diff options
author | Ben Cheng <bccheng@google.com> | 2014-05-17 17:03:43 -0700 |
---|---|---|
committer | Ben Cheng <bccheng@google.com> | 2014-05-17 17:12:35 -0700 |
commit | 8c493ead6366b552adee796de296936b78f34c5a (patch) | |
tree | 4936e52fb9b84edbcd9293bd321027413d1835bf /gcc-4.9/gcc/config/rs6000 | |
parent | 9750bde7e561731ce8a07cdbd0165a688e74a696 (diff) | |
download | toolchain_gcc-8c493ead6366b552adee796de296936b78f34c5a.tar.gz toolchain_gcc-8c493ead6366b552adee796de296936b78f34c5a.tar.bz2 toolchain_gcc-8c493ead6366b552adee796de296936b78f34c5a.zip |
[4.9] Refresh GCC 4.9 to the 20140514 snapshot.
For critical bug fixes including devirtualization and codegen.
Change-Id: I8138d3dc408fc12db5eecb01d2753d39219712f2
Diffstat (limited to 'gcc-4.9/gcc/config/rs6000')
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/altivec.md | 112 | ||||
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/dfp.md | 69 | ||||
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/predicates.md | 9 | ||||
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/rs6000-builtin.def | 135 | ||||
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/rs6000.c | 116 | ||||
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/rs6000.h | 15 | ||||
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/rs6000.md | 255 | ||||
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/sync.md | 105 | ||||
-rw-r--r-- | gcc-4.9/gcc/config/rs6000/vsx.md | 14 |
9 files changed, 793 insertions, 37 deletions
diff --git a/gcc-4.9/gcc/config/rs6000/altivec.md b/gcc-4.9/gcc/config/rs6000/altivec.md index 674cb40bf..a8cfcb739 100644 --- a/gcc-4.9/gcc/config/rs6000/altivec.md +++ b/gcc-4.9/gcc/config/rs6000/altivec.md @@ -143,6 +143,9 @@ UNSPEC_VSUBEUQM UNSPEC_VSUBECUQ UNSPEC_VBPERMQ + UNSPEC_BCDADD + UNSPEC_BCDSUB + UNSPEC_BCD_OVERFLOW ]) (define_c_enum "unspecv" @@ -3334,3 +3337,112 @@ "vbpermq %0,%1,%2" [(set_attr "length" "4") (set_attr "type" "vecsimple")]) + +;; Decimal Integer operations +(define_int_iterator UNSPEC_BCD_ADD_SUB [UNSPEC_BCDADD UNSPEC_BCDSUB]) + +(define_int_attr bcd_add_sub [(UNSPEC_BCDADD "add") + (UNSPEC_BCDSUB "sub")]) + +(define_code_iterator BCD_TEST [eq lt gt unordered]) + +(define_insn "bcd<bcd_add_sub>" + [(set (match_operand:V1TI 0 "register_operand" "") + (unspec:V1TI [(match_operand:V1TI 1 "register_operand" "") + (match_operand:V1TI 2 "register_operand" "") + (match_operand:QI 3 "const_0_to_1_operand" "")] + UNSPEC_BCD_ADD_SUB)) + (clobber (reg:CCFP 74))] + "TARGET_P8_VECTOR" + "bcd<bcd_add_sub>. %0,%1,%2,%3" + [(set_attr "length" "4") + (set_attr "type" "vecsimple")]) + +;; Use a floating point type (V2DFmode) for the compare to set CR6 so that we +;; can use the unordered test for BCD nans and add/subtracts that overflow. An +;; UNORDERED test on an integer type (like V1TImode) is not defined. The type +;; probably should be one that can go in the VMX (Altivec) registers, so we +;; can't use DDmode or DFmode. +(define_insn "*bcd<bcd_add_sub>_test" + [(set (reg:CCFP 74) + (compare:CCFP + (unspec:V2DF [(match_operand:V1TI 1 "register_operand" "v") + (match_operand:V1TI 2 "register_operand" "v") + (match_operand:QI 3 "const_0_to_1_operand" "i")] + UNSPEC_BCD_ADD_SUB) + (match_operand:V2DF 4 "zero_constant" "j"))) + (clobber (match_scratch:V1TI 0 "=v"))] + "TARGET_P8_VECTOR" + "bcd<bcd_add_sub>. %0,%1,%2,%3" + [(set_attr "length" "4") + (set_attr "type" "vecsimple")]) + +(define_insn "*bcd<bcd_add_sub>_test2" + [(set (match_operand:V1TI 0 "register_operand" "=v") + (unspec:V1TI [(match_operand:V1TI 1 "register_operand" "v") + (match_operand:V1TI 2 "register_operand" "v") + (match_operand:QI 3 "const_0_to_1_operand" "i")] + UNSPEC_BCD_ADD_SUB)) + (set (reg:CCFP 74) + (compare:CCFP + (unspec:V2DF [(match_dup 1) + (match_dup 2) + (match_dup 3)] + UNSPEC_BCD_ADD_SUB) + (match_operand:V2DF 4 "zero_constant" "j")))] + "TARGET_P8_VECTOR" + "bcd<bcd_add_sub>. %0,%1,%2,%3" + [(set_attr "length" "4") + (set_attr "type" "vecsimple")]) + +(define_expand "bcd<bcd_add_sub>_<code>" + [(parallel [(set (reg:CCFP 74) + (compare:CCFP + (unspec:V2DF [(match_operand:V1TI 1 "register_operand" "") + (match_operand:V1TI 2 "register_operand" "") + (match_operand:QI 3 "const_0_to_1_operand" "")] + UNSPEC_BCD_ADD_SUB) + (match_dup 4))) + (clobber (match_scratch:V1TI 5 ""))]) + (set (match_operand:SI 0 "register_operand" "") + (BCD_TEST:SI (reg:CCFP 74) + (const_int 0)))] + "TARGET_P8_VECTOR" +{ + operands[4] = CONST0_RTX (V2DFmode); +}) + +;; Peephole2 pattern to combine a bcdadd/bcdsub that calculates the value and +;; the bcdadd/bcdsub that tests the value. The combiner won't work since +;; CR6 is a hard coded register. Unfortunately, all of the Altivec predicate +;; support is hard coded to use the fixed register CR6 instead of creating +;; a register class for CR6. + +(define_peephole2 + [(parallel [(set (match_operand:V1TI 0 "register_operand" "") + (unspec:V1TI [(match_operand:V1TI 1 "register_operand" "") + (match_operand:V1TI 2 "register_operand" "") + (match_operand:QI 3 "const_0_to_1_operand" "")] + UNSPEC_BCD_ADD_SUB)) + (clobber (reg:CCFP 74))]) + (parallel [(set (reg:CCFP 74) + (compare:CCFP + (unspec:V2DF [(match_dup 1) + (match_dup 2) + (match_dup 3)] + UNSPEC_BCD_ADD_SUB) + (match_operand:V2DF 4 "zero_constant" ""))) + (clobber (match_operand:V1TI 5 "register_operand" ""))])] + "TARGET_P8_VECTOR" + [(parallel [(set (match_dup 0) + (unspec:V1TI [(match_dup 1) + (match_dup 2) + (match_dup 3)] + UNSPEC_BCD_ADD_SUB)) + (set (reg:CCFP 74) + (compare:CCFP + (unspec:V2DF [(match_dup 1) + (match_dup 2) + (match_dup 3)] + UNSPEC_BCD_ADD_SUB) + (match_dup 4)))])]) diff --git a/gcc-4.9/gcc/config/rs6000/dfp.md b/gcc-4.9/gcc/config/rs6000/dfp.md index 8e99bc0d7..40e27e77d 100644 --- a/gcc-4.9/gcc/config/rs6000/dfp.md +++ b/gcc-4.9/gcc/config/rs6000/dfp.md @@ -322,3 +322,72 @@ "TARGET_DFP" "dctfixq %0,%1" [(set_attr "type" "fp")]) + + +;; Decimal builtin support + +(define_c_enum "unspec" + [UNSPEC_DDEDPD + UNSPEC_DENBCD + UNSPEC_DXEX + UNSPEC_DIEX + UNSPEC_DSCLI + UNSPEC_DSCRI]) + +(define_mode_iterator D64_D128 [DD TD]) + +(define_mode_attr dfp_suffix [(DD "") + (TD "q")]) + +(define_insn "dfp_ddedpd_<mode>" + [(set (match_operand:D64_D128 0 "gpc_reg_operand" "=d") + (unspec:D64_D128 [(match_operand:QI 1 "const_0_to_3_operand" "i") + (match_operand:D64_D128 2 "gpc_reg_operand" "d")] + UNSPEC_DDEDPD))] + "TARGET_DFP" + "ddedpd<dfp_suffix> %1,%0,%2" + [(set_attr "type" "fp")]) + +(define_insn "dfp_denbcd_<mode>" + [(set (match_operand:D64_D128 0 "gpc_reg_operand" "=d") + (unspec:D64_D128 [(match_operand:QI 1 "const_0_to_1_operand" "i") + (match_operand:D64_D128 2 "gpc_reg_operand" "d")] + UNSPEC_DENBCD))] + "TARGET_DFP" + "denbcd<dfp_suffix> %1,%0,%2" + [(set_attr "type" "fp")]) + +(define_insn "dfp_dxex_<mode>" + [(set (match_operand:D64_D128 0 "gpc_reg_operand" "=d") + (unspec:D64_D128 [(match_operand:D64_D128 1 "gpc_reg_operand" "d")] + UNSPEC_DXEX))] + "TARGET_DFP" + "dxex<dfp_suffix> %0,%1" + [(set_attr "type" "fp")]) + +(define_insn "dfp_diex_<mode>" + [(set (match_operand:D64_D128 0 "gpc_reg_operand" "=d") + (unspec:D64_D128 [(match_operand:D64_D128 1 "gpc_reg_operand" "d") + (match_operand:D64_D128 2 "gpc_reg_operand" "d")] + UNSPEC_DXEX))] + "TARGET_DFP" + "diex<dfp_suffix> %0,%1,%2" + [(set_attr "type" "fp")]) + +(define_insn "dfp_dscli_<mode>" + [(set (match_operand:D64_D128 0 "gpc_reg_operand" "=d") + (unspec:D64_D128 [(match_operand:D64_D128 1 "gpc_reg_operand" "d") + (match_operand:QI 2 "immediate_operand" "i")] + UNSPEC_DSCLI))] + "TARGET_DFP" + "dscli<dfp_suffix> %0,%1,%2" + [(set_attr "type" "fp")]) + +(define_insn "dfp_dscri_<mode>" + [(set (match_operand:D64_D128 0 "gpc_reg_operand" "=d") + (unspec:D64_D128 [(match_operand:D64_D128 1 "gpc_reg_operand" "d") + (match_operand:QI 2 "immediate_operand" "i")] + UNSPEC_DSCRI))] + "TARGET_DFP" + "dscri<dfp_suffix> %0,%1,%2" + [(set_attr "type" "fp")]) diff --git a/gcc-4.9/gcc/config/rs6000/predicates.md b/gcc-4.9/gcc/config/rs6000/predicates.md index 28f4f5d98..8c384b380 100644 --- a/gcc-4.9/gcc/config/rs6000/predicates.md +++ b/gcc-4.9/gcc/config/rs6000/predicates.md @@ -171,6 +171,11 @@ (and (match_code "const_int") (match_test "IN_RANGE (INTVAL (op), 0, 1)"))) +;; Match op = 0..3. +(define_predicate "const_0_to_3_operand" + (and (match_code "const_int") + (match_test "IN_RANGE (INTVAL (op), 0, 3)"))) + ;; Match op = 2 or op = 3. (define_predicate "const_2_to_3_operand" (and (match_code "const_int") @@ -624,14 +629,14 @@ (match_test "offsettable_nonstrict_memref_p (op)"))) ;; Return 1 if the operand is suitable for load/store quad memory. -;; This predicate only checks for non-atomic loads/stores. +;; This predicate only checks for non-atomic loads/stores (not lqarx/stqcx). (define_predicate "quad_memory_operand" (match_code "mem") { rtx addr, op0, op1; int ret; - if (!TARGET_QUAD_MEMORY) + if (!TARGET_QUAD_MEMORY && !TARGET_SYNC_TI) ret = 0; else if (!memory_operand (op, mode)) diff --git a/gcc-4.9/gcc/config/rs6000/rs6000-builtin.def b/gcc-4.9/gcc/config/rs6000/rs6000-builtin.def index 83351691f..8e15bdf16 100644 --- a/gcc-4.9/gcc/config/rs6000/rs6000-builtin.def +++ b/gcc-4.9/gcc/config/rs6000/rs6000-builtin.def @@ -570,6 +570,75 @@ MASK, /* MASK */ \ (ATTR | RS6000_BTC_SPECIAL), /* ATTR */ \ CODE_FOR_nothing) /* ICODE */ + + +/* Decimal floating point builtins for instructions. */ +#define BU_DFP_MISC_1(ENUM, NAME, ATTR, ICODE) \ + RS6000_BUILTIN_1 (MISC_BUILTIN_ ## ENUM, /* ENUM */ \ + "__builtin_" NAME, /* NAME */ \ + RS6000_BTM_DFP, /* MASK */ \ + (RS6000_BTC_ ## ATTR /* ATTR */ \ + | RS6000_BTC_UNARY), \ + CODE_FOR_ ## ICODE) /* ICODE */ + +#define BU_DFP_MISC_2(ENUM, NAME, ATTR, ICODE) \ + RS6000_BUILTIN_2 (MISC_BUILTIN_ ## ENUM, /* ENUM */ \ + "__builtin_" NAME, /* NAME */ \ + RS6000_BTM_DFP, /* MASK */ \ + (RS6000_BTC_ ## ATTR /* ATTR */ \ + | RS6000_BTC_BINARY), \ + CODE_FOR_ ## ICODE) /* ICODE */ + + +/* Miscellaneous builtins for instructions added in ISA 2.06. These + instructions don't require either the DFP or VSX options, just the basic ISA + 2.06 (popcntd) enablement since they operate on general purpose + registers. */ +#define BU_P7_MISC_1(ENUM, NAME, ATTR, ICODE) \ + RS6000_BUILTIN_1 (MISC_BUILTIN_ ## ENUM, /* ENUM */ \ + "__builtin_" NAME, /* NAME */ \ + RS6000_BTM_POPCNTD, /* MASK */ \ + (RS6000_BTC_ ## ATTR /* ATTR */ \ + | RS6000_BTC_UNARY), \ + CODE_FOR_ ## ICODE) /* ICODE */ + +#define BU_P7_MISC_2(ENUM, NAME, ATTR, ICODE) \ + RS6000_BUILTIN_2 (MISC_BUILTIN_ ## ENUM, /* ENUM */ \ + "__builtin_" NAME, /* NAME */ \ + RS6000_BTM_POPCNTD, /* MASK */ \ + (RS6000_BTC_ ## ATTR /* ATTR */ \ + | RS6000_BTC_BINARY), \ + CODE_FOR_ ## ICODE) /* ICODE */ + + +/* Miscellaneous builtins for instructions added in ISA 2.07. These + instructions do require the ISA 2.07 vector support, but they aren't vector + instructions. */ +#define BU_P8V_MISC_3(ENUM, NAME, ATTR, ICODE) \ + RS6000_BUILTIN_3 (MISC_BUILTIN_ ## ENUM, /* ENUM */ \ + "__builtin_" NAME, /* NAME */ \ + RS6000_BTM_P8_VECTOR, /* MASK */ \ + (RS6000_BTC_ ## ATTR /* ATTR */ \ + | RS6000_BTC_TERNARY), \ + CODE_FOR_ ## ICODE) /* ICODE */ + +/* Miscellaneous builtins. */ +#define BU_MISC_1(ENUM, NAME, ATTR, ICODE) \ + RS6000_BUILTIN_2 (MISC_BUILTIN_ ## ENUM, /* ENUM */ \ + "__builtin_" NAME, /* NAME */ \ + RS6000_BTM_HARD_FLOAT, /* MASK */ \ + (RS6000_BTC_ ## ATTR /* ATTR */ \ + | RS6000_BTC_UNARY), \ + CODE_FOR_ ## ICODE) /* ICODE */ + +#define BU_MISC_2(ENUM, NAME, ATTR, ICODE) \ + RS6000_BUILTIN_2 (MISC_BUILTIN_ ## ENUM, /* ENUM */ \ + "__builtin_" NAME, /* NAME */ \ + RS6000_BTM_HARD_FLOAT, /* MASK */ \ + (RS6000_BTC_ ## ATTR /* ATTR */ \ + | RS6000_BTC_BINARY), \ + CODE_FOR_ ## ICODE) /* ICODE */ + #endif /* Insure 0 is not a legitimate index. */ @@ -1412,10 +1481,10 @@ BU_P8V_AV_2 (ORC_V4SF, "orc_v4sf", CONST, orcv4sf3) BU_P8V_AV_2 (ORC_V2DF, "orc_v2df", CONST, orcv2df3) /* 3 argument altivec instructions added in ISA 2.07. */ -BU_P8V_AV_3 (VADDEUQM, "vaddeuqm", CONST, altivec_vaddeuqm) -BU_P8V_AV_3 (VADDECUQ, "vaddecuq", CONST, altivec_vaddecuq) -BU_P8V_AV_3 (VSUBEUQM, "vsubeuqm", CONST, altivec_vsubeuqm) -BU_P8V_AV_3 (VSUBECUQ, "vsubecuq", CONST, altivec_vsubecuq) +BU_P8V_AV_3 (VADDEUQM, "vaddeuqm", CONST, altivec_vaddeuqm) +BU_P8V_AV_3 (VADDECUQ, "vaddecuq", CONST, altivec_vaddecuq) +BU_P8V_AV_3 (VSUBEUQM, "vsubeuqm", CONST, altivec_vsubeuqm) +BU_P8V_AV_3 (VSUBECUQ, "vsubecuq", CONST, altivec_vsubecuq) /* Vector comparison instructions added in ISA 2.07. */ BU_P8V_AV_2 (VCMPEQUD, "vcmpequd", CONST, vector_eqv2di) @@ -1475,6 +1544,64 @@ BU_P8V_OVERLOAD_3 (VSUBECUQ, "vsubecuq") BU_P8V_OVERLOAD_3 (VSUBEUQM, "vsubeuqm") +/* 2 argument extended divide functions added in ISA 2.06. */ +BU_P7_MISC_2 (DIVWE, "divwe", CONST, dive_si) +BU_P7_MISC_2 (DIVWEO, "divweo", CONST, diveo_si) +BU_P7_MISC_2 (DIVWEU, "divweu", CONST, diveu_si) +BU_P7_MISC_2 (DIVWEUO, "divweuo", CONST, diveuo_si) +BU_P7_MISC_2 (DIVDE, "divde", CONST, dive_di) +BU_P7_MISC_2 (DIVDEO, "divdeo", CONST, diveo_di) +BU_P7_MISC_2 (DIVDEU, "divdeu", CONST, diveu_di) +BU_P7_MISC_2 (DIVDEUO, "divdeuo", CONST, diveuo_di) + +/* 1 argument DFP (decimal floating point) functions added in ISA 2.05. */ +BU_DFP_MISC_1 (DXEX, "dxex", CONST, dfp_dxex_dd) +BU_DFP_MISC_1 (DXEXQ, "dxexq", CONST, dfp_dxex_td) + +/* 2 argument DFP (decimal floating point) functions added in ISA 2.05. */ +BU_DFP_MISC_2 (DDEDPD, "ddedpd", CONST, dfp_ddedpd_dd) +BU_DFP_MISC_2 (DDEDPDQ, "ddedpdq", CONST, dfp_ddedpd_td) +BU_DFP_MISC_2 (DENBCD, "denbcd", CONST, dfp_denbcd_dd) +BU_DFP_MISC_2 (DENBCDQ, "denbcdq", CONST, dfp_denbcd_td) +BU_DFP_MISC_2 (DIEX, "diex", CONST, dfp_diex_dd) +BU_DFP_MISC_2 (DIEXQ, "diexq", CONST, dfp_diex_td) +BU_DFP_MISC_2 (DSCLI, "dscli", CONST, dfp_dscli_dd) +BU_DFP_MISC_2 (DSCLIQ, "dscliq", CONST, dfp_dscli_td) +BU_DFP_MISC_2 (DSCRI, "dscri", CONST, dfp_dscri_dd) +BU_DFP_MISC_2 (DSCRIQ, "dscriq", CONST, dfp_dscri_td) + +/* 1 argument BCD functions added in ISA 2.06. */ +BU_P7_MISC_1 (CDTBCD, "cdtbcd", CONST, cdtbcd) +BU_P7_MISC_1 (CBCDTD, "cbcdtd", CONST, cbcdtd) + +/* 2 argument BCD functions added in ISA 2.06. */ +BU_P7_MISC_2 (ADDG6S, "addg6s", CONST, addg6s) + +/* 3 argument BCD functions added in ISA 2.07. */ +BU_P8V_MISC_3 (BCDADD, "bcdadd", CONST, bcdadd) +BU_P8V_MISC_3 (BCDADD_LT, "bcdadd_lt", CONST, bcdadd_lt) +BU_P8V_MISC_3 (BCDADD_EQ, "bcdadd_eq", CONST, bcdadd_eq) +BU_P8V_MISC_3 (BCDADD_GT, "bcdadd_gt", CONST, bcdadd_gt) +BU_P8V_MISC_3 (BCDADD_OV, "bcdadd_ov", CONST, bcdadd_unordered) +BU_P8V_MISC_3 (BCDSUB, "bcdsub", CONST, bcdsub) +BU_P8V_MISC_3 (BCDSUB_LT, "bcdsub_lt", CONST, bcdsub_lt) +BU_P8V_MISC_3 (BCDSUB_EQ, "bcdsub_eq", CONST, bcdsub_eq) +BU_P8V_MISC_3 (BCDSUB_GT, "bcdsub_gt", CONST, bcdsub_gt) +BU_P8V_MISC_3 (BCDSUB_OV, "bcdsub_ov", CONST, bcdsub_unordered) + +/* 2 argument pack/unpack 128-bit floating point types. */ +BU_DFP_MISC_2 (PACK_TD, "pack_dec128", CONST, packtd) +BU_DFP_MISC_2 (UNPACK_TD, "unpack_dec128", CONST, unpacktd) + +BU_MISC_2 (PACK_TF, "pack_longdouble", CONST, packtf) +BU_MISC_2 (UNPACK_TF, "unpack_longdouble", CONST, unpacktf) +BU_MISC_1 (UNPACK_TF_0, "longdouble_dw0", CONST, unpacktf_0) +BU_MISC_1 (UNPACK_TF_1, "longdouble_dw1", CONST, unpacktf_1) + +BU_P7_MISC_2 (PACK_V1TI, "pack_vector_int128", CONST, packv1ti) +BU_P7_MISC_2 (UNPACK_V1TI, "unpack_vector_int128", CONST, unpackv1ti) + + /* 1 argument crypto functions. */ BU_CRYPTO_1 (VSBOX, "vsbox", CONST, crypto_vsbox) diff --git a/gcc-4.9/gcc/config/rs6000/rs6000.c b/gcc-4.9/gcc/config/rs6000/rs6000.c index 494efc562..4f1a399a2 100644 --- a/gcc-4.9/gcc/config/rs6000/rs6000.c +++ b/gcc-4.9/gcc/config/rs6000/rs6000.c @@ -2283,6 +2283,24 @@ rs6000_debug_reg_global (void) if (rs6000_float_gprs) fprintf (stderr, DEBUG_FMT_S, "float_gprs", "true"); + fprintf (stderr, DEBUG_FMT_S, "fprs", + (TARGET_FPRS ? "true" : "false")); + + fprintf (stderr, DEBUG_FMT_S, "single_float", + (TARGET_SINGLE_FLOAT ? "true" : "false")); + + fprintf (stderr, DEBUG_FMT_S, "double_float", + (TARGET_DOUBLE_FLOAT ? "true" : "false")); + + fprintf (stderr, DEBUG_FMT_S, "soft_float", + (TARGET_SOFT_FLOAT ? "true" : "false")); + + fprintf (stderr, DEBUG_FMT_S, "e500_single", + (TARGET_E500_SINGLE ? "true" : "false")); + + fprintf (stderr, DEBUG_FMT_S, "e500_double", + (TARGET_E500_DOUBLE ? "true" : "false")); + if (TARGET_LINK_STACK) fprintf (stderr, DEBUG_FMT_S, "link_stack", "true"); @@ -3017,7 +3035,9 @@ rs6000_builtin_mask_calculate (void) | ((rs6000_cpu == PROCESSOR_CELL) ? RS6000_BTM_CELL : 0) | ((TARGET_P8_VECTOR) ? RS6000_BTM_P8_VECTOR : 0) | ((TARGET_CRYPTO) ? RS6000_BTM_CRYPTO : 0) - | ((TARGET_HTM) ? RS6000_BTM_HTM : 0)); + | ((TARGET_HTM) ? RS6000_BTM_HTM : 0) + | ((TARGET_DFP) ? RS6000_BTM_DFP : 0) + | ((TARGET_HARD_FLOAT) ? RS6000_BTM_HARD_FLOAT : 0)); } /* Override command line options. Mostly we process the processor type and @@ -3374,6 +3394,13 @@ rs6000_option_override_internal (bool global_init_p) rs6000_isa_flags &= ~OPTION_MASK_VSX_TIMODE; } + if (TARGET_DFP && !TARGET_HARD_FLOAT) + { + if (rs6000_isa_flags_explicit & OPTION_MASK_DFP) + error ("-mhard-dfp requires -mhard-float"); + rs6000_isa_flags &= ~OPTION_MASK_DFP; + } + /* The quad memory instructions only works in 64-bit mode. In 32-bit mode, silently turn off quad memory mode. */ if ((TARGET_QUAD_MEMORY || TARGET_QUAD_MEMORY_ATOMIC) && !TARGET_POWERPC64) @@ -12381,7 +12408,15 @@ rs6000_expand_ternop_builtin (enum insn_code icode, tree exp, rtx target) } } else if (icode == CODE_FOR_vsx_set_v2df - || icode == CODE_FOR_vsx_set_v2di) + || icode == CODE_FOR_vsx_set_v2di + || icode == CODE_FOR_bcdadd + || icode == CODE_FOR_bcdadd_lt + || icode == CODE_FOR_bcdadd_eq + || icode == CODE_FOR_bcdadd_gt + || icode == CODE_FOR_bcdsub + || icode == CODE_FOR_bcdsub_lt + || icode == CODE_FOR_bcdsub_eq + || icode == CODE_FOR_bcdsub_gt) { /* Only allow 1-bit unsigned literals. */ STRIP_NOPS (arg2); @@ -12392,6 +12427,44 @@ rs6000_expand_ternop_builtin (enum insn_code icode, tree exp, rtx target) return const0_rtx; } } + else if (icode == CODE_FOR_dfp_ddedpd_dd + || icode == CODE_FOR_dfp_ddedpd_td) + { + /* Only allow 2-bit unsigned literals where the value is 0 or 2. */ + STRIP_NOPS (arg0); + if (TREE_CODE (arg0) != INTEGER_CST + || TREE_INT_CST_LOW (arg2) & ~0x3) + { + error ("argument 1 must be 0 or 2"); + return const0_rtx; + } + } + else if (icode == CODE_FOR_dfp_denbcd_dd + || icode == CODE_FOR_dfp_denbcd_td) + { + /* Only allow 1-bit unsigned literals. */ + STRIP_NOPS (arg0); + if (TREE_CODE (arg0) != INTEGER_CST + || TREE_INT_CST_LOW (arg0) & ~0x1) + { + error ("argument 1 must be a 1-bit unsigned literal"); + return const0_rtx; + } + } + else if (icode == CODE_FOR_dfp_dscli_dd + || icode == CODE_FOR_dfp_dscli_td + || icode == CODE_FOR_dfp_dscri_dd + || icode == CODE_FOR_dfp_dscri_td) + { + /* Only allow 6-bit unsigned literals. */ + STRIP_NOPS (arg1); + if (TREE_CODE (arg1) != INTEGER_CST + || TREE_INT_CST_LOW (arg1) & ~0x3f) + { + error ("argument 2 must be a 6-bit unsigned literal"); + return const0_rtx; + } + } else if (icode == CODE_FOR_crypto_vshasigmaw || icode == CODE_FOR_crypto_vshasigmad) { @@ -13483,6 +13556,16 @@ rs6000_invalid_builtin (enum rs6000_builtins fncode) error ("Builtin function %s requires the -mpaired option", name); else if ((fnmask & RS6000_BTM_SPE) != 0) error ("Builtin function %s requires the -mspe option", name); + else if ((fnmask & (RS6000_BTM_DFP | RS6000_BTM_P8_VECTOR)) + == (RS6000_BTM_DFP | RS6000_BTM_P8_VECTOR)) + error ("Builtin function %s requires the -mhard-dfp and" + "-mpower8-vector options", name); + else if ((fnmask & RS6000_BTM_DFP) != 0) + error ("Builtin function %s requires the -mhard-dfp option", name); + else if ((fnmask & RS6000_BTM_P8_VECTOR) != 0) + error ("Builtin function %s requires the -mpower8-vector option", name); + else if ((fnmask & RS6000_BTM_HARD_FLOAT) != 0) + error ("Builtin function %s requires the -mhard-float option", name); else error ("Builtin function %s is not supported with the current options", name); @@ -13671,7 +13754,10 @@ rs6000_expand_builtin (tree exp, rtx target, rtx subtarget ATTRIBUTE_UNUSED, return ret; } - gcc_assert (TARGET_ALTIVEC || TARGET_VSX || TARGET_SPE || TARGET_PAIRED_FLOAT); + unsigned attr = rs6000_builtin_info[uns_fcode].attr & RS6000_BTC_TYPE_MASK; + gcc_assert (attr == RS6000_BTC_UNARY + || attr == RS6000_BTC_BINARY + || attr == RS6000_BTC_TERNARY); /* Handle simple unary operations. */ d = bdesc_1arg; @@ -13762,6 +13848,9 @@ rs6000_init_builtins (void) uintTI_type_internal_node = unsigned_intTI_type_node; float_type_internal_node = float_type_node; double_type_internal_node = double_type_node; + long_double_type_internal_node = long_double_type_node; + dfloat64_type_internal_node = dfloat64_type_node; + dfloat128_type_internal_node = dfloat128_type_node; void_type_internal_node = void_type_node; /* Initialize the modes for builtin_function_type, mapping a machine mode to @@ -13776,6 +13865,9 @@ rs6000_init_builtins (void) builtin_mode_to_type[TImode][1] = unsigned_intTI_type_node; builtin_mode_to_type[SFmode][0] = float_type_node; builtin_mode_to_type[DFmode][0] = double_type_node; + builtin_mode_to_type[TFmode][0] = long_double_type_node; + builtin_mode_to_type[DDmode][0] = dfloat64_type_node; + builtin_mode_to_type[TDmode][0] = dfloat128_type_node; builtin_mode_to_type[V1TImode][0] = V1TI_type_node; builtin_mode_to_type[V1TImode][1] = unsigned_V1TI_type_node; builtin_mode_to_type[V2SImode][0] = V2SI_type_node; @@ -14868,6 +14960,8 @@ builtin_function_type (enum machine_mode mode_ret, enum machine_mode mode_arg0, /* unsigned 1 argument functions. */ case CRYPTO_BUILTIN_VSBOX: case P8V_BUILTIN_VGBBD: + case MISC_BUILTIN_CDTBCD: + case MISC_BUILTIN_CBCDTD: h.uns_p[0] = 1; h.uns_p[1] = 1; break; @@ -14886,6 +14980,11 @@ builtin_function_type (enum machine_mode mode_ret, enum machine_mode mode_arg0, case CRYPTO_BUILTIN_VPMSUMW: case CRYPTO_BUILTIN_VPMSUMD: case CRYPTO_BUILTIN_VPMSUM: + case MISC_BUILTIN_ADDG6S: + case MISC_BUILTIN_DIVWEU: + case MISC_BUILTIN_DIVWEUO: + case MISC_BUILTIN_DIVDEU: + case MISC_BUILTIN_DIVDEUO: h.uns_p[0] = 1; h.uns_p[1] = 1; h.uns_p[2] = 1; @@ -14947,9 +15046,18 @@ builtin_function_type (enum machine_mode mode_ret, enum machine_mode mode_arg0, /* signed args, unsigned return. */ case VSX_BUILTIN_XVCVDPUXDS_UNS: case ALTIVEC_BUILTIN_FIXUNS_V4SF_V4SI: + case MISC_BUILTIN_UNPACK_TD: + case MISC_BUILTIN_UNPACK_V1TI: h.uns_p[0] = 1; break; + /* unsigned arguments for 128-bit pack instructions. */ + case MISC_BUILTIN_PACK_TD: + case MISC_BUILTIN_PACK_V1TI: + h.uns_p[1] = 1; + h.uns_p[2] = 1; + break; + default: break; } @@ -31203,6 +31311,8 @@ static struct rs6000_opt_mask const rs6000_builtin_mask_names[] = { "power8-vector", RS6000_BTM_P8_VECTOR, false, false }, { "crypto", RS6000_BTM_CRYPTO, false, false }, { "htm", RS6000_BTM_HTM, false, false }, + { "hard-dfp", RS6000_BTM_DFP, false, false }, + { "hard-float", RS6000_BTM_HARD_FLOAT, false, false }, }; /* Option variables that we want to support inside attribute((target)) and diff --git a/gcc-4.9/gcc/config/rs6000/rs6000.h b/gcc-4.9/gcc/config/rs6000/rs6000.h index 9ec3647fe..21330dc65 100644 --- a/gcc-4.9/gcc/config/rs6000/rs6000.h +++ b/gcc-4.9/gcc/config/rs6000/rs6000.h @@ -624,7 +624,8 @@ extern int rs6000_vector_align[]; || TARGET_CMPB /* ISA 2.05 */ \ || TARGET_POPCNTD /* ISA 2.06 */ \ || TARGET_ALTIVEC \ - || TARGET_VSX))) + || TARGET_VSX \ + || TARGET_HARD_FLOAT))) /* E500 cores only support plain "sync", not lwsync. */ #define TARGET_NO_LWSYNC (rs6000_cpu == PROCESSOR_PPC8540 \ @@ -2516,6 +2517,8 @@ extern int frame_pointer_needed; #define RS6000_BTM_FRSQRTES MASK_POPCNTB /* FRSQRTES instruction. */ #define RS6000_BTM_POPCNTD MASK_POPCNTD /* Target supports ISA 2.06. */ #define RS6000_BTM_CELL MASK_FPRND /* Target is cell powerpc. */ +#define RS6000_BTM_DFP MASK_DFP /* Decimal floating point. */ +#define RS6000_BTM_HARD_FLOAT MASK_SOFT_FLOAT /* Hardware floating point. */ #define RS6000_BTM_COMMON (RS6000_BTM_ALTIVEC \ | RS6000_BTM_VSX \ @@ -2527,7 +2530,9 @@ extern int frame_pointer_needed; | RS6000_BTM_FRSQRTES \ | RS6000_BTM_HTM \ | RS6000_BTM_POPCNTD \ - | RS6000_BTM_CELL) + | RS6000_BTM_CELL \ + | RS6000_BTM_DFP \ + | RS6000_BTM_HARD_FLOAT) /* Define builtin enum index. */ @@ -2622,6 +2627,9 @@ enum rs6000_builtin_type_index RS6000_BTI_UINTTI, /* unsigned_intTI_type_node */ RS6000_BTI_float, /* float_type_node */ RS6000_BTI_double, /* double_type_node */ + RS6000_BTI_long_double, /* long_double_type_node */ + RS6000_BTI_dfloat64, /* dfloat64_type_node */ + RS6000_BTI_dfloat128, /* dfloat128_type_node */ RS6000_BTI_void, /* void_type_node */ RS6000_BTI_MAX }; @@ -2673,6 +2681,9 @@ enum rs6000_builtin_type_index #define uintTI_type_internal_node (rs6000_builtin_types[RS6000_BTI_UINTTI]) #define float_type_internal_node (rs6000_builtin_types[RS6000_BTI_float]) #define double_type_internal_node (rs6000_builtin_types[RS6000_BTI_double]) +#define long_double_type_internal_node (rs6000_builtin_types[RS6000_BTI_long_double]) +#define dfloat64_type_internal_node (rs6000_builtin_types[RS6000_BTI_dfloat64]) +#define dfloat128_type_internal_node (rs6000_builtin_types[RS6000_BTI_dfloat128]) #define void_type_internal_node (rs6000_builtin_types[RS6000_BTI_void]) extern GTY(()) tree rs6000_builtin_types[RS6000_BTI_MAX]; diff --git a/gcc-4.9/gcc/config/rs6000/rs6000.md b/gcc-4.9/gcc/config/rs6000/rs6000.md index 64c9e7c10..e853bc4f9 100644 --- a/gcc-4.9/gcc/config/rs6000/rs6000.md +++ b/gcc-4.9/gcc/config/rs6000/rs6000.md @@ -125,6 +125,16 @@ UNSPEC_P8V_MTVSRD UNSPEC_P8V_XXPERMDI UNSPEC_P8V_RELOAD_FROM_VSX + UNSPEC_ADDG6S + UNSPEC_CDTBCD + UNSPEC_CBCDTD + UNSPEC_DIVE + UNSPEC_DIVEO + UNSPEC_DIVEU + UNSPEC_DIVEUO + UNSPEC_UNPACK_128BIT + UNSPEC_PACK_128BIT + UNSPEC_LSQ ]) ;; @@ -481,6 +491,10 @@ (V2DF "X,X,X,X,X") (V1TI "X,X,X,X,X")]) +;; Mode attribute to give the correct type for integer divides +(define_mode_attr idiv_ldiv [(SI "idiv") + (DI "ldiv")]) + ;; Start with fixed-point load and store insns. Here we put only the more ;; complex forms. Basic data transfer is done later. @@ -2755,10 +2769,7 @@ (match_operand:GPR 2 "gpc_reg_operand" "r")))] "" "div<wd>u %0,%1,%2" - [(set (attr "type") - (cond [(match_operand:SI 0 "" "") - (const_string "idiv")] - (const_string "ldiv")))]) + [(set_attr "type" "<idiv_ldiv>")]) ;; For powers of two we can do srai/aze for divide and then adjust for @@ -2782,10 +2793,7 @@ (match_operand:GPR 2 "gpc_reg_operand" "r")))] "" "div<wd> %0,%1,%2" - [(set (attr "type") - (cond [(match_operand:SI 0 "" "") - (const_string "idiv")] - (const_string "ldiv")))]) + [(set_attr "type" "<idiv_ldiv>")]) (define_expand "mod<mode>3" [(use (match_operand:GPR 0 "gpc_reg_operand" "")) @@ -6534,6 +6542,49 @@ [(set_attr "length" "8") (set_attr "type" "fpload")]) +;; Define the TImode operations that can be done in a small number +;; of instructions. The & constraints are to prevent the register +;; allocator from allocating registers that overlap with the inputs +;; (for example, having an input in 7,8 and an output in 6,7). We +;; also allow for the output being the same as one of the inputs. + +(define_insn "addti3" + [(set (match_operand:TI 0 "gpc_reg_operand" "=&r,&r,r,r") + (plus:TI (match_operand:TI 1 "gpc_reg_operand" "%r,r,0,0") + (match_operand:TI 2 "reg_or_short_operand" "r,I,r,I")))] + "TARGET_64BIT" +{ + if (WORDS_BIG_ENDIAN) + return (GET_CODE (operands[2])) != CONST_INT + ? \"addc %L0,%L1,%L2\;adde %0,%1,%2\" + : \"addic %L0,%L1,%2\;add%G2e %0,%1\"; + else + return (GET_CODE (operands[2])) != CONST_INT + ? \"addc %0,%1,%2\;adde %L0,%L1,%L2\" + : \"addic %0,%1,%2\;add%G2e %L0,%L1\"; +} + [(set_attr "type" "two") + (set_attr "length" "8")]) + +(define_insn "subti3" + [(set (match_operand:TI 0 "gpc_reg_operand" "=&r,&r,r,r,r") + (minus:TI (match_operand:TI 1 "reg_or_short_operand" "r,I,0,r,I") + (match_operand:TI 2 "gpc_reg_operand" "r,r,r,0,0")))] + "TARGET_64BIT" +{ + if (WORDS_BIG_ENDIAN) + return (GET_CODE (operands[1]) != CONST_INT) + ? \"subfc %L0,%L2,%L1\;subfe %0,%2,%1\" + : \"subfic %L0,%L2,%1\;subf%G1e %0,%2\"; + else + return (GET_CODE (operands[1]) != CONST_INT) + ? \"subfc %0,%2,%1\;subfe %L0,%L2,%L1\" + : \"subfic %0,%2,%1\;subf%G1e %L0,%L2\"; +} + [(set_attr "type" "two") + (set_attr "length" "8")]) + + ;; Define the DImode operations that can be done in a small number ;; of instructions. The & constraints are to prevent the register ;; allocator from allocating registers that overlap with the inputs @@ -9395,7 +9446,8 @@ (match_operand:FMOVE64 1 "input_operand" "r,Y,r,G,H,F"))] "! TARGET_POWERPC64 && ((TARGET_FPRS && TARGET_SINGLE_FLOAT) - || TARGET_SOFT_FLOAT || TARGET_E500_SINGLE) + || TARGET_SOFT_FLOAT || TARGET_E500_SINGLE + || (<MODE>mode == DDmode && TARGET_E500_DOUBLE)) && (gpc_reg_operand (operands[0], <MODE>mode) || gpc_reg_operand (operands[1], <MODE>mode))" "#" @@ -15691,6 +15743,191 @@ }) +;; Miscellaneous ISA 2.06 (power7) instructions +(define_insn "addg6s" + [(set (match_operand:SI 0 "register_operand" "=r") + (unspec:SI [(match_operand:SI 1 "register_operand" "r") + (match_operand:SI 2 "register_operand" "r")] + UNSPEC_ADDG6S))] + "TARGET_POPCNTD" + "addg6s %0,%1,%2" + [(set_attr "type" "integer") + (set_attr "length" "4")]) + +(define_insn "cdtbcd" + [(set (match_operand:SI 0 "register_operand" "=r") + (unspec:SI [(match_operand:SI 1 "register_operand" "r")] + UNSPEC_CDTBCD))] + "TARGET_POPCNTD" + "cdtbcd %0,%1" + [(set_attr "type" "integer") + (set_attr "length" "4")]) + +(define_insn "cbcdtd" + [(set (match_operand:SI 0 "register_operand" "=r") + (unspec:SI [(match_operand:SI 1 "register_operand" "r")] + UNSPEC_CBCDTD))] + "TARGET_POPCNTD" + "cbcdtd %0,%1" + [(set_attr "type" "integer") + (set_attr "length" "4")]) + +(define_int_iterator UNSPEC_DIV_EXTEND [UNSPEC_DIVE + UNSPEC_DIVEO + UNSPEC_DIVEU + UNSPEC_DIVEUO]) + +(define_int_attr div_extend [(UNSPEC_DIVE "e") + (UNSPEC_DIVEO "eo") + (UNSPEC_DIVEU "eu") + (UNSPEC_DIVEUO "euo")]) + +(define_insn "div<div_extend>_<mode>" + [(set (match_operand:GPR 0 "register_operand" "=r") + (unspec:GPR [(match_operand:GPR 1 "register_operand" "r") + (match_operand:GPR 2 "register_operand" "r")] + UNSPEC_DIV_EXTEND))] + "TARGET_POPCNTD" + "div<wd><div_extend> %0,%1,%2" + [(set_attr "type" "<idiv_ldiv>")]) + + +;; Pack/unpack 128-bit floating point types that take 2 scalar registers + +; Type of the 64-bit part when packing/unpacking 128-bit floating point types +(define_mode_attr FP128_64 [(TF "DF") (TD "DI")]) + +(define_expand "unpack<mode>" + [(set (match_operand:<FP128_64> 0 "nonimmediate_operand" "") + (unspec:<FP128_64> + [(match_operand:FMOVE128 1 "register_operand" "") + (match_operand:QI 2 "const_0_to_1_operand" "")] + UNSPEC_UNPACK_128BIT))] + "" + "") + +;; The Advance Toolchain 7.0-3 added private builtins: __builtin_longdouble_dw0 +;; and __builtin_longdouble_dw1 to optimize glibc. Add support for these +;; builtins here. + +(define_expand "unpacktf_0" + [(set (match_operand:DF 0 "nonimmediate_operand" "") + (unspec:DF [(match_operand:TF 1 "register_operand" "") + (const_int 0)] + UNSPEC_UNPACK_128BIT))] + "" + "") + +(define_expand "unpacktf_1" + [(set (match_operand:DF 0 "nonimmediate_operand" "") + (unspec:DF [(match_operand:TF 1 "register_operand" "") + (const_int 1)] + UNSPEC_UNPACK_128BIT))] + "" + "") + +(define_insn_and_split "unpack<mode>_dm" + [(set (match_operand:<FP128_64> 0 "nonimmediate_operand" "=d,m,d,r,m") + (unspec:<FP128_64> + [(match_operand:FMOVE128 1 "register_operand" "d,d,r,d,r") + (match_operand:QI 2 "const_0_to_1_operand" "i,i,i,i,i")] + UNSPEC_UNPACK_128BIT))] + "TARGET_POWERPC64 && TARGET_DIRECT_MOVE" + "#" + "&& reload_completed" + [(set (match_dup 0) (match_dup 3))] +{ + unsigned fp_regno = REGNO (operands[1]) + UINTVAL (operands[2]); + + if (REG_P (operands[0]) && REGNO (operands[0]) == fp_regno) + { + emit_note (NOTE_INSN_DELETED); + DONE; + } + + operands[3] = gen_rtx_REG (<FP128_64>mode, fp_regno); +} + [(set_attr "type" "fp,fpstore,mffgpr,mftgpr,store") + (set_attr "length" "4")]) + +(define_insn_and_split "unpack<mode>_nodm" + [(set (match_operand:<FP128_64> 0 "nonimmediate_operand" "=d,m") + (unspec:<FP128_64> + [(match_operand:FMOVE128 1 "register_operand" "d,d") + (match_operand:QI 2 "const_0_to_1_operand" "i,i")] + UNSPEC_UNPACK_128BIT))] + "!TARGET_POWERPC64 || !TARGET_DIRECT_MOVE" + "#" + "&& reload_completed" + [(set (match_dup 0) (match_dup 3))] +{ + unsigned fp_regno = REGNO (operands[1]) + UINTVAL (operands[2]); + + if (REG_P (operands[0]) && REGNO (operands[0]) == fp_regno) + { + emit_note (NOTE_INSN_DELETED); + DONE; + } + + operands[3] = gen_rtx_REG (<FP128_64>mode, fp_regno); +} + [(set_attr "type" "fp,fpstore") + (set_attr "length" "4")]) + +(define_insn_and_split "pack<mode>" + [(set (match_operand:FMOVE128 0 "register_operand" "=d,&d") + (unspec:FMOVE128 + [(match_operand:<FP128_64> 1 "register_operand" "0,d") + (match_operand:<FP128_64> 2 "register_operand" "d,d")] + UNSPEC_PACK_128BIT))] + "" + "@ + fmr %L0,%2 + #" + "&& reload_completed && REGNO (operands[0]) != REGNO (operands[1])" + [(set (match_dup 3) (match_dup 1)) + (set (match_dup 4) (match_dup 2))] +{ + unsigned dest_hi = REGNO (operands[0]); + unsigned dest_lo = dest_hi + 1; + + gcc_assert (!IN_RANGE (REGNO (operands[1]), dest_hi, dest_lo)); + gcc_assert (!IN_RANGE (REGNO (operands[2]), dest_hi, dest_lo)); + + operands[3] = gen_rtx_REG (<FP128_64>mode, dest_hi); + operands[4] = gen_rtx_REG (<FP128_64>mode, dest_lo); +} + [(set_attr "type" "fp,fp") + (set_attr "length" "4,8")]) + +(define_insn "unpackv1ti" + [(set (match_operand:DI 0 "register_operand" "=d,d") + (unspec:DI [(match_operand:V1TI 1 "register_operand" "0,wa") + (match_operand:QI 2 "const_0_to_1_operand" "O,i")] + UNSPEC_UNPACK_128BIT))] + "TARGET_VSX" +{ + if (REGNO (operands[0]) == REGNO (operands[1]) && INTVAL (operands[2]) == 0) + return ASM_COMMENT_START " xxpermdi to same register"; + + operands[3] = GEN_INT (INTVAL (operands[2]) == 0 ? 0 : 3); + return "xxpermdi %x0,%x1,%x1,%3"; +} + [(set_attr "type" "vecperm") + (set_attr "length" "4")]) + +(define_insn "packv1ti" + [(set (match_operand:V1TI 0 "register_operand" "=wa") + (unspec:V1TI + [(match_operand:DI 1 "register_operand" "d") + (match_operand:DI 2 "register_operand" "d")] + UNSPEC_PACK_128BIT))] + "TARGET_VSX" + "xxpermdi %x0,%x1,%x2,0" + [(set_attr "type" "vecperm") + (set_attr "length" "4")]) + + (include "sync.md") (include "vector.md") diff --git a/gcc-4.9/gcc/config/rs6000/sync.md b/gcc-4.9/gcc/config/rs6000/sync.md index 7db439074..63152ed04 100644 --- a/gcc-4.9/gcc/config/rs6000/sync.md +++ b/gcc-4.9/gcc/config/rs6000/sync.md @@ -107,10 +107,17 @@ "isync" [(set_attr "type" "isync")]) +;; Types that we should provide atomic instructions for. +(define_mode_iterator AINT [QI + HI + SI + (DI "TARGET_POWERPC64") + (TI "TARGET_SYNC_TI")]) + ;; The control dependency used for load dependency described ;; in B.2.3 of the Power ISA 2.06B. (define_insn "loadsync_<mode>" - [(unspec_volatile:BLK [(match_operand:INT1 0 "register_operand" "r")] + [(unspec_volatile:BLK [(match_operand:AINT 0 "register_operand" "r")] UNSPECV_ISYNC) (clobber (match_scratch:CC 1 "=y"))] "" @@ -118,18 +125,56 @@ [(set_attr "type" "isync") (set_attr "length" "12")]) +(define_insn "load_quadpti" + [(set (match_operand:PTI 0 "quad_int_reg_operand" "=&r") + (unspec:PTI + [(match_operand:TI 1 "quad_memory_operand" "wQ")] UNSPEC_LSQ))] + "TARGET_SYNC_TI + && !reg_mentioned_p (operands[0], operands[1])" + "lq %0,%1" + [(set_attr "type" "load") + (set_attr "length" "4")]) + (define_expand "atomic_load<mode>" - [(set (match_operand:INT1 0 "register_operand" "") ;; output - (match_operand:INT1 1 "memory_operand" "")) ;; memory + [(set (match_operand:AINT 0 "register_operand" "") ;; output + (match_operand:AINT 1 "memory_operand" "")) ;; memory (use (match_operand:SI 2 "const_int_operand" ""))] ;; model "" { + if (<MODE>mode == TImode && !TARGET_SYNC_TI) + FAIL; + enum memmodel model = (enum memmodel) INTVAL (operands[2]); if (model == MEMMODEL_SEQ_CST) emit_insn (gen_hwsync ()); - emit_move_insn (operands[0], operands[1]); + if (<MODE>mode != TImode) + emit_move_insn (operands[0], operands[1]); + else + { + rtx op0 = operands[0]; + rtx op1 = operands[1]; + rtx pti_reg = gen_reg_rtx (PTImode); + + // Can't have indexed address for 'lq' + if (indexed_address (XEXP (op1, 0), TImode)) + { + rtx old_addr = XEXP (op1, 0); + rtx new_addr = force_reg (Pmode, old_addr); + operands[1] = op1 = replace_equiv_address (op1, new_addr); + } + + emit_insn (gen_load_quadpti (pti_reg, op1)); + + if (WORDS_BIG_ENDIAN) + emit_move_insn (op0, gen_lowpart (TImode, pti_reg)); + else + { + emit_move_insn (gen_lowpart (DImode, op0), gen_highpart (DImode, pti_reg)); + emit_move_insn (gen_highpart (DImode, op0), gen_lowpart (DImode, pti_reg)); + } + } switch (model) { @@ -146,12 +191,24 @@ DONE; }) +(define_insn "store_quadpti" + [(set (match_operand:PTI 0 "quad_memory_operand" "=wQ") + (unspec:PTI + [(match_operand:PTI 1 "quad_int_reg_operand" "r")] UNSPEC_LSQ))] + "TARGET_SYNC_TI" + "stq %1,%0" + [(set_attr "type" "store") + (set_attr "length" "4")]) + (define_expand "atomic_store<mode>" - [(set (match_operand:INT1 0 "memory_operand" "") ;; memory - (match_operand:INT1 1 "register_operand" "")) ;; input + [(set (match_operand:AINT 0 "memory_operand" "") ;; memory + (match_operand:AINT 1 "register_operand" "")) ;; input (use (match_operand:SI 2 "const_int_operand" ""))] ;; model "" { + if (<MODE>mode == TImode && !TARGET_SYNC_TI) + FAIL; + enum memmodel model = (enum memmodel) INTVAL (operands[2]); switch (model) { @@ -166,7 +223,33 @@ default: gcc_unreachable (); } - emit_move_insn (operands[0], operands[1]); + if (<MODE>mode != TImode) + emit_move_insn (operands[0], operands[1]); + else + { + rtx op0 = operands[0]; + rtx op1 = operands[1]; + rtx pti_reg = gen_reg_rtx (PTImode); + + // Can't have indexed address for 'stq' + if (indexed_address (XEXP (op0, 0), TImode)) + { + rtx old_addr = XEXP (op0, 0); + rtx new_addr = force_reg (Pmode, old_addr); + operands[0] = op0 = replace_equiv_address (op0, new_addr); + } + + if (WORDS_BIG_ENDIAN) + emit_move_insn (pti_reg, gen_lowpart (PTImode, op1)); + else + { + emit_move_insn (gen_lowpart (DImode, pti_reg), gen_highpart (DImode, op1)); + emit_move_insn (gen_highpart (DImode, pti_reg), gen_lowpart (DImode, op1)); + } + + emit_insn (gen_store_quadpti (gen_lowpart (PTImode, op0), pti_reg)); + } + DONE; }) @@ -180,14 +263,6 @@ SI (DI "TARGET_POWERPC64")]) -;; Types that we should provide atomic instructions for. - -(define_mode_iterator AINT [QI - HI - SI - (DI "TARGET_POWERPC64") - (TI "TARGET_SYNC_TI")]) - (define_insn "load_locked<mode>" [(set (match_operand:ATOMIC 0 "int_reg_operand" "=r") (unspec_volatile:ATOMIC diff --git a/gcc-4.9/gcc/config/rs6000/vsx.md b/gcc-4.9/gcc/config/rs6000/vsx.md index d83cdc3df..23d85ab06 100644 --- a/gcc-4.9/gcc/config/rs6000/vsx.md +++ b/gcc-4.9/gcc/config/rs6000/vsx.md @@ -1891,7 +1891,12 @@ (parallel [(const_int 0) (const_int 4) (const_int 1) (const_int 5)])))] "VECTOR_MEM_VSX_P (<MODE>mode)" - "xxmrghw %x0,%x1,%x2" +{ + if (BYTES_BIG_ENDIAN) + return "xxmrghw %x0,%x1,%x2"; + else + return "xxmrglw %x0,%x2,%x1"; +} [(set_attr "type" "vecperm")]) (define_insn "vsx_xxmrglw_<mode>" @@ -1903,7 +1908,12 @@ (parallel [(const_int 2) (const_int 6) (const_int 3) (const_int 7)])))] "VECTOR_MEM_VSX_P (<MODE>mode)" - "xxmrglw %x0,%x1,%x2" +{ + if (BYTES_BIG_ENDIAN) + return "xxmrglw %x0,%x1,%x2"; + else + return "xxmrghw %x0,%x2,%x1"; +} [(set_attr "type" "vecperm")]) ;; Shift left double by word immediate |