2008年12月9日火曜日

mipsel-nec-elf

Toppers/JSP の KZ-Vr4131PCI-01 ターゲットは、GCC の mipsel-nec-elf というターゲットを前提にしているらしいので、とりあえず作ってみた。

http://cvs.sourceforge.jp/cgi-bin/viewcvs.cgi/jsp4cm3/jsp/doc/mips3.txt?rev=1.1
http://cvs.sourceforge.jp/cgi-bin/viewcvs.cgi/jsp4cm3/jsp/doc/gnu_install.txt?rev=1.1

(ただし、binutils-2.13.2.1、gcc-3.3.2、newlib-1.11.0 の環境らしいので、以下の環境で通るかどうかはわからない。)
$ tar zxvf binutils-2.19.tar.gz
$ mkdir build-binutils-mipsel-nec-elf
$ cd build-binutils-mipsel-nec-elf
$ ../binutils-2.19/configure --prefix=/usr/local/mips --target=mipsel-nec-elf --disable-nls
$ make -i
$ make -i install
$ tar zxvf gcc-4.3.2.tar.gz
$ tar zxvf newlib-1.16.0.tar.gz
$ cd gcc-4.3.2.tar.gz
$ ln -s ../newlib-1.16.0/newlib
$ ln -s ../newlib-1.16.0/libgloss
$ cd ..
$ mkdir build-gcc-mipsel-nec-elf
$ cd build-gcc-mipsel-nec-elf
$ ../gcc-4.3.2/configure --with-gmp=/usr/local --with-mpfr=/usr/local --enable-languages=c,c++ --disable-nls --disable-win32-regist
ry --disable-shared --disable-libssp --prefix=/usr/local/mips/ --target=mipsel-nec-elf --disable-multilib --without-fp --disable-li
bmudflap --disable-libstdcxx-pch --with-newlib --with-headers=/home/aloha/work/mips/build/gcc-4.3.2/newlib/libc/include/
$ make
$ make install

重要なのは configure オプションだけなので、展開のところとかは適宜読み替えて適当に。
texinfo 絡みでエラーが出るので、make -i で強制的にビルドとインストールを進めている (他に致命的なエラーがあった場合、ひどい事になる場合もあるので、本当は随時確認しながら無視した方が良い。今回はコケることが最初からわかっているので、最初からエラーを無視 (-i) してビルド & インストールしてる)

2008年12月6日土曜日

mips-unknown-elf gcc with newlib

以前は mips-linux-elf で作ったので、linux に依存するヘッダを要求されてコケた。

linux kernel をビルドすることに拘らなければ、mips-unknown-elf で普通に通った。

newlib-1.16.0 を展開して、gcc-4.3.2/ 以下に newlib-1.6.0/newlib と newlib-1.6.0/libgloss へのシンボリックリンクを張る。

binutils -> gcc -> newlib の順に、以下のように configure ; make ; make install する (別途 build-*** のようなディレクトリを掘って、そこで行う)。

途中で makeinfo がどうこうというエラーが出たら、make -i で無視してビルドを続行する。
$ ../binutils-2.19/configure --prefix=/usr/local/mips --target=mipsel-unknown-elf --disable-nls --without-included-gette
xt

$ ../gcc-4.3.2/configure --with-gmp=/usr/local --with-mpfr=/usr/local --enable-languages=c,c++ --disable-nls --disable-w
in32-registry --disable-shared --without-included-gettext --disable-libssp --prefix=/usr/local/mips/ --target=mipsel-unk
nown-elf --disable-multilib --without-fp --disable-libmudflap --disable-libstdcxx-pch --with-newlib --with-headers=/home
/aloha/work/mips/build/gcc-4.3.2/newlib/libc/include/

$ ../newlib-1.16.0/configure --target=mipsel-unknown-elf --prefix=/usr/loca/mips

(newlib のビルドは gcc と同時に行われるから不要かもしれない。未確認)

できた gcc で、-EB オプションを付けると
$ /usr/local/mips/bin/mipsel-unknown-elf-gcc.exe -EB test.c
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crti.o: compiled for a little endian system and target is big endian
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crti.o: endianness incompatible with that of the selected emulation
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: failed to merge target specific data of file c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crti.o
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtbegin.o: compiled for a little endian system and target is big endian
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtbegin.o: endianness incompatible with that of the selected emulation
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: failed to merge target specific data of file c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtbegin.o
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtend.o: compiled for a little endian system and target is big endian
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtend.o: endianness incompatible with that of the selected emulation
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: failed to merge target specific data of file c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtend.o
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtn.o: compiled for a little endian system and target is big endian
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtn.o: endianness incompatible with that of the selected emulation
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: failed to merge target specific data of file c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/crtn.o
c:/msys/1.0/local/mips/bin/../lib/gcc/mipsel-unknown-elf/4.3.2/../../../../mipsel-unknown-elf/bin/ld.exe: warning: cannot find entry symbol _start; defaulting to 0000000000400050
collect2: ld returned 1 exit status

エラーになってしまう (最後の警告は、crt0.o と ld script を用意してないから当たり前)

これを両方サポートできるようにするために、enable-targets とかでなんとかできるかと思ったけど、結局上手く行かなかった。

2008年12月5日金曜日

VEC

gcc の VEC (vec.[ch]) データ構造は、プリプロセッサを使って C++ の template みたいなことをやってるらしい (C++ 言語のカラクリ とかでも触れられてるけど、cfront 時代の template の実現手法としては比較的ポピュラー)

ほんと GCC は C++ の機能を C で無理やり実装するのが得意だよなぁ。

2008年12月4日木曜日

GCC の MIPS ターゲット

newlib で作ろうとしたら、asm/unistd.h が無いよ!と、明らかに Linux Kernel の header らしきファイルを要求されて上手く行かなかった。

しかたがないので、--disable-libstdc と --disable-libstdcxx で作ってみた (MinGW + MSYS + flex + bison)。
$ ../binutils-2.19/configure --prefix=/usr/local/mips --target=mips-elf --disable-nls --without-included-gettext
...
$ ../gcc-4.3.2/configure --with-gmp=/usr/local --with-mpfr=/usr/local --enable-languages=c,c++ --disable-nls --disable-win32-registry --di
sable-shared --without-x --without-included-gettext --disable-libssp --prefix=/usr/local/mips/ --target=mips-elf --disable-multilib --with
out-fp --disable-libmudflap --disable-libstdc --disable-libstdcxx --disable-libstdcxx-pch --enable-threads --enable-hash-synchronization
...

enable-threads とか enable-hash-synchronizasion は、特に意味が無いようだ。disable-libstdcxx-pch も、たぶんいらない気がする (MinGW では、pch (PreCompiled Header) のテストが通らないから、いつまでたっても gcc が 4 系にならないらしい。ssp (StackSmashingProtector) と pch は disable にしないと、普通の gcc もビルドできない。
$ /usr/local/mips/bin/mips-elf-gcc-4.3.2.exe -v
Using built-in specs.
Target: mips-elf
Configured with: ../gcc-4.3.2/configure --with-gmp=/usr/local --with-mpfr=/usr/local --enable-languages=c,c++ --disable-nls --disable-win32-registry --disable-shared --without-x --without-included-gettext --disable-libssp --prefix=/usr/local/mips/ --target=mips-elf --disable-multilib --without-fp --disable-libmudflap --disable-libstdc --disable-libstdcxx --disable-libstdcxx-pch --enable-threads --enable-hash-synchronization
Thread model: single
gcc version 4.3.2 (GCC)

libgcc の target ごとに使われる関数

libgcc は、乗算や除算など C の言語仕様のうち、CPU が直接サポートしてない機能を補助するために、GCC が勝手にくっ付けるライブラリ。

gcc-4.3.2/gcc/config/arm/t-linux とかを見ると。
LIB1ASMSRC = arm/lib1funcs.asm
LIB1ASMFUNCS = _udivsi3 _divsi3 _umodsi3 _modsi3 _dvmd_lnx

のように、呼ばれる関数が書いてある。

target ごとの設定は t-* とかに書いてあるらしい。

2008年12月3日水曜日

GCC のサンプルバックエンド

Incremental Machine Descriptions for Spim を参考に、gcc-4.3.2 に対応させて作ってみた。
http://github.com/alohakun/gcc-sample-backend/tree/master
最小限の C コードだけに対応したバックエンド。
$ ../bin/gcc-spim.exe -v
Using built-in specs.
Target: spim
Configured with: ../gcc-4.3.2/configure --with-gmp=/usr/local
--with-mpfr=/usr/local --enable-languages=c
--prefix=/home/aloha/test/gcc --target=spim --disable-libgcc
--enable-threads --disable-nls --disable-win32-registry --disable-shared
--without-x --enable-hash-synchronization
--enable-version-specific-runtime-libs --without-included-gettext
--disable-bootstrap --disable-libssp --program-suffix=-spim
--with-as=/usr/local/bin/as --with-ld=/usr/local/bin/ld
Thread model: single
gcc version 4.3.2 (GCC)

$ cat test0.c
void f()
{
L:
goto L;
}

void g()
{
}

$ ../bin/gcc-spim.exe -S -O2 test0.c

$ cat test0.s
.text
.align 2
.globl f
f:
L2: sw $ra, 0($sp)
sw $sp, -4($sp)
sw $fp, -8($sp)
move $fp,$sp
addi $sp, $fp, -44
L3: j L3
.align 2
.globl g
g:
sw $ra, 0($sp)
sw $sp, -4($sp)
sw $fp, -8($sp)
move $fp,$sp
sw $s0, 0($fp)
sw $s1, -4($fp)
sw $s2, -8($fp)
sw $s3, -12($fp)
sw $s4, -16($fp)
sw $s5, -20($fp)
sw $s6, -24($fp)
sw $s7, -28($fp)
addi $sp, $fp, -44
lw $s0, 0($fp)
lw $s1, -4($fp)
lw $s2, -8($fp)
lw $s3, -12($fp)
lw $s4, -16($fp)
lw $s5, -20($fp)
lw $s6, -24($fp)
lw $s7, -28($fp)
move $sp,$fp
lw $fp, -8($sp)
lw $ra, 0($sp)
jr $ra

こんだけでも move や addi が生成されるので、けっこう大きい。

変更ファイル : (spim ターゲットを追加しただけ)

gcc-4.3.2/config.sub
gcc-4.3.2/gcc/config.gcc

追加ファイル :

gcc-4.3.2/gcc/config/spim/spim.md (138 行)
gcc-43.2/gcc/config/spim/spim.h (370 行)
gcc-4.3.2/gcc/config/spim/spim.c (737 行)
gcc-4.3.2/gcc/config/spim/spim-protos.h (52 行)

展開して、--target=spim を指定して、通常通りビルドすればいけるはず (prefix を適当に指定)。
libgcc と as と ld を disable にしてあるので、binutils のインストールもたぶん不要 (もちろん、-S 付けてのアセンブリコード生成までしかできない)

2008年12月2日火曜日

RTL 式

RTL 式は、常にポインタ経由で扱われる C 構造体 struct rtx 型。

整数、ワイド整数(HOST_WIDE_INT)、C 文字列(char*)、配列を含む S 式。

RTL 式内の C 文字列 "..." は、基本的に C の文字列リテラルと同じだけど、複数行に渡って記述できるように拡張されている。

配列は [ ... ] という式表現。任意個のポインタを含む。
長さ 0 の配列は作れない。代わりにヌルポインタを使う。
ヌルポインタは (nil) と書く。

GET_CODE (x) で取得、PUT_CODE (x, newcode) で変更できる RTL コードにより、オペランドの数などが決まる。
rtl.def で定義されている列挙体。

gcc-4.3.2/gcc/rtl.def

/* This file contains the definitions and documentation for the
Register Transfer Expressions (rtx's) that make up the
Register Transfer Language (rtl) used in the Back End of the GNU compiler.
Copyright (C) 1987, 1988, 1992, 1994, 1995, 1997, 1998, 1999, 2000, 2004,
2005, 2006, 2007
Free Software Foundation, Inc.

This file is part of GCC.

GCC 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 3, or (at your option) any later
version.

GCC 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 GCC; see the file COPYING3. If not see
. */


/* Expression definitions and descriptions for all targets are in this file.
Some will not be used for some targets.

The fields in the cpp macro call "DEF_RTL_EXPR()"
are used to create declarations in the C source of the compiler.

The fields are:

1. The internal name of the rtx used in the C source.
It is a tag in the enumeration "enum rtx_code" defined in "rtl.h".
By convention these are in UPPER_CASE.

2. The name of the rtx in the external ASCII format read by
read_rtx(), and printed by print_rtx().
These names are stored in rtx_name[].
By convention these are the internal (field 1) names in lower_case.

3. The print format, and type of each rtx->u.fld[] (field) in this rtx.
These formats are stored in rtx_format[].
The meaning of the formats is documented in front of this array in rtl.c

4. The class of the rtx. These are stored in rtx_class and are accessed
via the GET_RTX_CLASS macro. They are defined as follows:

RTX_CONST_OBJ
an rtx code that can be used to represent a constant object
(e.g, CONST_INT)
RTX_OBJ
an rtx code that can be used to represent an object (e.g, REG, MEM)
RTX_COMPARE
an rtx code for a comparison (e.g, LT, GT)
RTX_COMM_COMPARE
an rtx code for a commutative comparison (e.g, EQ, NE, ORDERED)
RTX_UNARY
an rtx code for a unary arithmetic expression (e.g, NEG, NOT)
RTX_COMM_ARITH
an rtx code for a commutative binary operation (e.g,, PLUS, MULT)
RTX_TERNARY
an rtx code for a non-bitfield three input operation (IF_THEN_ELSE)
RTX_BIN_ARITH
an rtx code for a non-commutative binary operation (e.g., MINUS, DIV)
RTX_BITFIELD_OPS
an rtx code for a bit-field operation (ZERO_EXTRACT, SIGN_EXTRACT)
RTX_INSN
an rtx code for a machine insn (INSN, JUMP_INSN, CALL_INSN)
RTX_MATCH
an rtx code for something that matches in insns (e.g, MATCH_DUP)
RTX_AUTOINC
an rtx code for autoincrement addressing modes (e.g. POST_DEC)
RTX_EXTRA
everything else

All of the expressions that appear only in machine descriptions,
not in RTL used by the compiler itself, are at the end of the file. */

/* Unknown, or no such operation; the enumeration constant should have
value zero. */
DEF_RTL_EXPR(UNKNOWN, "UnKnown", "*", RTX_EXTRA)

/* ---------------------------------------------------------------------
Expressions used in constructing lists.
--------------------------------------------------------------------- */

/* a linked list of expressions */
DEF_RTL_EXPR(EXPR_LIST, "expr_list", "ee", RTX_EXTRA)

/* a linked list of instructions.
The insns are represented in print by their uids. */
DEF_RTL_EXPR(INSN_LIST, "insn_list", "ue", RTX_EXTRA)

/* SEQUENCE appears in the result of a `gen_...' function
for a DEFINE_EXPAND that wants to make several insns.
Its elements are the bodies of the insns that should be made.
`emit_insn' takes the SEQUENCE apart and makes separate insns. */
DEF_RTL_EXPR(SEQUENCE, "sequence", "E", RTX_EXTRA)

/* Refers to the address of its argument. This is only used in alias.c. */
DEF_RTL_EXPR(ADDRESS, "address", "e", RTX_MATCH)

/* ----------------------------------------------------------------------
Expression types used for things in the instruction chain.

All formats must start with "iuu" to handle the chain.
Each insn expression holds an rtl instruction and its semantics
during back-end processing.
See macros's in "rtl.h" for the meaning of each rtx->u.fld[].

---------------------------------------------------------------------- */

/* An instruction that cannot jump. */
DEF_RTL_EXPR(INSN, "insn", "iuuBieie", RTX_INSN)

/* An instruction that can possibly jump.
Fields ( rtx->u.fld[] ) have exact same meaning as INSN's. */
DEF_RTL_EXPR(JUMP_INSN, "jump_insn", "iuuBieie0", RTX_INSN)

/* An instruction that can possibly call a subroutine
but which will not change which instruction comes next
in the current function.
Field ( rtx->u.fld[8] ) is CALL_INSN_FUNCTION_USAGE.
All other fields ( rtx->u.fld[] ) have exact same meaning as INSN's. */
DEF_RTL_EXPR(CALL_INSN, "call_insn", "iuuBieiee", RTX_INSN)

/* A marker that indicates that control will not flow through. */
DEF_RTL_EXPR(BARRIER, "barrier", "iuu00000", RTX_EXTRA)

/* Holds a label that is followed by instructions.
Operand:
4: is used in jump.c for the use-count of the label.
5: is used in the sh backend.
6: is a number that is unique in the entire compilation.
7: is the user-given name of the label, if any. */
DEF_RTL_EXPR(CODE_LABEL, "code_label", "iuuB00is", RTX_EXTRA)

/* Say where in the code a source line starts, for symbol table's sake.
Operand:
4: note-specific data
5: enum insn_note
6: unique number if insn_note == note_insn_deleted_label. */
DEF_RTL_EXPR(NOTE, "note", "iuuB0ni", RTX_EXTRA)

/* ----------------------------------------------------------------------
Top level constituents of INSN, JUMP_INSN and CALL_INSN.
---------------------------------------------------------------------- */

/* Conditionally execute code.
Operand 0 is the condition that if true, the code is executed.
Operand 1 is the code to be executed (typically a SET).

Semantics are that there are no side effects if the condition
is false. This pattern is created automatically by the if_convert
pass run after reload or by target-specific splitters. */
DEF_RTL_EXPR(COND_EXEC, "cond_exec", "ee", RTX_EXTRA)

/* Several operations to be done in parallel (perhaps under COND_EXEC). */
DEF_RTL_EXPR(PARALLEL, "parallel", "E", RTX_EXTRA)

#ifdef USE_MAPPED_LOCATION
/* A string that is passed through to the assembler as input.
One can obviously pass comments through by using the
assembler comment syntax.
These occur in an insn all by themselves as the PATTERN.
They also appear inside an ASM_OPERANDS
as a convenient way to hold a string. */
DEF_RTL_EXPR(ASM_INPUT, "asm_input", "si", RTX_EXTRA)

/* An assembler instruction with operands.
1st operand is the instruction template.
2nd operand is the constraint for the output.
3rd operand is the number of the output this expression refers to.
When an insn stores more than one value, a separate ASM_OPERANDS
is made for each output; this integer distinguishes them.
4th is a vector of values of input operands.
5th is a vector of modes and constraints for the input operands.
Each element is an ASM_INPUT containing a constraint string
and whose mode indicates the mode of the input operand.
6th is the source line number. */
DEF_RTL_EXPR(ASM_OPERANDS, "asm_operands", "ssiEEi", RTX_EXTRA)
#else
/* A string that is passed through to the assembler as input.
One can obviously pass comments through by using the
assembler comment syntax.
These occur in an insn all by themselves as the PATTERN.
They also appear inside an ASM_OPERANDS
as a convenient way to hold a string. */
DEF_RTL_EXPR(ASM_INPUT, "asm_input", "ssi", RTX_EXTRA)

/* An assembler instruction with operands.
1st operand is the instruction template.
2nd operand is the constraint for the output.
3rd operand is the number of the output this expression refers to.
When an insn stores more than one value, a separate ASM_OPERANDS
is made for each output; this integer distinguishes them.
4th is a vector of values of input operands.
5th is a vector of modes and constraints for the input operands.
Each element is an ASM_INPUT containing a constraint string
and whose mode indicates the mode of the input operand.
6th is the name of the containing source file.
7th is the source line number. */
DEF_RTL_EXPR(ASM_OPERANDS, "asm_operands", "ssiEEsi", RTX_EXTRA)
#endif

/* A machine-specific operation.
1st operand is a vector of operands being used by the operation so that
any needed reloads can be done.
2nd operand is a unique value saying which of a number of machine-specific
operations is to be performed.
(Note that the vector must be the first operand because of the way that
genrecog.c record positions within an insn.)

UNSPEC can occur all by itself in a PATTERN, as a component of a PARALLEL,
or inside an expression.
UNSPEC by itself or as a component of a PARALLEL
is currently considered not deletable.

FIXME: Replace all uses of UNSPEC that appears by itself or as a component
of a PARALLEL with USE.
*/
DEF_RTL_EXPR(UNSPEC, "unspec", "Ei", RTX_EXTRA)

/* Similar, but a volatile operation and one which may trap. */
DEF_RTL_EXPR(UNSPEC_VOLATILE, "unspec_volatile", "Ei", RTX_EXTRA)

/* Vector of addresses, stored as full words. */
/* Each element is a LABEL_REF to a CODE_LABEL whose address we want. */
DEF_RTL_EXPR(ADDR_VEC, "addr_vec", "E", RTX_EXTRA)

/* Vector of address differences X0 - BASE, X1 - BASE, ...
First operand is BASE; the vector contains the X's.
The machine mode of this rtx says how much space to leave
for each difference and is adjusted by branch shortening if
CASE_VECTOR_SHORTEN_MODE is defined.
The third and fourth operands store the target labels with the
minimum and maximum addresses respectively.
The fifth operand stores flags for use by branch shortening.
Set at the start of shorten_branches:
min_align: the minimum alignment for any of the target labels.
base_after_vec: true iff BASE is after the ADDR_DIFF_VEC.
min_after_vec: true iff minimum addr target label is after the ADDR_DIFF_VEC.
max_after_vec: true iff maximum addr target label is after the ADDR_DIFF_VEC.
min_after_base: true iff minimum address target label is after BASE.
max_after_base: true iff maximum address target label is after BASE.
Set by the actual branch shortening process:
offset_unsigned: true iff offsets have to be treated as unsigned.
scale: scaling that is necessary to make offsets fit into the mode.

The third, fourth and fifth operands are only valid when
CASE_VECTOR_SHORTEN_MODE is defined, and only in an optimizing
compilations. */

DEF_RTL_EXPR(ADDR_DIFF_VEC, "addr_diff_vec", "eEee0", RTX_EXTRA)

/* Memory prefetch, with attributes supported on some targets.
Operand 1 is the address of the memory to fetch.
Operand 2 is 1 for a write access, 0 otherwise.
Operand 3 is the level of temporal locality; 0 means there is no
temporal locality and 1, 2, and 3 are for increasing levels of temporal
locality.

The attributes specified by operands 2 and 3 are ignored for targets
whose prefetch instructions do not support them. */
DEF_RTL_EXPR(PREFETCH, "prefetch", "eee", RTX_EXTRA)

/* ----------------------------------------------------------------------
At the top level of an instruction (perhaps under PARALLEL).
---------------------------------------------------------------------- */

/* Assignment.
Operand 1 is the location (REG, MEM, PC, CC0 or whatever) assigned to.
Operand 2 is the value stored there.
ALL assignment must use SET.
Instructions that do multiple assignments must use multiple SET,
under PARALLEL. */
DEF_RTL_EXPR(SET, "set", "ee", RTX_EXTRA)

/* Indicate something is used in a way that we don't want to explain.
For example, subroutine calls will use the register
in which the static chain is passed.

USE can not appear as an operand of other rtx except for PARALLEL.
USE is not deletable, as it indicates that the operand
is used in some unknown way. */
DEF_RTL_EXPR(USE, "use", "e", RTX_EXTRA)

/* Indicate something is clobbered in a way that we don't want to explain.
For example, subroutine calls will clobber some physical registers
(the ones that are by convention not saved).

CLOBBER can not appear as an operand of other rtx except for PARALLEL.
CLOBBER of a hard register appearing by itself (not within PARALLEL)
is considered undeletable before reload. */
DEF_RTL_EXPR(CLOBBER, "clobber", "e", RTX_EXTRA)

/* Call a subroutine.
Operand 1 is the address to call.
Operand 2 is the number of arguments. */

DEF_RTL_EXPR(CALL, "call", "ee", RTX_EXTRA)

/* Return from a subroutine. */

DEF_RTL_EXPR(RETURN, "return", "", RTX_EXTRA)

/* Conditional trap.
Operand 1 is the condition.
Operand 2 is the trap code.
For an unconditional trap, make the condition (const_int 1). */
DEF_RTL_EXPR(TRAP_IF, "trap_if", "ee", RTX_EXTRA)

/* Placeholder for _Unwind_Resume before we know if a function call
or a branch is needed. Operand 1 is the exception region from
which control is flowing. */
DEF_RTL_EXPR(RESX, "resx", "i", RTX_EXTRA)

/* ----------------------------------------------------------------------
Primitive values for use in expressions.
---------------------------------------------------------------------- */

/* numeric integer constant */
DEF_RTL_EXPR(CONST_INT, "const_int", "w", RTX_CONST_OBJ)

/* fixed-point constant */
DEF_RTL_EXPR(CONST_FIXED, "const_fixed", "www", RTX_CONST_OBJ)

/* numeric floating point constant.
Operands hold the value. They are all 'w' and there may be from 2 to 6;
see real.h. */
DEF_RTL_EXPR(CONST_DOUBLE, "const_double", CONST_DOUBLE_FORMAT, RTX_CONST_OBJ)

/* Describes a vector constant. */
DEF_RTL_EXPR(CONST_VECTOR, "const_vector", "E", RTX_CONST_OBJ)

/* String constant. Used for attributes in machine descriptions and
for special cases in DWARF2 debug output. NOT used for source-
language string constants. */
DEF_RTL_EXPR(CONST_STRING, "const_string", "s", RTX_OBJ)

/* This is used to encapsulate an expression whose value is constant
(such as the sum of a SYMBOL_REF and a CONST_INT) so that it will be
recognized as a constant operand rather than by arithmetic instructions. */

DEF_RTL_EXPR(CONST, "const", "e", RTX_CONST_OBJ)

/* program counter. Ordinary jumps are represented
by a SET whose first operand is (PC). */
DEF_RTL_EXPR(PC, "pc", "", RTX_OBJ)

/* Used in the cselib routines to describe a value. Objects of this
kind are only allocated in cselib.c, in an alloc pool instead of
in GC memory. The only operand of a VALUE is a cselib_val_struct. */
DEF_RTL_EXPR(VALUE, "value", "0", RTX_OBJ)

/* A register. The "operand" is the register number, accessed with
the REGNO macro. If this number is less than FIRST_PSEUDO_REGISTER
than a hardware register is being referred to. The second operand
holds the original register number - this will be different for a
pseudo register that got turned into a hard register. The third
operand points to a reg_attrs structure.
This rtx needs to have as many (or more) fields as a MEM, since we
can change REG rtx's into MEMs during reload. */
DEF_RTL_EXPR(REG, "reg", "i00", RTX_OBJ)

/* A scratch register. This represents a register used only within a
single insn. It will be turned into a REG during register allocation
or reload unless the constraint indicates that the register won't be
needed, in which case it can remain a SCRATCH. This code is
marked as having one operand so it can be turned into a REG. */
DEF_RTL_EXPR(SCRATCH, "scratch", "0", RTX_OBJ)

/* A reference to a part of another value. The first operand is the
complete value and the second is the byte offset of the selected part. */
DEF_RTL_EXPR(SUBREG, "subreg", "ei", RTX_EXTRA)

/* This one-argument rtx is used for move instructions
that are guaranteed to alter only the low part of a destination.
Thus, (SET (SUBREG:HI (REG...)) (MEM:HI ...))
has an unspecified effect on the high part of REG,
but (SET (STRICT_LOW_PART (SUBREG:HI (REG...))) (MEM:HI ...))
is guaranteed to alter only the bits of REG that are in HImode.

The actual instruction used is probably the same in both cases,
but the register constraints may be tighter when STRICT_LOW_PART
is in use. */

DEF_RTL_EXPR(STRICT_LOW_PART, "strict_low_part", "e", RTX_EXTRA)

/* (CONCAT a b) represents the virtual concatenation of a and b
to make a value that has as many bits as a and b put together.
This is used for complex values. Normally it appears only
in DECL_RTLs and during RTL generation, but not in the insn chain. */
DEF_RTL_EXPR(CONCAT, "concat", "ee", RTX_OBJ)

/* (CONCATN [a1 a2 ... an]) represents the virtual concatenation of
all An to make a value. This is an extension of CONCAT to larger
number of components. Like CONCAT, it should not appear in the
insn chain. Every element of the CONCATN is the same size. */
DEF_RTL_EXPR(CONCATN, "concatn", "E", RTX_OBJ)

/* A memory location; operand is the address. The second operand is the
alias set to which this MEM belongs. We use `0' instead of `w' for this
field so that the field need not be specified in machine descriptions. */
DEF_RTL_EXPR(MEM, "mem", "e0", RTX_OBJ)

/* Reference to an assembler label in the code for this function.
The operand is a CODE_LABEL found in the insn chain. */
DEF_RTL_EXPR(LABEL_REF, "label_ref", "u", RTX_CONST_OBJ)

/* Reference to a named label:
Operand 0: label name
Operand 1: flags (see SYMBOL_FLAG_* in rtl.h)
Operand 2: tree from which this symbol is derived, or null.
This is either a DECL node, or some kind of constant. */
DEF_RTL_EXPR(SYMBOL_REF, "symbol_ref", "s00", RTX_CONST_OBJ)

/* The condition code register is represented, in our imagination,
as a register holding a value that can be compared to zero.
In fact, the machine has already compared them and recorded the
results; but instructions that look at the condition code
pretend to be looking at the entire value and comparing it. */
DEF_RTL_EXPR(CC0, "cc0", "", RTX_OBJ)

/* ----------------------------------------------------------------------
Expressions for operators in an rtl pattern
---------------------------------------------------------------------- */

/* if_then_else. This is used in representing ordinary
conditional jump instructions.
Operand:
0: condition
1: then expr
2: else expr */
DEF_RTL_EXPR(IF_THEN_ELSE, "if_then_else", "eee", RTX_TERNARY)

/* Comparison, produces a condition code result. */
DEF_RTL_EXPR(COMPARE, "compare", "ee", RTX_BIN_ARITH)

/* plus */
DEF_RTL_EXPR(PLUS, "plus", "ee", RTX_COMM_ARITH)

/* Operand 0 minus operand 1. */
DEF_RTL_EXPR(MINUS, "minus", "ee", RTX_BIN_ARITH)

/* Minus operand 0. */
DEF_RTL_EXPR(NEG, "neg", "e", RTX_UNARY)

DEF_RTL_EXPR(MULT, "mult", "ee", RTX_COMM_ARITH)

/* Multiplication with signed saturation */
DEF_RTL_EXPR(SS_MULT, "ss_mult", "ee", RTX_COMM_ARITH)
/* Multiplication with unsigned saturation */
DEF_RTL_EXPR(US_MULT, "us_mult", "ee", RTX_COMM_ARITH)

/* Operand 0 divided by operand 1. */
DEF_RTL_EXPR(DIV, "div", "ee", RTX_BIN_ARITH)
/* Division with signed saturation */
DEF_RTL_EXPR(SS_DIV, "ss_div", "ee", RTX_BIN_ARITH)
/* Division with unsigned saturation */
DEF_RTL_EXPR(US_DIV, "us_div", "ee", RTX_BIN_ARITH)

/* Remainder of operand 0 divided by operand 1. */
DEF_RTL_EXPR(MOD, "mod", "ee", RTX_BIN_ARITH)

/* Unsigned divide and remainder. */
DEF_RTL_EXPR(UDIV, "udiv", "ee", RTX_BIN_ARITH)
DEF_RTL_EXPR(UMOD, "umod", "ee", RTX_BIN_ARITH)

/* Bitwise operations. */
DEF_RTL_EXPR(AND, "and", "ee", RTX_COMM_ARITH)
DEF_RTL_EXPR(IOR, "ior", "ee", RTX_COMM_ARITH)
DEF_RTL_EXPR(XOR, "xor", "ee", RTX_COMM_ARITH)
DEF_RTL_EXPR(NOT, "not", "e", RTX_UNARY)

/* Operand:
0: value to be shifted.
1: number of bits. */
DEF_RTL_EXPR(ASHIFT, "ashift", "ee", RTX_BIN_ARITH) /* shift left */
DEF_RTL_EXPR(ROTATE, "rotate", "ee", RTX_BIN_ARITH) /* rotate left */
DEF_RTL_EXPR(ASHIFTRT, "ashiftrt", "ee", RTX_BIN_ARITH) /* arithmetic shift right */
DEF_RTL_EXPR(LSHIFTRT, "lshiftrt", "ee", RTX_BIN_ARITH) /* logical shift right */
DEF_RTL_EXPR(ROTATERT, "rotatert", "ee", RTX_BIN_ARITH) /* rotate right */

/* Minimum and maximum values of two operands. We need both signed and
unsigned forms. (We cannot use MIN for SMIN because it conflicts
with a macro of the same name.) The signed variants should be used
with floating point. Further, if both operands are zeros, or if either
operand is NaN, then it is unspecified which of the two operands is
returned as the result. */

DEF_RTL_EXPR(SMIN, "smin", "ee", RTX_COMM_ARITH)
DEF_RTL_EXPR(SMAX, "smax", "ee", RTX_COMM_ARITH)
DEF_RTL_EXPR(UMIN, "umin", "ee", RTX_COMM_ARITH)
DEF_RTL_EXPR(UMAX, "umax", "ee", RTX_COMM_ARITH)

/* These unary operations are used to represent incrementation
and decrementation as they occur in memory addresses.
The amount of increment or decrement are not represented
because they can be understood from the machine-mode of the
containing MEM. These operations exist in only two cases:
1. pushes onto the stack.
2. created automatically by the life_analysis pass in flow.c. */
DEF_RTL_EXPR(PRE_DEC, "pre_dec", "e", RTX_AUTOINC)
DEF_RTL_EXPR(PRE_INC, "pre_inc", "e", RTX_AUTOINC)
DEF_RTL_EXPR(POST_DEC, "post_dec", "e", RTX_AUTOINC)
DEF_RTL_EXPR(POST_INC, "post_inc", "e", RTX_AUTOINC)

/* These binary operations are used to represent generic address
side-effects in memory addresses, except for simple incrementation
or decrementation which use the above operations. They are
created automatically by the life_analysis pass in flow.c.
The first operand is a REG which is used as the address.
The second operand is an expression that is assigned to the
register, either before (PRE_MODIFY) or after (POST_MODIFY)
evaluating the address.
Currently, the compiler can only handle second operands of the
form (plus (reg) (reg)) and (plus (reg) (const_int)), where
the first operand of the PLUS has to be the same register as
the first operand of the *_MODIFY. */
DEF_RTL_EXPR(PRE_MODIFY, "pre_modify", "ee", RTX_AUTOINC)
DEF_RTL_EXPR(POST_MODIFY, "post_modify", "ee", RTX_AUTOINC)

/* Comparison operations. The ordered comparisons exist in two
flavors, signed and unsigned. */
DEF_RTL_EXPR(NE, "ne", "ee", RTX_COMM_COMPARE)
DEF_RTL_EXPR(EQ, "eq", "ee", RTX_COMM_COMPARE)
DEF_RTL_EXPR(GE, "ge", "ee", RTX_COMPARE)
DEF_RTL_EXPR(GT, "gt", "ee", RTX_COMPARE)
DEF_RTL_EXPR(LE, "le", "ee", RTX_COMPARE)
DEF_RTL_EXPR(LT, "lt", "ee", RTX_COMPARE)
DEF_RTL_EXPR(GEU, "geu", "ee", RTX_COMPARE)
DEF_RTL_EXPR(GTU, "gtu", "ee", RTX_COMPARE)
DEF_RTL_EXPR(LEU, "leu", "ee", RTX_COMPARE)
DEF_RTL_EXPR(LTU, "ltu", "ee", RTX_COMPARE)

/* Additional floating point unordered comparison flavors. */
DEF_RTL_EXPR(UNORDERED, "unordered", "ee", RTX_COMM_COMPARE)
DEF_RTL_EXPR(ORDERED, "ordered", "ee", RTX_COMM_COMPARE)

/* These are equivalent to unordered or ... */
DEF_RTL_EXPR(UNEQ, "uneq", "ee", RTX_COMM_COMPARE)
DEF_RTL_EXPR(UNGE, "unge", "ee", RTX_COMPARE)
DEF_RTL_EXPR(UNGT, "ungt", "ee", RTX_COMPARE)
DEF_RTL_EXPR(UNLE, "unle", "ee", RTX_COMPARE)
DEF_RTL_EXPR(UNLT, "unlt", "ee", RTX_COMPARE)

/* This is an ordered NE, ie !UNEQ, ie false for NaN. */
DEF_RTL_EXPR(LTGT, "ltgt", "ee", RTX_COMM_COMPARE)

/* Represents the result of sign-extending the sole operand.
The machine modes of the operand and of the SIGN_EXTEND expression
determine how much sign-extension is going on. */
DEF_RTL_EXPR(SIGN_EXTEND, "sign_extend", "e", RTX_UNARY)

/* Similar for zero-extension (such as unsigned short to int). */
DEF_RTL_EXPR(ZERO_EXTEND, "zero_extend", "e", RTX_UNARY)

/* Similar but here the operand has a wider mode. */
DEF_RTL_EXPR(TRUNCATE, "truncate", "e", RTX_UNARY)

/* Similar for extending floating-point values (such as SFmode to DFmode). */
DEF_RTL_EXPR(FLOAT_EXTEND, "float_extend", "e", RTX_UNARY)
DEF_RTL_EXPR(FLOAT_TRUNCATE, "float_truncate", "e", RTX_UNARY)

/* Conversion of fixed point operand to floating point value. */
DEF_RTL_EXPR(FLOAT, "float", "e", RTX_UNARY)

/* With fixed-point machine mode:
Conversion of floating point operand to fixed point value.
Value is defined only when the operand's value is an integer.
With floating-point machine mode (and operand with same mode):
Operand is rounded toward zero to produce an integer value
represented in floating point. */
DEF_RTL_EXPR(FIX, "fix", "e", RTX_UNARY)

/* Conversion of unsigned fixed point operand to floating point value. */
DEF_RTL_EXPR(UNSIGNED_FLOAT, "unsigned_float", "e", RTX_UNARY)

/* With fixed-point machine mode:
Conversion of floating point operand to *unsigned* fixed point value.
Value is defined only when the operand's value is an integer. */
DEF_RTL_EXPR(UNSIGNED_FIX, "unsigned_fix", "e", RTX_UNARY)

/* Conversions involving fractional fixed-point types without saturation,
including:
fractional to fractional (of different precision),
signed integer to fractional,
fractional to signed integer,
floating point to fractional,
fractional to floating point.
NOTE: fractional can be either signed or unsigned for conversions. */
DEF_RTL_EXPR(FRACT_CONVERT, "fract_convert", "e", RTX_UNARY)

/* Conversions involving fractional fixed-point types and unsigned integer
without saturation, including:
unsigned integer to fractional,
fractional to unsigned integer.
NOTE: fractional can be either signed or unsigned for conversions. */
DEF_RTL_EXPR(UNSIGNED_FRACT_CONVERT, "unsigned_fract_convert", "e", RTX_UNARY)

/* Conversions involving fractional fixed-point types with saturation,
including:
fractional to fractional (of different precision),
signed integer to fractional,
floating point to fractional.
NOTE: fractional can be either signed or unsigned for conversions. */
DEF_RTL_EXPR(SAT_FRACT, "sat_fract", "e", RTX_UNARY)

/* Conversions involving fractional fixed-point types and unsigned integer
with saturation, including:
unsigned integer to fractional.
NOTE: fractional can be either signed or unsigned for conversions. */
DEF_RTL_EXPR(UNSIGNED_SAT_FRACT, "unsigned_sat_fract", "e", RTX_UNARY)

/* Absolute value */
DEF_RTL_EXPR(ABS, "abs", "e", RTX_UNARY)

/* Square root */
DEF_RTL_EXPR(SQRT, "sqrt", "e", RTX_UNARY)

/* Swap bytes. */
DEF_RTL_EXPR(BSWAP, "bswap", "e", RTX_UNARY)

/* Find first bit that is set.
Value is 1 + number of trailing zeros in the arg.,
or 0 if arg is 0. */
DEF_RTL_EXPR(FFS, "ffs", "e", RTX_UNARY)

/* Count leading zeros. */
DEF_RTL_EXPR(CLZ, "clz", "e", RTX_UNARY)

/* Count trailing zeros. */
DEF_RTL_EXPR(CTZ, "ctz", "e", RTX_UNARY)

/* Population count (number of 1 bits). */
DEF_RTL_EXPR(POPCOUNT, "popcount", "e", RTX_UNARY)

/* Population parity (number of 1 bits modulo 2). */
DEF_RTL_EXPR(PARITY, "parity", "e", RTX_UNARY)

/* Reference to a signed bit-field of specified size and position.
Operand 0 is the memory unit (usually SImode or QImode) which
contains the field's first bit. Operand 1 is the width, in bits.
Operand 2 is the number of bits in the memory unit before the
first bit of this field.
If BITS_BIG_ENDIAN is defined, the first bit is the msb and
operand 2 counts from the msb of the memory unit.
Otherwise, the first bit is the lsb and operand 2 counts from
the lsb of the memory unit.
This kind of expression can not appear as an lvalue in RTL. */
DEF_RTL_EXPR(SIGN_EXTRACT, "sign_extract", "eee", RTX_BITFIELD_OPS)

/* Similar for unsigned bit-field.
But note! This kind of expression _can_ appear as an lvalue. */
DEF_RTL_EXPR(ZERO_EXTRACT, "zero_extract", "eee", RTX_BITFIELD_OPS)

/* For RISC machines. These save memory when splitting insns. */

/* HIGH are the high-order bits of a constant expression. */
DEF_RTL_EXPR(HIGH, "high", "e", RTX_CONST_OBJ)

/* LO_SUM is the sum of a register and the low-order bits
of a constant expression. */
DEF_RTL_EXPR(LO_SUM, "lo_sum", "ee", RTX_OBJ)

/* Describes a merge operation between two vector values.
Operands 0 and 1 are the vectors to be merged, operand 2 is a bitmask
that specifies where the parts of the result are taken from. Set bits
indicate operand 0, clear bits indicate operand 1. The parts are defined
by the mode of the vectors. */
DEF_RTL_EXPR(VEC_MERGE, "vec_merge", "eee", RTX_TERNARY)

/* Describes an operation that selects parts of a vector.
Operands 0 is the source vector, operand 1 is a PARALLEL that contains
a CONST_INT for each of the subparts of the result vector, giving the
number of the source subpart that should be stored into it. */
DEF_RTL_EXPR(VEC_SELECT, "vec_select", "ee", RTX_BIN_ARITH)

/* Describes a vector concat operation. Operands 0 and 1 are the source
vectors, the result is a vector that is as long as operands 0 and 1
combined and is the concatenation of the two source vectors. */
DEF_RTL_EXPR(VEC_CONCAT, "vec_concat", "ee", RTX_BIN_ARITH)

/* Describes an operation that converts a small vector into a larger one by
duplicating the input values. The output vector mode must have the same
submodes as the input vector mode, and the number of output parts must be
an integer multiple of the number of input parts. */
DEF_RTL_EXPR(VEC_DUPLICATE, "vec_duplicate", "e", RTX_UNARY)

/* Addition with signed saturation */
DEF_RTL_EXPR(SS_PLUS, "ss_plus", "ee", RTX_COMM_ARITH)

/* Addition with unsigned saturation */
DEF_RTL_EXPR(US_PLUS, "us_plus", "ee", RTX_COMM_ARITH)

/* Operand 0 minus operand 1, with signed saturation. */
DEF_RTL_EXPR(SS_MINUS, "ss_minus", "ee", RTX_BIN_ARITH)

/* Negation with signed saturation. */
DEF_RTL_EXPR(SS_NEG, "ss_neg", "e", RTX_UNARY)
/* Negation with unsigned saturation. */
DEF_RTL_EXPR(US_NEG, "us_neg", "e", RTX_UNARY)

/* Absolute value with signed saturation. */
DEF_RTL_EXPR(SS_ABS, "ss_abs", "e", RTX_UNARY)

/* Shift left with signed saturation. */
DEF_RTL_EXPR(SS_ASHIFT, "ss_ashift", "ee", RTX_BIN_ARITH)

/* Shift left with unsigned saturation. */
DEF_RTL_EXPR(US_ASHIFT, "us_ashift", "ee", RTX_BIN_ARITH)

/* Operand 0 minus operand 1, with unsigned saturation. */
DEF_RTL_EXPR(US_MINUS, "us_minus", "ee", RTX_BIN_ARITH)

/* Signed saturating truncate. */
DEF_RTL_EXPR(SS_TRUNCATE, "ss_truncate", "e", RTX_UNARY)

/* Unsigned saturating truncate. */
DEF_RTL_EXPR(US_TRUNCATE, "us_truncate", "e", RTX_UNARY)

/* Information about the variable and its location. */
/* Changed 'te' to 'tei'; the 'i' field is for recording
initialization status of variables. */
DEF_RTL_EXPR(VAR_LOCATION, "var_location", "tei", RTX_EXTRA)

/* All expressions from this point forward appear only in machine
descriptions. */
#ifdef GENERATOR_FILE

/* Include a secondary machine-description file at this point. */
DEF_RTL_EXPR(INCLUDE, "include", "s", RTX_EXTRA)

/* Pattern-matching operators: */

/* Use the function named by the second arg (the string)
as a predicate; if matched, store the structure that was matched
in the operand table at index specified by the first arg (the integer).
If the second arg is the null string, the structure is just stored.

A third string argument indicates to the register allocator restrictions
on where the operand can be allocated.

If the target needs no restriction on any instruction this field should
be the null string.

The string is prepended by:
'=' to indicate the operand is only written to.
'+' to indicate the operand is both read and written to.

Each character in the string represents an allocable class for an operand.
'g' indicates the operand can be any valid class.
'i' indicates the operand can be immediate (in the instruction) data.
'r' indicates the operand can be in a register.
'm' indicates the operand can be in memory.
'o' a subset of the 'm' class. Those memory addressing modes that
can be offset at compile time (have a constant added to them).

Other characters indicate target dependent operand classes and
are described in each target's machine description.

For instructions with more than one operand, sets of classes can be
separated by a comma to indicate the appropriate multi-operand constraints.
There must be a 1 to 1 correspondence between these sets of classes in
all operands for an instruction.
*/
DEF_RTL_EXPR(MATCH_OPERAND, "match_operand", "iss", RTX_MATCH)

/* Match a SCRATCH or a register. When used to generate rtl, a
SCRATCH is generated. As for MATCH_OPERAND, the mode specifies
the desired mode and the first argument is the operand number.
The second argument is the constraint. */
DEF_RTL_EXPR(MATCH_SCRATCH, "match_scratch", "is", RTX_MATCH)

/* Apply a predicate, AND match recursively the operands of the rtx.
Operand 0 is the operand-number, as in match_operand.
Operand 1 is a predicate to apply (as a string, a function name).
Operand 2 is a vector of expressions, each of which must match
one subexpression of the rtx this construct is matching. */
DEF_RTL_EXPR(MATCH_OPERATOR, "match_operator", "isE", RTX_MATCH)

/* Match a PARALLEL of arbitrary length. The predicate is applied
to the PARALLEL and the initial expressions in the PARALLEL are matched.
Operand 0 is the operand-number, as in match_operand.
Operand 1 is a predicate to apply to the PARALLEL.
Operand 2 is a vector of expressions, each of which must match the
corresponding element in the PARALLEL. */
DEF_RTL_EXPR(MATCH_PARALLEL, "match_parallel", "isE", RTX_MATCH)

/* Match only something equal to what is stored in the operand table
at the index specified by the argument. Use with MATCH_OPERAND. */
DEF_RTL_EXPR(MATCH_DUP, "match_dup", "i", RTX_MATCH)

/* Match only something equal to what is stored in the operand table
at the index specified by the argument. Use with MATCH_OPERATOR. */
DEF_RTL_EXPR(MATCH_OP_DUP, "match_op_dup", "iE", RTX_MATCH)

/* Match only something equal to what is stored in the operand table
at the index specified by the argument. Use with MATCH_PARALLEL. */
DEF_RTL_EXPR(MATCH_PAR_DUP, "match_par_dup", "iE", RTX_MATCH)

/* Appears only in define_predicate/define_special_predicate
expressions. Evaluates true only if the operand has an RTX code
from the set given by the argument (a comma-separated list). If the
second argument is present and nonempty, it is a sequence of digits
and/or letters which indicates the subexpression to test, using the
same syntax as genextract/genrecog's location strings: 0-9 for
XEXP (op, n), a-z for XVECEXP (op, 0, n); each character applies to
the result of the one before it. */
DEF_RTL_EXPR(MATCH_CODE, "match_code", "ss", RTX_MATCH)

/* Appears only in define_predicate/define_special_predicate
expressions. The argument is a C expression to be injected at this
point in the predicate formula. */
DEF_RTL_EXPR(MATCH_TEST, "match_test", "s", RTX_MATCH)

/* Insn (and related) definitions. */

/* Definition of the pattern for one kind of instruction.
Operand:
0: names this instruction.
If the name is the null string, the instruction is in the
machine description just to be recognized, and will never be emitted by
the tree to rtl expander.
1: is the pattern.
2: is a string which is a C expression
giving an additional condition for recognizing this pattern.
A null string means no extra condition.
3: is the action to execute if this pattern is matched.
If this assembler code template starts with a * then it is a fragment of
C code to run to decide on a template to use. Otherwise, it is the
template to use.
4: optionally, a vector of attributes for this insn.
*/
DEF_RTL_EXPR(DEFINE_INSN, "define_insn", "sEsTV", RTX_EXTRA)

/* Definition of a peephole optimization.
1st operand: vector of insn patterns to match
2nd operand: C expression that must be true
3rd operand: template or C code to produce assembler output.
4: optionally, a vector of attributes for this insn.

This form is deprecated; use define_peephole2 instead. */
DEF_RTL_EXPR(DEFINE_PEEPHOLE, "define_peephole", "EsTV", RTX_EXTRA)

/* Definition of a split operation.
1st operand: insn pattern to match
2nd operand: C expression that must be true
3rd operand: vector of insn patterns to place into a SEQUENCE
4th operand: optionally, some C code to execute before generating the
insns. This might, for example, create some RTX's and store them in
elements of `recog_data.operand' for use by the vector of
insn-patterns.
(`operands' is an alias here for `recog_data.operand'). */
DEF_RTL_EXPR(DEFINE_SPLIT, "define_split", "EsES", RTX_EXTRA)

/* Definition of an insn and associated split.
This is the concatenation, with a few modifications, of a define_insn
and a define_split which share the same pattern.
Operand:
0: names this instruction.
If the name is the null string, the instruction is in the
machine description just to be recognized, and will never be emitted by
the tree to rtl expander.
1: is the pattern.
2: is a string which is a C expression
giving an additional condition for recognizing this pattern.
A null string means no extra condition.
3: is the action to execute if this pattern is matched.
If this assembler code template starts with a * then it is a fragment of
C code to run to decide on a template to use. Otherwise, it is the
template to use.
4: C expression that must be true for split. This may start with "&&"
in which case the split condition is the logical and of the insn
condition and what follows the "&&" of this operand.
5: vector of insn patterns to place into a SEQUENCE
6: optionally, some C code to execute before generating the
insns. This might, for example, create some RTX's and store them in
elements of `recog_data.operand' for use by the vector of
insn-patterns.
(`operands' is an alias here for `recog_data.operand').
7: optionally, a vector of attributes for this insn. */
DEF_RTL_EXPR(DEFINE_INSN_AND_SPLIT, "define_insn_and_split", "sEsTsESV", RTX_EXTRA)

/* Definition of an RTL peephole operation.
Follows the same arguments as define_split. */
DEF_RTL_EXPR(DEFINE_PEEPHOLE2, "define_peephole2", "EsES", RTX_EXTRA)

/* Define how to generate multiple insns for a standard insn name.
1st operand: the insn name.
2nd operand: vector of insn-patterns.
Use match_operand to substitute an element of `recog_data.operand'.
3rd operand: C expression that must be true for this to be available.
This may not test any operands.
4th operand: Extra C code to execute before generating the insns.
This might, for example, create some RTX's and store them in
elements of `recog_data.operand' for use by the vector of
insn-patterns.
(`operands' is an alias here for `recog_data.operand'). */
DEF_RTL_EXPR(DEFINE_EXPAND, "define_expand", "sEss", RTX_EXTRA)

/* Define a requirement for delay slots.
1st operand: Condition involving insn attributes that, if true,
indicates that the insn requires the number of delay slots
shown.
2nd operand: Vector whose length is the three times the number of delay
slots required.
Each entry gives three conditions, each involving attributes.
The first must be true for an insn to occupy that delay slot
location. The second is true for all insns that can be
annulled if the branch is true and the third is true for all
insns that can be annulled if the branch is false.

Multiple DEFINE_DELAYs may be present. They indicate differing
requirements for delay slots. */
DEF_RTL_EXPR(DEFINE_DELAY, "define_delay", "eE", RTX_EXTRA)

/* Define attribute computation for `asm' instructions. */
DEF_RTL_EXPR(DEFINE_ASM_ATTRIBUTES, "define_asm_attributes", "V", RTX_EXTRA)

/* Definition of a conditional execution meta operation. Automatically
generates new instances of DEFINE_INSN, selected by having attribute
"predicable" true. The new pattern will contain a COND_EXEC and the
predicate at top-level.

Operand:
0: The predicate pattern. The top-level form should match a
relational operator. Operands should have only one alternative.
1: A C expression giving an additional condition for recognizing
the generated pattern.
2: A template or C code to produce assembler output. */
DEF_RTL_EXPR(DEFINE_COND_EXEC, "define_cond_exec", "Ess", RTX_EXTRA)

/* Definition of an operand predicate. The difference between
DEFINE_PREDICATE and DEFINE_SPECIAL_PREDICATE is that genrecog will
not warn about a match_operand with no mode if it has a predicate
defined with DEFINE_SPECIAL_PREDICATE.

Operand:
0: The name of the predicate.
1: A boolean expression which computes whether or not the predicate
matches. This expression can use IOR, AND, NOT, MATCH_OPERAND,
MATCH_CODE, and MATCH_TEST. It must be specific enough that genrecog
can calculate the set of RTX codes that can possibly match.
2: A C function body which must return true for the predicate to match.
Optional. Use this when the test is too complicated to fit into a
match_test expression. */
DEF_RTL_EXPR(DEFINE_PREDICATE, "define_predicate", "ses", RTX_EXTRA)
DEF_RTL_EXPR(DEFINE_SPECIAL_PREDICATE, "define_special_predicate", "ses", RTX_EXTRA)

/* Definition of a register operand constraint. This simply maps the
constraint string to a register class.

Operand:
0: The name of the constraint (often, but not always, a single letter).
1: A C expression which evaluates to the appropriate register class for
this constraint. If this is not just a constant, it should look only
at -m switches and the like.
2: A docstring for this constraint, in Texinfo syntax; not currently
used, in future will be incorporated into the manual's list of
machine-specific operand constraints. */
DEF_RTL_EXPR(DEFINE_REGISTER_CONSTRAINT, "define_register_constraint", "sss", RTX_EXTRA)

/* Definition of a non-register operand constraint. These look at the
operand and decide whether it fits the constraint.

DEFINE_CONSTRAINT gets no special treatment if it fails to match.
It is appropriate for constant-only constraints, and most others.

DEFINE_MEMORY_CONSTRAINT tells reload that this constraint can be made
to match, if it doesn't already, by converting the operand to the form
(mem (reg X)) where X is a base register. It is suitable for constraints
that describe a subset of all memory references.

DEFINE_ADDRESS_CONSTRAINT tells reload that this constraint can be made
to match, if it doesn't already, by converting the operand to the form
(reg X) where X is a base register. It is suitable for constraints that
describe a subset of all address references.

When in doubt, use plain DEFINE_CONSTRAINT.

Operand:
0: The name of the constraint (often, but not always, a single letter).
1: A docstring for this constraint, in Texinfo syntax; not currently
used, in future will be incorporated into the manual's list of
machine-specific operand constraints.
2: A boolean expression which computes whether or not the constraint
matches. It should follow the same rules as a define_predicate
expression, including the bit about specifying the set of RTX codes
that could possibly match. MATCH_TEST subexpressions may make use of
these variables:
`op' - the RTL object defining the operand.
`mode' - the mode of `op'.
`ival' - INTVAL(op), if op is a CONST_INT.
`hval' - CONST_DOUBLE_HIGH(op), if op is an integer CONST_DOUBLE.
`lval' - CONST_DOUBLE_LOW(op), if op is an integer CONST_DOUBLE.
`rval' - CONST_DOUBLE_REAL_VALUE(op), if op is a floating-point
CONST_DOUBLE.
Do not use ival/hval/lval/rval if op is not the appropriate kind of
RTL object. */
DEF_RTL_EXPR(DEFINE_CONSTRAINT, "define_constraint", "sse", RTX_EXTRA)
DEF_RTL_EXPR(DEFINE_MEMORY_CONSTRAINT, "define_memory_constraint", "sse", RTX_EXTRA)
DEF_RTL_EXPR(DEFINE_ADDRESS_CONSTRAINT, "define_address_constraint", "sse", RTX_EXTRA)


/* Constructions for CPU pipeline description described by NDFAs. */

/* (define_cpu_unit string [string]) describes cpu functional
units (separated by comma).

1st operand: Names of cpu functional units.
2nd operand: Name of automaton (see comments for DEFINE_AUTOMATON).

All define_reservations, define_cpu_units, and
define_query_cpu_units should have unique names which may not be
"nothing". */
DEF_RTL_EXPR(DEFINE_CPU_UNIT, "define_cpu_unit", "sS", RTX_EXTRA)

/* (define_query_cpu_unit string [string]) describes cpu functional
units analogously to define_cpu_unit. The reservation of such
units can be queried for automaton state. */
DEF_RTL_EXPR(DEFINE_QUERY_CPU_UNIT, "define_query_cpu_unit", "sS", RTX_EXTRA)

/* (exclusion_set string string) means that each CPU functional unit
in the first string can not be reserved simultaneously with any
unit whose name is in the second string and vise versa. CPU units
in the string are separated by commas. For example, it is useful
for description CPU with fully pipelined floating point functional
unit which can execute simultaneously only single floating point
insns or only double floating point insns. All CPU functional
units in a set should belong to the same automaton. */
DEF_RTL_EXPR(EXCLUSION_SET, "exclusion_set", "ss", RTX_EXTRA)

/* (presence_set string string) means that each CPU functional unit in
the first string can not be reserved unless at least one of pattern
of units whose names are in the second string is reserved. This is
an asymmetric relation. CPU units or unit patterns in the strings
are separated by commas. Pattern is one unit name or unit names
separated by white-spaces.

For example, it is useful for description that slot1 is reserved
after slot0 reservation for a VLIW processor. We could describe it
by the following construction

(presence_set "slot1" "slot0")

Or slot1 is reserved only after slot0 and unit b0 reservation. In
this case we could write

(presence_set "slot1" "slot0 b0")

All CPU functional units in a set should belong to the same
automaton. */
DEF_RTL_EXPR(PRESENCE_SET, "presence_set", "ss", RTX_EXTRA)

/* (final_presence_set string string) is analogous to `presence_set'.
The difference between them is when checking is done. When an
instruction is issued in given automaton state reflecting all
current and planned unit reservations, the automaton state is
changed. The first state is a source state, the second one is a
result state. Checking for `presence_set' is done on the source
state reservation, checking for `final_presence_set' is done on the
result reservation. This construction is useful to describe a
reservation which is actually two subsequent reservations. For
example, if we use

(presence_set "slot1" "slot0")

the following insn will be never issued (because slot1 requires
slot0 which is absent in the source state).

(define_reservation "insn_and_nop" "slot0 + slot1")

but it can be issued if we use analogous `final_presence_set'. */
DEF_RTL_EXPR(FINAL_PRESENCE_SET, "final_presence_set", "ss", RTX_EXTRA)

/* (absence_set string string) means that each CPU functional unit in
the first string can be reserved only if each pattern of units
whose names are in the second string is not reserved. This is an
asymmetric relation (actually exclusion set is analogous to this
one but it is symmetric). CPU units or unit patterns in the string
are separated by commas. Pattern is one unit name or unit names
separated by white-spaces.

For example, it is useful for description that slot0 can not be
reserved after slot1 or slot2 reservation for a VLIW processor. We
could describe it by the following construction

(absence_set "slot2" "slot0, slot1")

Or slot2 can not be reserved if slot0 and unit b0 are reserved or
slot1 and unit b1 are reserved . In this case we could write

(absence_set "slot2" "slot0 b0, slot1 b1")

All CPU functional units in a set should to belong the same
automaton. */
DEF_RTL_EXPR(ABSENCE_SET, "absence_set", "ss", RTX_EXTRA)

/* (final_absence_set string string) is analogous to `absence_set' but
checking is done on the result (state) reservation. See comments
for `final_presence_set'. */
DEF_RTL_EXPR(FINAL_ABSENCE_SET, "final_absence_set", "ss", RTX_EXTRA)

/* (define_bypass number out_insn_names in_insn_names) names bypass
with given latency (the first number) from insns given by the first
string (see define_insn_reservation) into insns given by the second
string. Insn names in the strings are separated by commas. The
third operand is optional name of function which is additional
guard for the bypass. The function will get the two insns as
parameters. If the function returns zero the bypass will be
ignored for this case. Additional guard is necessary to recognize
complicated bypasses, e.g. when consumer is load address. */
DEF_RTL_EXPR(DEFINE_BYPASS, "define_bypass", "issS", RTX_EXTRA)

/* (define_automaton string) describes names of automata generated and
used for pipeline hazards recognition. The names are separated by
comma. Actually it is possibly to generate the single automaton
but unfortunately it can be very large. If we use more one
automata, the summary size of the automata usually is less than the
single one. The automaton name is used in define_cpu_unit and
define_query_cpu_unit. All automata should have unique names. */
DEF_RTL_EXPR(DEFINE_AUTOMATON, "define_automaton", "s", RTX_EXTRA)

/* (automata_option string) describes option for generation of
automata. Currently there are the following options:

o "no-minimization" which makes no minimization of automata. This
is only worth to do when we are debugging the description and
need to look more accurately at reservations of states.

o "time" which means printing additional time statistics about
generation of automata.

o "v" which means generation of file describing the result
automata. The file has suffix `.dfa' and can be used for the
description verification and debugging.

o "w" which means generation of warning instead of error for
non-critical errors.

o "ndfa" which makes nondeterministic finite state automata.

o "progress" which means output of a progress bar showing how many
states were generated so far for automaton being processed. */
DEF_RTL_EXPR(AUTOMATA_OPTION, "automata_option", "s", RTX_EXTRA)

/* (define_reservation string string) names reservation (the first
string) of cpu functional units (the 2nd string). Sometimes unit
reservations for different insns contain common parts. In such
case, you can describe common part and use its name (the 1st
parameter) in regular expression in define_insn_reservation. All
define_reservations, define_cpu_units, and define_query_cpu_units
should have unique names which may not be "nothing". */
DEF_RTL_EXPR(DEFINE_RESERVATION, "define_reservation", "ss", RTX_EXTRA)

/* (define_insn_reservation name default_latency condition regexpr)
describes reservation of cpu functional units (the 3nd operand) for
instruction which is selected by the condition (the 2nd parameter).
The first parameter is used for output of debugging information.
The reservations are described by a regular expression according
the following syntax:

regexp = regexp "," oneof
| oneof

oneof = oneof "|" allof
| allof

allof = allof "+" repeat
| repeat

repeat = element "*" number
| element

element = cpu_function_unit_name
| reservation_name
| result_name
| "nothing"
| "(" regexp ")"

1. "," is used for describing start of the next cycle in
reservation.

2. "|" is used for describing the reservation described by the
first regular expression *or* the reservation described by the
second regular expression *or* etc.

3. "+" is used for describing the reservation described by the
first regular expression *and* the reservation described by the
second regular expression *and* etc.

4. "*" is used for convenience and simply means sequence in
which the regular expression are repeated NUMBER times with
cycle advancing (see ",").

5. cpu functional unit name which means its reservation.

6. reservation name -- see define_reservation.

7. string "nothing" means no units reservation. */

DEF_RTL_EXPR(DEFINE_INSN_RESERVATION, "define_insn_reservation", "sies", RTX_EXTRA)

/* Expressions used for insn attributes. */

/* Definition of an insn attribute.
1st operand: name of the attribute
2nd operand: comma-separated list of possible attribute values
3rd operand: expression for the default value of the attribute. */
DEF_RTL_EXPR(DEFINE_ATTR, "define_attr", "sse", RTX_EXTRA)

/* Marker for the name of an attribute. */
DEF_RTL_EXPR(ATTR, "attr", "s", RTX_EXTRA)

/* For use in the last (optional) operand of DEFINE_INSN or DEFINE_PEEPHOLE and
in DEFINE_ASM_INSN to specify an attribute to assign to insns matching that
pattern.

(set_attr "name" "value") is equivalent to
(set (attr "name") (const_string "value")) */
DEF_RTL_EXPR(SET_ATTR, "set_attr", "ss", RTX_EXTRA)

/* In the last operand of DEFINE_INSN and DEFINE_PEEPHOLE, this can be used to
specify that attribute values are to be assigned according to the
alternative matched.

The following three expressions are equivalent:

(set (attr "att") (cond [(eq_attrq "alternative" "1") (const_string "a1")
(eq_attrq "alternative" "2") (const_string "a2")]
(const_string "a3")))
(set_attr_alternative "att" [(const_string "a1") (const_string "a2")
(const_string "a3")])
(set_attr "att" "a1,a2,a3")
*/
DEF_RTL_EXPR(SET_ATTR_ALTERNATIVE, "set_attr_alternative", "sE", RTX_EXTRA)

/* A conditional expression true if the value of the specified attribute of
the current insn equals the specified value. The first operand is the
attribute name and the second is the comparison value. */
DEF_RTL_EXPR(EQ_ATTR, "eq_attr", "ss", RTX_EXTRA)

/* A special case of the above representing a set of alternatives. The first
operand is bitmap of the set, the second one is the default value. */
DEF_RTL_EXPR(EQ_ATTR_ALT, "eq_attr_alt", "ii", RTX_EXTRA)

/* A conditional expression which is true if the specified flag is
true for the insn being scheduled in reorg.

genattr.c defines the following flags which can be tested by
(attr_flag "foo") expressions in eligible_for_delay.

forward, backward, very_likely, likely, very_unlikely, and unlikely. */

DEF_RTL_EXPR (ATTR_FLAG, "attr_flag", "s", RTX_EXTRA)

/* General conditional. The first operand is a vector composed of pairs of
expressions. The first element of each pair is evaluated, in turn.
The value of the conditional is the second expression of the first pair
whose first expression evaluates nonzero. If none of the expressions is
true, the second operand will be used as the value of the conditional. */
DEF_RTL_EXPR(COND, "cond", "Ee", RTX_EXTRA)

#endif /* GENERATOR_FILE */

/*
Local variables:
mode:c
End:
*/

RTL 式の中では "const_int" のような小文字だけの文字列。
C 式の中では CONST_INT のような大文字の定数が使われる。

RTX

rtx ってのは RTL eXpression の略らしい。

http://gcc.gnu.org/onlinedocs/gccint/RTL-Objects.html#RTL-Objects