APUOP(M_CHD,           RI7,    0x1f5,  "chd",          _A4(A_T,A_U7,A_P,A_A),  00012,  SHUF)   /* genCtl%%insD  RT<-sta(Ra+I4,siz) */
+APUOP(M_CWD,           RI7,    0x1f6,  "cwd",          _A4(A_T,A_U7,A_P,A_A),  00012,  SHUF)   /* genCtl%%insD  RT<-sta(Ra+I4,siz) */
+APUOP(M_CDD,           RI7,    0x1f7,  "cdd",          _A4(A_T,A_U7,A_P,A_A),  00012,  SHUF)   /* genCtl%%insD  RT<-sta(Ra+I4,siz) */
+APUOP(M_ROTQBII,       RI7,    0x1f8,  "rotqbii",      _A3(A_T,A_A,A_U3),      00012,  SHUF)   /* ROTQBII       RT<-RA<<<I7 */
+APUOP(M_ROTQBYI,       RI7,    0x1fc,  "rotqbyi",      _A3(A_T,A_A,A_S7N),     00012,  SHUF)   /* ROTQBYI       RT<-RA<<<(I7*8) */
+APUOP(M_ROTQMBII,      RI7,    0x1f9,  "rotqmbii",     _A3(A_T,A_A,A_S3),      00012,  SHUF)   /* ROTQMBII      RT<-RA<<I7 */
+APUOP(M_ROTQMBYI,      RI7,    0x1fd,  "rotqmbyi",     _A3(A_T,A_A,A_S6),      00012,  SHUF)   /* ROTQMBYI      RT<-RA<<I7 */
+APUOP(M_SHLQBII,       RI7,    0x1fb,  "shlqbii",      _A3(A_T,A_A,A_U3),      00012,  SHUF)   /* SHLQBII       RT<-RA<<I7 */
+APUOP(M_SHLQBYI,       RI7,    0x1ff,  "shlqbyi",      _A3(A_T,A_A,A_U5),      00012,  SHUF)   /* SHLQBYI       RT<-RA<<I7 */
+APUOP(M_STQD,          RI10,   0x120,  "stqd",         _A4(A_T,A_S14,A_P,A_A), 00011,  LS)     /* SToreQDisp    M[Ra+I10]<-RT */
+APUOP(M_BIHNZ,         RR,     0x12b,  "bihnz",        _A2(A_T,A_A),   00011,  BR)     /* BIHNZ         IP<-RA_if(RT) */
+APUOP(M_BIHZ,          RR,     0x12a,  "bihz",         _A2(A_T,A_A),   00011,  BR)     /* BIHZ          IP<-RA_if(RT) */
+APUOP(M_BINZ,          RR,     0x129,  "binz",         _A2(A_T,A_A),   00011,  BR)     /* BINZ          IP<-RA_if(RT) */
+APUOP(M_BIZ,           RR,     0x128,  "biz",          _A2(A_T,A_A),   00011,  BR)     /* BIZ           IP<-RA_if(RT) */
+APUOP(M_CBX,           RR,     0x1d4,  "cbx",          _A3(A_T,A_A,A_B),               00112,  SHUF)   /* genCtl%%insX  RT<-sta(Ra+Rb,siz) */
+APUOP(M_CHX,           RR,     0x1d5,  "chx",          _A3(A_T,A_A,A_B),               00112,  SHUF)   /* genCtl%%insX  RT<-sta(Ra+Rb,siz) */
+APUOP(M_CWX,           RR,     0x1d6,  "cwx",          _A3(A_T,A_A,A_B),               00112,  SHUF)   /* genCtl%%insX  RT<-sta(Ra+Rb,siz) */
+APUOP(M_CDX,           RR,     0x1d7,  "cdx",          _A3(A_T,A_A,A_B),               00112,  SHUF)   /* genCtl%%insX  RT<-sta(Ra+Rb,siz) */
+APUOP(M_LQX,           RR,     0x1c4,  "lqx",          _A3(A_T,A_A,A_B),               00112,  LS)     /* LoadQindeX    RT<-M[Ra+Rb] */
+APUOP(M_ROTQBI,                RR,     0x1d8,  "rotqbi",       _A3(A_T,A_A,A_B),               00112,  SHUF)   /* ROTQBI        RT<-RA<<<Rb */
+APUOP(M_ROTQMBI,       RR,     0x1d9,  "rotqmbi",      _A3(A_T,A_A,A_B),               00112,  SHUF)   /* ROTQMBI       RT<-RA<<Rb */
+APUOP(M_SHLQBI,                RR,     0x1db,  "shlqbi",       _A3(A_T,A_A,A_B),               00112,  SHUF)   /* SHLQBI        RT<-RA<<Rb */
+APUOP(M_ROTQBY,                RR,     0x1dc,  "rotqby",       _A3(A_T,A_A,A_B),               00112,          SHUF)   /* ROTQBY        RT<-RA<<<(Rb*8) */
+APUOP(M_ROTQMBY,       RR,     0x1dd,  "rotqmby",      _A3(A_T,A_A,A_B),               00112,          SHUF)   /* ROTQMBY       RT<-RA<<Rb */
+APUOP(M_SHLQBY,                RR,     0x1df,  "shlqby",       _A3(A_T,A_A,A_B),               00112,  SHUF)   /* SHLQBY        RT<-RA<<Rb */
+APUOP(M_ROTQBYBI,      RR,     0x1cc,  "rotqbybi",     _A3(A_T,A_A,A_B),               00112,          SHUF)   /* ROTQBYBI      RT<-RA<<Rb */
+APUOP(M_ROTQMBYBI,     RR,     0x1cd,  "rotqmbybi",    _A3(A_T,A_A,A_B),               00112,          SHUF)   /* ROTQMBYBI     RT<-RA<<Rb */
+APUOP(M_SHLQBYBI,      RR,     0x1cf,  "shlqbybi",     _A3(A_T,A_A,A_B),               00112,  SHUF)   /* SHLQBYBI      RT<-RA<<Rb */
+APUOP(M_STQX,          RR,     0x144,  "stqx",         _A3(A_T,A_A,A_B),               00111,  LS)     /* SToreQindeX   M[Ra+Rb]<-RT */
+APUOP(M_SHUFB,         RRR,    0x580,  "shufb",        _A4(A_C,A_A,A_B,A_T),   02111,  SHUF)   /* SHUFfleBytes  RC<-f(RA,RB,RT) */
+APUOP(M_IL,            RI16,   0x204,  "il",           _A2(A_T,A_S16), 00002,  FX2)    /* ImmLoad       RT<-sxt(I16) */
+APUOP(M_ILH,           RI16,   0x20c,  "ilh",          _A2(A_T,A_X16), 00002,  FX2)    /* ImmLoadH      RT<-I16 */
+APUOP(M_ILHU,          RI16,   0x208,  "ilhu",         _A2(A_T,A_X16), 00002,  FX2)    /* ImmLoadHUpper RT<-I16<<16 */
+APUOP(M_ILA,           RI18,   0x210,  "ila",          _A2(A_T,A_U18), 00002,  FX2)    /* ImmLoadAddr   RT<-zxt(I18) */
+APUOP(M_NOP,           RR,     0x201,  "nop",          _A1(A_T),               00000,  NOP)    /* XNOP          no_operation */
+APUOP(M_NOP2,          RR,     0x201,  "nop",          _A0(),          00000,  NOP)    /* XNOP          no_operation */
+APUOP(M_IOHL,          RI16,   0x304,  "iohl",         _A2(A_T,A_X16), 00003,  FX2)    /* AddImmeXt     RT<-RT+sxt(I16) */
+APUOP(M_ANDBI,         RI10,   0x0b0,  "andbi",        _A3(A_T,A_A,A_S10B),    00012,  FX2)    /* AND%I         RT<-RA&I10 */
+APUOP(M_ANDHI,         RI10,   0x0a8,  "andhi",        _A3(A_T,A_A,A_S10),     00012,  FX2)    /* AND%I         RT<-RA&I10 */
+APUOP(M_ANDI,          RI10,   0x0a0,  "andi",         _A3(A_T,A_A,A_S10),     00012,  FX2)    /* AND%I         RT<-RA&I10 */
+APUOP(M_ORBI,          RI10,   0x030,  "orbi",         _A3(A_T,A_A,A_S10B),    00012,  FX2)    /* OR%I          RT<-RA|I10 */
+APUOP(M_ORHI,          RI10,   0x028,  "orhi",         _A3(A_T,A_A,A_S10),     00012,  FX2)    /* OR%I          RT<-RA|I10 */
+APUOP(M_ORI,           RI10,   0x020,  "ori",          _A3(A_T,A_A,A_S10),     00012,  FX2)    /* OR%I          RT<-RA|I10 */
+APUOP(M_ORX,           RR,     0x1f0,  "orx",          _A2(A_T,A_A),           00012,  BR)     /* ORX           RT<-RA.w0|RA.w1|RA.w2|RA.w3 */
+APUOP(M_XORBI,         RI10,   0x230,  "xorbi",        _A3(A_T,A_A,A_S10B),    00012,  FX2)    /* XOR%I         RT<-RA^I10 */
+APUOP(M_XORHI,         RI10,   0x228,  "xorhi",        _A3(A_T,A_A,A_S10),     00012,  FX2)    /* XOR%I         RT<-RA^I10 */
+APUOP(M_XORI,          RI10,   0x220,  "xori",         _A3(A_T,A_A,A_S10),     00012,  FX2)    /* XOR%I         RT<-RA^I10 */
+APUOP(M_AHI,           RI10,   0x0e8,  "ahi",          _A3(A_T,A_A,A_S10),     00012,  FX2)    /* Add%Immed     RT<-RA+I10 */
+APUOP(M_AI,            RI10,   0x0e0,  "ai",           _A3(A_T,A_A,A_S10),     00012,  FX2)    /* Add%Immed     RT<-RA+I10 */
+APUOP(M_SFHI,          RI10,   0x068,  "sfhi",         _A3(A_T,A_A,A_S10),     00012,  FX2)    /* SubFrom%Imm   RT<-I10-RA */
+APUOP(M_SFI,           RI10,   0x060,  "sfi",          _A3(A_T,A_A,A_S10),     00012,  FX2)    /* SubFrom%Imm   RT<-I10-RA */
+APUOP(M_CGTBI,         RI10,   0x270,  "cgtbi",        _A3(A_T,A_A,A_S10B),    00012,  FX2)    /* CGT%I         RT<-(RA>I10) */
+APUOP(M_CGTHI,         RI10,   0x268,  "cgthi",        _A3(A_T,A_A,A_S10),     00012,  FX2)    /* CGT%I         RT<-(RA>I10) */
+APUOP(M_CGTI,          RI10,   0x260,  "cgti",         _A3(A_T,A_A,A_S10),     00012,  FX2)    /* CGT%I         RT<-(RA>I10) */
+APUOP(M_CLGTBI,                RI10,   0x2f0,  "clgtbi",       _A3(A_T,A_A,A_S10B),    00012,  FX2)    /* CLGT%I        RT<-(RA>I10) */
+APUOP(M_CLGTHI,                RI10,   0x2e8,  "clgthi",       _A3(A_T,A_A,A_S10),     00012,  FX2)    /* CLGT%I        RT<-(RA>I10) */
+APUOP(M_CLGTI,         RI10,   0x2e0,  "clgti",        _A3(A_T,A_A,A_S10),     00012,  FX2)    /* CLGT%I        RT<-(RA>I10) */
+APUOP(M_CEQBI,         RI10,   0x3f0,  "ceqbi",        _A3(A_T,A_A,A_S10B),    00012,  FX2)    /* CEQ%I         RT<-(RA=I10) */
+APUOP(M_CEQHI,         RI10,   0x3e8,  "ceqhi",        _A3(A_T,A_A,A_S10),     00012,  FX2)    /* CEQ%I         RT<-(RA=I10) */
+APUOP(M_CEQI,          RI10,   0x3e0,  "ceqi",         _A3(A_T,A_A,A_S10),     00012,  FX2)    /* CEQ%I         RT<-(RA=I10) */
+APUOP(M_HGTI,          RI10,   0x278,  "hgti",         _A3(A_T,A_A,A_S10),     00010,  FX2)    /* HaltGTI       halt_if(RA>I10) */
+APUOP(M_HGTI2,         RI10,   0x278,  "hgti",         _A2(A_A,A_S10), 00010,  FX2)    /* HaltGTI       halt_if(RA>I10) */
+APUOP(M_HLGTI,         RI10,   0x2f8,  "hlgti",        _A3(A_T,A_A,A_S10),     00010,  FX2)    /* HaltLGTI      halt_if(RA>I10) */
+APUOP(M_HLGTI2,                RI10,   0x2f8,  "hlgti",        _A2(A_A,A_S10), 00010,  FX2)    /* HaltLGTI      halt_if(RA>I10) */
+APUOP(M_HEQI,          RI10,   0x3f8,  "heqi",         _A3(A_T,A_A,A_S10),     00010,  FX2)    /* HaltEQImm     halt_if(RA=I10) */
+APUOP(M_HEQI2,         RI10,   0x3f8,  "heqi",         _A2(A_A,A_S10), 00010,  FX2)    /* HaltEQImm     halt_if(RA=I10) */
+APUOP(M_MPYI,          RI10,   0x3a0,  "mpyi",         _A3(A_T,A_A,A_S10),     00012,  FP7)    /* MPYI          RT<-RA*I10 */
+APUOP(M_MPYUI,         RI10,   0x3a8,  "mpyui",        _A3(A_T,A_A,A_S10),     00012,  FP7)    /* MPYUI         RT<-RA*I10 */
+APUOP(M_CFLTS,         RI8,    0x3b0,  "cflts",        _A3(A_T,A_A,A_U7A),     00012,  FP7)    /* CFLTS         RT<-int(RA,I8) */
+APUOP(M_CFLTU,         RI8,    0x3b2,  "cfltu",        _A3(A_T,A_A,A_U7A),     00012,  FP7)    /* CFLTU         RT<-int(RA,I8) */
+APUOP(M_CSFLT,         RI8,    0x3b4,  "csflt",        _A3(A_T,A_A,A_U7B),     00012,  FP7)    /* CSFLT         RT<-flt(RA,I8) */
+APUOP(M_CUFLT,         RI8,    0x3b6,  "cuflt",        _A3(A_T,A_A,A_U7B),     00012,  FP7)    /* CUFLT         RT<-flt(RA,I8) */
+APUOP(M_FESD,          RR,     0x3b8,  "fesd",         _A2(A_T,A_A),   00012,  FPD)    /* FESD          RT<-double(RA) */
+APUOP(M_FRDS,          RR,     0x3b9,  "frds",         _A2(A_T,A_A),   00012,  FPD)    /* FRDS          RT<-single(RA) */
+APUOP(M_FSCRRD,                RR,     0x398,  "fscrrd",       _A1(A_T),               00002,  FPD)    /* FSCRRD        RT<-FP_status */
+APUOP(M_FSCRWR,                RR,     0x3ba,  "fscrwr",       _A2(A_T,A_A),   00010,  FP7)    /* FSCRWR        FP_status<-RA */
+APUOP(M_FSCRWR2,       RR,     0x3ba,  "fscrwr",       _A1(A_A),               00010,  FP7)    /* FSCRWR        FP_status<-RA */
+APUOP(M_CLZ,           RR,     0x2a5,  "clz",          _A2(A_T,A_A),   00012,  FX2)    /* CLZ           RT<-clz(RA) */
+APUOP(M_CNTB,          RR,     0x2b4,  "cntb",         _A2(A_T,A_A),   00012,  FXB)    /* CNT           RT<-pop(RA) */
+APUOP(M_XSBH,          RR,     0x2b6,  "xsbh",         _A2(A_T,A_A),   00012,  FX2)    /* eXtSignBtoH   RT<-sign_ext(RA) */
+APUOP(M_XSHW,          RR,     0x2ae,  "xshw",         _A2(A_T,A_A),   00012,  FX2)    /* eXtSignHtoW   RT<-sign_ext(RA) */
+APUOP(M_XSWD,          RR,     0x2a6,  "xswd",         _A2(A_T,A_A),   00012,  FX2)    /* eXtSignWtoD   RT<-sign_ext(RA) */
+APUOP(M_ROTI,          RI7,    0x078,  "roti",         _A3(A_T,A_A,A_S7N),     00012,  FX3)    /* ROT%I         RT<-RA<<<I7 */
+APUOP(M_ROTMI,         RI7,    0x079,  "rotmi",        _A3(A_T,A_A,A_S7),      00012,  FX3)    /* ROT%MI        RT<-RA<<I7 */
+APUOP(M_ROTMAI,                RI7,    0x07a,  "rotmai",       _A3(A_T,A_A,A_S7),      00012,  FX3)    /* ROTMA%I       RT<-RA<<I7 */
+APUOP(M_SHLI,          RI7,    0x07b,  "shli",         _A3(A_T,A_A,A_U6),      00012,  FX3)    /* SHL%I         RT<-RA<<I7 */
+APUOP(M_ROTHI,         RI7,    0x07c,  "rothi",        _A3(A_T,A_A,A_S7N),     00012,  FX3)    /* ROT%I         RT<-RA<<<I7 */
+APUOP(M_ROTHMI,                RI7,    0x07d,  "rothmi",       _A3(A_T,A_A,A_S6),      00012,  FX3)    /* ROT%MI        RT<-RA<<I7 */
+APUOP(M_ROTMAHI,       RI7,    0x07e,  "rotmahi",      _A3(A_T,A_A,A_S6),      00012,  FX3)    /* ROTMA%I       RT<-RA<<I7 */
+APUOP(M_SHLHI,         RI7,    0x07f,  "shlhi",        _A3(A_T,A_A,A_U5),      00012,  FX3)    /* SHL%I         RT<-RA<<I7 */
+APUOP(M_A,             RR,     0x0c0,  "a",            _A3(A_T,A_A,A_B),               00112,  FX2)    /* Add%          RT<-RA+RB */
+APUOP(M_AH,            RR,     0x0c8,  "ah",           _A3(A_T,A_A,A_B),               00112,  FX2)    /* Add%          RT<-RA+RB */
+APUOP(M_SF,            RR,     0x040,  "sf",           _A3(A_T,A_A,A_B),               00112,  FX2)    /* SubFrom%      RT<-RB-RA */
+APUOP(M_SFH,           RR,     0x048,  "sfh",          _A3(A_T,A_A,A_B),               00112,  FX2)    /* SubFrom%      RT<-RB-RA */
+APUOP(M_CGT,           RR,     0x240,  "cgt",          _A3(A_T,A_A,A_B),               00112,  FX2)    /* CGT%          RT<-(RA>RB) */
+APUOP(M_CGTB,          RR,     0x250,  "cgtb",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* CGT%          RT<-(RA>RB) */
+APUOP(M_CGTH,          RR,     0x248,  "cgth",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* CGT%          RT<-(RA>RB) */
+APUOP(M_CLGT,          RR,     0x2c0,  "clgt",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* CLGT%         RT<-(RA>RB) */
+APUOP(M_CLGTB,         RR,     0x2d0,  "clgtb",        _A3(A_T,A_A,A_B),               00112,  FX2)    /* CLGT%         RT<-(RA>RB) */
+APUOP(M_CLGTH,         RR,     0x2c8,  "clgth",        _A3(A_T,A_A,A_B),               00112,  FX2)    /* CLGT%         RT<-(RA>RB) */
+APUOP(M_CEQ,           RR,     0x3c0,  "ceq",          _A3(A_T,A_A,A_B),               00112,  FX2)    /* CEQ%          RT<-(RA=RB) */
+APUOP(M_CEQB,          RR,     0x3d0,  "ceqb",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* CEQ%          RT<-(RA=RB) */
+APUOP(M_CEQH,          RR,     0x3c8,  "ceqh",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* CEQ%          RT<-(RA=RB) */
+APUOP(M_HGT,           RR,     0x258,  "hgt",          _A3(A_T,A_A,A_B),               00110,  FX2)    /* HaltGT        halt_if(RA>RB) */
+APUOP(M_HGT2,          RR,     0x258,  "hgt",          _A2(A_A,A_B),   00110,  FX2)    /* HaltGT        halt_if(RA>RB) */
+APUOP(M_HLGT,          RR,     0x2d8,  "hlgt",         _A3(A_T,A_A,A_B),               00110,  FX2)    /* HaltLGT       halt_if(RA>RB) */
+APUOP(M_HLGT2,         RR,     0x2d8,  "hlgt",         _A2(A_A,A_B),   00110,  FX2)    /* HaltLGT       halt_if(RA>RB) */
+APUOP(M_HEQ,           RR,     0x3d8,  "heq",          _A3(A_T,A_A,A_B),               00110,  FX2)    /* HaltEQ        halt_if(RA=RB) */
+APUOP(M_HEQ2,          RR,     0x3d8,  "heq",          _A2(A_A,A_B),   00110,  FX2)    /* HaltEQ        halt_if(RA=RB) */
+APUOP(M_FCEQ,          RR,     0x3c2,  "fceq",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* FCEQ          RT<-(RA=RB) */
+APUOP(M_FCMEQ,         RR,     0x3ca,  "fcmeq",        _A3(A_T,A_A,A_B),               00112,  FX2)    /* FCMEQ         RT<-(|RA|=|RB|) */
+APUOP(M_FCGT,          RR,     0x2c2,  "fcgt",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* FCGT          RT<-(RA<RB) */
+APUOP(M_FCMGT,         RR,     0x2ca,  "fcmgt",        _A3(A_T,A_A,A_B),               00112,  FX2)    /* FCMGT         RT<-(|RA|<|RB|) */
+APUOP(M_AND,           RR,     0x0c1,  "and",          _A3(A_T,A_A,A_B),               00112,  FX2)    /* AND           RT<-RA&RB */
+APUOP(M_NAND,          RR,     0x0c9,  "nand",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* NAND          RT<-!(RA&RB) */
+APUOP(M_OR,            RR,     0x041,  "or",           _A3(A_T,A_A,A_B),               00112,  FX2)    /* OR            RT<-RA|RB */
+APUOP(M_NOR,           RR,     0x049,  "nor",          _A3(A_T,A_A,A_B),               00112,  FX2)    /* NOR           RT<-!(RA&RB) */
+APUOP(M_XOR,           RR,     0x241,  "xor",          _A3(A_T,A_A,A_B),               00112,  FX2)    /* XOR           RT<-RA^RB */
+APUOP(M_EQV,           RR,     0x249,  "eqv",          _A3(A_T,A_A,A_B),               00112,  FX2)    /* EQuiValent    RT<-!(RA^RB) */
+APUOP(M_ANDC,          RR,     0x2c1,  "andc",         _A3(A_T,A_A,A_B),               00112,  FX2)    /* ANDComplement RT<-RA&!RB */
+APUOP(M_ORC,           RR,     0x2c9,  "orc",          _A3(A_T,A_A,A_B),               00112,  FX2)    /* ORComplement  RT<-RA|!RB */
+APUOP(M_ABSDB,         RR,     0x053,  "absdb",        _A3(A_T,A_A,A_B),               00112,  FXB)    /* ABSoluteDiff  RT<-|RA-RB| */
+APUOP(M_AVGB,          RR,     0x0d3,  "avgb",         _A3(A_T,A_A,A_B),               00112,  FXB)    /* AVG%          RT<-(RA+RB+1)/2 */
+APUOP(M_SUMB,          RR,     0x253,  "sumb",         _A3(A_T,A_A,A_B),               00112,  FXB)    /* SUM%          RT<-f(RA,RB) */
+APUOP(M_DFA,           RR,     0x2cc,  "dfa",          _A3(A_T,A_A,A_B),               00112,  FPD)    /* DFAdd         RT<-RA+RB */
+APUOP(M_DFM,           RR,     0x2ce,  "dfm",          _A3(A_T,A_A,A_B),               00112,  FPD)    /* DFMul         RT<-RA*RB */
+APUOP(M_DFS,           RR,     0x2cd,  "dfs",          _A3(A_T,A_A,A_B),               00112,  FPD)    /* DFSub         RT<-RA-RB */
+APUOP(M_FA,            RR,     0x2c4,  "fa",           _A3(A_T,A_A,A_B),               00112,  FP6)    /* FAdd          RT<-RA+RB */
+APUOP(M_FM,            RR,     0x2c6,  "fm",           _A3(A_T,A_A,A_B),               00112,  FP6)    /* FMul          RT<-RA*RB */
+APUOP(M_FS,            RR,     0x2c5,  "fs",           _A3(A_T,A_A,A_B),               00112,  FP6)    /* FSub          RT<-RA-RB */
+APUOP(M_MPY,           RR,     0x3c4,  "mpy",          _A3(A_T,A_A,A_B),               00112,  FP7)    /* MPY           RT<-RA*RB */
+APUOP(M_MPYH,          RR,     0x3c5,  "mpyh",         _A3(A_T,A_A,A_B),               00112,  FP7)    /* MPYH          RT<-(RAh*RB)<<16 */
+APUOP(M_MPYHH,         RR,     0x3c6,  "mpyhh",        _A3(A_T,A_A,A_B),               00112,  FP7)    /* MPYHH         RT<-RAh*RBh */
+APUOP(M_MPYHHU,                RR,     0x3ce,  "mpyhhu",       _A3(A_T,A_A,A_B),               00112,  FP7)    /* MPYHHU        RT<-RAh*RBh */
+APUOP(M_MPYS,          RR,     0x3c7,  "mpys",         _A3(A_T,A_A,A_B),               00112,  FP7)    /* MPYS          RT<-(RA*RB)>>16 */
+APUOP(M_MPYU,          RR,     0x3cc,  "mpyu",         _A3(A_T,A_A,A_B),               00112,  FP7)    /* MPYU          RT<-RA*RB */
+APUOP(M_FI,            RR,     0x3d4,  "fi",           _A3(A_T,A_A,A_B),               00112,  FP7)    /* FInterpolate  RT<-f(RA,RB) */
+APUOP(M_ROT,           RR,     0x058,  "rot",          _A3(A_T,A_A,A_B),               00112,  FX3)    /* ROT%          RT<-RA<<<RB */
+APUOP(M_ROTM,          RR,     0x059,  "rotm",         _A3(A_T,A_A,A_B),               00112,  FX3)    /* ROT%M         RT<-RA<<Rb */
+APUOP(M_ROTMA,         RR,     0x05a,  "rotma",        _A3(A_T,A_A,A_B),               00112,  FX3)    /* ROTMA%        RT<-RA<<Rb */
+APUOP(M_SHL,           RR,     0x05b,  "shl",          _A3(A_T,A_A,A_B),               00112,  FX3)    /* SHL%          RT<-RA<<Rb */
+APUOP(M_ROTH,          RR,     0x05c,  "roth",         _A3(A_T,A_A,A_B),               00112,  FX3)    /* ROT%          RT<-RA<<<RB */
+APUOP(M_ROTHM,         RR,     0x05d,  "rothm",        _A3(A_T,A_A,A_B),               00112,  FX3)    /* ROT%M         RT<-RA<<Rb */
+APUOP(M_ROTMAH,                RR,     0x05e,  "rotmah",       _A3(A_T,A_A,A_B),               00112,  FX3)    /* ROTMA%        RT<-RA<<Rb */
+APUOP(M_SHLH,          RR,     0x05f,  "shlh",         _A3(A_T,A_A,A_B),               00112,  FX3)    /* SHL%          RT<-RA<<Rb */
+APUOP(M_MPYHHA,                RR,     0x346,  "mpyhha",       _A3(A_T,A_A,A_B),               00113,  FP7)    /* MPYHHA        RT<-RAh*RBh+RT */
+APUOP(M_MPYHHAU,       RR,     0x34e,  "mpyhhau",      _A3(A_T,A_A,A_B),               00113,  FP7)    /* MPYHHAU       RT<-RAh*RBh+RT */
+APUOP(M_DFMA,          RR,     0x35c,  "dfma",         _A3(A_T,A_A,A_B),               00113,  FPD)    /* DFMAdd        RT<-RT+RA*RB */
+APUOP(M_DFMS,          RR,     0x35d,  "dfms",         _A3(A_T,A_A,A_B),               00113,  FPD)    /* DFMSub        RT<-RA*RB-RT */
+APUOP(M_DFNMS,         RR,     0x35e,  "dfnms",        _A3(A_T,A_A,A_B),               00113,  FPD)    /* DFNMSub       RT<-RT-RA*RB */
+APUOP(M_DFNMA,         RR,     0x35f,  "dfnma",        _A3(A_T,A_A,A_B),               00113,  FPD)    /* DFNMAdd       RT<-(-RT)-RA*RB */
+APUOP(M_FMA,           RRR,    0x700,  "fma",          _A4(A_C,A_A,A_B,A_T),   02111,  FP6)    /* FMAdd         RC<-RT+RA*RB */
+APUOP(M_FMS,           RRR,    0x780,  "fms",          _A4(A_C,A_A,A_B,A_T),   02111,  FP6)    /* FMSub         RC<-RA*RB-RT */
+APUOP(M_FNMS,          RRR,    0x680,  "fnms",         _A4(A_C,A_A,A_B,A_T),   02111,  FP6)    /* FNMSub        RC<-RT-RA*RB */
+APUOP(M_MPYA,          RRR,    0x600,  "mpya",         _A4(A_C,A_A,A_B,A_T),   02111,  FP7)    /* MPYA          RC<-RA*RB+RT */
+APUOP(M_SELB,          RRR,    0x400,  "selb",         _A4(A_C,A_A,A_B,A_T),   02111,  FX2)    /* SELectBits    RC<-RA&RT|RB&!RT */
+/* for system function call, this uses op-code of mtspr */
+APUOP(M_SYSCALL,       RI7,    0x10c,  "syscall",      _A3(A_T,A_A,A_S7N),     00002,  SPR)        /* System Call */
+/*
+pseudo instruction:
+system call
+value of I9    operation
+0      halt
+1              rt[0] = open(MEM[ra[0]],        ra[1])
+2              rt[0] = close(ra[0])
+3              rt[0] = read(ra[0],     MEM[ra[1]],     ra[2])
+4              rt[0] = write(ra[0],    MEM[ra[1]],     ra[2])
+5              printf(MEM[ra[0]],      ra[1],  ra[2],  ra[3])
+42             rt[0] = clock()
+52             rt[0] = lseek(ra0,      ra1,    ra2)
+
+*/
+
+
+/* new multiprecision add/sub */
+APUOP(M_ADDX,          RR,     0x340,  "addx",         _A3(A_T,A_A,A_B),               00113,          FX2)    /* Add_eXtended  RT<-RA+RB+RT */
+APUOP(M_CG,            RR,     0x0c2,  "cg",           _A3(A_T,A_A,A_B),               00112,          FX2)    /* CarryGenerate RT<-cout(RA+RB) */
+APUOP(M_CGX,           RR,     0x342,  "cgx",          _A3(A_T,A_A,A_B),               00113,          FX2)    /* CarryGen_eXtd RT<-cout(RA+RB+RT) */
+APUOP(M_SFX,           RR,     0x341,  "sfx",          _A3(A_T,A_A,A_B),               00113,          FX2)    /* Add_eXtended  RT<-RA+RB+RT */
+APUOP(M_BG,            RR,     0x042,  "bg",           _A3(A_T,A_A,A_B),               00112,          FX2)    /* CarryGenerate RT<-cout(RA+RB) */
+APUOP(M_BGX,           RR,     0x343,  "bgx",          _A3(A_T,A_A,A_B),               00113,          FX2)    /* CarryGen_eXtd RT<-cout(RA+RB+RT) */
+
+/*
+
+The following ops are a subset of above except with feature bits set.
+Feature bits are bits 11-17 of the instruction:
+
+  11 - C & P feature bit
+  12 - disable interrupts
+  13 - enable interrupts
+
+*/
+APUOPFB(M_BID,         RR,     0x1a8,  0x20,   "bid",          _A1(A_A),               00010,  BR)     /* BI            IP<-RA */
+APUOPFB(M_BIE,         RR,     0x1a8,  0x10,   "bie",          _A1(A_A),               00010,  BR)     /* BI            IP<-RA */
+APUOPFB(M_BISLD,       RR,     0x1a9,  0x20,   "bisld",        _A2(A_T,A_A),   00012,  BR)     /* BISL          RT,IP<-IP,RA */
+APUOPFB(M_BISLE,       RR,     0x1a9,  0x10,   "bisle",        _A2(A_T,A_A),   00012,  BR)     /* BISL          RT,IP<-IP,RA */
+APUOPFB(M_IRETD,       RR,     0x1aa,  0x20,   "iretd",        _A1(A_A),       00010,  BR)     /* IRET          IP<-SRR0 */
+APUOPFB(M_IRETD2,      RR,     0x1aa,  0x20,   "iretd",        _A0(),          00010,  BR)     /* IRET          IP<-SRR0 */
+APUOPFB(M_IRETE,       RR,     0x1aa,  0x10,   "irete",        _A1(A_A),       00010,  BR)     /* IRET          IP<-SRR0 */
+APUOPFB(M_IRETE2,      RR,     0x1aa,  0x10,   "irete",        _A0(),          00010,  BR)     /* IRET          IP<-SRR0 */
+APUOPFB(M_BISLEDD,     RR,     0x1ab,  0x20,   "bisledd",      _A2(A_T,A_A),   00012,  BR)     /* BISLED        RT,IP<-IP,RA_if(ext) */
+APUOPFB(M_BISLEDE,     RR,     0x1ab,  0x10,   "bislede",      _A2(A_T,A_A),   00012,  BR)     /* BISLED        RT,IP<-IP,RA_if(ext) */
+APUOPFB(M_BIHNZD,      RR,     0x12b,  0x20,   "bihnzd",       _A2(A_T,A_A),   00011,  BR)     /* BIHNZ         IP<-RA_if(RT) */
+APUOPFB(M_BIHNZE,      RR,     0x12b,  0x10,   "bihnze",       _A2(A_T,A_A),   00011,  BR)     /* BIHNZ         IP<-RA_if(RT) */
+APUOPFB(M_BIHZD,       RR,     0x12a,  0x20,   "bihzd",        _A2(A_T,A_A),   00011,  BR)     /* BIHZ          IP<-RA_if(RT) */
+APUOPFB(M_BIHZE,       RR,     0x12a,  0x10,   "bihze",        _A2(A_T,A_A),   00011,  BR)     /* BIHZ          IP<-RA_if(RT) */
+APUOPFB(M_BINZD,       RR,     0x129,  0x20,   "binzd",        _A2(A_T,A_A),   00011,  BR)     /* BINZ          IP<-RA_if(RT) */
+APUOPFB(M_BINZE,       RR,     0x129,  0x10,   "binze",        _A2(A_T,A_A),   00011,  BR)     /* BINZ          IP<-RA_if(RT) */
+APUOPFB(M_BIZD,                RR,     0x128,  0x20,   "bizd",         _A2(A_T,A_A),   00011,  BR)     /* BIZ           IP<-RA_if(RT) */
+APUOPFB(M_BIZE,                RR,     0x128,  0x10,   "bize",         _A2(A_T,A_A),   00011,  BR)     /* BIZ           IP<-RA_if(RT) */
+APUOPFB(M_SYNCC,       RR,     0x002,  0x40,   "syncc",        _A0(),          00000,  BR)     /* SYNCC          flush_pipe */
+APUOPFB(M_HBRP,                LBTI,   0x1ac,  0x40,   "hbrp",         _A0(),          00010,  LS)     /* HBR           BTB[B9]<-M[Ra] */
+
+/* Synonyms required by the AS manual. */
+APUOP(M_LR,            RI10,   0x020,  "lr",           _A2(A_T,A_A),   00012,  FX2)    /* OR%I          RT<-RA|I10 */
+APUOP(M_BIHT,          RR,     0x12b,  "biht",         _A2(A_T,A_A),   00011,  BR)     /* BIHNZ         IP<-RA_if(RT) */
+APUOP(M_BIHF,          RR,     0x12a,  "bihf",         _A2(A_T,A_A),   00011,  BR)     /* BIHZ          IP<-RA_if(RT) */
+APUOP(M_BIT,           RR,     0x129,  "bit",          _A2(A_T,A_A),   00011,  BR)     /* BINZ          IP<-RA_if(RT) */
+APUOP(M_BIF,           RR,     0x128,  "bif",          _A2(A_T,A_A),   00011,  BR)     /* BIZ           IP<-RA_if(RT) */
+APUOPFB(M_BIHTD,       RR,     0x12b,  0x20,   "bihtd",        _A2(A_T,A_A),   00011,  BR)     /* BIHNF         IP<-RA_if(RT) */
+APUOPFB(M_BIHTE,       RR,     0x12b,  0x10,   "bihte",        _A2(A_T,A_A),   00011,  BR)     /* BIHNF         IP<-RA_if(RT) */
+APUOPFB(M_BIHFD,       RR,     0x12a,  0x20,   "bihfd",        _A2(A_T,A_A),   00011,  BR)     /* BIHZ          IP<-RA_if(RT) */
+APUOPFB(M_BIHFE,       RR,     0x12a,  0x10,   "bihfe",        _A2(A_T,A_A),   00011,  BR)     /* BIHZ          IP<-RA_if(RT) */
+APUOPFB(M_BITD,        RR,     0x129,  0x20,   "bitd",         _A2(A_T,A_A),   00011,  BR)     /* BINF          IP<-RA_if(RT) */
+APUOPFB(M_BITE,        RR,     0x129,  0x10,   "bite",         _A2(A_T,A_A),   00011,  BR)     /* BINF          IP<-RA_if(RT) */
+APUOPFB(M_BIFD,                RR,     0x128,  0x20,   "bifd",         _A2(A_T,A_A),   00011,  BR)     /* BIZ           IP<-RA_if(RT) */
+APUOPFB(M_BIFE,                RR,     0x128,  0x10,   "bife",         _A2(A_T,A_A),   00011,  BR)     /* BIZ           IP<-RA_if(RT) */
+
+#undef _A0
+#undef _A1
+#undef _A2
+#undef _A3
+#undef _A4
diff --git a/arch/powerpc/xmon/spu-opc.c b/arch/powerpc/xmon/spu-opc.c
new file mode 100644 (file)
index 0000000..efffde9
--- /dev/null
@@ -0,0 +1,44 @@
+/* SPU opcode list
+
+   Copyright 2006 Free Software Foundation, Inc.
+
+   This file is part of GDB, GAS, and the GNU binutils.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License along
+   with this program; if not, write to the Free Software Foundation, Inc.,
+   51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.  */
+
+#include "spu.h"
+
+/* This file holds the Spu opcode table */
+\f
+
+/*
+   Example contents of spu-insn.h
+      id_tag   mode    mode    type    opcode  mnemonic        asmtype     dependency          FPU     L/S?    branch? instruction   
+                QUAD   WORD                                               (0,RC,RB,RA,RT)    latency                                           
+   APUOP(M_LQD,        1,      0,      RI9,    0x1f8,  "lqd",          ASM_RI9IDX,     00012,          FXU,    1,      0)      Load Quadword d-form 
+ */
+
+const struct spu_opcode spu_opcodes[] = {
+#define APUOP(TAG,MACFORMAT,OPCODE,MNEMONIC,ASMFORMAT,DEP,PIPE) \
+       { MACFORMAT, OPCODE, MNEMONIC, ASMFORMAT },
+#define APUOPFB(TAG,MACFORMAT,OPCODE,FB,MNEMONIC,ASMFORMAT,DEP,PIPE) \
+       { MACFORMAT, OPCODE, MNEMONIC, ASMFORMAT },
+#include "spu-insns.h"
+#undef APUOP
+#undef APUOPFB
+};
+
+const int spu_num_opcodes =
+  sizeof (spu_opcodes) / sizeof (spu_opcodes[0]);
diff --git a/arch/powerpc/xmon/spu.h b/arch/powerpc/xmon/spu.h
new file mode 100644 (file)
index 0000000..c761fc8
--- /dev/null
@@ -0,0 +1,126 @@
+/* SPU ELF support for BFD.
+
+   Copyright 2006 Free Software Foundation, Inc.
+
+   This file is part of GDB, GAS, and the GNU binutils.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program; if not, write to the Free Software Foundation,
+   Inc., 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.  */
+
+
+/* These two enums are from rel_apu/common/spu_asm_format.h */
+/* definition of instruction format */
+typedef enum {
+  RRR,
+  RI18,
+  RI16,
+  RI10,
+  RI8,
+  RI7,
+  RR,
+  LBT,
+  LBTI,
+  IDATA,
+  UNKNOWN_IFORMAT
+} spu_iformat;
+
+/* These values describe assembly instruction arguments.  They indicate
+ * how to encode, range checking and which relocation to use. */
+typedef enum {
+  A_T,  /* register at pos 0 */
+  A_A,  /* register at pos 7 */
+  A_B,  /* register at pos 14 */
+  A_C,  /* register at pos 21 */
+  A_S,  /* special purpose register at pos 7 */
+  A_H,  /* channel register at pos 7 */
+  A_P,  /* parenthesis, this has to separate regs from immediates */
+  A_S3,
+  A_S6,
+  A_S7N,
+  A_S7,
+  A_U7A,
+  A_U7B,
+  A_S10B,
+  A_S10,
+  A_S11,
+  A_S11I,
+  A_S14,
+  A_S16,
+  A_S18,
+  A_R18,
+  A_U3,
+  A_U5,
+  A_U6,
+  A_U7,
+  A_U14,
+  A_X16,
+  A_U18,
+  A_MAX
+} spu_aformat;
+
+enum spu_insns {
+#define APUOP(TAG,MACFORMAT,OPCODE,MNEMONIC,ASMFORMAT,DEP,PIPE) \
+       TAG,
+#define APUOPFB(TAG,MACFORMAT,OPCODE,FB,MNEMONIC,ASMFORMAT,DEP,PIPE) \
+       TAG,
+#include "spu-insns.h"
+#undef APUOP
+#undef APUOPFB
+        M_SPU_MAX
+};
+
+struct spu_opcode
+{
+   spu_iformat insn_type;
+   unsigned int opcode;
+   char *mnemonic;
+   int arg[5];
+};
+
+#define SIGNED_EXTRACT(insn,size,pos) (((int)((insn) << (32-size-pos))) >> (32-size))
+#define UNSIGNED_EXTRACT(insn,size,pos) (((insn) >> pos) & ((1 << size)-1))
+
+#define DECODE_INSN_RT(insn) (insn & 0x7f)
+#define DECODE_INSN_RA(insn) ((insn >> 7) & 0x7f)
+#define DECODE_INSN_RB(insn) ((insn >> 14) & 0x7f)
+#define DECODE_INSN_RC(insn) ((insn >> 21) & 0x7f)
+
+#define DECODE_INSN_I10(insn) SIGNED_EXTRACT(insn,10,14)
+#define DECODE_INSN_U10(insn) UNSIGNED_EXTRACT(insn,10,14)
+
+/* For branching, immediate loads, hbr and  lqa/stqa. */
+#define DECODE_INSN_I16(insn) SIGNED_EXTRACT(insn,16,7)
+#define DECODE_INSN_U16(insn) UNSIGNED_EXTRACT(insn,16,7)
+
+/* for stop */
+#define DECODE_INSN_U14(insn) UNSIGNED_EXTRACT(insn,14,0)
+
+/* For ila */
+#define DECODE_INSN_I18(insn) SIGNED_EXTRACT(insn,18,7)
+#define DECODE_INSN_U18(insn) UNSIGNED_EXTRACT(insn,18,7)
+
+/* For rotate and shift and generate control mask */
+#define DECODE_INSN_I7(insn) SIGNED_EXTRACT(insn,7,14)
+#define DECODE_INSN_U7(insn) UNSIGNED_EXTRACT(insn,7,14)
+
+/* For float <-> int conversion */
+#define DECODE_INSN_I8(insn)  SIGNED_EXTRACT(insn,8,14)
+#define DECODE_INSN_U8(insn) UNSIGNED_EXTRACT(insn,8,14)
+
+/* For hbr  */
+#define DECODE_INSN_I9a(insn) ((SIGNED_EXTRACT(insn,2,23) << 7) | UNSIGNED_EXTRACT(insn,7,0))
+#define DECODE_INSN_I9b(insn) ((SIGNED_EXTRACT(insn,2,14) << 7) | UNSIGNED_EXTRACT(insn,7,0))
+#define DECODE_INSN_U9a(insn) ((UNSIGNED_EXTRACT(insn,2,23) << 7) | UNSIGNED_EXTRACT(insn,7,0))
+#define DECODE_INSN_U9b(insn) ((UNSIGNED_EXTRACT(insn,2,14) << 7) | UNSIGNED_EXTRACT(insn,7,0))
+