Date: Sat, 30 Apr 2011 22:07:04 +0000 (UTC) From: Martin Matuska <mm@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-8@freebsd.org Subject: svn commit: r221274 - in stable/8: contrib/gcc contrib/gcc/config/i386 contrib/gcc/config/mips contrib/gcc/config/rs6000 contrib/gcc/config/s390 contrib/gcc/cp contrib/gcc/doc contrib/libstdc++ con... Message-ID: <201104302207.p3UM74JH064153@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: mm Date: Sat Apr 30 22:07:04 2011 New Revision: 221274 URL: http://svn.freebsd.org/changeset/base/221274 Log: MFC 219374, 219376, 219639, 219640, 219697, 219711, 220150: MFC r219374: Backport Intel Core 2 and AMD Geode CPU types from gcc-4.3 (GPLv2) These options are supported in this shape in all newer GCC versions. Source: gcc-4_3-branch (rev. 118090, 118973, 120846; GPLv2) MFC r219376: Add AMD Geode CPU type to bsd.cpu.mk and examples/etc/make.conf For CPUTYPE=core2 use -march=core2 MFC r219639: Backport SSSE3 instruction set support to base gcc. Enabled by default for -march=core2 Source: gcc-4_3-branch (rev. 117958, 121687, 121726, 123639; GPLv2) MFC r219640: Add ssse3 capability for CPUTYPE=core2 to MACHINE_CPU in bsd.cpu.mk MFC r219697: Fix -march/-mtune=native autodetection for Intel Core 2 CPUs Source: gcc-4_3-branch (partial rev. 119454; GPLv2) MFC r219711: Backport missing tunings for -march=core2: - enable extra 80387 mathematical constants (ext_80387_constants) - enable compare and exchange 16 bytes (cmpxchg16b) Verified against llvm-gcc (and apple gcc) Source: gcc-4_3-branch (ref. svn revs. 119260, 121140; GPLv2) MFC r220150: Upgrade of base gcc and libstdc++ to the last GPLv2-licensed revision (rev. 127959 of gcc-4_2-branch). Resolved GCC bugs: c++: 17763, 29365, 30535, 30917, 31337, 31941, 32108, 32112, 32346, 32898, 32992 debug: 32610, 32914 libstdc++: 33084, 33128 middle-end: 32563 rtl-optimization: 33148 tree-optimization: 25413, 32723 target: 32218 Source: gcc-4_2-branch (up to rev. 127959) Obtained from: gcc (var. revs of gcc-4_2-branch and gcc-4_3-branch; GPLv2) PR: gnu/153298, gnu/153959, gnu/154385, gnu/155308 Added: stable/8/contrib/gcc/config/i386/geode.md - copied unchanged from r219374, head/contrib/gcc/config/i386/geode.md stable/8/contrib/gcc/config/i386/tmmintrin.h - copied unchanged from r219639, head/contrib/gcc/config/i386/tmmintrin.h Modified: stable/8/contrib/gcc/BASE-VER stable/8/contrib/gcc/ChangeLog stable/8/contrib/gcc/DATESTAMP stable/8/contrib/gcc/DEV-PHASE stable/8/contrib/gcc/config.gcc stable/8/contrib/gcc/config/i386/driver-i386.c stable/8/contrib/gcc/config/i386/i386.c stable/8/contrib/gcc/config/i386/i386.h stable/8/contrib/gcc/config/i386/i386.md stable/8/contrib/gcc/config/i386/i386.opt stable/8/contrib/gcc/config/i386/sse.md stable/8/contrib/gcc/config/i386/xmmintrin.h stable/8/contrib/gcc/config/mips/predicates.md stable/8/contrib/gcc/config/rs6000/rs6000.c stable/8/contrib/gcc/config/s390/s390.md stable/8/contrib/gcc/cp/ChangeLog stable/8/contrib/gcc/cp/call.c stable/8/contrib/gcc/cp/cp-tree.h stable/8/contrib/gcc/cp/cxx-pretty-print.c stable/8/contrib/gcc/cp/decl.c stable/8/contrib/gcc/cp/decl2.c stable/8/contrib/gcc/cp/error.c stable/8/contrib/gcc/cp/lex.c stable/8/contrib/gcc/cp/name-lookup.c stable/8/contrib/gcc/cp/pt.c stable/8/contrib/gcc/cp/semantics.c stable/8/contrib/gcc/cp/typeck.c stable/8/contrib/gcc/doc/contrib.texi stable/8/contrib/gcc/doc/extend.texi stable/8/contrib/gcc/doc/gcc.1 stable/8/contrib/gcc/doc/invoke.texi stable/8/contrib/gcc/dwarf2out.c stable/8/contrib/gcc/expr.c stable/8/contrib/gcc/fold-const.c stable/8/contrib/gcc/gimplify.c stable/8/contrib/gcc/reload1.c stable/8/contrib/gcc/simplify-rtx.c stable/8/contrib/gcc/target-def.h stable/8/contrib/gcc/target.h stable/8/contrib/gcc/targhooks.c stable/8/contrib/gcc/targhooks.h stable/8/contrib/gcc/tree-if-conv.c stable/8/contrib/gcc/tree-ssa-structalias.c stable/8/contrib/gcc/tree-vect-analyze.c stable/8/contrib/gcc/tree-vect-patterns.c stable/8/contrib/gcc/tree.c stable/8/contrib/gcc/tree.h stable/8/contrib/libstdc++/ChangeLog stable/8/contrib/libstdc++/include/std/std_valarray.h stable/8/contrib/libstdc++/include/tr1/random stable/8/share/examples/etc/make.conf stable/8/share/mk/bsd.cpu.mk Directory Properties: stable/8/contrib/gcc/ (props changed) stable/8/contrib/libstdc++/ (props changed) stable/8/share/examples/ (props changed) stable/8/share/examples/cvsup/ (props changed) stable/8/share/mk/ (props changed) Modified: stable/8/contrib/gcc/BASE-VER ============================================================================== --- stable/8/contrib/gcc/BASE-VER Sat Apr 30 21:55:08 2011 (r221273) +++ stable/8/contrib/gcc/BASE-VER Sat Apr 30 22:07:04 2011 (r221274) @@ -1 +1 @@ -4.2.1 +4.2.2 Modified: stable/8/contrib/gcc/ChangeLog ============================================================================== --- stable/8/contrib/gcc/ChangeLog Sat Apr 30 21:55:08 2011 (r221273) +++ stable/8/contrib/gcc/ChangeLog Sat Apr 30 22:07:04 2011 (r221274) @@ -1,3 +1,121 @@ +2007-08-31 Jakub Jelinek <jakub@redhat.com> + + PR rtl-optimization/33148 + * simplify-rtx.c (simplify_unary_operation_1): Only optimize + (neg (lt X 0)) if X has scalar int mode. + + PR debug/32914 + * dwarf2out.c (rtl_for_decl_init): If vector decl has CONSTRUCTOR + initializer, use build_vector_from_ctor if possible to create + VECTOR_CST out of it. If vector initializer is not VECTOR_CST + even after this, return NULL. + +2007-08-27 Jason Merrill <jason@redhat.com> + + PR c++/31337 + * gimplify.c (gimplify_modify_expr): Discard the assignment of + zero-sized types after calling gimplify_modify_expr_rhs. + +2007-08-24 Jakub Jelinek <jakub@redhat.com> + + PR debug/32610 + * dwarf2out.c (gen_decl_die): Don't call + gen_tagged_type_instantiation_die if decl doesn't have tagged type. + +2007-08-24 Richard Guenther <rguenther@suse.de> + + * expr.c (get_inner_reference): Remove unused variable. + +2007-08-24 Richard Guenther <rguenther@suse.de> + + * expr.c (get_inner_reference): Do computation of bitoffset + from offset in a way we can detect overflow reliably. + +2007-08-22 Richard Guenther <rguenther@suse.de> + + PR middle-end/32563 + * tree.c (host_integerp): Treat sizetype as signed as it is + sign-extended. + +2007-08-20 Adam Nemet <anemet@caviumnetworks.com> + + * config/mips/predicates.md (const_call_insn_operand): Invoke + SYMBOL_REF_LONG_CALL_P only on SYMBOL_REFs. + +2007-08-17 Chen liqin <liqin@sunnorth.com.cn> + + * config/score/score.md : Update pattern tablejump. + * config/score/score.c : Update score_initialize_trampoline + function. + * config/score/score.h (TRAMPOLINE_TEMPLATE): Added macro. + (TRAMPOLINE_INSNS, TRAMPOLINE_SIZE) Update macro. + * doc/contrib.texi: Add my entry. + +2007-08-02 Andreas Krebbel <krebbel1@de.ibm.com> + + * config/s390/s390.md ("*xordi3_cconly"): Change xr to xg. + +2007-08-01 Andreas Krebbel <krebbel1@de.ibm.com> + + * config/s390/s390.md (TF in GPR splitter): Change operand_subword + parameter to TFmode. + +2007-07-30 Mark Mitchell <mark@codesourcery.com> + + * BASE-VER: Bump. + * DEV-PHASE: Mark as prerelease. + +2007-07-25 Steve Ellcey <sje@cup.hp.com> + + PR target/32218 + * tree-vect-patterns.c (vect_pattern_recog_1): Check for valid type. + +2007-07-25 Dorit Nuzman <dorit@il.ibm.com> + Devang Patel <dpatel@apple.com> + + PR tree-optimization/25413 + * targhooks.c (default_builtin_vector_alignment_reachable): New. + * targhooks.h (default_builtin_vector_alignment_reachable): New. + * tree.h (contains_packed_reference): New. + * expr.c (contains_packed_reference): New. + * tree-vect-analyze.c (vector_alignment_reachable_p): New. + (vect_enhance_data_refs_alignment): Call + vector_alignment_reachable_p. + * target.h (vector_alignment_reachable): New builtin. + * target-def.h (TARGET_VECTOR_ALIGNMENT_REACHABLE): New. + * config/rs6000/rs6000.c (rs6000_vector_alignment_reachable): New. + (TARGET_VECTOR_ALIGNMENT_REACHABLE): Define. + +2007-07-24 Richard Guenther <rguenther@suse.de> + + Backport from mainline: + 2007-07-16 Richard Guenther <rguenther@suse.de> + Uros Bizjak <ubizjak@gmail.com> + + * tree-if-conv.c (find_phi_replacement_condition): Unshare "*cond" + before forcing it to gimple operand. + +2007-07-24 Richard Guenther <rguenther@suse.de> + + PR tree-optimization/32723 + Backport from mainline: + 2007-03-09 Daniel Berlin <dberlin@dberlin.org> + + * tree-ssa-structalias.c (shared_bitmap_info_t): New structure. + (shared_bitmap_table): New variable. + (shared_bitmap_hash): New function. + (shared_bitmap_eq): Ditto + (shared_bitmap_lookup): Ditto. + (shared_bitmap_add): Ditto. + (find_what_p_points_to): Rewrite to use shared bitmap hashtable. + (init_alias_vars): Init shared bitmap hashtable. + (delete_points_to_sets): Delete shared bitmap hashtable. + +2007-07-23 Bernd Schmidt <bernd.schmidt@analog.com> + + * reload1.c (choose_reload_regs): Set reload_spill_index for regs + chosen during find_reloads. + 2007-07-19 Release Manager * GCC 4.2.1 released. Modified: stable/8/contrib/gcc/DATESTAMP ============================================================================== --- stable/8/contrib/gcc/DATESTAMP Sat Apr 30 21:55:08 2011 (r221273) +++ stable/8/contrib/gcc/DATESTAMP Sat Apr 30 22:07:04 2011 (r221274) @@ -1 +1 @@ -20070719 +20070831 Modified: stable/8/contrib/gcc/DEV-PHASE ============================================================================== --- stable/8/contrib/gcc/DEV-PHASE Sat Apr 30 21:55:08 2011 (r221273) +++ stable/8/contrib/gcc/DEV-PHASE Sat Apr 30 22:07:04 2011 (r221274) @@ -0,0 +1 @@ +prerelease Modified: stable/8/contrib/gcc/config.gcc ============================================================================== --- stable/8/contrib/gcc/config.gcc Sat Apr 30 21:55:08 2011 (r221273) +++ stable/8/contrib/gcc/config.gcc Sat Apr 30 22:07:04 2011 (r221274) @@ -268,11 +268,13 @@ xscale-*-*) ;; i[34567]86-*-*) cpu_type=i386 - extra_headers="mmintrin.h mm3dnow.h xmmintrin.h emmintrin.h pmmintrin.h" + extra_headers="mmintrin.h mm3dnow.h xmmintrin.h emmintrin.h + pmmintrin.h tmmintrin.h" ;; x86_64-*-*) cpu_type=i386 - extra_headers="mmintrin.h mm3dnow.h xmmintrin.h emmintrin.h pmmintrin.h" + extra_headers="mmintrin.h mm3dnow.h xmmintrin.h emmintrin.h + pmmintrin.h tmmintrin.h" need_64bit_hwint=yes ;; ia64-*-*) @@ -1207,14 +1209,14 @@ i[34567]86-*-solaris2*) # FIXME: -m64 for i[34567]86-*-* should be allowed just # like -m32 for x86_64-*-*. case X"${with_cpu}" in - Xgeneric|Xnocona|Xx86-64|Xk8|Xopteron|Xathlon64|Xathlon-fx) + Xgeneric|Xcore2|Xnocona|Xx86-64|Xk8|Xopteron|Xathlon64|Xathlon-fx) ;; X) with_cpu=generic ;; *) echo "Unsupported CPU used in --with-cpu=$with_cpu, supported values:" 1>&2 - echo "generic nocona x86-64 k8 opteron athlon64 athlon-fx" 1>&2 + echo "generic core2 nocona x86-64 k8 opteron athlon64 athlon-fx" 1>&2 exit 1 ;; esac @@ -2537,6 +2539,9 @@ if test x$with_cpu = x ; then nocona-*) with_cpu=nocona ;; + core2-*) + with_cpu=core2 + ;; pentium_m-*) with_cpu=pentium-m ;; @@ -2556,6 +2561,9 @@ if test x$with_cpu = x ; then nocona-*) with_cpu=nocona ;; + core2-*) + with_cpu=core2 + ;; *) with_cpu=generic ;; @@ -2787,7 +2795,7 @@ case "${target}" in esac # OK ;; - "" | k8 | opteron | athlon64 | athlon-fx | nocona | generic) + "" | k8 | opteron | athlon64 | athlon-fx | nocona | core2 | generic) # OK ;; *) Modified: stable/8/contrib/gcc/config/i386/driver-i386.c ============================================================================== --- stable/8/contrib/gcc/config/i386/driver-i386.c Sat Apr 30 21:55:08 2011 (r221273) +++ stable/8/contrib/gcc/config/i386/driver-i386.c Sat Apr 30 22:07:04 2011 (r221274) @@ -39,6 +39,7 @@ const char *host_detect_local_cpu (int a #define bit_SSE2 (1 << 26) #define bit_SSE3 (1 << 0) +#define bit_SSSE3 (1 << 9) #define bit_CMPXCHG16B (1 << 13) #define bit_3DNOW (1 << 31) @@ -66,7 +67,7 @@ const char *host_detect_local_cpu (int a unsigned int vendor; unsigned int ext_level; unsigned char has_mmx = 0, has_3dnow = 0, has_3dnowp = 0, has_sse = 0; - unsigned char has_sse2 = 0, has_sse3 = 0, has_cmov = 0; + unsigned char has_sse2 = 0, has_sse3 = 0, has_ssse3 = 0, has_cmov = 0; unsigned char has_longmode = 0, has_cmpxchg8b = 0; unsigned char is_amd = 0; unsigned int family = 0; @@ -107,6 +108,7 @@ const char *host_detect_local_cpu (int a has_sse = !!(edx & bit_SSE); has_sse2 = !!(edx & bit_SSE2); has_sse3 = !!(ecx & bit_SSE3); + has_ssse3 = !!(ecx & bit_SSSE3); /* We don't care for extended family. */ family = (eax >> 8) & ~(1 << 4); @@ -148,7 +150,9 @@ const char *host_detect_local_cpu (int a /* We have no idea. Use something reasonable. */ if (arch) { - if (has_sse3) + if (has_ssse3) + cpu = "core2"; + else if (has_sse3) { if (has_longmode) cpu = "nocona"; @@ -230,6 +234,9 @@ const char *host_detect_local_cpu (int a cpu = "generic"; } break; + case PROCESSOR_GEODE: + cpu = "geode"; + break; case PROCESSOR_K6: if (has_3dnow) cpu = "k6-3"; Copied: stable/8/contrib/gcc/config/i386/geode.md (from r219374, head/contrib/gcc/config/i386/geode.md) ============================================================================== --- /dev/null 00:00:00 1970 (empty, because file is newly added) +++ stable/8/contrib/gcc/config/i386/geode.md Sat Apr 30 22:07:04 2011 (r221274, copy of r219374, head/contrib/gcc/config/i386/geode.md) @@ -0,0 +1,153 @@ +;; Geode Scheduling +;; Copyright (C) 2006 +;; 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 2, 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 COPYING. If not, write to +;; the Free Software Foundation, 51 Franklin Street, Fifth Floor, +;; Boston, MA 02110-1301, USA. +;; +;; The Geode architecture is one insn issue processor. +;; +;; This description is based on data from the following documents: +;; +;; "AMD Geode GX Processor Data Book" +;; Advanced Micro Devices, Inc., Aug 2005. +;; +;; "AMD Geode LX Processor Data Book" +;; Advanced Micro Devices, Inc., Jan 2006. +;; +;; +;; CPU execution units of the Geode: +;; +;; issue describes the issue pipeline. +;; alu describes the Integer unit +;; fpu describes the FP unit +;; +;; The fp unit is out of order execution unit with register renaming. +;; There is also memory management unit and execution pipeline for +;; load/store operations. We ignore it and difference between insns +;; using memory and registers. + +(define_automaton "geode") + +(define_cpu_unit "geode_issue,geode_alu,geode_fpu" "geode") + +(define_insn_reservation "alu" 1 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "alu,alu1,negnot,icmp,lea,test,imov,imovx,icmov,incdec,setcc")) + "geode_issue,geode_alu") + +(define_insn_reservation "shift" 2 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "ishift,ishift1,rotate,rotate1,cld")) + "geode_issue,geode_alu*2") + +(define_insn_reservation "imul" 7 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "imul")) + "geode_issue,geode_alu*7") + +(define_insn_reservation "idiv" 40 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "idiv")) + "geode_issue,geode_alu*40") + +;; The branch unit. +(define_insn_reservation "call" 2 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "call,callv")) + "geode_issue,geode_alu*2") + +(define_insn_reservation "geode_branch" 1 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "ibr")) + "geode_issue,geode_alu") + +(define_insn_reservation "geode_pop_push" 1 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "pop,push")) + "geode_issue,geode_alu") + +(define_insn_reservation "geode_leave" 2 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "leave")) + "geode_issue,geode_alu*2") + +(define_insn_reservation "geode_load_str" 4 + (and (eq_attr "cpu" "geode") + (and (eq_attr "type" "str") + (eq_attr "memory" "load,both"))) + "geode_issue,geode_alu*4") + +(define_insn_reservation "geode_store_str" 2 + (and (eq_attr "cpu" "geode") + (and (eq_attr "type" "str") + (eq_attr "memory" "store"))) + "geode_issue,geode_alu*2") + +;; Be optimistic +(define_insn_reservation "geode_unknown" 1 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "multi,other")) + "geode_issue,geode_alu") + +;; FPU + +(define_insn_reservation "geode_fop" 6 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "fop,fcmp")) + "geode_issue,geode_fpu*6") + +(define_insn_reservation "geode_fsimple" 1 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "fmov,fcmov,fsgn,fxch")) + "geode_issue,geode_fpu") + +(define_insn_reservation "geode_fist" 4 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "fistp,fisttp")) + "geode_issue,geode_fpu*4") + +(define_insn_reservation "geode_fmul" 10 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "fmul")) + "geode_issue,geode_fpu*10") + +(define_insn_reservation "geode_fdiv" 47 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "fdiv")) + "geode_issue,geode_fpu*47") + +;; We use minimal latency (fsin) here +(define_insn_reservation "geode_fpspc" 54 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "fpspc")) + "geode_issue,geode_fpu*54") + +(define_insn_reservation "geode_frndint" 12 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "frndint")) + "geode_issue,geode_fpu*12") + +(define_insn_reservation "geode_mmxmov" 1 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "mmxmov")) + "geode_issue,geode_fpu") + +(define_insn_reservation "geode_mmx" 2 + (and (eq_attr "cpu" "geode") + (eq_attr "type" "mmx,mmxadd,mmxmul,mmxcmp,mmxcvt,mmxshft")) + "geode_issue,geode_fpu*2") Modified: stable/8/contrib/gcc/config/i386/i386.c ============================================================================== --- stable/8/contrib/gcc/config/i386/i386.c Sat Apr 30 21:55:08 2011 (r221273) +++ stable/8/contrib/gcc/config/i386/i386.c Sat Apr 30 22:07:04 2011 (r221274) @@ -336,6 +336,60 @@ struct processor_costs pentiumpro_cost = }; static const +struct processor_costs geode_cost = { + COSTS_N_INSNS (1), /* cost of an add instruction */ + COSTS_N_INSNS (1), /* cost of a lea instruction */ + COSTS_N_INSNS (2), /* variable shift costs */ + COSTS_N_INSNS (1), /* constant shift costs */ + {COSTS_N_INSNS (3), /* cost of starting multiply for QI */ + COSTS_N_INSNS (4), /* HI */ + COSTS_N_INSNS (7), /* SI */ + COSTS_N_INSNS (7), /* DI */ + COSTS_N_INSNS (7)}, /* other */ + 0, /* cost of multiply per each bit set */ + {COSTS_N_INSNS (15), /* cost of a divide/mod for QI */ + COSTS_N_INSNS (23), /* HI */ + COSTS_N_INSNS (39), /* SI */ + COSTS_N_INSNS (39), /* DI */ + COSTS_N_INSNS (39)}, /* other */ + COSTS_N_INSNS (1), /* cost of movsx */ + COSTS_N_INSNS (1), /* cost of movzx */ + 8, /* "large" insn */ + 4, /* MOVE_RATIO */ + 1, /* cost for loading QImode using movzbl */ + {1, 1, 1}, /* cost of loading integer registers + in QImode, HImode and SImode. + Relative to reg-reg move (2). */ + {1, 1, 1}, /* cost of storing integer registers */ + 1, /* cost of reg,reg fld/fst */ + {1, 1, 1}, /* cost of loading fp registers + in SFmode, DFmode and XFmode */ + {4, 6, 6}, /* cost of storing fp registers + in SFmode, DFmode and XFmode */ + + 1, /* cost of moving MMX register */ + {1, 1}, /* cost of loading MMX registers + in SImode and DImode */ + {1, 1}, /* cost of storing MMX registers + in SImode and DImode */ + 1, /* cost of moving SSE register */ + {1, 1, 1}, /* cost of loading SSE registers + in SImode, DImode and TImode */ + {1, 1, 1}, /* cost of storing SSE registers + in SImode, DImode and TImode */ + 1, /* MMX or SSE register to integer */ + 32, /* size of prefetch block */ + 1, /* number of parallel prefetches */ + 1, /* Branch cost */ + COSTS_N_INSNS (6), /* cost of FADD and FSUB insns. */ + COSTS_N_INSNS (11), /* cost of FMUL instruction. */ + COSTS_N_INSNS (47), /* cost of FDIV instruction. */ + COSTS_N_INSNS (1), /* cost of FABS instruction. */ + COSTS_N_INSNS (1), /* cost of FCHS instruction. */ + COSTS_N_INSNS (54), /* cost of FSQRT instruction. */ +}; + +static const struct processor_costs k6_cost = { COSTS_N_INSNS (1), /* cost of an add instruction */ COSTS_N_INSNS (2), /* cost of a lea instruction */ @@ -600,6 +654,58 @@ struct processor_costs nocona_cost = { COSTS_N_INSNS (44), /* cost of FSQRT instruction. */ }; +static const +struct processor_costs core2_cost = { + COSTS_N_INSNS (1), /* cost of an add instruction */ + COSTS_N_INSNS (1) + 1, /* cost of a lea instruction */ + COSTS_N_INSNS (1), /* variable shift costs */ + COSTS_N_INSNS (1), /* constant shift costs */ + {COSTS_N_INSNS (3), /* cost of starting multiply for QI */ + COSTS_N_INSNS (3), /* HI */ + COSTS_N_INSNS (3), /* SI */ + COSTS_N_INSNS (3), /* DI */ + COSTS_N_INSNS (3)}, /* other */ + 0, /* cost of multiply per each bit set */ + {COSTS_N_INSNS (22), /* cost of a divide/mod for QI */ + COSTS_N_INSNS (22), /* HI */ + COSTS_N_INSNS (22), /* SI */ + COSTS_N_INSNS (22), /* DI */ + COSTS_N_INSNS (22)}, /* other */ + COSTS_N_INSNS (1), /* cost of movsx */ + COSTS_N_INSNS (1), /* cost of movzx */ + 8, /* "large" insn */ + 16, /* MOVE_RATIO */ + 2, /* cost for loading QImode using movzbl */ + {6, 6, 6}, /* cost of loading integer registers + in QImode, HImode and SImode. + Relative to reg-reg move (2). */ + {4, 4, 4}, /* cost of storing integer registers */ + 2, /* cost of reg,reg fld/fst */ + {6, 6, 6}, /* cost of loading fp registers + in SFmode, DFmode and XFmode */ + {4, 4, 4}, /* cost of loading integer registers */ + 2, /* cost of moving MMX register */ + {6, 6}, /* cost of loading MMX registers + in SImode and DImode */ + {4, 4}, /* cost of storing MMX registers + in SImode and DImode */ + 2, /* cost of moving SSE register */ + {6, 6, 6}, /* cost of loading SSE registers + in SImode, DImode and TImode */ + {4, 4, 4}, /* cost of storing SSE registers + in SImode, DImode and TImode */ + 2, /* MMX or SSE register to integer */ + 128, /* size of prefetch block */ + 8, /* number of parallel prefetches */ + 3, /* Branch cost */ + COSTS_N_INSNS (3), /* cost of FADD and FSUB insns. */ + COSTS_N_INSNS (5), /* cost of FMUL instruction. */ + COSTS_N_INSNS (32), /* cost of FDIV instruction. */ + COSTS_N_INSNS (1), /* cost of FABS instruction. */ + COSTS_N_INSNS (1), /* cost of FCHS instruction. */ + COSTS_N_INSNS (58), /* cost of FSQRT instruction. */ +}; + /* Generic64 should produce code tuned for Nocona and K8. */ static const struct processor_costs generic64_cost = { @@ -721,38 +827,41 @@ const struct processor_costs *ix86_cost #define m_486 (1<<PROCESSOR_I486) #define m_PENT (1<<PROCESSOR_PENTIUM) #define m_PPRO (1<<PROCESSOR_PENTIUMPRO) +#define m_GEODE (1<<PROCESSOR_GEODE) +#define m_K6_GEODE (m_K6 | m_GEODE) #define m_K6 (1<<PROCESSOR_K6) #define m_ATHLON (1<<PROCESSOR_ATHLON) #define m_PENT4 (1<<PROCESSOR_PENTIUM4) #define m_K8 (1<<PROCESSOR_K8) #define m_ATHLON_K8 (m_K8 | m_ATHLON) #define m_NOCONA (1<<PROCESSOR_NOCONA) +#define m_CORE2 (1<<PROCESSOR_CORE2) #define m_GENERIC32 (1<<PROCESSOR_GENERIC32) #define m_GENERIC64 (1<<PROCESSOR_GENERIC64) #define m_GENERIC (m_GENERIC32 | m_GENERIC64) /* Generic instruction choice should be common subset of supported CPUs - (PPro/PENT4/NOCONA/Athlon/K8). */ + (PPro/PENT4/NOCONA/CORE2/Athlon/K8). */ /* Leave is not affecting Nocona SPEC2000 results negatively, so enabling for Generic64 seems like good code size tradeoff. We can't enable it for 32bit generic because it is not working well with PPro base chips. */ -const int x86_use_leave = m_386 | m_K6 | m_ATHLON_K8 | m_GENERIC64; -const int x86_push_memory = m_386 | m_K6 | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_GENERIC; +const int x86_use_leave = m_386 | m_K6_GEODE | m_ATHLON_K8 | m_CORE2 | m_GENERIC64; +const int x86_push_memory = m_386 | m_K6_GEODE | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; const int x86_zero_extend_with_and = m_486 | m_PENT; -const int x86_movx = m_ATHLON_K8 | m_PPRO | m_PENT4 | m_NOCONA | m_GENERIC /* m_386 | m_K6 */; +const int x86_movx = m_ATHLON_K8 | m_PPRO | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC | m_GEODE /* m_386 | m_K6 */; const int x86_double_with_add = ~m_386; const int x86_use_bit_test = m_386; -const int x86_unroll_strlen = m_486 | m_PENT | m_PPRO | m_ATHLON_K8 | m_K6 | m_GENERIC; -const int x86_cmove = m_PPRO | m_ATHLON_K8 | m_PENT4 | m_NOCONA; +const int x86_unroll_strlen = m_486 | m_PENT | m_PPRO | m_ATHLON_K8 | m_K6 | m_CORE2 | m_GENERIC; +const int x86_cmove = m_PPRO | m_GEODE | m_ATHLON_K8 | m_PENT4 | m_NOCONA; const int x86_3dnow_a = m_ATHLON_K8; -const int x86_deep_branch = m_PPRO | m_K6 | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_GENERIC; +const int x86_deep_branch = m_PPRO | m_K6_GEODE | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; /* Branch hints were put in P4 based on simulation result. But after P4 was made, no performance benefit was observed with branch hints. It also increases the code size. As the result, icc never generates branch hints. */ const int x86_branch_hints = 0; -const int x86_use_sahf = m_PPRO | m_K6 | m_PENT4 | m_NOCONA | m_GENERIC32; /*m_GENERIC | m_ATHLON_K8 ? */ +const int x86_use_sahf = m_PPRO | m_K6_GEODE | m_PENT4 | m_NOCONA | m_GENERIC32; /*m_GENERIC | m_ATHLON_K8 ? */ /* We probably ought to watch for partial register stalls on Generic32 compilation setting as well. However in current implementation the partial register stalls are not eliminated very well - they can @@ -762,15 +871,15 @@ const int x86_use_sahf = m_PPRO | m_K6 | with partial reg. dependencies used by Athlon/P4 based chips, it is better to leave it off for generic32 for now. */ const int x86_partial_reg_stall = m_PPRO; -const int x86_partial_flag_reg_stall = m_GENERIC; -const int x86_use_himode_fiop = m_386 | m_486 | m_K6; -const int x86_use_simode_fiop = ~(m_PPRO | m_ATHLON_K8 | m_PENT | m_GENERIC); +const int x86_partial_flag_reg_stall = m_CORE2 | m_GENERIC; +const int x86_use_himode_fiop = m_386 | m_486 | m_K6_GEODE; +const int x86_use_simode_fiop = ~(m_PPRO | m_ATHLON_K8 | m_PENT | m_CORE2 | m_GENERIC); const int x86_use_mov0 = m_K6; -const int x86_use_cltd = ~(m_PENT | m_K6 | m_GENERIC); +const int x86_use_cltd = ~(m_PENT | m_K6 | m_CORE2 | m_GENERIC); const int x86_read_modify_write = ~m_PENT; const int x86_read_modify = ~(m_PENT | m_PPRO); const int x86_split_long_moves = m_PPRO; -const int x86_promote_QImode = m_K6 | m_PENT | m_386 | m_486 | m_ATHLON_K8 | m_GENERIC; /* m_PENT4 ? */ +const int x86_promote_QImode = m_K6_GEODE | m_PENT | m_386 | m_486 | m_ATHLON_K8 | m_CORE2 | m_GENERIC; /* m_PENT4 ? */ const int x86_fast_prefix = ~(m_PENT | m_486 | m_386); const int x86_single_stringop = m_386 | m_PENT4 | m_NOCONA; const int x86_qimode_math = ~(0); @@ -780,18 +889,18 @@ const int x86_promote_qi_regs = 0; if our scheme for avoiding partial stalls was more effective. */ const int x86_himode_math = ~(m_PPRO); const int x86_promote_hi_regs = m_PPRO; -const int x86_sub_esp_4 = m_ATHLON_K8 | m_PPRO | m_PENT4 | m_NOCONA | m_GENERIC; -const int x86_sub_esp_8 = m_ATHLON_K8 | m_PPRO | m_386 | m_486 | m_PENT4 | m_NOCONA | m_GENERIC; -const int x86_add_esp_4 = m_ATHLON_K8 | m_K6 | m_PENT4 | m_NOCONA | m_GENERIC; -const int x86_add_esp_8 = m_ATHLON_K8 | m_PPRO | m_K6 | m_386 | m_486 | m_PENT4 | m_NOCONA | m_GENERIC; -const int x86_integer_DFmode_moves = ~(m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_PPRO | m_GENERIC); -const int x86_partial_reg_dependency = m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_GENERIC; -const int x86_memory_mismatch_stall = m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_GENERIC; -const int x86_accumulate_outgoing_args = m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_PPRO | m_GENERIC; -const int x86_prologue_using_move = m_ATHLON_K8 | m_PPRO | m_GENERIC; -const int x86_epilogue_using_move = m_ATHLON_K8 | m_PPRO | m_GENERIC; +const int x86_sub_esp_4 = m_ATHLON_K8 | m_PPRO | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; +const int x86_sub_esp_8 = m_ATHLON_K8 | m_PPRO | m_386 | m_486 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; +const int x86_add_esp_4 = m_ATHLON_K8 | m_K6_GEODE | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; +const int x86_add_esp_8 = m_ATHLON_K8 | m_PPRO | m_K6_GEODE | m_386 | m_486 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; +const int x86_integer_DFmode_moves = ~(m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_PPRO | m_CORE2 | m_GENERIC | m_GEODE); +const int x86_partial_reg_dependency = m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; +const int x86_memory_mismatch_stall = m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; +const int x86_accumulate_outgoing_args = m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_PPRO | m_CORE2 | m_GENERIC; +const int x86_prologue_using_move = m_ATHLON_K8 | m_PPRO | m_CORE2 | m_GENERIC; +const int x86_epilogue_using_move = m_ATHLON_K8 | m_PPRO | m_CORE2 | m_GENERIC; const int x86_shift1 = ~m_486; -const int x86_arch_always_fancy_math_387 = m_PENT | m_PPRO | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_GENERIC; +const int x86_arch_always_fancy_math_387 = m_PENT | m_PPRO | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; /* In Generic model we have an conflict here in between PPro/Pentium4 based chips that thread 128bit SSE registers as single units versus K8 based chips that divide SSE registers to two 64bit halves. @@ -801,7 +910,7 @@ const int x86_arch_always_fancy_math_387 this option on P4 brings over 20% SPECfp regression, while enabling it on K8 brings roughly 2.4% regression that can be partly masked by careful scheduling of moves. */ -const int x86_sse_partial_reg_dependency = m_PENT4 | m_NOCONA | m_PPRO | m_GENERIC; +const int x86_sse_partial_reg_dependency = m_PENT4 | m_NOCONA | m_PPRO | m_CORE2 | m_GENERIC; /* Set for machines where the type and dependencies are resolved on SSE register parts instead of whole registers, so we may maintain just lower part of scalar values in proper format leaving the upper part @@ -810,28 +919,28 @@ const int x86_sse_split_regs = m_ATHLON_ const int x86_sse_typeless_stores = m_ATHLON_K8; const int x86_sse_load0_by_pxor = m_PPRO | m_PENT4 | m_NOCONA; const int x86_use_ffreep = m_ATHLON_K8; -const int x86_rep_movl_optimal = m_386 | m_PENT | m_PPRO | m_K6; -const int x86_use_incdec = ~(m_PENT4 | m_NOCONA | m_GENERIC); +const int x86_rep_movl_optimal = m_386 | m_PENT | m_PPRO | m_K6_GEODE | m_CORE2; +const int x86_use_incdec = ~(m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC); /* ??? Allowing interunit moves makes it all too easy for the compiler to put integer data in xmm registers. Which results in pretty abysmal code. */ const int x86_inter_unit_moves = 0 /* ~(m_ATHLON_K8) */; -const int x86_ext_80387_constants = m_K6 | m_ATHLON | m_PENT4 | m_NOCONA | m_PPRO | m_GENERIC32; +const int x86_ext_80387_constants = m_K6_GEODE | m_ATHLON | m_PENT4 | m_NOCONA | m_CORE2 | m_PPRO | m_GENERIC32; /* Some CPU cores are not able to predict more than 4 branch instructions in the 16 byte window. */ -const int x86_four_jump_limit = m_PPRO | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_GENERIC; -const int x86_schedule = m_PPRO | m_ATHLON_K8 | m_K6 | m_PENT | m_GENERIC; +const int x86_four_jump_limit = m_PPRO | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC; +const int x86_schedule = m_PPRO | m_ATHLON_K8 | m_K6_GEODE | m_PENT | m_CORE2 | m_GENERIC; const int x86_use_bt = m_ATHLON_K8; /* Compare and exchange was added for 80486. */ const int x86_cmpxchg = ~m_386; /* Compare and exchange 8 bytes was added for pentium. */ const int x86_cmpxchg8b = ~(m_386 | m_486); /* Compare and exchange 16 bytes was added for nocona. */ -const int x86_cmpxchg16b = m_NOCONA; +const int x86_cmpxchg16b = m_NOCONA | m_CORE2; /* Exchange and add was added for 80486. */ const int x86_xadd = ~m_386; -const int x86_pad_returns = m_ATHLON_K8 | m_GENERIC; +const int x86_pad_returns = m_ATHLON_K8 | m_CORE2 | m_GENERIC; /* In case the average insn count for single function invocation is lower than this constant, emit fast (but longer) prologue and @@ -1402,16 +1511,24 @@ ix86_handle_option (size_t code, const c case OPT_msse: if (!value) { - target_flags &= ~(MASK_SSE2 | MASK_SSE3); - target_flags_explicit |= MASK_SSE2 | MASK_SSE3; + target_flags &= ~(MASK_SSE2 | MASK_SSE3 | MASK_SSSE3); + target_flags_explicit |= MASK_SSE2 | MASK_SSE3 | MASK_SSSE3; } return true; case OPT_msse2: if (!value) { - target_flags &= ~MASK_SSE3; - target_flags_explicit |= MASK_SSE3; + target_flags &= ~(MASK_SSE3 | MASK_SSSE3); + target_flags_explicit |= MASK_SSE3 | MASK_SSSE3; + } + return true; + + case OPT_msse3: + if (!value) + { + target_flags &= ~MASK_SSSE3; + target_flags_explicit |= MASK_SSSE3; } return true; @@ -1455,11 +1572,13 @@ override_options (void) {&i486_cost, 0, 0, 16, 15, 16, 15, 16}, {&pentium_cost, 0, 0, 16, 7, 16, 7, 16}, {&pentiumpro_cost, 0, 0, 16, 15, 16, 7, 16}, + {&geode_cost, 0, 0, 0, 0, 0, 0, 0}, {&k6_cost, 0, 0, 32, 7, 32, 7, 32}, {&athlon_cost, 0, 0, 16, 7, 16, 7, 16}, {&pentium4_cost, 0, 0, 0, 0, 0, 0, 0}, {&k8_cost, 0, 0, 16, 7, 16, 7, 16}, {&nocona_cost, 0, 0, 0, 0, 0, 0, 0}, + {&core2_cost, 0, 0, 16, 7, 16, 7, 16}, {&generic32_cost, 0, 0, 16, 7, 16, 7, 16}, {&generic64_cost, 0, 0, 16, 7, 16, 7, 16} }; @@ -1478,7 +1597,8 @@ override_options (void) PTA_PREFETCH_SSE = 16, PTA_3DNOW = 32, PTA_3DNOW_A = 64, - PTA_64BIT = 128 + PTA_64BIT = 128, + PTA_SSSE3 = 256 } flags; } const processor_alias_table[] = @@ -1506,6 +1626,11 @@ override_options (void) | PTA_MMX | PTA_PREFETCH_SSE}, {"nocona", PROCESSOR_NOCONA, PTA_SSE | PTA_SSE2 | PTA_SSE3 | PTA_64BIT | PTA_MMX | PTA_PREFETCH_SSE}, + {"core2", PROCESSOR_CORE2, PTA_SSE | PTA_SSE2 | PTA_SSE3 | PTA_SSSE3 + | PTA_64BIT | PTA_MMX + | PTA_PREFETCH_SSE}, + {"geode", PROCESSOR_GEODE, PTA_MMX | PTA_PREFETCH_SSE | PTA_3DNOW + | PTA_3DNOW_A}, {"k6", PROCESSOR_K6, PTA_MMX}, {"k6-2", PROCESSOR_K6, PTA_MMX | PTA_3DNOW}, {"k6-3", PROCESSOR_K6, PTA_MMX | PTA_3DNOW}, @@ -1695,6 +1820,9 @@ override_options (void) if (processor_alias_table[i].flags & PTA_SSE3 && !(target_flags_explicit & MASK_SSE3)) target_flags |= MASK_SSE3; + if (processor_alias_table[i].flags & PTA_SSSE3 + && !(target_flags_explicit & MASK_SSSE3)) + target_flags |= MASK_SSSE3; if (processor_alias_table[i].flags & PTA_PREFETCH_SSE) x86_prefetch_sse = true; if (TARGET_64BIT && !(processor_alias_table[i].flags & PTA_64BIT)) @@ -1871,6 +1999,10 @@ override_options (void) if (!TARGET_80387) target_flags |= MASK_NO_FANCY_MATH_387; + /* Turn on SSE3 builtins for -mssse3. */ + if (TARGET_SSSE3) + target_flags |= MASK_SSE3; + /* Turn on SSE2 builtins for -msse3. */ if (TARGET_SSE3) target_flags |= MASK_SSE2; @@ -13706,6 +13838,9 @@ ix86_issue_rate (void) case PROCESSOR_GENERIC64: return 3; + case PROCESSOR_CORE2: + return 4; + default: return 1; } @@ -14574,6 +14709,41 @@ enum ix86_builtins IX86_BUILTIN_MONITOR, IX86_BUILTIN_MWAIT, + /* SSSE3. */ + IX86_BUILTIN_PHADDW, + IX86_BUILTIN_PHADDD, + IX86_BUILTIN_PHADDSW, + IX86_BUILTIN_PHSUBW, + IX86_BUILTIN_PHSUBD, + IX86_BUILTIN_PHSUBSW, + IX86_BUILTIN_PMADDUBSW, + IX86_BUILTIN_PMULHRSW, + IX86_BUILTIN_PSHUFB, + IX86_BUILTIN_PSIGNB, + IX86_BUILTIN_PSIGNW, + IX86_BUILTIN_PSIGND, + IX86_BUILTIN_PALIGNR, + IX86_BUILTIN_PABSB, + IX86_BUILTIN_PABSW, + IX86_BUILTIN_PABSD, + + IX86_BUILTIN_PHADDW128, + IX86_BUILTIN_PHADDD128, + IX86_BUILTIN_PHADDSW128, + IX86_BUILTIN_PHSUBW128, + IX86_BUILTIN_PHSUBD128, + IX86_BUILTIN_PHSUBSW128, + IX86_BUILTIN_PMADDUBSW128, + IX86_BUILTIN_PMULHRSW128, + IX86_BUILTIN_PSHUFB128, + IX86_BUILTIN_PSIGNB128, + IX86_BUILTIN_PSIGNW128, + IX86_BUILTIN_PSIGND128, + IX86_BUILTIN_PALIGNR128, + IX86_BUILTIN_PABSB128, + IX86_BUILTIN_PABSW128, + IX86_BUILTIN_PABSD128, + IX86_BUILTIN_VEC_INIT_V2SI, IX86_BUILTIN_VEC_INIT_V4HI, IX86_BUILTIN_VEC_INIT_V8QI, @@ -14915,7 +15085,33 @@ static const struct builtin_description { MASK_SSE3, CODE_FOR_sse3_haddv4sf3, "__builtin_ia32_haddps", IX86_BUILTIN_HADDPS, 0, 0 }, { MASK_SSE3, CODE_FOR_sse3_haddv2df3, "__builtin_ia32_haddpd", IX86_BUILTIN_HADDPD, 0, 0 }, { MASK_SSE3, CODE_FOR_sse3_hsubv4sf3, "__builtin_ia32_hsubps", IX86_BUILTIN_HSUBPS, 0, 0 }, - { MASK_SSE3, CODE_FOR_sse3_hsubv2df3, "__builtin_ia32_hsubpd", IX86_BUILTIN_HSUBPD, 0, 0 } + { MASK_SSE3, CODE_FOR_sse3_hsubv2df3, "__builtin_ia32_hsubpd", IX86_BUILTIN_HSUBPD, 0, 0 }, + + /* SSSE3 */ + { MASK_SSSE3, CODE_FOR_ssse3_phaddwv8hi3, "__builtin_ia32_phaddw128", IX86_BUILTIN_PHADDW128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phaddwv4hi3, "__builtin_ia32_phaddw", IX86_BUILTIN_PHADDW, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phadddv4si3, "__builtin_ia32_phaddd128", IX86_BUILTIN_PHADDD128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phadddv2si3, "__builtin_ia32_phaddd", IX86_BUILTIN_PHADDD, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phaddswv8hi3, "__builtin_ia32_phaddsw128", IX86_BUILTIN_PHADDSW128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phaddswv4hi3, "__builtin_ia32_phaddsw", IX86_BUILTIN_PHADDSW, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phsubwv8hi3, "__builtin_ia32_phsubw128", IX86_BUILTIN_PHSUBW128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phsubwv4hi3, "__builtin_ia32_phsubw", IX86_BUILTIN_PHSUBW, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phsubdv4si3, "__builtin_ia32_phsubd128", IX86_BUILTIN_PHSUBD128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phsubdv2si3, "__builtin_ia32_phsubd", IX86_BUILTIN_PHSUBD, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phsubswv8hi3, "__builtin_ia32_phsubsw128", IX86_BUILTIN_PHSUBSW128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_phsubswv4hi3, "__builtin_ia32_phsubsw", IX86_BUILTIN_PHSUBSW, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_pmaddubswv8hi3, "__builtin_ia32_pmaddubsw128", IX86_BUILTIN_PMADDUBSW128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_pmaddubswv4hi3, "__builtin_ia32_pmaddubsw", IX86_BUILTIN_PMADDUBSW, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_pmulhrswv8hi3, "__builtin_ia32_pmulhrsw128", IX86_BUILTIN_PMULHRSW128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_pmulhrswv4hi3, "__builtin_ia32_pmulhrsw", IX86_BUILTIN_PMULHRSW, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_pshufbv16qi3, "__builtin_ia32_pshufb128", IX86_BUILTIN_PSHUFB128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_pshufbv8qi3, "__builtin_ia32_pshufb", IX86_BUILTIN_PSHUFB, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_psignv16qi3, "__builtin_ia32_psignb128", IX86_BUILTIN_PSIGNB128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_psignv8qi3, "__builtin_ia32_psignb", IX86_BUILTIN_PSIGNB, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_psignv8hi3, "__builtin_ia32_psignw128", IX86_BUILTIN_PSIGNW128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_psignv4hi3, "__builtin_ia32_psignw", IX86_BUILTIN_PSIGNW, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_psignv4si3, "__builtin_ia32_psignd128", IX86_BUILTIN_PSIGND128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_ssse3_psignv2si3, "__builtin_ia32_psignd", IX86_BUILTIN_PSIGND, 0, 0 } }; static const struct builtin_description bdesc_1arg[] = @@ -14962,6 +15158,14 @@ static const struct builtin_description /* SSE3 */ { MASK_SSE3, CODE_FOR_sse3_movshdup, 0, IX86_BUILTIN_MOVSHDUP, 0, 0 }, { MASK_SSE3, CODE_FOR_sse3_movsldup, 0, IX86_BUILTIN_MOVSLDUP, 0, 0 }, + + /* SSSE3 */ + { MASK_SSSE3, CODE_FOR_absv16qi2, "__builtin_ia32_pabsb128", IX86_BUILTIN_PABSB128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_absv8qi2, "__builtin_ia32_pabsb", IX86_BUILTIN_PABSB, 0, 0 }, + { MASK_SSSE3, CODE_FOR_absv8hi2, "__builtin_ia32_pabsw128", IX86_BUILTIN_PABSW128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_absv4hi2, "__builtin_ia32_pabsw", IX86_BUILTIN_PABSW, 0, 0 }, + { MASK_SSSE3, CODE_FOR_absv4si2, "__builtin_ia32_pabsd128", IX86_BUILTIN_PABSD128, 0, 0 }, + { MASK_SSSE3, CODE_FOR_absv2si2, "__builtin_ia32_pabsd", IX86_BUILTIN_PABSD, 0, 0 }, }; static void @@ -15096,6 +15300,16 @@ ix86_init_mmx_sse_builtins (void) /* Normal vector unops. */ tree v4sf_ftype_v4sf = build_function_type_list (V4SF_type_node, V4SF_type_node, NULL_TREE); + tree v16qi_ftype_v16qi + = build_function_type_list (V16QI_type_node, V16QI_type_node, NULL_TREE); + tree v8hi_ftype_v8hi + = build_function_type_list (V8HI_type_node, V8HI_type_node, NULL_TREE); + tree v4si_ftype_v4si + = build_function_type_list (V4SI_type_node, V4SI_type_node, NULL_TREE); + tree v8qi_ftype_v8qi + = build_function_type_list (V8QI_type_node, V8QI_type_node, NULL_TREE); + tree v4hi_ftype_v4hi + = build_function_type_list (V4HI_type_node, V4HI_type_node, NULL_TREE); /* Normal vector binops. */ tree v4sf_ftype_v4sf_v4sf @@ -15115,6 +15329,12 @@ ix86_init_mmx_sse_builtins (void) long_long_unsigned_type_node, long_long_unsigned_type_node, NULL_TREE); + tree di_ftype_di_di_int + = build_function_type_list (long_long_unsigned_type_node, + long_long_unsigned_type_node, + long_long_unsigned_type_node, + integer_type_node, NULL_TREE); + tree v2si_ftype_v2sf = build_function_type_list (V2SI_type_node, V2SF_type_node, NULL_TREE); tree v2sf_ftype_v2si @@ -15216,6 +15436,9 @@ ix86_init_mmx_sse_builtins (void) tree v2di_ftype_v2di_int = build_function_type_list (V2DI_type_node, V2DI_type_node, integer_type_node, NULL_TREE); + tree v2di_ftype_v2di_v2di_int + = build_function_type_list (V2DI_type_node, V2DI_type_node, + V2DI_type_node, integer_type_node, NULL_TREE); tree v4si_ftype_v4si_int = build_function_type_list (V4SI_type_node, V4SI_type_node, integer_type_node, NULL_TREE); @@ -15332,6 +15555,50 @@ ix86_init_mmx_sse_builtins (void) def_builtin (d->mask, d->name, type, d->code); } + /* Add all builtins that are more or less simple operations on 1 operand. */ + for (i = 0, d = bdesc_1arg; i < ARRAY_SIZE (bdesc_1arg); i++, d++) + { + enum machine_mode mode; + tree type; + + if (d->name == 0) + continue; + mode = insn_data[d->icode].operand[1].mode; + + switch (mode) + { + case V16QImode: + type = v16qi_ftype_v16qi; + break; + case V8HImode: + type = v8hi_ftype_v8hi; + break; + case V4SImode: + type = v4si_ftype_v4si; + break; + case V2DFmode: + type = v2df_ftype_v2df; + break; + case V4SFmode: + type = v4sf_ftype_v4sf; + break; + case V8QImode: + type = v8qi_ftype_v8qi; + break; + case V4HImode: + type = v4hi_ftype_v4hi; + break; + case V2SImode: + type = v2si_ftype_v2si; + break; + + default: + abort (); + } + + def_builtin (d->mask, d->name, type, d->code); + } + /* Add the remaining MMX insns with somewhat more complicated types. */ def_builtin (MASK_MMX, "__builtin_ia32_emms", void_ftype_void, IX86_BUILTIN_EMMS); def_builtin (MASK_MMX, "__builtin_ia32_psllw", v4hi_ftype_v4hi_di, IX86_BUILTIN_PSLLW); @@ -15531,6 +15798,12 @@ ix86_init_mmx_sse_builtins (void) def_builtin (MASK_SSE3, "__builtin_ia32_lddqu", v16qi_ftype_pcchar, IX86_BUILTIN_LDDQU); + /* SSSE3. */ + def_builtin (MASK_SSSE3, "__builtin_ia32_palignr128", + v2di_ftype_v2di_v2di_int, IX86_BUILTIN_PALIGNR128); + def_builtin (MASK_SSSE3, "__builtin_ia32_palignr", di_ftype_di_di_int, + IX86_BUILTIN_PALIGNR); + /* Access to the vec_init patterns. */ ftype = build_function_type_list (V2SI_type_node, integer_type_node, integer_type_node, NULL_TREE); @@ -16029,7 +16302,7 @@ ix86_expand_builtin (tree exp, rtx targe tree arglist = TREE_OPERAND (exp, 1); tree arg0, arg1, arg2; rtx op0, op1, op2, pat; - enum machine_mode tmode, mode0, mode1, mode2; + enum machine_mode tmode, mode0, mode1, mode2, mode3; unsigned int fcode = DECL_FUNCTION_CODE (fndecl); switch (fcode) @@ -16499,6 +16772,52 @@ ix86_expand_builtin (tree exp, rtx targe return ix86_expand_unop_builtin (CODE_FOR_sse3_lddqu, arglist, target, 1); + case IX86_BUILTIN_PALIGNR: + case IX86_BUILTIN_PALIGNR128: + if (fcode == IX86_BUILTIN_PALIGNR) + { + icode = CODE_FOR_ssse3_palignrdi; + mode = DImode; + } + else + { + icode = CODE_FOR_ssse3_palignrti; + mode = V2DImode; + } + arg0 = TREE_VALUE (arglist); + arg1 = TREE_VALUE (TREE_CHAIN (arglist)); + arg2 = TREE_VALUE (TREE_CHAIN (TREE_CHAIN (arglist))); + op0 = expand_expr (arg0, NULL_RTX, VOIDmode, 0); + op1 = expand_expr (arg1, NULL_RTX, VOIDmode, 0); + op2 = expand_expr (arg2, NULL_RTX, VOIDmode, 0); + tmode = insn_data[icode].operand[0].mode; + mode1 = insn_data[icode].operand[1].mode; + mode2 = insn_data[icode].operand[2].mode; *** DIFF OUTPUT TRUNCATED AT 1000 LINES ***
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201104302207.p3UM74JH064153>