diff options
Diffstat (limited to 'gcc/config/i386/i386.md')
-rw-r--r-- | gcc/config/i386/i386.md | 202 |
1 files changed, 147 insertions, 55 deletions
diff --git a/gcc/config/i386/i386.md b/gcc/config/i386/i386.md index c7ae4ac..e733a40 100644 --- a/gcc/config/i386/i386.md +++ b/gcc/config/i386/i386.md @@ -5730,16 +5730,17 @@ (symbol_ref "!TARGET_PARTIAL_REG_STALL")] (symbol_ref "true")))]) -(define_insn "*add<mode>_1_slp" - [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>")) - (plus:SWI12 (match_operand:SWI12 1 "nonimmediate_operand" "%0") - (match_operand:SWI12 2 "general_operand" "<r>mn"))) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*add<mode>_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (plus:SWI12 (match_operand:SWI12 1 "nonimmediate_operand" "%0,!<r>") + (match_operand:SWI12 2 "general_operand" "<r>mn,<r>mn"))) (clobber (reg:CC FLAGS_REG))] - "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)) - /* FIXME: without this LRA can't reload this pattern, see PR82524. */ - && (rtx_equal_p (operands[0], operands[1]) - || rtx_equal_p (operands[0], operands[2]))" + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" { + if (which_alternative) + return "#"; + switch (get_attr_type (insn)) { case TYPE_INCDEC: @@ -5758,6 +5759,13 @@ return "add{<imodesuffix>}\t{%2, %0|%0, %2}"; } } + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (parallel + [(set (strict_low_part (match_dup 0)) + (plus:SWI12 (match_dup 0) (match_dup 2))) + (clobber (reg:CC FLAGS_REG))])] + "" [(set (attr "type") (if_then_else (match_operand:QI 2 "incdec_operand") (const_string "incdec") @@ -6676,15 +6684,23 @@ [(set_attr "type" "alu") (set_attr "mode" "SI")]) -(define_insn "*sub<mode>_1_slp" - [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>")) - (minus:SWI12 (match_operand:SWI12 1 "register_operand" "0") - (match_operand:SWI12 2 "general_operand" "<r>mn"))) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*sub<mode>_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (minus:SWI12 (match_operand:SWI12 1 "register_operand" "0,!<r>") + (match_operand:SWI12 2 "general_operand" "<r>mn,<r>mn"))) (clobber (reg:CC FLAGS_REG))] - "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)) - /* FIXME: without this LRA can't reload this pattern, see PR82524. */ - && rtx_equal_p (operands[0], operands[1])" - "sub{<imodesuffix>}\t{%2, %0|%0, %2}" + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" + "@ + sub{<imodesuffix>}\t{%2, %0|%0, %2} + #" + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (parallel + [(set (strict_low_part (match_dup 0)) + (minus:SWI12 (match_dup 0) (match_dup 2))) + (clobber (reg:CC FLAGS_REG))])] + "" [(set_attr "type" "alu") (set_attr "mode" "<MODE>")]) @@ -9606,16 +9622,23 @@ (symbol_ref "!TARGET_PARTIAL_REG_STALL")] (symbol_ref "true")))]) -(define_insn "*and<mode>_1_slp" - [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>")) - (and:SWI12 (match_operand:SWI12 1 "nonimmediate_operand" "%0") - (match_operand:SWI12 2 "general_operand" "<r>mn"))) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*and<mode>_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (and:SWI12 (match_operand:SWI12 1 "nonimmediate_operand" "%0,!<r>") + (match_operand:SWI12 2 "general_operand" "<r>mn,<r>mn"))) (clobber (reg:CC FLAGS_REG))] - "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)) - /* FIXME: without this LRA can't reload this pattern, see PR82524. */ - && (rtx_equal_p (operands[0], operands[1]) - || rtx_equal_p (operands[0], operands[2]))" - "and{<imodesuffix>}\t{%2, %0|%0, %2}" + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" + "@ + and{<imodesuffix>}\t{%2, %0|%0, %2} + #" + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (parallel + [(set (strict_low_part (match_dup 0)) + (and:SWI12 (match_dup 0) (match_dup 2))) + (clobber (reg:CC FLAGS_REG))])] + "" [(set_attr "type" "alu") (set_attr "mode" "<MODE>")]) @@ -10199,16 +10222,23 @@ (symbol_ref "!TARGET_PARTIAL_REG_STALL")] (symbol_ref "true")))]) -(define_insn "*<code><mode>_1_slp" - [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>")) - (any_or:SWI12 (match_operand:SWI12 1 "nonimmediate_operand" "%0") - (match_operand:SWI12 2 "general_operand" "<r>mn"))) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*<code><mode>_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (any_or:SWI12 (match_operand:SWI12 1 "nonimmediate_operand" "%0,!<r>") + (match_operand:SWI12 2 "general_operand" "<r>mn,<r>mn"))) (clobber (reg:CC FLAGS_REG))] - "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)) - /* FIXME: without this LRA can't reload this pattern, see PR82524. */ - && (rtx_equal_p (operands[0], operands[1]) - || rtx_equal_p (operands[0], operands[2]))" - "<logic>{<imodesuffix>}\t{%2, %0|%0, %2}" + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" + "@ + <logic>{<imodesuffix>}\t{%2, %0|%0, %2} + #" + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (parallel + [(set (strict_low_part (match_dup 0)) + (any_or:SWI12 (match_dup 0) (match_dup 2))) + (clobber (reg:CC FLAGS_REG))])] + "" [(set_attr "type" "alu") (set_attr "mode" "<MODE>")]) @@ -10505,6 +10535,25 @@ [(set_attr "type" "negnot") (set_attr "mode" "SI")]) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*neg<mode>_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (neg:SWI12 (match_operand:SWI12 1 "register_operand" "0,!<r>"))) + (clobber (reg:CC FLAGS_REG))] + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" + "@ + neg{<imodesuffix>}\t%0 + #" + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (parallel + [(set (strict_low_part (match_dup 0)) + (neg:SWI12 (match_dup 0))) + (clobber (reg:CC FLAGS_REG))])] + "" + [(set_attr "type" "negnot") + (set_attr "mode" "<MODE>")]) + (define_insn "*neg<mode>_2" [(set (reg FLAGS_REG) (compare @@ -11060,6 +11109,22 @@ (symbol_ref "!TARGET_PARTIAL_REG_STALL")] (symbol_ref "true")))]) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*one_cmpl<mode>_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (not:SWI12 (match_operand:SWI12 1 "register_operand" "0,!<r>")))] + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" + "@ + not{<imodesuffix>}\t%0 + #" + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (set (strict_low_part (match_dup 0)) + (not:SWI12 (match_dup 0)))] + "" + [(set_attr "type" "negnot") + (set_attr "mode" "<MODE>")]) + (define_insn "*one_cmpl<mode>2_2" [(set (reg FLAGS_REG) (compare (not:SWI (match_operand:SWI 1 "nonimmediate_operand" "0")) @@ -11638,15 +11703,17 @@ (symbol_ref "!TARGET_PARTIAL_REG_STALL")] (symbol_ref "true")))]) -(define_insn "*ashl<mode>3_1_slp" - [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>")) - (ashift:SWI12 (match_operand:SWI12 1 "register_operand" "0") - (match_operand:QI 2 "nonmemory_operand" "cI"))) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*ashl<mode>3_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (ashift:SWI12 (match_operand:SWI12 1 "register_operand" "0,!<r>") + (match_operand:QI 2 "nonmemory_operand" "cI,cI"))) (clobber (reg:CC FLAGS_REG))] - "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)) - /* FIXME: without this LRA can't reload this pattern, see PR82524. */ - && rtx_equal_p (operands[0], operands[1])" + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" { + if (which_alternative) + return "#"; + switch (get_attr_type (insn)) { case TYPE_ALU: @@ -11661,6 +11728,13 @@ return "sal{<imodesuffix>}\t{%2, %0|%0, %2}"; } } + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (parallel + [(set (strict_low_part (match_dup 0)) + (ashift:SWI12 (match_dup 0) (match_dup 2))) + (clobber (reg:CC FLAGS_REG))])] + "" [(set (attr "type") (cond [(and (match_test "TARGET_DOUBLE_WITH_ADD") (match_operand 2 "const1_operand")) @@ -12356,21 +12430,30 @@ (const_string "*"))) (set_attr "mode" "HI")]) -(define_insn "*<insn><mode>3_1_slp" - [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>")) - (any_shiftrt:SWI12 (match_operand:SWI12 1 "register_operand" "0") - (match_operand:QI 2 "nonmemory_operand" "cI"))) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*<insn><mode>3_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (any_shiftrt:SWI12 (match_operand:SWI12 1 "register_operand" "0,!<r>") + (match_operand:QI 2 "nonmemory_operand" "cI,cI"))) (clobber (reg:CC FLAGS_REG))] - "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)) - /* FIXME: without this LRA can't reload this pattern, see PR82524. */ - && rtx_equal_p (operands[0], operands[1])" + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" { + if (which_alternative) + return "#"; + if (operands[2] == const1_rtx && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun))) return "<shift>{<imodesuffix>}\t%0"; else return "<shift>{<imodesuffix>}\t{%2, %0|%0, %2}"; } + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (parallel + [(set (strict_low_part (match_dup 0)) + (any_shiftrt:SWI12 (match_dup 0) (match_dup 2))) + (clobber (reg:CC FLAGS_REG))])] + "" [(set_attr "type" "ishift") (set (attr "length_immediate") (if_then_else @@ -12814,21 +12897,30 @@ (const_string "*"))) (set_attr "mode" "<MODE>")]) -(define_insn "*<insn><mode>3_1_slp" - [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>")) - (any_rotate:SWI12 (match_operand:SWI12 1 "register_operand" "0") - (match_operand:QI 2 "nonmemory_operand" "cI"))) +;; Alternative 1 is needed to work around LRA limitation, see PR82524. +(define_insn_and_split "*<insn><mode>3_1_slp" + [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>,<r>")) + (any_rotate:SWI12 (match_operand:SWI12 1 "register_operand" "0,!<r>") + (match_operand:QI 2 "nonmemory_operand" "cI,cI"))) (clobber (reg:CC FLAGS_REG))] - "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)) - /* FIXME: without this LRA can't reload this pattern, see PR82524. */ - && rtx_equal_p (operands[0], operands[1])" + "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)" { + if (which_alternative) + return "#"; + if (operands[2] == const1_rtx && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun))) return "<rotate>{<imodesuffix>}\t%0"; else return "<rotate>{<imodesuffix>}\t{%2, %0|%0, %2}"; } + "&& reload_completed" + [(set (strict_low_part (match_dup 0)) (match_dup 1)) + (parallel + [(set (strict_low_part (match_dup 0)) + (any_rotate:SWI12 (match_dup 0) (match_dup 2))) + (clobber (reg:CC FLAGS_REG))])] + "" [(set_attr "type" "rotate") (set (attr "length_immediate") (if_then_else |