Merge branch 'master' into v2.1

diff --cc dynasm/dasm_mips.h
Simple merge

diff --cc dynasm/dasm_ppc.h
Simple merge

diff --cc dynasm/dasm_x86.h
index f598da9afddf228ad18f43011606d167424283e5,8ae911df185c7b484e088289a0a4f61da835a333..dc14d8837ea84f414da8221a82b54dc31e266e87
--- 1/dynasm/dasm_x86.h
--- 2/dynasm/dasm_x86.h
+++ b/dynasm/dasm_x86.h
@@@ -193,8 -193,9 +193,9 @@@ void dasm_put(Dst_DECL, int start, ...
        b[pos++] = n;
        switch (action) {
        case DASM_DISP:
 -      if (n == 0) { if ((mrm&7) == 4) mrm = p[-2]; if ((mrm&7) != 5) break; }
 +      if (n == 0) { if (mrm < 0) mrm = p[-2]; if ((mrm&7) != 5) break; }
-       case DASM_IMM_DB: if (((n+128)&-256) == 0) goto ob;
+       /* fallthrough */
+       case DASM_IMM_DB: if (((n+128)&-256) == 0) goto ob; /* fallthrough */
        case DASM_REL_A: /* Assumes ptrdiff_t is int. !x64 */
        case DASM_IMM_D: ofs += 4; break;
        case DASM_IMM_S: CK(((n+128)&-256) == 0, RANGE_I); goto ob;
@@@ -203,17 -204,11 +204,17 @@@
        case DASM_IMM_W: CK((n&-65536) == 0, RANGE_I); ofs += 2; break;
        case DASM_SPACE: p++; ofs += n; break;
        case DASM_SETLABEL: b[pos-2] = -0x40000000; break;  /* Neg. label ofs. */
 -      case DASM_VREG: CK((n&-8) == 0 && (n != 4 || (*p&1) == 0), RANGE_VREG);
 -      if (*p++ == 1 && *p == DASM_DISP) mrm = n;
 +      case DASM_VREG: CK((n&-16) == 0 && (n != 4 || (*p>>5) != 2), RANGE_VREG);
 +      if (*p < 0x40 && p[1] == DASM_DISP) mrm = n;
 +      if (*p < 0x20 && (n&7) == 4) ofs++;
 +      switch ((*p++ >> 3) & 3) {
-       case 3: n |= b[pos-3];
-       case 2: n |= b[pos-2];
++      case 3: n |= b[pos-3]; /* fallthrough */
++      case 2: n |= b[pos-2]; /* fallthrough */
 +      case 1: if (n <= 7) { b[pos-1] |= 0x10; ofs--; }
 +      }
        continue;
        }
 -      mrm = 4;
 +      mrm = -1;
      } else {
        int *pl, n;
        switch (action) {
@@@ -395,32 -394,15 +400,35 @@@ int dasm_encode(Dst_DECL, void *buffer
        case DASM_IMM_DB: if (((n+128)&-256) == 0) {
            db: if (!mark) mark = cp; mark[-2] += 2; mark = NULL; goto wb;
          } else mark = NULL;
+         /* fallthrough */
        case DASM_IMM_D: wd: dasmd(n); break;
        case DASM_IMM_WB: if (((n+128)&-256) == 0) goto db; else mark = NULL;
+         /* fallthrough */
        case DASM_IMM_W: dasmw(n); break;
 -      case DASM_VREG: { int t = *p++; if (t >= 2) n<<=3; cp[-1] |= n; break; }
 +      case DASM_VREG: {
 +        int t = *p++;
 +        unsigned char *ex = cp - (t&7);
 +        if ((n & 8) && t < 0xa0) {
 +          if (*ex & 0x80) ex[1] ^= 0x20 << (t>>6); else *ex ^= 1 << (t>>6);
 +          n &= 7;
 +        } else if (n & 0x10) {
 +          if (*ex & 0x80) {
 +            *ex = 0xc5; ex[1] = (ex[1] & 0x80) | ex[2]; ex += 2;
 +          }
 +          while (++ex < cp) ex[-1] = *ex;
 +          if (mark) mark--;
 +          cp--;
 +          n &= 7;
 +        }
 +        if (t >= 0xc0) n <<= 4;
 +        else if (t >= 0x40) n <<= 3;
 +        else if (n == 4 && t < 0x20) { cp[-1] ^= n; *cp++ = 0x20; }
 +        cp[-1] ^= n;
 +        break;
 +      }
        case DASM_REL_LG: p++; if (n >= 0) goto rel_pc;
          b++; n = (int)(ptrdiff_t)D->globals[-n];
+         /* fallthrough */
        case DASM_REL_A: rel_a:
          n -= (unsigned int)(ptrdiff_t)(cp+4); goto wd; /* !x64 */
        case DASM_REL_PC: rel_pc: {

diff --cc src/lj_asm.c
index 5f83779e143e922b565a8f9a85b40496e65306ca,dd7186f6de2810d68084d87ca751afafc4508780..992dcf59bf8e28f9fd8d036db9f7640f59a4b111
--- 1/src/lj_asm.c
--- 2/src/lj_asm.c
+++ b/src/lj_asm.c
@@@ -2133,20 -1722,11 +2133,23 @@@ static void asm_setup_regsp(ASMState *a
        /* fallthrough */
  #endif
      /* C calls evict all scratch regs and return results in RID_RET. */
 -    case IR_SNEW: case IR_XSNEW: case IR_NEWREF:
 +    case IR_SNEW: case IR_XSNEW: case IR_NEWREF: case IR_BUFPUT:
        if (REGARG_NUMGPR < 3 && as->evenspill < 3)
        as->evenspill = 3;  /* lj_str_new and lj_tab_newkey need 3 args. */
 -    case IR_TNEW: case IR_TDUP: case IR_CNEW: case IR_CNEWI: case IR_TOSTR:
 +#if LJ_TARGET_X86 && LJ_HASFFI
 +      if (0) {
 +    case IR_CNEW:
 +      if (ir->op2 != REF_NIL && as->evenspill < 4)
 +        as->evenspill = 4;  /* lj_cdata_newv needs 4 args. */
 +      }
++      /* fallthrough */
 +#else
++      /* fallthrough */
 +    case IR_CNEW:
 +#endif
+       /* fallthrough */
 +    case IR_TNEW: case IR_TDUP: case IR_CNEWI: case IR_TOSTR:
 +    case IR_BUFSTR:
        ir->prev = REGSP_HINT(RID_RET);
        if (inloop)
        as->modset = RSET_SCRATCH;
@@@ -2155,28 -1735,23 +2158,29 @@@
        if (inloop)
        as->modset = RSET_SCRATCH;
        break;
 -#if !LJ_TARGET_X86ORX64 && !LJ_SOFTFP
 -    case IR_ATAN2: case IR_LDEXP:
 +#if !LJ_SOFTFP
 +    case IR_ATAN2:
 +#if LJ_TARGET_X86
 +      if (as->evenspill < 4)  /* Leave room to call atan2(). */
 +      as->evenspill = 4;
 +#endif
 +#if !LJ_TARGET_X86ORX64
 +    case IR_LDEXP:
 +#endif
  #endif
++      /* fallthrough */
      case IR_POW:
        if (!LJ_SOFTFP && irt_isnum(ir->t)) {
 -#if LJ_TARGET_X86ORX64
 -      ir->prev = REGSP_HINT(RID_XMM0);
        if (inloop)
 -        as->modset |= RSET_RANGE(RID_XMM0, RID_XMM1+1)|RID2RSET(RID_EAX);
 +        as->modset |= RSET_SCRATCH;
 +#if LJ_TARGET_X86
 +      break;
  #else
        ir->prev = REGSP_HINT(RID_FPRET);
 -      if (inloop)
 -        as->modset |= RSET_SCRATCH;
 -#endif
        continue;
 +#endif
        }
-       /* fallthrough for integer POW */
+       /* fallthrough */ /* for integer POW */
      case IR_DIV: case IR_MOD:
        if (!irt_isnum(ir->t)) {
        ir->prev = REGSP_HINT(RID_RET);
@@@ -2210,10 -1786,7 +2214,11 @@@
  #endif
  #if LJ_TARGET_X86ORX64
      /* Non-constant shift counts need to be in RID_ECX on x86/x64. */
 -    case IR_BSHL: case IR_BSHR: case IR_BSAR: case IR_BROL: case IR_BROR:
 +    case IR_BSHL: case IR_BSHR: case IR_BSAR:
 +      if ((as->flags & JIT_F_BMI2))  /* Except if BMI2 is available. */
 +      break;
++      /* fallthrough */
 +    case IR_BROL: case IR_BROR:
        if (!irref_isk(ir->op2) && !ra_hashint(IR(ir->op2)->r)) {
        IR(ir->op2)->r = REGSP_HINT(RID_ECX);
        if (inloop)

diff --cc src/lj_cparse.c
Simple merge

diff --cc src/lj_err.c
index b6be357e79182bf821d06b838133bed81b3b86f7,13a1ded7ce8b6a7858ee2ecd6dca972935cd9063..abf176e5c93f6ad1a3fd59dda16b604b3d13ebe1
--- 1/src/lj_err.c
--- 2/src/lj_err.c
+++ b/src/lj_err.c
@@@ -148,8 -149,11 +148,9 @@@ static void *err_unwind(lua_State *L, v
        }
        return cf;
      case FRAME_CONT:  /* Continuation frame. */
 -#if LJ_HASFFI
 -      if ((frame-1)->u32.lo == LJ_CONT_FFI_CALLBACK)
 +      if (frame_iscont_fficb(frame))
        goto unwind_c;
 -#endif
+       /* fallthrough */
      case FRAME_VARG:  /* Vararg frame. */
        frame = frame_prevd(frame);
        break;

diff --cc src/lj_opt_sink.c
Simple merge

diff --cc src/lj_parse.c
Simple merge

diff --cc src/luajit.c
Simple merge