13 files changed, 151 insertions, 92 deletions

diff --git a/src/lj_asm_arm.h b/src/lj_asm_arm.h
index 9d055c81..d2579349 100644
--- a/src/lj_asm_arm.h
+++ b/src/lj_asm_arm.h
@@ -1659,8 +1659,8 @@ static void asm_min_max(ASMState *as, IRIns *ir, int cc, int fcc)
     asm_intmin_max(as, ir, cc);
 }
 
-#define asm_min(as, ir)         asm_min_max(as, ir, CC_GT, CC_HI)
-#define asm_max(as, ir)         asm_min_max(as, ir, CC_LT, CC_LO)
+#define asm_min(as, ir)         asm_min_max(as, ir, CC_GT, CC_PL)
+#define asm_max(as, ir)         asm_min_max(as, ir, CC_LT, CC_LE)
 
 /* -- Comparisons --------------------------------------------------------- */
 
@@ -1852,7 +1852,7 @@ static void asm_hiop(ASMState *as, IRIns *ir)
   } else if ((ir-1)->o == IR_MIN || (ir-1)->o == IR_MAX) {
     as->curins--;  /* Always skip the loword min/max. */
     if (uselo || usehi)
-      asm_sfpmin_max(as, ir-1, (ir-1)->o == IR_MIN ? CC_HI : CC_LO);
+      asm_sfpmin_max(as, ir-1, (ir-1)->o == IR_MIN ? CC_PL : CC_LE);
     return;
 #elif LJ_HASFFI
   } else if ((ir-1)->o == IR_CONV) {
diff --git a/src/lj_asm_arm64.h b/src/lj_asm_arm64.h
index 624cc2da..f640b91b 100644
--- a/src/lj_asm_arm64.h
+++ b/src/lj_asm_arm64.h
@@ -1598,7 +1598,7 @@ static void asm_fpmin_max(ASMState *as, IRIns *ir, A64CC fcc)
   Reg dest = (ra_dest(as, ir, RSET_FPR) & 31);
   Reg right, left = ra_alloc2(as, ir, RSET_FPR);
   right = ((left >> 8) & 31); left &= 31;
-  emit_dnm(as, A64I_FCSELd | A64F_CC(fcc), dest, left, right);
+  emit_dnm(as, A64I_FCSELd | A64F_CC(fcc), dest, right, left);
   emit_nm(as, A64I_FCMPd, left, right);
 }
 
@@ -1610,8 +1610,8 @@ static void asm_min_max(ASMState *as, IRIns *ir, A64CC cc, A64CC fcc)
     asm_intmin_max(as, ir, cc);
 }
 
-#define asm_max(as, ir)         asm_min_max(as, ir, CC_GT, CC_HI)
-#define asm_min(as, ir)         asm_min_max(as, ir, CC_LT, CC_LO)
+#define asm_min(as, ir)         asm_min_max(as, ir, CC_LT, CC_PL)
+#define asm_max(as, ir)         asm_min_max(as, ir, CC_GT, CC_LE)
 
 /* -- Comparisons --------------------------------------------------------- */
 
diff --git a/src/lj_asm_mips.h b/src/lj_asm_mips.h
index 9309b781..a242904e 100644
--- a/src/lj_asm_mips.h
+++ b/src/lj_asm_mips.h
@@ -2121,12 +2121,12 @@ static void asm_min_max(ASMState *as, IRIns *ir, int ismax)
     right = (left >> 8); left &= 255;
 #if !LJ_TARGET_MIPSR6
     if (dest == left) {
-      emit_fg(as, MIPSI_MOVT_D, dest, right);
+      emit_fg(as, MIPSI_MOVF_D, dest, right);
     } else {
-      emit_fg(as, MIPSI_MOVF_D, dest, left);
+      emit_fg(as, MIPSI_MOVT_D, dest, left);
       if (dest != right) emit_fg(as, MIPSI_MOV_D, dest, right);
     }
-    emit_fgh(as, MIPSI_C_OLT_D, 0, ismax ? left : right, ismax ? right : left);
+    emit_fgh(as, MIPSI_C_OLT_D, 0, ismax ? right : left, ismax ? left : right);
 #else
     emit_fgh(as, ismax ? MIPSI_MAX_D : MIPSI_MIN_D, dest, left, right);
 #endif
diff --git a/src/lj_asm_ppc.h b/src/lj_asm_ppc.h
index 8fa8c8ef..afcd6b7a 100644
--- a/src/lj_asm_ppc.h
+++ b/src/lj_asm_ppc.h
@@ -1724,9 +1724,8 @@ static void asm_min_max(ASMState *as, IRIns *ir, int ismax)
     if (tmp == left || tmp == right)
       tmp = ra_scratch(as, rset_exclude(rset_exclude(rset_exclude(RSET_FPR,
                                         dest), left), right));
-    emit_facb(as, PPCI_FSEL, dest, tmp,
-              ismax ? left : right, ismax ? right : left);
-    emit_fab(as, PPCI_FSUB, tmp, left, right);
+    emit_facb(as, PPCI_FSEL, dest, tmp, left, right);
+    emit_fab(as, PPCI_FSUB, tmp, ismax ? left : right, ismax ? right : left);
   } else {
     Reg dest = ra_dest(as, ir, RSET_GPR);
     Reg tmp1 = RID_TMP, tmp2 = dest;
diff --git a/src/lj_opt_fold.c b/src/lj_opt_fold.c
index b4d05a26..cefd69c8 100644
--- a/src/lj_opt_fold.c
+++ b/src/lj_opt_fold.c
@@ -1774,8 +1774,6 @@ LJFOLDF(reassoc_intarith_k64)
 #endif
 }
 
-LJFOLD(MIN MIN any)
-LJFOLD(MAX MAX any)
 LJFOLD(BAND BAND any)
 LJFOLD(BOR BOR any)
 LJFOLDF(reassoc_dup)
@@ -1785,6 +1783,15 @@ LJFOLDF(reassoc_dup)
   return NEXTFOLD;
 }
 
+LJFOLD(MIN MIN any)
+LJFOLD(MAX MAX any)
+LJFOLDF(reassoc_dup_minmax)
+{
+  if (fins->op2 == fleft->op2)
+    return LEFTFOLD;  /* (a o b) o b ==> a o b */
+  return NEXTFOLD;
+}
+
 LJFOLD(BXOR BXOR any)
 LJFOLDF(reassoc_bxor)
 {
@@ -1823,23 +1830,12 @@ LJFOLDF(reassoc_shift)
   return NEXTFOLD;
 }
 
-LJFOLD(MIN MIN KNUM)
-LJFOLD(MAX MAX KNUM)
 LJFOLD(MIN MIN KINT)
 LJFOLD(MAX MAX KINT)
 LJFOLDF(reassoc_minmax_k)
 {
   IRIns *irk = IR(fleft->op2);
-  if (irk->o == IR_KNUM) {
-    lua_Number a = ir_knum(irk)->n;
-    lua_Number y = lj_vm_foldarith(a, knumright, fins->o - IR_ADD);
-    if (a == y)  /* (x o k1) o k2 ==> x o k1, if (k1 o k2) == k1. */
-      return LEFTFOLD;
-    PHIBARRIER(fleft);
-    fins->op1 = fleft->op1;
-    fins->op2 = (IRRef1)lj_ir_knum(J, y);
-    return RETRYFOLD;  /* (x o k1) o k2 ==> x o (k1 o k2) */
-  } else if (irk->o == IR_KINT) {
+  if (irk->o == IR_KINT) {
     int32_t a = irk->i;
     int32_t y = kfold_intop(a, fright->i, fins->o);
     if (a == y)  /* (x o k1) o k2 ==> x o k1, if (k1 o k2) == k1. */
@@ -1852,24 +1848,6 @@ LJFOLDF(reassoc_minmax_k)
   return NEXTFOLD;
 }
 
-LJFOLD(MIN MAX any)
-LJFOLD(MAX MIN any)
-LJFOLDF(reassoc_minmax_left)
-{
-  if (fins->op2 == fleft->op1 || fins->op2 == fleft->op2)
-    return RIGHTFOLD;  /* (b o1 a) o2 b ==> b; (a o1 b) o2 b ==> b */
-  return NEXTFOLD;
-}
-
-LJFOLD(MIN any MAX)
-LJFOLD(MAX any MIN)
-LJFOLDF(reassoc_minmax_right)
-{
-  if (fins->op1 == fright->op1 || fins->op1 == fright->op2)
-    return LEFTFOLD;  /* a o2 (a o1 b) ==> a; a o2 (b o1 a) ==> a */
-  return NEXTFOLD;
-}
-
 /* -- Array bounds check elimination -------------------------------------- */
 
 /* Eliminate ABC across PHIs to handle t[i-1] forwarding case.
@@ -1995,8 +1973,6 @@ LJFOLDF(comm_comp)
 
 LJFOLD(BAND any any)
 LJFOLD(BOR any any)
-LJFOLD(MIN any any)
-LJFOLD(MAX any any)
 LJFOLDF(comm_dup)
 {
   if (fins->op1 == fins->op2)  /* x o x ==> x */
@@ -2004,6 +1980,15 @@ LJFOLDF(comm_dup)
   return fold_comm_swap(J);
 }
 
+LJFOLD(MIN any any)
+LJFOLD(MAX any any)
+LJFOLDF(comm_dup_minmax)
+{
+  if (fins->op1 == fins->op2)  /* x o x ==> x */
+    return LEFTFOLD;
+  return NEXTFOLD;
+}
+
 LJFOLD(BXOR any any)
 LJFOLDF(comm_bxor)
 {
diff --git a/src/lj_vmmath.c b/src/lj_vmmath.c
index 2a41bcaa..e89405d7 100644
--- a/src/lj_vmmath.c
+++ b/src/lj_vmmath.c
@@ -50,8 +50,8 @@ double lj_vm_foldarith(double x, double y, int op)
 #if LJ_HASJIT
   case IR_ATAN2 - IR_ADD: return atan2(x, y); break;
   case IR_LDEXP - IR_ADD: return ldexp(x, (int)y); break;
-  case IR_MIN - IR_ADD: return x > y ? y : x; break;
-  case IR_MAX - IR_ADD: return x < y ? y : x; break;
+  case IR_MIN - IR_ADD: return x < y ? x : y; break;
+  case IR_MAX - IR_ADD: return x > y ? x : y; break;
 #endif
   default: return x;
   }
diff --git a/src/vm_arm.dasc b/src/vm_arm.dasc
index 5bbdbbff..013688fb 100644
--- a/src/vm_arm.dasc
+++ b/src/vm_arm.dasc
@@ -1716,8 +1716,8 @@ static void build_subroutines(BuildCtx *ctx)
   |.endif
   |.endmacro
   |
-  |  math_minmax math_min, gt, hi
-  |  math_minmax math_max, lt, lo
+  |  math_minmax math_min, gt, pl
+  |  math_minmax math_max, lt, le
   |
   |//-- String library -----------------------------------------------------
   |
diff --git a/src/vm_arm64.dasc b/src/vm_arm64.dasc
index 62946373..c157696c 100644
--- a/src/vm_arm64.dasc
+++ b/src/vm_arm64.dasc
@@ -1489,8 +1489,8 @@ static void build_subroutines(BuildCtx *ctx)
   |  b <6
   |.endmacro
   |
-  |  math_minmax math_min, gt, hi
-  |  math_minmax math_max, lt, lo
+  |  math_minmax math_min, gt, pl
+  |  math_minmax math_max, lt, le
   |
   |//-- String library -----------------------------------------------------
   |
diff --git a/src/vm_mips.dasc b/src/vm_mips.dasc
index 37506139..0c84c13b 100644
--- a/src/vm_mips.dasc
+++ b/src/vm_mips.dasc
@@ -1768,7 +1768,7 @@ static void build_subroutines(BuildCtx *ctx)
   |  b ->fff_res
   |.  li RD, (2+1)*8
   |
-  |.macro math_minmax, name, intins, fpins
+  |.macro math_minmax, name, intins, ismax
   |  .ffunc_1 name
   |  addu TMP3, BASE, NARGS8:RC
   |  bne SFARG1HI, TISNUM, >5
@@ -1822,13 +1822,21 @@ static void build_subroutines(BuildCtx *ctx)
   |.endif
   |7:
   |.if FPU
+  |.if ismax
+  |  c.olt.d FARG1, FRET1
+  |.else
   |  c.olt.d FRET1, FARG1
-  |  fpins FRET1, FARG1
+  |.endif
+  |  movf.d FRET1, FARG1
+  |.else
+  |.if ismax
+  |  bal ->vm_sfcmpogt
   |.else
   |  bal ->vm_sfcmpolt
+  |.endif
   |.  nop
-  |  intins SFARG1LO, SFARG2LO, CRET1
-  |  intins SFARG1HI, SFARG2HI, CRET1
+  |  movz SFARG1LO, SFARG2LO, CRET1
+  |  movz SFARG1HI, SFARG2HI, CRET1
   |.endif
   |  b <6
   |.  addiu TMP2, TMP2, 8
@@ -1849,8 +1857,8 @@ static void build_subroutines(BuildCtx *ctx)
   |
   |.endmacro
   |
-  |  math_minmax math_min, movz, movf.d
-  |  math_minmax math_max, movn, movt.d
+  |  math_minmax math_min, movz, 0
+  |  math_minmax math_max, movn, 1
   |
   |//-- String library -----------------------------------------------------
   |
@@ -2692,6 +2700,43 @@ static void build_subroutines(BuildCtx *ctx)
   |.  move CRET1, CRET2
   |.endif
   |
+  |->vm_sfcmpogt:
+  |.if not FPU
+  |  sll AT, SFARG2HI, 1
+  |  sll TMP0, SFARG1HI, 1
+  |  or CRET1, SFARG2LO, SFARG1LO
+  |  or TMP1, AT, TMP0
+  |  or TMP1, TMP1, CRET1
+  |  beqz TMP1, >8                      // Both args +-0: return 0.
+  |.  sltu CRET1, r0, SFARG2LO
+  |  lui TMP1, 0xffe0
+  |  addu AT, AT, CRET1
+  |   sltu CRET1, r0, SFARG1LO
+  |  sltu AT, TMP1, AT
+  |   addu TMP0, TMP0, CRET1
+  |   sltu TMP0, TMP1, TMP0
+  |  or TMP1, AT, TMP0
+  |  bnez TMP1, >9                      // Either arg is NaN: return 0 or 1;
+  |.  and AT, SFARG2HI, SFARG1HI
+  |  bltz AT, >5                        // Both args negative?
+  |.  nop
+  |  beq SFARG2HI, SFARG1HI, >8
+  |.  sltu CRET1, SFARG2LO, SFARG1LO
+  |  jr ra
+  |.  slt CRET1, SFARG2HI, SFARG1HI
+  |5:  // Swap conditions if both operands are negative.
+  |  beq SFARG2HI, SFARG1HI, >8
+  |.  sltu CRET1, SFARG1LO, SFARG2LO
+  |  jr ra
+  |.  slt CRET1, SFARG1HI, SFARG2HI
+  |8:
+  |  jr ra
+  |.  nop
+  |9:
+  |  jr ra
+  |.  li CRET1, 0
+  |.endif
+  |
   |// Soft-float comparison. Equivalent to c.ole.d a, b or c.ole.d b, a.
   |// Input: SFARG*, TMP3. Output: CRET1. Temporaries: AT, TMP0, TMP1.
   |->vm_sfcmpolex:
@@ -2734,24 +2779,24 @@ static void build_subroutines(BuildCtx *ctx)
   |.  li CRET1, 0
   |.endif
   |
-  |.macro sfmin_max, name, intins
+  |.macro sfmin_max, name, fpcall
   |->vm_sf .. name:
   |.if JIT and not FPU
   |  move TMP2, ra
-  |  bal ->vm_sfcmpolt
+  |  bal ->fpcall
   |.  nop
   |  move TMP0, CRET1
   |  move SFRETHI, SFARG1HI
   |   move SFRETLO, SFARG1LO
   |  move ra, TMP2
-  |  intins SFRETHI, SFARG2HI, TMP0
+  |  movz SFRETHI, SFARG2HI, TMP0
   |  jr ra
-  |.  intins SFRETLO, SFARG2LO, TMP0
+  |.  movz SFRETLO, SFARG2LO, TMP0
   |.endif
   |.endmacro
   |
-  |  sfmin_max min, movz
-  |  sfmin_max max, movn
+  |  sfmin_max min, vm_sfcmpolt
+  |  sfmin_max max, vm_sfcmpogt
   |
   |//-----------------------------------------------------------------------
   |//-- Miscellaneous functions --------------------------------------------
diff --git a/src/vm_mips64.dasc b/src/vm_mips64.dasc
index 4ae19b7d..dac143a4 100644
--- a/src/vm_mips64.dasc
+++ b/src/vm_mips64.dasc
@@ -1852,18 +1852,26 @@ static void build_subroutines(BuildCtx *ctx)
   |.if MIPSR6
   |  fpins FRET1, FRET1, FARG1
   |.else
+  |.if fpins  // ismax
+  |  c.olt.d FARG1, FRET1
+  |.else
   |  c.olt.d FRET1, FARG1
-  |  fpins FRET1, FARG1
   |.endif
+  |  movf.d FRET1, FARG1
+  |.endif
+  |.else
+  |.if fpins  // ismax
+  |  bal ->vm_sfcmpogt
   |.else
   |  bal ->vm_sfcmpolt
+  |.endif
   |.  nop
   |.if MIPSR6
-  |  intins AT, CARG2, CRET1
-  |  intinsc CARG1, CARG1, CRET1
+  |  seleqz AT, CARG2, CRET1
+  |  selnez CARG1, CARG1, CRET1
   |  or CARG1, CARG1, AT
   |.else
-  |  intins CARG1, CARG2, CRET1
+  |  movz CARG1, CARG2, CRET1
   |.endif
   |.endif
   |  b <6
@@ -1889,8 +1897,8 @@ static void build_subroutines(BuildCtx *ctx)
   |  math_minmax math_min, seleqz, selnez, min.d
   |  math_minmax math_max, selnez, seleqz, max.d
   |.else
-  |  math_minmax math_min, movz, _, movf.d
-  |  math_minmax math_max, movn, _, movt.d
+  |  math_minmax math_min, movz, _, 0
+  |  math_minmax math_max, movn, _, 1
   |.endif
   |
   |//-- String library -----------------------------------------------------
@@ -2108,7 +2116,6 @@ static void build_subroutines(BuildCtx *ctx)
   |  dsllv CRET2, CRET2, TMP0   // Integer check.
   |  sextw AT, CRET1
   |  xor AT, CRET1, AT          // Range check.
-  |  jr ra
   |.if MIPSR6
   |  seleqz AT, AT, CRET2
   |  selnez CRET2, CRET2, CRET2
@@ -2809,6 +2816,34 @@ static void build_subroutines(BuildCtx *ctx)
   |.  move CRET1, CRET2
   |.endif
   |
+  |->vm_sfcmpogt:
+  |.if not FPU
+  |  dsll AT, CARG2, 1
+  |  dsll TMP0, CARG1, 1
+  |  or TMP1, AT, TMP0
+  |  beqz TMP1, >8                      // Both args +-0: return 0.
+  |.  lui TMP1, 0xffe0
+  |  dsll TMP1, TMP1, 32
+  |   sltu AT, TMP1, AT
+  |   sltu TMP0, TMP1, TMP0
+  |  or TMP1, AT, TMP0
+  |  bnez TMP1, >9                      // Either arg is NaN: return 0 or 1;
+  |.  and AT, CARG2, CARG1
+  |  bltz AT, >5                        // Both args negative?
+  |.  nop
+  |  jr ra
+  |.  slt CRET1, CARG2, CARG1
+  |5:  // Swap conditions if both operands are negative.
+  |  jr ra
+  |.  slt CRET1, CARG1, CARG2
+  |8:
+  |  jr ra
+  |.  li CRET1, 0
+  |9:
+  |  jr ra
+  |.  li CRET1, 0
+  |.endif
+  |
   |// Soft-float comparison. Equivalent to c.ole.d a, b or c.ole.d b, a.
   |// Input: CARG1, CARG2, TMP3. Output: CRET1. Temporaries: AT, TMP0, TMP1.
   |->vm_sfcmpolex:
@@ -2840,34 +2875,29 @@ static void build_subroutines(BuildCtx *ctx)
   |.  li CRET1, 0
   |.endif
   |
-  |.macro sfmin_max, name, intins, intinsc
+  |.macro sfmin_max, name, fpcall
   |->vm_sf .. name:
   |.if JIT and not FPU
   |  move TMP2, ra
-  |  bal ->vm_sfcmpolt
+  |  bal ->fpcall
   |.  nop
   |  move ra, TMP2
   |  move TMP0, CRET1
   |  move CRET1, CARG1
   |.if MIPSR6
-  |  intins CRET1, CRET1, TMP0
-  |  intinsc TMP0, CARG2, TMP0
+  |  selnez CRET1, CRET1, TMP0
+  |  seleqz TMP0, CARG2, TMP0
   |  jr ra
   |.  or CRET1, CRET1, TMP0
   |.else
   |  jr ra
-  |.  intins CRET1, CARG2, TMP0
+  |.  movz CRET1, CARG2, TMP0
   |.endif
   |.endif
   |.endmacro
   |
-  |.if MIPSR6
-  |  sfmin_max min, selnez, seleqz
-  |  sfmin_max max, seleqz, selnez
-  |.else
-  |  sfmin_max min, movz, _
-  |  sfmin_max max, movn, _
-  |.endif
+  |  sfmin_max min, vm_sfcmpolt
+  |  sfmin_max max, vm_sfcmpogt
   |
   |//-----------------------------------------------------------------------
   |//-- Miscellaneous functions --------------------------------------------
diff --git a/src/vm_ppc.dasc b/src/vm_ppc.dasc
index a66e30b5..7a2d321e 100644
--- a/src/vm_ppc.dasc
+++ b/src/vm_ppc.dasc
@@ -2309,12 +2309,12 @@ static void build_subroutines(BuildCtx *ctx)
   |6:
   |   addi SAVE0, SAVE0, 8
   |.if FPU
-  |  fsub f0, FARG1, FARG2
   |.if ismax
-  |  fsel FARG1, f0, FARG1, FARG2
+  |  fsub f0, FARG1, FARG2
   |.else
-  |  fsel FARG1, f0, FARG2, FARG1
+  |  fsub f0, FARG2, FARG1
   |.endif
+  |  fsel FARG1, f0, FARG1, FARG2
   |.else
   |  stw CARG1, SFSAVE_1
   |  stw CARG2, SFSAVE_2
@@ -2354,13 +2354,13 @@ static void build_subroutines(BuildCtx *ctx)
   |   checknum CARG2
   |  bge cr1, ->fff_resn
   |   bge ->fff_fallback
-  |  fsub f0, FARG1, FARG2
-  |   addi TMP1, TMP1, 8
   |.if ismax
-  |  fsel FARG1, f0, FARG1, FARG2
+  |  fsub f0, FARG1, FARG2
   |.else
-  |  fsel FARG1, f0, FARG2, FARG1
+  |  fsub f0, FARG2, FARG1
   |.endif
+  |   addi TMP1, TMP1, 8
+  |  fsel FARG1, f0, FARG1, FARG2
   |  b <1
   |.endif
   |.endmacro
diff --git a/src/vm_x64.dasc b/src/vm_x64.dasc
index a5749b17..c714f4c7 100644
--- a/src/vm_x64.dasc
+++ b/src/vm_x64.dasc
@@ -1840,7 +1840,7 @@ static void build_subroutines(BuildCtx *ctx)
   |  jmp ->fff_res
   |
   |.macro math_minmax, name, cmovop, sseop
-  |  .ffunc name
+  |  .ffunc_1 name
   |  mov RAd, 2
   |.if DUALNUM
   |  mov RB, [BASE]
diff --git a/src/vm_x86.dasc b/src/vm_x86.dasc
index 1965b06b..c3999e7c 100644
--- a/src/vm_x86.dasc
+++ b/src/vm_x86.dasc
@@ -2233,7 +2233,7 @@ static void build_subroutines(BuildCtx *ctx)
   |  xorps xmm4, xmm4; jmp <1                   // Return +-Inf and +-0.
   |
   |.macro math_minmax, name, cmovop, sseop
-  |  .ffunc name
+  |  .ffunc_1 name
   |  mov RA, 2
   |  cmp dword [BASE+4], LJ_TISNUM
   |.if DUALNUM