summaryrefslogtreecommitdiff
path: root/libs/luajit-cmake/luajit/src/vm_mips.dasc
diff options
context:
space:
mode:
authorsanine <sanine.not@pm.me>2023-03-11 15:58:20 -0600
committersanine <sanine.not@pm.me>2023-03-11 15:58:20 -0600
commitebc50b387ab209c9f9a0d92e340ac293d5697274 (patch)
treeea8c8b3677a18c994d2b9d33dbef3461dcf18113 /libs/luajit-cmake/luajit/src/vm_mips.dasc
parentc2329b4c8258baa9429c77566c9def97d00e96d7 (diff)
build & link with luajit instead of lua5.1
Diffstat (limited to 'libs/luajit-cmake/luajit/src/vm_mips.dasc')
-rw-r--r--libs/luajit-cmake/luajit/src/vm_mips.dasc5392
1 files changed, 5392 insertions, 0 deletions
diff --git a/libs/luajit-cmake/luajit/src/vm_mips.dasc b/libs/luajit-cmake/luajit/src/vm_mips.dasc
new file mode 100644
index 0000000..34645bf
--- /dev/null
+++ b/libs/luajit-cmake/luajit/src/vm_mips.dasc
@@ -0,0 +1,5392 @@
+|// Low-level VM code for MIPS CPUs.
+|// Bytecode interpreter, fast functions and helper functions.
+|// Copyright (C) 2005-2022 Mike Pall. See Copyright Notice in luajit.h
+|//
+|// MIPS soft-float support contributed by Djordje Kovacevic and
+|// Stefan Pejic from RT-RK.com, sponsored by Cisco Systems, Inc.
+|
+|.arch mips
+|.section code_op, code_sub
+|
+|.actionlist build_actionlist
+|.globals GLOB_
+|.globalnames globnames
+|.externnames extnames
+|
+|// Note: The ragged indentation of the instructions is intentional.
+|// The starting columns indicate data dependencies.
+|
+|//-----------------------------------------------------------------------
+|
+|// Fixed register assignments for the interpreter.
+|// Don't use: r0 = 0, r26/r27 = reserved, r28 = gp, r29 = sp, r31 = ra
+|
+|.macro .FPU, a, b
+|.if FPU
+| a, b
+|.endif
+|.endmacro
+|
+|// The following must be C callee-save (but BASE is often refetched).
+|.define BASE, r16 // Base of current Lua stack frame.
+|.define KBASE, r17 // Constants of current Lua function.
+|.define PC, r18 // Next PC.
+|.define DISPATCH, r19 // Opcode dispatch table.
+|.define LREG, r20 // Register holding lua_State (also in SAVE_L).
+|.define MULTRES, r21 // Size of multi-result: (nresults+1)*8.
+|
+|.define JGL, r30 // On-trace: global_State + 32768.
+|
+|// Constants for type-comparisons, stores and conversions. C callee-save.
+|.define TISNUM, r22
+|.define TISNIL, r30
+|.if FPU
+|.define TOBIT, f30 // 2^52 + 2^51.
+|.endif
+|
+|// The following temporaries are not saved across C calls, except for RA.
+|.define RA, r23 // Callee-save.
+|.define RB, r8
+|.define RC, r9
+|.define RD, r10
+|.define INS, r11
+|
+|.define AT, r1 // Assembler temporary.
+|.define TMP0, r12
+|.define TMP1, r13
+|.define TMP2, r14
+|.define TMP3, r15
+|
+|// MIPS o32 calling convention.
+|.define CFUNCADDR, r25
+|.define CARG1, r4
+|.define CARG2, r5
+|.define CARG3, r6
+|.define CARG4, r7
+|
+|.define CRET1, r2
+|.define CRET2, r3
+|
+|.if ENDIAN_LE
+|.define SFRETLO, CRET1
+|.define SFRETHI, CRET2
+|.define SFARG1LO, CARG1
+|.define SFARG1HI, CARG2
+|.define SFARG2LO, CARG3
+|.define SFARG2HI, CARG4
+|.else
+|.define SFRETLO, CRET2
+|.define SFRETHI, CRET1
+|.define SFARG1LO, CARG2
+|.define SFARG1HI, CARG1
+|.define SFARG2LO, CARG4
+|.define SFARG2HI, CARG3
+|.endif
+|
+|.if FPU
+|.define FARG1, f12
+|.define FARG2, f14
+|
+|.define FRET1, f0
+|.define FRET2, f2
+|.endif
+|
+|// Stack layout while in interpreter. Must match with lj_frame.h.
+|.if FPU // MIPS32 hard-float.
+|
+|.define CFRAME_SPACE, 112 // Delta for sp.
+|
+|.define SAVE_ERRF, 124(sp) // 32 bit C frame info.
+|.define SAVE_NRES, 120(sp)
+|.define SAVE_CFRAME, 116(sp)
+|.define SAVE_L, 112(sp)
+|//----- 8 byte aligned, ^^^^ 16 byte register save area, owned by interpreter.
+|.define SAVE_GPR_, 72 // .. 72+10*4: 32 bit GPR saves.
+|.define SAVE_FPR_, 24 // .. 24+6*8: 64 bit FPR saves.
+|
+|.else // MIPS32 soft-float
+|
+|.define CFRAME_SPACE, 64 // Delta for sp.
+|
+|.define SAVE_ERRF, 76(sp) // 32 bit C frame info.
+|.define SAVE_NRES, 72(sp)
+|.define SAVE_CFRAME, 68(sp)
+|.define SAVE_L, 64(sp)
+|//----- 8 byte aligned, ^^^^ 16 byte register save area, owned by interpreter.
+|.define SAVE_GPR_, 24 // .. 24+10*4: 32 bit GPR saves.
+|
+|.endif
+|
+|.define SAVE_PC, 20(sp)
+|.define ARG5, 16(sp)
+|.define CSAVE_4, 12(sp)
+|.define CSAVE_3, 8(sp)
+|.define CSAVE_2, 4(sp)
+|.define CSAVE_1, 0(sp)
+|//----- 8 byte aligned, ^^^^ 16 byte register save area, owned by callee.
+|
+|.define ARG5_OFS, 16
+|.define SAVE_MULTRES, ARG5
+|
+|//-----------------------------------------------------------------------
+|
+|.macro saveregs
+| addiu sp, sp, -CFRAME_SPACE
+| sw ra, SAVE_GPR_+9*4(sp)
+| sw r30, SAVE_GPR_+8*4(sp)
+| .FPU sdc1 f30, SAVE_FPR_+5*8(sp)
+| sw r23, SAVE_GPR_+7*4(sp)
+| sw r22, SAVE_GPR_+6*4(sp)
+| .FPU sdc1 f28, SAVE_FPR_+4*8(sp)
+| sw r21, SAVE_GPR_+5*4(sp)
+| sw r20, SAVE_GPR_+4*4(sp)
+| .FPU sdc1 f26, SAVE_FPR_+3*8(sp)
+| sw r19, SAVE_GPR_+3*4(sp)
+| sw r18, SAVE_GPR_+2*4(sp)
+| .FPU sdc1 f24, SAVE_FPR_+2*8(sp)
+| sw r17, SAVE_GPR_+1*4(sp)
+| sw r16, SAVE_GPR_+0*4(sp)
+| .FPU sdc1 f22, SAVE_FPR_+1*8(sp)
+| .FPU sdc1 f20, SAVE_FPR_+0*8(sp)
+|.endmacro
+|
+|.macro restoreregs_ret
+| lw ra, SAVE_GPR_+9*4(sp)
+| lw r30, SAVE_GPR_+8*4(sp)
+| .FPU ldc1 f30, SAVE_FPR_+5*8(sp)
+| lw r23, SAVE_GPR_+7*4(sp)
+| lw r22, SAVE_GPR_+6*4(sp)
+| .FPU ldc1 f28, SAVE_FPR_+4*8(sp)
+| lw r21, SAVE_GPR_+5*4(sp)
+| lw r20, SAVE_GPR_+4*4(sp)
+| .FPU ldc1 f26, SAVE_FPR_+3*8(sp)
+| lw r19, SAVE_GPR_+3*4(sp)
+| lw r18, SAVE_GPR_+2*4(sp)
+| .FPU ldc1 f24, SAVE_FPR_+2*8(sp)
+| lw r17, SAVE_GPR_+1*4(sp)
+| lw r16, SAVE_GPR_+0*4(sp)
+| .FPU ldc1 f22, SAVE_FPR_+1*8(sp)
+| .FPU ldc1 f20, SAVE_FPR_+0*8(sp)
+| jr ra
+| addiu sp, sp, CFRAME_SPACE
+|.endmacro
+|
+|// Type definitions. Some of these are only used for documentation.
+|.type L, lua_State, LREG
+|.type GL, global_State
+|.type TVALUE, TValue
+|.type GCOBJ, GCobj
+|.type STR, GCstr
+|.type TAB, GCtab
+|.type LFUNC, GCfuncL
+|.type CFUNC, GCfuncC
+|.type PROTO, GCproto
+|.type UPVAL, GCupval
+|.type NODE, Node
+|.type NARGS8, int
+|.type TRACE, GCtrace
+|.type SBUF, SBuf
+|
+|//-----------------------------------------------------------------------
+|
+|// Trap for not-yet-implemented parts.
+|.macro NYI; .long 0xec1cf0f0; .endmacro
+|
+|// Macros to mark delay slots.
+|.macro ., a; a; .endmacro
+|.macro ., a,b; a,b; .endmacro
+|.macro ., a,b,c; a,b,c; .endmacro
+|
+|//-----------------------------------------------------------------------
+|
+|// Endian-specific defines.
+|.if ENDIAN_LE
+|.define FRAME_PC, -4
+|.define FRAME_FUNC, -8
+|.define HI, 4
+|.define LO, 0
+|.define OFS_RD, 2
+|.define OFS_RA, 1
+|.define OFS_OP, 0
+|.else
+|.define FRAME_PC, -8
+|.define FRAME_FUNC, -4
+|.define HI, 0
+|.define LO, 4
+|.define OFS_RD, 0
+|.define OFS_RA, 2
+|.define OFS_OP, 3
+|.endif
+|
+|// Instruction decode.
+|.macro decode_OP1, dst, ins; andi dst, ins, 0xff; .endmacro
+|.macro decode_OP4a, dst, ins; andi dst, ins, 0xff; .endmacro
+|.macro decode_OP4b, dst; sll dst, dst, 2; .endmacro
+|.macro decode_RC4a, dst, ins; srl dst, ins, 14; .endmacro
+|.macro decode_RC4b, dst; andi dst, dst, 0x3fc; .endmacro
+|.macro decode_RD4b, dst; sll dst, dst, 2; .endmacro
+|.macro decode_RA8a, dst, ins; srl dst, ins, 5; .endmacro
+|.macro decode_RA8b, dst; andi dst, dst, 0x7f8; .endmacro
+|.macro decode_RB8a, dst, ins; srl dst, ins, 21; .endmacro
+|.macro decode_RB8b, dst; andi dst, dst, 0x7f8; .endmacro
+|.macro decode_RD8a, dst, ins; srl dst, ins, 16; .endmacro
+|.macro decode_RD8b, dst; sll dst, dst, 3; .endmacro
+|.macro decode_RDtoRC8, dst, src; andi dst, src, 0x7f8; .endmacro
+|
+|// Instruction fetch.
+|.macro ins_NEXT1
+| lw INS, 0(PC)
+| addiu PC, PC, 4
+|.endmacro
+|// Instruction decode+dispatch.
+|.macro ins_NEXT2
+| decode_OP4a TMP1, INS
+| decode_OP4b TMP1
+| addu TMP0, DISPATCH, TMP1
+| decode_RD8a RD, INS
+| lw AT, 0(TMP0)
+| decode_RA8a RA, INS
+| decode_RD8b RD
+| jr AT
+| decode_RA8b RA
+|.endmacro
+|.macro ins_NEXT
+| ins_NEXT1
+| ins_NEXT2
+|.endmacro
+|
+|// Instruction footer.
+|.if 1
+| // Replicated dispatch. Less unpredictable branches, but higher I-Cache use.
+| .define ins_next, ins_NEXT
+| .define ins_next_, ins_NEXT
+| .define ins_next1, ins_NEXT1
+| .define ins_next2, ins_NEXT2
+|.else
+| // Common dispatch. Lower I-Cache use, only one (very) unpredictable branch.
+| // Affects only certain kinds of benchmarks (and only with -j off).
+| .macro ins_next
+| b ->ins_next
+| .endmacro
+| .macro ins_next1
+| .endmacro
+| .macro ins_next2
+| b ->ins_next
+| .endmacro
+| .macro ins_next_
+| ->ins_next:
+| ins_NEXT
+| .endmacro
+|.endif
+|
+|// Call decode and dispatch.
+|.macro ins_callt
+| // BASE = new base, RB = LFUNC/CFUNC, RC = nargs*8, FRAME_PC(BASE) = PC
+| lw PC, LFUNC:RB->pc
+| lw INS, 0(PC)
+| addiu PC, PC, 4
+| decode_OP4a TMP1, INS
+| decode_RA8a RA, INS
+| decode_OP4b TMP1
+| decode_RA8b RA
+| addu TMP0, DISPATCH, TMP1
+| lw TMP0, 0(TMP0)
+| jr TMP0
+| addu RA, RA, BASE
+|.endmacro
+|
+|.macro ins_call
+| // BASE = new base, RB = LFUNC/CFUNC, RC = nargs*8, PC = caller PC
+| sw PC, FRAME_PC(BASE)
+| ins_callt
+|.endmacro
+|
+|//-----------------------------------------------------------------------
+|
+|.macro branch_RD
+| srl TMP0, RD, 1
+| lui AT, (-(BCBIAS_J*4 >> 16) & 65535)
+| addu TMP0, TMP0, AT
+| addu PC, PC, TMP0
+|.endmacro
+|
+|// Assumes DISPATCH is relative to GL.
+#define DISPATCH_GL(field) (GG_DISP2G + (int)offsetof(global_State, field))
+#define DISPATCH_J(field) (GG_DISP2J + (int)offsetof(jit_State, field))
+#define GG_DISP2GOT (GG_OFS(got) - GG_OFS(dispatch))
+#define DISPATCH_GOT(name) (GG_DISP2GOT + 4*LJ_GOT_##name)
+|
+#define PC2PROTO(field) ((int)offsetof(GCproto, field)-(int)sizeof(GCproto))
+|
+|.macro load_got, func
+| lw CFUNCADDR, DISPATCH_GOT(func)(DISPATCH)
+|.endmacro
+|// Much faster. Sadly, there's no easy way to force the required code layout.
+|// .macro call_intern, func; bal extern func; .endmacro
+|.macro call_intern, func; jalr CFUNCADDR; .endmacro
+|.macro call_extern; jalr CFUNCADDR; .endmacro
+|.macro jmp_extern; jr CFUNCADDR; .endmacro
+|
+|.macro hotcheck, delta, target
+| srl TMP1, PC, 1
+| andi TMP1, TMP1, 126
+| addu TMP1, TMP1, DISPATCH
+| lhu TMP2, GG_DISP2HOT(TMP1)
+| addiu TMP2, TMP2, -delta
+| bltz TMP2, target
+|. sh TMP2, GG_DISP2HOT(TMP1)
+|.endmacro
+|
+|.macro hotloop
+| hotcheck HOTCOUNT_LOOP, ->vm_hotloop
+|.endmacro
+|
+|.macro hotcall
+| hotcheck HOTCOUNT_CALL, ->vm_hotcall
+|.endmacro
+|
+|// Set current VM state. Uses TMP0.
+|.macro li_vmstate, st; li TMP0, ~LJ_VMST_..st; .endmacro
+|.macro st_vmstate; sw TMP0, DISPATCH_GL(vmstate)(DISPATCH); .endmacro
+|
+|// Move table write barrier back. Overwrites mark and tmp.
+|.macro barrierback, tab, mark, tmp, target
+| lw tmp, DISPATCH_GL(gc.grayagain)(DISPATCH)
+| andi mark, mark, ~LJ_GC_BLACK & 255 // black2gray(tab)
+| sw tab, DISPATCH_GL(gc.grayagain)(DISPATCH)
+| sb mark, tab->marked
+| b target
+|. sw tmp, tab->gclist
+|.endmacro
+|
+|//-----------------------------------------------------------------------
+
+/* Generate subroutines used by opcodes and other parts of the VM. */
+/* The .code_sub section should be last to help static branch prediction. */
+static void build_subroutines(BuildCtx *ctx)
+{
+ |.code_sub
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Return handling ----------------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |->vm_returnp:
+ | // See vm_return. Also: TMP2 = previous base.
+ | andi AT, PC, FRAME_P
+ | beqz AT, ->cont_dispatch
+ |. li TMP1, LJ_TTRUE
+ |
+ | // Return from pcall or xpcall fast func.
+ | lw PC, FRAME_PC(TMP2) // Fetch PC of previous frame.
+ | move BASE, TMP2 // Restore caller base.
+ | // Prepending may overwrite the pcall frame, so do it at the end.
+ | sw TMP1, FRAME_PC(RA) // Prepend true to results.
+ | addiu RA, RA, -8
+ |
+ |->vm_returnc:
+ | addiu RD, RD, 8 // RD = (nresults+1)*8.
+ | andi TMP0, PC, FRAME_TYPE
+ | beqz RD, ->vm_unwind_c_eh
+ |. li CRET1, LUA_YIELD
+ | beqz TMP0, ->BC_RET_Z // Handle regular return to Lua.
+ |. move MULTRES, RD
+ |
+ |->vm_return:
+ | // BASE = base, RA = resultptr, RD/MULTRES = (nresults+1)*8, PC = return
+ | // TMP0 = PC & FRAME_TYPE
+ | li TMP2, -8
+ | xori AT, TMP0, FRAME_C
+ | and TMP2, PC, TMP2
+ | bnez AT, ->vm_returnp
+ |. subu TMP2, BASE, TMP2 // TMP2 = previous base.
+ |
+ | addiu TMP1, RD, -8
+ | sw TMP2, L->base
+ | li_vmstate C
+ | lw TMP2, SAVE_NRES
+ | addiu BASE, BASE, -8
+ | st_vmstate
+ | beqz TMP1, >2
+ |. sll TMP2, TMP2, 3
+ |1:
+ | addiu TMP1, TMP1, -8
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ | addiu RA, RA, 8
+ | sw SFRETHI, HI(BASE)
+ | sw SFRETLO, LO(BASE)
+ | bnez TMP1, <1
+ |. addiu BASE, BASE, 8
+ |
+ |2:
+ | bne TMP2, RD, >6
+ |3:
+ |. sw BASE, L->top // Store new top.
+ |
+ |->vm_leave_cp:
+ | lw TMP0, SAVE_CFRAME // Restore previous C frame.
+ | move CRET1, r0 // Ok return status for vm_pcall.
+ | sw TMP0, L->cframe
+ |
+ |->vm_leave_unw:
+ | restoreregs_ret
+ |
+ |6:
+ | lw TMP1, L->maxstack
+ | slt AT, TMP2, RD
+ | bnez AT, >7 // Less results wanted?
+ | // More results wanted. Check stack size and fill up results with nil.
+ |. slt AT, BASE, TMP1
+ | beqz AT, >8
+ |. nop
+ | sw TISNIL, HI(BASE)
+ | addiu RD, RD, 8
+ | b <2
+ |. addiu BASE, BASE, 8
+ |
+ |7: // Less results wanted.
+ | subu TMP0, RD, TMP2
+ | subu TMP0, BASE, TMP0 // Either keep top or shrink it.
+ | b <3
+ |. movn BASE, TMP0, TMP2 // LUA_MULTRET+1 case?
+ |
+ |8: // Corner case: need to grow stack for filling up results.
+ | // This can happen if:
+ | // - A C function grows the stack (a lot).
+ | // - The GC shrinks the stack in between.
+ | // - A return back from a lua_call() with (high) nresults adjustment.
+ | load_got lj_state_growstack
+ | move MULTRES, RD
+ | srl CARG2, TMP2, 3
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | lw TMP2, SAVE_NRES
+ | lw BASE, L->top // Need the (realloced) L->top in BASE.
+ | move RD, MULTRES
+ | b <2
+ |. sll TMP2, TMP2, 3
+ |
+ |->vm_unwind_c: // Unwind C stack, return from vm_pcall.
+ | // (void *cframe, int errcode)
+ | move sp, CARG1
+ | move CRET1, CARG2
+ |->vm_unwind_c_eh: // Landing pad for external unwinder.
+ | lw L, SAVE_L
+ | li TMP0, ~LJ_VMST_C
+ | lw GL:TMP1, L->glref
+ | b ->vm_leave_unw
+ |. sw TMP0, GL:TMP1->vmstate
+ |
+ |->vm_unwind_ff: // Unwind C stack, return from ff pcall.
+ | // (void *cframe)
+ | li AT, -4
+ | and sp, CARG1, AT
+ |->vm_unwind_ff_eh: // Landing pad for external unwinder.
+ | lw L, SAVE_L
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | li TISNUM, LJ_TISNUM // Setup type comparison constants.
+ | li TISNIL, LJ_TNIL
+ | lw BASE, L->base
+ | lw DISPATCH, L->glref // Setup pointer to dispatch table.
+ | .FPU mtc1 TMP3, TOBIT
+ | li TMP1, LJ_TFALSE
+ | li_vmstate INTERP
+ | lw PC, FRAME_PC(BASE) // Fetch PC of previous frame.
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | addiu RA, BASE, -8 // Results start at BASE-8.
+ | addiu DISPATCH, DISPATCH, GG_G2DISP
+ | sw TMP1, HI(RA) // Prepend false to error message.
+ | st_vmstate
+ | b ->vm_returnc
+ |. li RD, 16 // 2 results: false + error message.
+ |
+ |->vm_unwind_stub: // Jump to exit stub from unwinder.
+ | jr CARG1
+ |. move ra, CARG2
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Grow stack for calls -----------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |->vm_growstack_c: // Grow stack for C function.
+ | b >2
+ |. li CARG2, LUA_MINSTACK
+ |
+ |->vm_growstack_l: // Grow stack for Lua function.
+ | // BASE = new base, RA = BASE+framesize*8, RC = nargs*8, PC = first PC
+ | addu RC, BASE, RC
+ | subu RA, RA, BASE
+ | sw BASE, L->base
+ | addiu PC, PC, 4 // Must point after first instruction.
+ | sw RC, L->top
+ | srl CARG2, RA, 3
+ |2:
+ | // L->base = new base, L->top = top
+ | load_got lj_state_growstack
+ | sw PC, SAVE_PC
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | lw BASE, L->base
+ | lw RC, L->top
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | subu RC, RC, BASE
+ | // BASE = new base, RB = LFUNC/CFUNC, RC = nargs*8, FRAME_PC(BASE) = PC
+ | ins_callt // Just retry the call.
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Entry points into the assembler VM ---------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |->vm_resume: // Setup C frame and resume thread.
+ | // (lua_State *L, TValue *base, int nres1 = 0, ptrdiff_t ef = 0)
+ | saveregs
+ | move L, CARG1
+ | lw DISPATCH, L->glref // Setup pointer to dispatch table.
+ | move BASE, CARG2
+ | lbu TMP1, L->status
+ | sw L, SAVE_L
+ | li PC, FRAME_CP
+ | addiu TMP0, sp, CFRAME_RESUME
+ | addiu DISPATCH, DISPATCH, GG_G2DISP
+ | sw r0, SAVE_NRES
+ | sw r0, SAVE_ERRF
+ | sw CARG1, SAVE_PC // Any value outside of bytecode is ok.
+ | sw r0, SAVE_CFRAME
+ | beqz TMP1, >3
+ |. sw TMP0, L->cframe
+ |
+ | // Resume after yield (like a return).
+ | sw L, DISPATCH_GL(cur_L)(DISPATCH)
+ | move RA, BASE
+ | lw BASE, L->base
+ | li TISNUM, LJ_TISNUM // Setup type comparison constants.
+ | lw TMP1, L->top
+ | lw PC, FRAME_PC(BASE)
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | subu RD, TMP1, BASE
+ | .FPU mtc1 TMP3, TOBIT
+ | sb r0, L->status
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | li_vmstate INTERP
+ | addiu RD, RD, 8
+ | st_vmstate
+ | move MULTRES, RD
+ | andi TMP0, PC, FRAME_TYPE
+ | beqz TMP0, ->BC_RET_Z
+ |. li TISNIL, LJ_TNIL
+ | b ->vm_return
+ |. nop
+ |
+ |->vm_pcall: // Setup protected C frame and enter VM.
+ | // (lua_State *L, TValue *base, int nres1, ptrdiff_t ef)
+ | saveregs
+ | sw CARG4, SAVE_ERRF
+ | b >1
+ |. li PC, FRAME_CP
+ |
+ |->vm_call: // Setup C frame and enter VM.
+ | // (lua_State *L, TValue *base, int nres1)
+ | saveregs
+ | li PC, FRAME_C
+ |
+ |1: // Entry point for vm_pcall above (PC = ftype).
+ | lw TMP1, L:CARG1->cframe
+ | move L, CARG1
+ | sw CARG3, SAVE_NRES
+ | lw DISPATCH, L->glref // Setup pointer to dispatch table.
+ | sw CARG1, SAVE_L
+ | move BASE, CARG2
+ | addiu DISPATCH, DISPATCH, GG_G2DISP
+ | sw CARG1, SAVE_PC // Any value outside of bytecode is ok.
+ | sw TMP1, SAVE_CFRAME
+ | sw sp, L->cframe // Add our C frame to cframe chain.
+ |
+ |3: // Entry point for vm_cpcall/vm_resume (BASE = base, PC = ftype).
+ | sw L, DISPATCH_GL(cur_L)(DISPATCH)
+ | lw TMP2, L->base // TMP2 = old base (used in vmeta_call).
+ | li TISNUM, LJ_TISNUM // Setup type comparison constants.
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | lw TMP1, L->top
+ | .FPU mtc1 TMP3, TOBIT
+ | addu PC, PC, BASE
+ | subu NARGS8:RC, TMP1, BASE
+ | subu PC, PC, TMP2 // PC = frame delta + frame type
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | li_vmstate INTERP
+ | li TISNIL, LJ_TNIL
+ | st_vmstate
+ |
+ |->vm_call_dispatch:
+ | // TMP2 = old base, BASE = new base, RC = nargs*8, PC = caller PC
+ | lw TMP0, FRAME_PC(BASE)
+ | li AT, LJ_TFUNC
+ | bne TMP0, AT, ->vmeta_call
+ |. lw LFUNC:RB, FRAME_FUNC(BASE)
+ |
+ |->vm_call_dispatch_f:
+ | ins_call
+ | // BASE = new base, RB = func, RC = nargs*8, PC = caller PC
+ |
+ |->vm_cpcall: // Setup protected C frame, call C.
+ | // (lua_State *L, lua_CFunction func, void *ud, lua_CPFunction cp)
+ | saveregs
+ | move L, CARG1
+ | lw TMP0, L:CARG1->stack
+ | sw CARG1, SAVE_L
+ | lw TMP1, L->top
+ | lw DISPATCH, L->glref // Setup pointer to dispatch table.
+ | sw CARG1, SAVE_PC // Any value outside of bytecode is ok.
+ | subu TMP0, TMP0, TMP1 // Compute -savestack(L, L->top).
+ | lw TMP1, L->cframe
+ | addiu DISPATCH, DISPATCH, GG_G2DISP
+ | sw TMP0, SAVE_NRES // Neg. delta means cframe w/o frame.
+ | sw r0, SAVE_ERRF // No error function.
+ | sw TMP1, SAVE_CFRAME
+ | sw sp, L->cframe // Add our C frame to cframe chain.
+ | sw L, DISPATCH_GL(cur_L)(DISPATCH)
+ | jalr CARG4 // (lua_State *L, lua_CFunction func, void *ud)
+ |. move CFUNCADDR, CARG4
+ | move BASE, CRET1
+ | bnez CRET1, <3 // Else continue with the call.
+ |. li PC, FRAME_CP
+ | b ->vm_leave_cp // No base? Just remove C frame.
+ |. nop
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Metamethod handling ------------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |// The lj_meta_* functions (except for lj_meta_cat) don't reallocate the
+ |// stack, so BASE doesn't need to be reloaded across these calls.
+ |
+ |//-- Continuation dispatch ----------------------------------------------
+ |
+ |->cont_dispatch:
+ | // BASE = meta base, RA = resultptr, RD = (nresults+1)*8
+ | lw TMP0, -16+LO(BASE) // Continuation.
+ | move RB, BASE
+ | move BASE, TMP2 // Restore caller BASE.
+ | lw LFUNC:TMP1, FRAME_FUNC(TMP2)
+ |.if FFI
+ | sltiu AT, TMP0, 2
+ |.endif
+ | lw PC, -16+HI(RB) // Restore PC from [cont|PC].
+ | addu TMP2, RA, RD
+ |.if FFI
+ | bnez AT, >1
+ |.endif
+ |. sw TISNIL, -8+HI(TMP2) // Ensure one valid arg.
+ | lw TMP1, LFUNC:TMP1->pc
+ | // BASE = base, RA = resultptr, RB = meta base
+ | jr TMP0 // Jump to continuation.
+ |. lw KBASE, PC2PROTO(k)(TMP1)
+ |
+ |.if FFI
+ |1:
+ | bnez TMP0, ->cont_ffi_callback // cont = 1: return from FFI callback.
+ | // cont = 0: tailcall from C function.
+ |. addiu TMP1, RB, -16
+ | b ->vm_call_tail
+ |. subu RC, TMP1, BASE
+ |.endif
+ |
+ |->cont_cat: // RA = resultptr, RB = meta base
+ | lw INS, -4(PC)
+ | addiu CARG2, RB, -16
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ | decode_RB8a MULTRES, INS
+ | decode_RA8a RA, INS
+ | decode_RB8b MULTRES
+ | decode_RA8b RA
+ | addu TMP1, BASE, MULTRES
+ | sw BASE, L->base
+ | subu CARG3, CARG2, TMP1
+ | sw SFRETHI, HI(CARG2)
+ | bne TMP1, CARG2, ->BC_CAT_Z
+ |. sw SFRETLO, LO(CARG2)
+ | addu RA, BASE, RA
+ | sw SFRETHI, HI(RA)
+ | b ->cont_nop
+ |. sw SFRETLO, LO(RA)
+ |
+ |//-- Table indexing metamethods -----------------------------------------
+ |
+ |->vmeta_tgets1:
+ | addiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | li TMP0, LJ_TSTR
+ | sw STR:RC, LO(CARG3)
+ | b >1
+ |. sw TMP0, HI(CARG3)
+ |
+ |->vmeta_tgets:
+ | addiu CARG2, DISPATCH, DISPATCH_GL(tmptv)
+ | li TMP0, LJ_TTAB
+ | sw TAB:RB, LO(CARG2)
+ | addiu CARG3, DISPATCH, DISPATCH_GL(tmptv2)
+ | sw TMP0, HI(CARG2)
+ | li TMP1, LJ_TSTR
+ | sw STR:RC, LO(CARG3)
+ | b >1
+ |. sw TMP1, HI(CARG3)
+ |
+ |->vmeta_tgetb: // TMP0 = index
+ | addiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | sw TMP0, LO(CARG3)
+ | sw TISNUM, HI(CARG3)
+ |
+ |->vmeta_tgetv:
+ |1:
+ | load_got lj_meta_tget
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | call_intern lj_meta_tget // (lua_State *L, TValue *o, TValue *k)
+ |. move CARG1, L
+ | // Returns TValue * (finished) or NULL (metamethod).
+ | beqz CRET1, >3
+ |. addiu TMP1, BASE, -FRAME_CONT
+ | lw SFARG1HI, HI(CRET1)
+ | lw SFARG2HI, LO(CRET1)
+ | ins_next1
+ | sw SFARG1HI, HI(RA)
+ | sw SFARG2HI, LO(RA)
+ | ins_next2
+ |
+ |3: // Call __index metamethod.
+ | // BASE = base, L->top = new base, stack = cont/func/t/k
+ | lw BASE, L->top
+ | sw PC, -16+HI(BASE) // [cont|PC]
+ | subu PC, BASE, TMP1
+ | lw LFUNC:RB, FRAME_FUNC(BASE) // Guaranteed to be a function here.
+ | b ->vm_call_dispatch_f
+ |. li NARGS8:RC, 16 // 2 args for func(t, k).
+ |
+ |->vmeta_tgetr:
+ | load_got lj_tab_getinth
+ | call_intern lj_tab_getinth // (GCtab *t, int32_t key)
+ |. nop
+ | // Returns cTValue * or NULL.
+ | beqz CRET1, ->BC_TGETR_Z
+ |. move SFARG2HI, TISNIL
+ | lw SFARG2HI, HI(CRET1)
+ | b ->BC_TGETR_Z
+ |. lw SFARG2LO, LO(CRET1)
+ |
+ |//-----------------------------------------------------------------------
+ |
+ |->vmeta_tsets1:
+ | addiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | li TMP0, LJ_TSTR
+ | sw STR:RC, LO(CARG3)
+ | b >1
+ |. sw TMP0, HI(CARG3)
+ |
+ |->vmeta_tsets:
+ | addiu CARG2, DISPATCH, DISPATCH_GL(tmptv)
+ | li TMP0, LJ_TTAB
+ | sw TAB:RB, LO(CARG2)
+ | addiu CARG3, DISPATCH, DISPATCH_GL(tmptv2)
+ | sw TMP0, HI(CARG2)
+ | li TMP1, LJ_TSTR
+ | sw STR:RC, LO(CARG3)
+ | b >1
+ |. sw TMP1, HI(CARG3)
+ |
+ |->vmeta_tsetb: // TMP0 = index
+ | addiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | sw TMP0, LO(CARG3)
+ | sw TISNUM, HI(CARG3)
+ |
+ |->vmeta_tsetv:
+ |1:
+ | load_got lj_meta_tset
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | call_intern lj_meta_tset // (lua_State *L, TValue *o, TValue *k)
+ |. move CARG1, L
+ | // Returns TValue * (finished) or NULL (metamethod).
+ | lw SFARG1HI, HI(RA)
+ | beqz CRET1, >3
+ |. lw SFARG1LO, LO(RA)
+ | // NOBARRIER: lj_meta_tset ensures the table is not black.
+ | ins_next1
+ | sw SFARG1HI, HI(CRET1)
+ | sw SFARG1LO, LO(CRET1)
+ | ins_next2
+ |
+ |3: // Call __newindex metamethod.
+ | // BASE = base, L->top = new base, stack = cont/func/t/k/(v)
+ | addiu TMP1, BASE, -FRAME_CONT
+ | lw BASE, L->top
+ | sw PC, -16+HI(BASE) // [cont|PC]
+ | subu PC, BASE, TMP1
+ | lw LFUNC:RB, FRAME_FUNC(BASE) // Guaranteed to be a function here.
+ | sw SFARG1HI, 16+HI(BASE) // Copy value to third argument.
+ | sw SFARG1LO, 16+LO(BASE)
+ | b ->vm_call_dispatch_f
+ |. li NARGS8:RC, 24 // 3 args for func(t, k, v)
+ |
+ |->vmeta_tsetr:
+ | load_got lj_tab_setinth
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | call_intern lj_tab_setinth // (lua_State *L, GCtab *t, int32_t key)
+ |. move CARG1, L
+ | // Returns TValue *.
+ | b ->BC_TSETR_Z
+ |. nop
+ |
+ |//-- Comparison metamethods ---------------------------------------------
+ |
+ |->vmeta_comp:
+ | // RA/RD point to o1/o2.
+ | move CARG2, RA
+ | move CARG3, RD
+ | load_got lj_meta_comp
+ | addiu PC, PC, -4
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | decode_OP1 CARG4, INS
+ | call_intern lj_meta_comp // (lua_State *L, TValue *o1, *o2, int op)
+ |. move CARG1, L
+ | // Returns 0/1 or TValue * (metamethod).
+ |3:
+ | sltiu AT, CRET1, 2
+ | beqz AT, ->vmeta_binop
+ | negu TMP2, CRET1
+ |4:
+ | lhu RD, OFS_RD(PC)
+ | addiu PC, PC, 4
+ | lui TMP1, (-(BCBIAS_J*4 >> 16) & 65535)
+ | sll RD, RD, 2
+ | addu RD, RD, TMP1
+ | and RD, RD, TMP2
+ | addu PC, PC, RD
+ |->cont_nop:
+ | ins_next
+ |
+ |->cont_ra: // RA = resultptr
+ | lbu TMP1, -4+OFS_RA(PC)
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ | sll TMP1, TMP1, 3
+ | addu TMP1, BASE, TMP1
+ | sw SFRETHI, HI(TMP1)
+ | b ->cont_nop
+ |. sw SFRETLO, LO(TMP1)
+ |
+ |->cont_condt: // RA = resultptr
+ | lw TMP0, HI(RA)
+ | sltiu AT, TMP0, LJ_TISTRUECOND
+ | b <4
+ |. negu TMP2, AT // Branch if result is true.
+ |
+ |->cont_condf: // RA = resultptr
+ | lw TMP0, HI(RA)
+ | sltiu AT, TMP0, LJ_TISTRUECOND
+ | b <4
+ |. addiu TMP2, AT, -1 // Branch if result is false.
+ |
+ |->vmeta_equal:
+ | // SFARG1LO/SFARG2LO point to o1/o2. TMP0 is set to 0/1.
+ | load_got lj_meta_equal
+ | move CARG2, SFARG1LO
+ | move CARG3, SFARG2LO
+ | move CARG4, TMP0
+ | addiu PC, PC, -4
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | call_intern lj_meta_equal // (lua_State *L, GCobj *o1, *o2, int ne)
+ |. move CARG1, L
+ | // Returns 0/1 or TValue * (metamethod).
+ | b <3
+ |. nop
+ |
+ |->vmeta_equal_cd:
+ |.if FFI
+ | load_got lj_meta_equal_cd
+ | move CARG2, INS
+ | addiu PC, PC, -4
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | call_intern lj_meta_equal_cd // (lua_State *L, BCIns op)
+ |. move CARG1, L
+ | // Returns 0/1 or TValue * (metamethod).
+ | b <3
+ |. nop
+ |.endif
+ |
+ |->vmeta_istype:
+ | load_got lj_meta_istype
+ | addiu PC, PC, -4
+ | sw BASE, L->base
+ | srl CARG2, RA, 3
+ | srl CARG3, RD, 3
+ | sw PC, SAVE_PC
+ | call_intern lj_meta_istype // (lua_State *L, BCReg ra, BCReg tp)
+ |. move CARG1, L
+ | b ->cont_nop
+ |. nop
+ |
+ |//-- Arithmetic metamethods ---------------------------------------------
+ |
+ |->vmeta_unm:
+ | move RC, RB
+ |
+ |->vmeta_arith:
+ | load_got lj_meta_arith
+ | decode_OP1 TMP0, INS
+ | sw BASE, L->base
+ | move CARG2, RA
+ | sw PC, SAVE_PC
+ | move CARG3, RB
+ | move CARG4, RC
+ | sw TMP0, ARG5
+ | call_intern lj_meta_arith // (lua_State *L, TValue *ra,*rb,*rc, BCReg op)
+ |. move CARG1, L
+ | // Returns NULL (finished) or TValue * (metamethod).
+ | beqz CRET1, ->cont_nop
+ |. nop
+ |
+ | // Call metamethod for binary op.
+ |->vmeta_binop:
+ | // BASE = old base, CRET1 = new base, stack = cont/func/o1/o2
+ | subu TMP1, CRET1, BASE
+ | sw PC, -16+HI(CRET1) // [cont|PC]
+ | move TMP2, BASE
+ | addiu PC, TMP1, FRAME_CONT
+ | move BASE, CRET1
+ | b ->vm_call_dispatch
+ |. li NARGS8:RC, 16 // 2 args for func(o1, o2).
+ |
+ |->vmeta_len:
+ | // CARG2 already set by BC_LEN.
+#if LJ_52
+ | move MULTRES, CARG1
+#endif
+ | load_got lj_meta_len
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | call_intern lj_meta_len // (lua_State *L, TValue *o)
+ |. move CARG1, L
+ | // Returns NULL (retry) or TValue * (metamethod base).
+#if LJ_52
+ | bnez CRET1, ->vmeta_binop // Binop call for compatibility.
+ |. nop
+ | b ->BC_LEN_Z
+ |. move CARG1, MULTRES
+#else
+ | b ->vmeta_binop // Binop call for compatibility.
+ |. nop
+#endif
+ |
+ |//-- Call metamethod ----------------------------------------------------
+ |
+ |->vmeta_call: // Resolve and call __call metamethod.
+ | // TMP2 = old base, BASE = new base, RC = nargs*8
+ | load_got lj_meta_call
+ | sw TMP2, L->base // This is the callers base!
+ | addiu CARG2, BASE, -8
+ | sw PC, SAVE_PC
+ | addu CARG3, BASE, RC
+ | move MULTRES, NARGS8:RC
+ | call_intern lj_meta_call // (lua_State *L, TValue *func, TValue *top)
+ |. move CARG1, L
+ | lw LFUNC:RB, FRAME_FUNC(BASE) // Guaranteed to be a function here.
+ | addiu NARGS8:RC, MULTRES, 8 // Got one more argument now.
+ | ins_call
+ |
+ |->vmeta_callt: // Resolve __call for BC_CALLT.
+ | // BASE = old base, RA = new base, RC = nargs*8
+ | load_got lj_meta_call
+ | sw BASE, L->base
+ | addiu CARG2, RA, -8
+ | sw PC, SAVE_PC
+ | addu CARG3, RA, RC
+ | move MULTRES, NARGS8:RC
+ | call_intern lj_meta_call // (lua_State *L, TValue *func, TValue *top)
+ |. move CARG1, L
+ | lw TMP1, FRAME_PC(BASE)
+ | lw LFUNC:RB, FRAME_FUNC(RA) // Guaranteed to be a function here.
+ | b ->BC_CALLT_Z
+ |. addiu NARGS8:RC, MULTRES, 8 // Got one more argument now.
+ |
+ |//-- Argument coercion for 'for' statement ------------------------------
+ |
+ |->vmeta_for:
+ | load_got lj_meta_for
+ | sw BASE, L->base
+ | move CARG2, RA
+ | sw PC, SAVE_PC
+ | move MULTRES, INS
+ | call_intern lj_meta_for // (lua_State *L, TValue *base)
+ |. move CARG1, L
+ |.if JIT
+ | decode_OP1 TMP0, MULTRES
+ | li AT, BC_JFORI
+ |.endif
+ | decode_RA8a RA, MULTRES
+ | decode_RD8a RD, MULTRES
+ | decode_RA8b RA
+ |.if JIT
+ | beq TMP0, AT, =>BC_JFORI
+ |. decode_RD8b RD
+ | b =>BC_FORI
+ |. nop
+ |.else
+ | b =>BC_FORI
+ |. decode_RD8b RD
+ |.endif
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Fast functions -----------------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |.macro .ffunc, name
+ |->ff_ .. name:
+ |.endmacro
+ |
+ |.macro .ffunc_1, name
+ |->ff_ .. name:
+ | lw SFARG1HI, HI(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ |. lw SFARG1LO, LO(BASE)
+ |.endmacro
+ |
+ |.macro .ffunc_2, name
+ |->ff_ .. name:
+ | sltiu AT, NARGS8:RC, 16
+ | lw SFARG1HI, HI(BASE)
+ | bnez AT, ->fff_fallback
+ |. lw SFARG2HI, 8+HI(BASE)
+ | lw SFARG1LO, LO(BASE)
+ | lw SFARG2LO, 8+LO(BASE)
+ |.endmacro
+ |
+ |.macro .ffunc_n, name // Caveat: has delay slot!
+ |->ff_ .. name:
+ | lw SFARG1HI, HI(BASE)
+ |.if FPU
+ | ldc1 FARG1, 0(BASE)
+ |.else
+ | lw SFARG1LO, LO(BASE)
+ |.endif
+ | beqz NARGS8:RC, ->fff_fallback
+ |. sltiu AT, SFARG1HI, LJ_TISNUM
+ | beqz AT, ->fff_fallback
+ |.endmacro
+ |
+ |.macro .ffunc_nn, name // Caveat: has delay slot!
+ |->ff_ .. name:
+ | sltiu AT, NARGS8:RC, 16
+ | lw SFARG1HI, HI(BASE)
+ | bnez AT, ->fff_fallback
+ |. lw SFARG2HI, 8+HI(BASE)
+ | sltiu TMP0, SFARG1HI, LJ_TISNUM
+ |.if FPU
+ | ldc1 FARG1, 0(BASE)
+ |.else
+ | lw SFARG1LO, LO(BASE)
+ |.endif
+ | sltiu TMP1, SFARG2HI, LJ_TISNUM
+ |.if FPU
+ | ldc1 FARG2, 8(BASE)
+ |.else
+ | lw SFARG2LO, 8+LO(BASE)
+ |.endif
+ | and TMP0, TMP0, TMP1
+ | beqz TMP0, ->fff_fallback
+ |.endmacro
+ |
+ |// Inlined GC threshold check. Caveat: uses TMP0 and TMP1 and has delay slot!
+ |.macro ffgccheck
+ | lw TMP0, DISPATCH_GL(gc.total)(DISPATCH)
+ | lw TMP1, DISPATCH_GL(gc.threshold)(DISPATCH)
+ | subu AT, TMP0, TMP1
+ | bgezal AT, ->fff_gcstep
+ |.endmacro
+ |
+ |//-- Base library: checks -----------------------------------------------
+ |
+ |.ffunc_1 assert
+ | sltiu AT, SFARG1HI, LJ_TISTRUECOND
+ | beqz AT, ->fff_fallback
+ |. addiu RA, BASE, -8
+ | lw PC, FRAME_PC(BASE)
+ | addiu RD, NARGS8:RC, 8 // Compute (nresults+1)*8.
+ | addu TMP2, RA, NARGS8:RC
+ | sw SFARG1HI, HI(RA)
+ | addiu TMP1, BASE, 8
+ | beq BASE, TMP2, ->fff_res // Done if exactly 1 argument.
+ |. sw SFARG1LO, LO(RA)
+ |1:
+ | lw SFRETHI, HI(TMP1)
+ | lw SFRETLO, LO(TMP1)
+ | sw SFRETHI, -8+HI(TMP1)
+ | sw SFRETLO, -8+LO(TMP1)
+ | bne TMP1, TMP2, <1
+ |. addiu TMP1, TMP1, 8
+ | b ->fff_res
+ |. nop
+ |
+ |.ffunc type
+ | lw SFARG1HI, HI(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ |. sltiu TMP0, SFARG1HI, LJ_TISNUM
+ | movn SFARG1HI, TISNUM, TMP0
+ | not TMP1, SFARG1HI
+ | sll TMP1, TMP1, 3
+ | addu TMP1, CFUNC:RB, TMP1
+ | lw SFARG1HI, CFUNC:TMP1->upvalue[0].u32.hi
+ | b ->fff_restv
+ |. lw SFARG1LO, CFUNC:TMP1->upvalue[0].u32.lo
+ |
+ |//-- Base library: getters and setters ---------------------------------
+ |
+ |.ffunc_1 getmetatable
+ | li AT, LJ_TTAB
+ | bne SFARG1HI, AT, >6
+ |. li AT, LJ_TUDATA
+ |1: // Field metatable must be at same offset for GCtab and GCudata!
+ | lw TAB:SFARG1LO, TAB:SFARG1LO->metatable
+ |2:
+ | lw STR:RC, DISPATCH_GL(gcroot[GCROOT_MMNAME+MM_metatable])(DISPATCH)
+ | beqz TAB:SFARG1LO, ->fff_restv
+ |. li SFARG1HI, LJ_TNIL
+ | lw TMP0, TAB:SFARG1LO->hmask
+ | li SFARG1HI, LJ_TTAB // Use metatable as default result.
+ | lw TMP1, STR:RC->sid
+ | lw NODE:TMP2, TAB:SFARG1LO->node
+ | and TMP1, TMP1, TMP0 // idx = str->sid & tab->hmask
+ | sll TMP0, TMP1, 5
+ | sll TMP1, TMP1, 3
+ | subu TMP1, TMP0, TMP1
+ | addu NODE:TMP2, NODE:TMP2, TMP1 // node = tab->node + (idx*32-idx*8)
+ | li AT, LJ_TSTR
+ |3: // Rearranged logic, because we expect _not_ to find the key.
+ | lw CARG4, offsetof(Node, key)+HI(NODE:TMP2)
+ | lw TMP0, offsetof(Node, key)+LO(NODE:TMP2)
+ | lw NODE:TMP3, NODE:TMP2->next
+ | bne CARG4, AT, >4
+ |. lw CARG3, offsetof(Node, val)+HI(NODE:TMP2)
+ | beq TMP0, STR:RC, >5
+ |. lw TMP1, offsetof(Node, val)+LO(NODE:TMP2)
+ |4:
+ | beqz NODE:TMP3, ->fff_restv // Not found, keep default result.
+ |. move NODE:TMP2, NODE:TMP3
+ | b <3
+ |. nop
+ |5:
+ | beq CARG3, TISNIL, ->fff_restv // Ditto for nil value.
+ |. nop
+ | move SFARG1HI, CARG3 // Return value of mt.__metatable.
+ | b ->fff_restv
+ |. move SFARG1LO, TMP1
+ |
+ |6:
+ | beq SFARG1HI, AT, <1
+ |. sltu AT, TISNUM, SFARG1HI
+ | movz SFARG1HI, TISNUM, AT
+ | not TMP1, SFARG1HI
+ | sll TMP1, TMP1, 2
+ | addu TMP1, DISPATCH, TMP1
+ | b <2
+ |. lw TAB:SFARG1LO, DISPATCH_GL(gcroot[GCROOT_BASEMT])(TMP1)
+ |
+ |.ffunc_2 setmetatable
+ | // Fast path: no mt for table yet and not clearing the mt.
+ | li AT, LJ_TTAB
+ | bne SFARG1HI, AT, ->fff_fallback
+ |. addiu SFARG2HI, SFARG2HI, -LJ_TTAB
+ | lw TAB:TMP1, TAB:SFARG1LO->metatable
+ | lbu TMP3, TAB:SFARG1LO->marked
+ | or AT, SFARG2HI, TAB:TMP1
+ | bnez AT, ->fff_fallback
+ |. andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ | beqz AT, ->fff_restv
+ |. sw TAB:SFARG2LO, TAB:SFARG1LO->metatable
+ | barrierback TAB:SFARG1LO, TMP3, TMP0, ->fff_restv
+ |
+ |.ffunc rawget
+ | lw CARG4, HI(BASE)
+ | sltiu AT, NARGS8:RC, 16
+ | lw TAB:CARG2, LO(BASE)
+ | load_got lj_tab_get
+ | addiu CARG4, CARG4, -LJ_TTAB
+ | or AT, AT, CARG4
+ | bnez AT, ->fff_fallback
+ | addiu CARG3, BASE, 8
+ | call_intern lj_tab_get // (lua_State *L, GCtab *t, cTValue *key)
+ |. move CARG1, L
+ | // Returns cTValue *.
+ | lw SFARG1HI, HI(CRET1)
+ | b ->fff_restv
+ |. lw SFARG1LO, LO(CRET1)
+ |
+ |//-- Base library: conversions ------------------------------------------
+ |
+ |.ffunc tonumber
+ | // Only handles the number case inline (without a base argument).
+ | lw CARG1, HI(BASE)
+ | xori AT, NARGS8:RC, 8 // Exactly one number argument.
+ | sltu TMP0, TISNUM, CARG1
+ | or AT, AT, TMP0
+ | bnez AT, ->fff_fallback
+ |. lw SFARG1HI, HI(BASE)
+ | b ->fff_restv
+ |. lw SFARG1LO, LO(BASE)
+ |
+ |.ffunc_1 tostring
+ | // Only handles the string or number case inline.
+ | li AT, LJ_TSTR
+ | // A __tostring method in the string base metatable is ignored.
+ | beq SFARG1HI, AT, ->fff_restv // String key?
+ | // Handle numbers inline, unless a number base metatable is present.
+ |. lw TMP1, DISPATCH_GL(gcroot[GCROOT_BASEMT_NUM])(DISPATCH)
+ | sltu TMP0, TISNUM, SFARG1HI
+ | or TMP0, TMP0, TMP1
+ | bnez TMP0, ->fff_fallback
+ |. sw BASE, L->base // Add frame since C call can throw.
+ | ffgccheck
+ |. sw PC, SAVE_PC // Redundant (but a defined value).
+ | load_got lj_strfmt_number
+ | move CARG1, L
+ | call_intern lj_strfmt_number // (lua_State *L, cTValue *o)
+ |. move CARG2, BASE
+ | // Returns GCstr *.
+ | li SFARG1HI, LJ_TSTR
+ | b ->fff_restv
+ |. move SFARG1LO, CRET1
+ |
+ |//-- Base library: iterators -------------------------------------------
+ |
+ |.ffunc next
+ | lw CARG2, HI(BASE)
+ | lw TAB:CARG1, LO(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ |. addu TMP2, BASE, NARGS8:RC
+ | li AT, LJ_TTAB
+ | sw TISNIL, HI(TMP2) // Set missing 2nd arg to nil.
+ | bne CARG2, AT, ->fff_fallback
+ |. lw PC, FRAME_PC(BASE)
+ | load_got lj_tab_next
+ | addiu CARG2, BASE, 8
+ | call_intern lj_tab_next // (GCtab *t, cTValue *key, TValue *o)
+ |. addiu CARG3, BASE, -8
+ | // Returns 1=found, 0=end, -1=error.
+ | addiu RA, BASE, -8
+ | bgtz CRET1, ->fff_res // Found key/value.
+ |. li RD, (2+1)*8
+ | beqz CRET1, ->fff_restv // End of traversal: return nil.
+ |. li SFARG1HI, LJ_TNIL
+ | lw CFUNC:RB, FRAME_FUNC(BASE)
+ | b ->fff_fallback // Invalid key.
+ |. li RC, 2*8
+ |
+ |.ffunc_1 pairs
+ | li AT, LJ_TTAB
+ | bne SFARG1HI, AT, ->fff_fallback
+ |. lw PC, FRAME_PC(BASE)
+#if LJ_52
+ | lw TAB:TMP2, TAB:SFARG1LO->metatable
+ | lw TMP0, CFUNC:RB->upvalue[0].u32.hi
+ | lw TMP1, CFUNC:RB->upvalue[0].u32.lo
+ | bnez TAB:TMP2, ->fff_fallback
+#else
+ | lw TMP0, CFUNC:RB->upvalue[0].u32.hi
+ | lw TMP1, CFUNC:RB->upvalue[0].u32.lo
+#endif
+ |. addiu RA, BASE, -8
+ | sw TISNIL, 8+HI(BASE)
+ | sw TMP0, HI(RA)
+ | sw TMP1, LO(RA)
+ | b ->fff_res
+ |. li RD, (3+1)*8
+ |
+ |.ffunc ipairs_aux
+ | sltiu AT, NARGS8:RC, 16
+ | lw CARG3, HI(BASE)
+ | lw TAB:CARG1, LO(BASE)
+ | lw CARG4, 8+HI(BASE)
+ | bnez AT, ->fff_fallback
+ |. addiu CARG3, CARG3, -LJ_TTAB
+ | xor CARG4, CARG4, TISNUM
+ | and AT, CARG3, CARG4
+ | bnez AT, ->fff_fallback
+ |. lw PC, FRAME_PC(BASE)
+ | lw TMP2, 8+LO(BASE)
+ | lw TMP0, TAB:CARG1->asize
+ | lw TMP1, TAB:CARG1->array
+ | addiu TMP2, TMP2, 1
+ | sw TISNUM, -8+HI(BASE)
+ | sltu AT, TMP2, TMP0
+ | sw TMP2, -8+LO(BASE)
+ | beqz AT, >2 // Not in array part?
+ |. addiu RA, BASE, -8
+ | sll TMP3, TMP2, 3
+ | addu TMP3, TMP1, TMP3
+ | lw TMP1, HI(TMP3)
+ | lw TMP2, LO(TMP3)
+ |1:
+ | beq TMP1, TISNIL, ->fff_res // End of iteration, return 0 results.
+ |. li RD, (0+1)*8
+ | sw TMP1, 8+HI(RA)
+ | sw TMP2, 8+LO(RA)
+ | b ->fff_res
+ |. li RD, (2+1)*8
+ |
+ |2: // Check for empty hash part first. Otherwise call C function.
+ | lw TMP0, TAB:CARG1->hmask
+ | load_got lj_tab_getinth
+ | beqz TMP0, ->fff_res
+ |. li RD, (0+1)*8
+ | call_intern lj_tab_getinth // (GCtab *t, int32_t key)
+ |. move CARG2, TMP2
+ | // Returns cTValue * or NULL.
+ | beqz CRET1, ->fff_res
+ |. li RD, (0+1)*8
+ | lw TMP1, HI(CRET1)
+ | b <1
+ |. lw TMP2, LO(CRET1)
+ |
+ |.ffunc_1 ipairs
+ | li AT, LJ_TTAB
+ | bne SFARG1HI, AT, ->fff_fallback
+ |. lw PC, FRAME_PC(BASE)
+#if LJ_52
+ | lw TAB:TMP2, TAB:SFARG1LO->metatable
+ | lw TMP0, CFUNC:RB->upvalue[0].u32.hi
+ | lw TMP1, CFUNC:RB->upvalue[0].u32.lo
+ | bnez TAB:TMP2, ->fff_fallback
+#else
+ | lw TMP0, CFUNC:RB->upvalue[0].u32.hi
+ | lw TMP1, CFUNC:RB->upvalue[0].u32.lo
+#endif
+ |. addiu RA, BASE, -8
+ | sw TISNUM, 8+HI(BASE)
+ | sw r0, 8+LO(BASE)
+ | sw TMP0, HI(RA)
+ | sw TMP1, LO(RA)
+ | b ->fff_res
+ |. li RD, (3+1)*8
+ |
+ |//-- Base library: catch errors ----------------------------------------
+ |
+ |.ffunc pcall
+ | lbu TMP3, DISPATCH_GL(hookmask)(DISPATCH)
+ | beqz NARGS8:RC, ->fff_fallback
+ | move TMP2, BASE
+ | addiu BASE, BASE, 8
+ | // Remember active hook before pcall.
+ | srl TMP3, TMP3, HOOK_ACTIVE_SHIFT
+ | andi TMP3, TMP3, 1
+ | addiu PC, TMP3, 8+FRAME_PCALL
+ | b ->vm_call_dispatch
+ |. addiu NARGS8:RC, NARGS8:RC, -8
+ |
+ |.ffunc xpcall
+ | sltiu AT, NARGS8:RC, 16
+ | lw CARG4, 8+HI(BASE)
+ | bnez AT, ->fff_fallback
+ |. lw CARG3, 8+LO(BASE)
+ | lw CARG1, LO(BASE)
+ | lw CARG2, HI(BASE)
+ | lbu TMP1, DISPATCH_GL(hookmask)(DISPATCH)
+ | li AT, LJ_TFUNC
+ | move TMP2, BASE
+ | bne CARG4, AT, ->fff_fallback // Traceback must be a function.
+ | addiu BASE, BASE, 16
+ | // Remember active hook before pcall.
+ | srl TMP3, TMP3, HOOK_ACTIVE_SHIFT
+ | sw CARG3, LO(TMP2) // Swap function and traceback.
+ | sw CARG4, HI(TMP2)
+ | andi TMP3, TMP3, 1
+ | sw CARG1, 8+LO(TMP2)
+ | sw CARG2, 8+HI(TMP2)
+ | addiu PC, TMP3, 16+FRAME_PCALL
+ | b ->vm_call_dispatch
+ |. addiu NARGS8:RC, NARGS8:RC, -16
+ |
+ |//-- Coroutine library --------------------------------------------------
+ |
+ |.macro coroutine_resume_wrap, resume
+ |.if resume
+ |.ffunc coroutine_resume
+ | lw CARG3, HI(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ |. lw CARG1, LO(BASE)
+ | li AT, LJ_TTHREAD
+ | bne CARG3, AT, ->fff_fallback
+ |.else
+ |.ffunc coroutine_wrap_aux
+ | lw L:CARG1, CFUNC:RB->upvalue[0].gcr
+ |.endif
+ | lbu TMP0, L:CARG1->status
+ | lw TMP1, L:CARG1->cframe
+ | lw CARG2, L:CARG1->top
+ | lw TMP2, L:CARG1->base
+ | addiu TMP3, TMP0, -LUA_YIELD
+ | bgtz TMP3, ->fff_fallback // st > LUA_YIELD?
+ |. xor TMP2, TMP2, CARG2
+ | bnez TMP1, ->fff_fallback // cframe != 0?
+ |. or AT, TMP2, TMP0
+ | lw TMP0, L:CARG1->maxstack
+ | beqz AT, ->fff_fallback // base == top && st == 0?
+ |. lw PC, FRAME_PC(BASE)
+ | addu TMP2, CARG2, NARGS8:RC
+ | sltu AT, TMP0, TMP2
+ | bnez AT, ->fff_fallback // Stack overflow?
+ |. sw PC, SAVE_PC
+ | sw BASE, L->base
+ |1:
+ |.if resume
+ | addiu BASE, BASE, 8 // Keep resumed thread in stack for GC.
+ | addiu NARGS8:RC, NARGS8:RC, -8
+ | addiu TMP2, TMP2, -8
+ |.endif
+ | sw TMP2, L:CARG1->top
+ | addu TMP1, BASE, NARGS8:RC
+ | move CARG3, CARG2
+ | sw BASE, L->top
+ |2: // Move args to coroutine.
+ | lw SFRETHI, HI(BASE)
+ | lw SFRETLO, LO(BASE)
+ | sltu AT, BASE, TMP1
+ | beqz AT, >3
+ |. addiu BASE, BASE, 8
+ | sw SFRETHI, HI(CARG3)
+ | sw SFRETLO, LO(CARG3)
+ | b <2
+ |. addiu CARG3, CARG3, 8
+ |3:
+ | bal ->vm_resume // (lua_State *L, TValue *base, 0, 0)
+ |. move L:RA, L:CARG1
+ | // Returns thread status.
+ |4:
+ | lw TMP2, L:RA->base
+ | sltiu AT, CRET1, LUA_YIELD+1
+ | lw TMP3, L:RA->top
+ | li_vmstate INTERP
+ | lw BASE, L->base
+ | sw L, DISPATCH_GL(cur_L)(DISPATCH)
+ | st_vmstate
+ | beqz AT, >8
+ |. subu RD, TMP3, TMP2
+ | lw TMP0, L->maxstack
+ | beqz RD, >6 // No results?
+ |. addu TMP1, BASE, RD
+ | sltu AT, TMP0, TMP1
+ | bnez AT, >9 // Need to grow stack?
+ |. addu TMP3, TMP2, RD
+ | sw TMP2, L:RA->top // Clear coroutine stack.
+ | move TMP1, BASE
+ |5: // Move results from coroutine.
+ | lw SFRETHI, HI(TMP2)
+ | lw SFRETLO, LO(TMP2)
+ | addiu TMP2, TMP2, 8
+ | sltu AT, TMP2, TMP3
+ | sw SFRETHI, HI(TMP1)
+ | sw SFRETLO, LO(TMP1)
+ | bnez AT, <5
+ |. addiu TMP1, TMP1, 8
+ |6:
+ | andi TMP0, PC, FRAME_TYPE
+ |.if resume
+ | li TMP1, LJ_TTRUE
+ | addiu RA, BASE, -8
+ | sw TMP1, -8+HI(BASE) // Prepend true to results.
+ | addiu RD, RD, 16
+ |.else
+ | move RA, BASE
+ | addiu RD, RD, 8
+ |.endif
+ |7:
+ | sw PC, SAVE_PC
+ | beqz TMP0, ->BC_RET_Z
+ |. move MULTRES, RD
+ | b ->vm_return
+ |. nop
+ |
+ |8: // Coroutine returned with error (at co->top-1).
+ |.if resume
+ | addiu TMP3, TMP3, -8
+ | li TMP1, LJ_TFALSE
+ | lw SFRETHI, HI(TMP3)
+ | lw SFRETLO, LO(TMP3)
+ | sw TMP3, L:RA->top // Remove error from coroutine stack.
+ | li RD, (2+1)*8
+ | sw TMP1, -8+HI(BASE) // Prepend false to results.
+ | addiu RA, BASE, -8
+ | sw SFRETHI, HI(BASE) // Copy error message.
+ | sw SFRETLO, LO(BASE)
+ | b <7
+ |. andi TMP0, PC, FRAME_TYPE
+ |.else
+ | load_got lj_ffh_coroutine_wrap_err
+ | move CARG2, L:RA
+ | call_intern lj_ffh_coroutine_wrap_err // (lua_State *L, lua_State *co)
+ |. move CARG1, L
+ |.endif
+ |
+ |9: // Handle stack expansion on return from yield.
+ | load_got lj_state_growstack
+ | srl CARG2, RD, 3
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | b <4
+ |. li CRET1, 0
+ |.endmacro
+ |
+ | coroutine_resume_wrap 1 // coroutine.resume
+ | coroutine_resume_wrap 0 // coroutine.wrap
+ |
+ |.ffunc coroutine_yield
+ | lw TMP0, L->cframe
+ | addu TMP1, BASE, NARGS8:RC
+ | sw BASE, L->base
+ | andi TMP0, TMP0, CFRAME_RESUME
+ | sw TMP1, L->top
+ | beqz TMP0, ->fff_fallback
+ |. li CRET1, LUA_YIELD
+ | sw r0, L->cframe
+ | b ->vm_leave_unw
+ |. sb CRET1, L->status
+ |
+ |//-- Math library -------------------------------------------------------
+ |
+ |.ffunc_1 math_abs
+ | bne SFARG1HI, TISNUM, >1
+ |. sra TMP0, SFARG1LO, 31
+ | xor TMP1, SFARG1LO, TMP0
+ | subu SFARG1LO, TMP1, TMP0
+ | bgez SFARG1LO, ->fff_restv
+ |. nop
+ | lui SFARG1HI, 0x41e0 // 2^31 as a double.
+ | b ->fff_restv
+ |. li SFARG1LO, 0
+ |1:
+ | sltiu AT, SFARG1HI, LJ_TISNUM
+ | beqz AT, ->fff_fallback
+ |. sll SFARG1HI, SFARG1HI, 1
+ | srl SFARG1HI, SFARG1HI, 1
+ |// fallthrough
+ |
+ |->fff_restv:
+ | // SFARG1LO/SFARG1HI = TValue result.
+ | lw PC, FRAME_PC(BASE)
+ | sw SFARG1HI, -8+HI(BASE)
+ | addiu RA, BASE, -8
+ | sw SFARG1LO, -8+LO(BASE)
+ |->fff_res1:
+ | // RA = results, PC = return.
+ | li RD, (1+1)*8
+ |->fff_res:
+ | // RA = results, RD = (nresults+1)*8, PC = return.
+ | andi TMP0, PC, FRAME_TYPE
+ | bnez TMP0, ->vm_return
+ |. move MULTRES, RD
+ | lw INS, -4(PC)
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ |5:
+ | sltu AT, RD, RB
+ | bnez AT, >6 // More results expected?
+ |. decode_RA8a TMP0, INS
+ | decode_RA8b TMP0
+ | ins_next1
+ | // Adjust BASE. KBASE is assumed to be set for the calling frame.
+ | subu BASE, RA, TMP0
+ | ins_next2
+ |
+ |6: // Fill up results with nil.
+ | addu TMP1, RA, RD
+ | addiu RD, RD, 8
+ | b <5
+ |. sw TISNIL, -8+HI(TMP1)
+ |
+ |.macro math_extern, func
+ | .ffunc math_ .. func
+ | lw SFARG1HI, HI(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ |. load_got func
+ | sltiu AT, SFARG1HI, LJ_TISNUM
+ | beqz AT, ->fff_fallback
+ |.if FPU
+ |. ldc1 FARG1, 0(BASE)
+ |.else
+ |. lw SFARG1LO, LO(BASE)
+ |.endif
+ | call_extern
+ |. nop
+ | b ->fff_resn
+ |. nop
+ |.endmacro
+ |
+ |.macro math_extern2, func
+ | .ffunc_nn math_ .. func
+ |. load_got func
+ | call_extern
+ |. nop
+ | b ->fff_resn
+ |. nop
+ |.endmacro
+ |
+ |// TODO: Return integer type if result is integer (own sf implementation).
+ |.macro math_round, func
+ |->ff_math_ .. func:
+ | lw SFARG1HI, HI(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ |. lw SFARG1LO, LO(BASE)
+ | beq SFARG1HI, TISNUM, ->fff_restv
+ |. sltu AT, SFARG1HI, TISNUM
+ | beqz AT, ->fff_fallback
+ |.if FPU
+ |. ldc1 FARG1, 0(BASE)
+ | bal ->vm_ .. func
+ |.else
+ |. load_got func
+ | call_extern
+ |.endif
+ |. nop
+ | b ->fff_resn
+ |. nop
+ |.endmacro
+ |
+ | math_round floor
+ | math_round ceil
+ |
+ |.ffunc math_log
+ | li AT, 8
+ | bne NARGS8:RC, AT, ->fff_fallback // Exactly 1 argument.
+ |. lw SFARG1HI, HI(BASE)
+ | sltiu AT, SFARG1HI, LJ_TISNUM
+ | beqz AT, ->fff_fallback
+ |. load_got log
+ |.if FPU
+ | call_extern
+ |. ldc1 FARG1, 0(BASE)
+ |.else
+ | call_extern
+ |. lw SFARG1LO, LO(BASE)
+ |.endif
+ | b ->fff_resn
+ |. nop
+ |
+ | math_extern log10
+ | math_extern exp
+ | math_extern sin
+ | math_extern cos
+ | math_extern tan
+ | math_extern asin
+ | math_extern acos
+ | math_extern atan
+ | math_extern sinh
+ | math_extern cosh
+ | math_extern tanh
+ | math_extern2 pow
+ | math_extern2 atan2
+ | math_extern2 fmod
+ |
+ |.if FPU
+ |.ffunc_n math_sqrt
+ |. sqrt.d FRET1, FARG1
+ |// fallthrough to ->fff_resn
+ |.else
+ | math_extern sqrt
+ |.endif
+ |
+ |->fff_resn:
+ | lw PC, FRAME_PC(BASE)
+ | addiu RA, BASE, -8
+ |.if FPU
+ | b ->fff_res1
+ |. sdc1 FRET1, -8(BASE)
+ |.else
+ | sw SFRETHI, -8+HI(BASE)
+ | b ->fff_res1
+ |. sw SFRETLO, -8+LO(BASE)
+ |.endif
+ |
+ |
+ |.ffunc math_ldexp
+ | sltiu AT, NARGS8:RC, 16
+ | lw SFARG1HI, HI(BASE)
+ | bnez AT, ->fff_fallback
+ |. lw CARG4, 8+HI(BASE)
+ | bne CARG4, TISNUM, ->fff_fallback
+ | load_got ldexp
+ |. sltu AT, SFARG1HI, TISNUM
+ | beqz AT, ->fff_fallback
+ |.if FPU
+ |. ldc1 FARG1, 0(BASE)
+ |.else
+ |. lw SFARG1LO, LO(BASE)
+ |.endif
+ | call_extern
+ |. lw CARG3, 8+LO(BASE)
+ | b ->fff_resn
+ |. nop
+ |
+ |.ffunc_n math_frexp
+ | load_got frexp
+ | lw PC, FRAME_PC(BASE)
+ | call_extern
+ |. addiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | lw TMP1, DISPATCH_GL(tmptv)(DISPATCH)
+ | addiu RA, BASE, -8
+ |.if FPU
+ | mtc1 TMP1, FARG2
+ | sdc1 FRET1, 0(RA)
+ | cvt.d.w FARG2, FARG2
+ | sdc1 FARG2, 8(RA)
+ |.else
+ | sw SFRETLO, LO(RA)
+ | sw SFRETHI, HI(RA)
+ | sw TMP1, 8+LO(RA)
+ | sw TISNUM, 8+HI(RA)
+ |.endif
+ | b ->fff_res
+ |. li RD, (2+1)*8
+ |
+ |.ffunc_n math_modf
+ | load_got modf
+ | lw PC, FRAME_PC(BASE)
+ | call_extern
+ |. addiu CARG3, BASE, -8
+ | addiu RA, BASE, -8
+ |.if FPU
+ | sdc1 FRET1, 0(BASE)
+ |.else
+ | sw SFRETLO, LO(BASE)
+ | sw SFRETHI, HI(BASE)
+ |.endif
+ | b ->fff_res
+ |. li RD, (2+1)*8
+ |
+ |.macro math_minmax, name, intins, ismax
+ | .ffunc_1 name
+ | addu TMP3, BASE, NARGS8:RC
+ | bne SFARG1HI, TISNUM, >5
+ |. addiu TMP2, BASE, 8
+ |1: // Handle integers.
+ |. lw SFARG2HI, HI(TMP2)
+ | beq TMP2, TMP3, ->fff_restv
+ |. lw SFARG2LO, LO(TMP2)
+ | bne SFARG2HI, TISNUM, >3
+ |. slt AT, SFARG1LO, SFARG2LO
+ | intins SFARG1LO, SFARG2LO, AT
+ | b <1
+ |. addiu TMP2, TMP2, 8
+ |
+ |3: // Convert intermediate result to number and continue with number loop.
+ | sltiu AT, SFARG2HI, LJ_TISNUM
+ | beqz AT, ->fff_fallback
+ |.if FPU
+ |. mtc1 SFARG1LO, FRET1
+ | cvt.d.w FRET1, FRET1
+ | b >7
+ |. ldc1 FARG1, 0(TMP2)
+ |.else
+ |. nop
+ | bal ->vm_sfi2d_1
+ |. nop
+ | b >7
+ |. nop
+ |.endif
+ |
+ |5:
+ |. sltiu AT, SFARG1HI, LJ_TISNUM
+ | beqz AT, ->fff_fallback
+ |.if FPU
+ |. ldc1 FRET1, 0(BASE)
+ |.endif
+ |
+ |6: // Handle numbers.
+ |. lw SFARG2HI, HI(TMP2)
+ |.if FPU
+ | beq TMP2, TMP3, ->fff_resn
+ |.else
+ | beq TMP2, TMP3, ->fff_restv
+ |.endif
+ |. sltiu AT, SFARG2HI, LJ_TISNUM
+ | beqz AT, >8
+ |.if FPU
+ |. ldc1 FARG1, 0(TMP2)
+ |.else
+ |. lw SFARG2LO, LO(TMP2)
+ |.endif
+ |7:
+ |.if FPU
+ |.if ismax
+ | c.olt.d FARG1, FRET1
+ |.else
+ | c.olt.d FRET1, FARG1
+ |.endif
+ | movf.d FRET1, FARG1
+ |.else
+ |.if ismax
+ | bal ->vm_sfcmpogt
+ |.else
+ | bal ->vm_sfcmpolt
+ |.endif
+ |. nop
+ | movz SFARG1LO, SFARG2LO, CRET1
+ | movz SFARG1HI, SFARG2HI, CRET1
+ |.endif
+ | b <6
+ |. addiu TMP2, TMP2, 8
+ |
+ |8: // Convert integer to number and continue with number loop.
+ | bne SFARG2HI, TISNUM, ->fff_fallback
+ |.if FPU
+ |. lwc1 FARG1, LO(TMP2)
+ | b <7
+ |. cvt.d.w FARG1, FARG1
+ |.else
+ |. nop
+ | bal ->vm_sfi2d_2
+ |. nop
+ | b <7
+ |. nop
+ |.endif
+ |
+ |.endmacro
+ |
+ | math_minmax math_min, movz, 0
+ | math_minmax math_max, movn, 1
+ |
+ |//-- String library -----------------------------------------------------
+ |
+ |.ffunc string_byte // Only handle the 1-arg case here.
+ | lw CARG3, HI(BASE)
+ | lw STR:CARG1, LO(BASE)
+ | xori AT, NARGS8:RC, 8
+ | addiu CARG3, CARG3, -LJ_TSTR
+ | or AT, AT, CARG3
+ | bnez AT, ->fff_fallback // Need exactly 1 string argument.
+ |. nop
+ | lw TMP0, STR:CARG1->len
+ | addiu RA, BASE, -8
+ | lw PC, FRAME_PC(BASE)
+ | sltu RD, r0, TMP0
+ | lbu TMP1, STR:CARG1[1] // Access is always ok (NUL at end).
+ | addiu RD, RD, 1
+ | sll RD, RD, 3 // RD = ((str->len != 0)+1)*8
+ | sw TISNUM, HI(RA)
+ | b ->fff_res
+ |. sw TMP1, LO(RA)
+ |
+ |.ffunc string_char // Only handle the 1-arg case here.
+ | ffgccheck
+ |. nop
+ | lw CARG3, HI(BASE)
+ | lw CARG1, LO(BASE)
+ | li TMP1, 255
+ | xori AT, NARGS8:RC, 8 // Exactly 1 argument.
+ | xor TMP0, CARG3, TISNUM // Integer.
+ | sltu TMP1, TMP1, CARG1 // !(255 < n).
+ | or AT, AT, TMP0
+ | or AT, AT, TMP1
+ | bnez AT, ->fff_fallback
+ |. li CARG3, 1
+ | addiu CARG2, sp, ARG5_OFS
+ | sb CARG1, ARG5
+ |->fff_newstr:
+ | load_got lj_str_new
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | call_intern lj_str_new // (lua_State *L, char *str, size_t l)
+ |. move CARG1, L
+ | // Returns GCstr *.
+ | lw BASE, L->base
+ |->fff_resstr:
+ | move SFARG1LO, CRET1
+ | b ->fff_restv
+ |. li SFARG1HI, LJ_TSTR
+ |
+ |.ffunc string_sub
+ | ffgccheck
+ |. nop
+ | addiu AT, NARGS8:RC, -16
+ | lw CARG3, 16+HI(BASE)
+ | lw TMP0, HI(BASE)
+ | lw STR:CARG1, LO(BASE)
+ | bltz AT, ->fff_fallback
+ |. lw CARG2, 8+HI(BASE)
+ | beqz AT, >1
+ |. li CARG4, -1
+ | bne CARG3, TISNUM, ->fff_fallback
+ |. lw CARG4, 16+LO(BASE)
+ |1:
+ | bne CARG2, TISNUM, ->fff_fallback
+ |. li AT, LJ_TSTR
+ | bne TMP0, AT, ->fff_fallback
+ |. lw CARG3, 8+LO(BASE)
+ | lw CARG2, STR:CARG1->len
+ | // STR:CARG1 = str, CARG2 = str->len, CARG3 = start, CARG4 = end
+ | slt AT, CARG4, r0
+ | addiu TMP0, CARG2, 1
+ | addu TMP1, CARG4, TMP0
+ | slt TMP3, CARG3, r0
+ | movn CARG4, TMP1, AT // if (end < 0) end += len+1
+ | addu TMP1, CARG3, TMP0
+ | movn CARG3, TMP1, TMP3 // if (start < 0) start += len+1
+ | li TMP2, 1
+ | slt AT, CARG4, r0
+ | slt TMP3, r0, CARG3
+ | movn CARG4, r0, AT // if (end < 0) end = 0
+ | movz CARG3, TMP2, TMP3 // if (start < 1) start = 1
+ | slt AT, CARG2, CARG4
+ | movn CARG4, CARG2, AT // if (end > len) end = len
+ | addu CARG2, STR:CARG1, CARG3
+ | subu CARG3, CARG4, CARG3 // len = end - start
+ | addiu CARG2, CARG2, sizeof(GCstr)-1
+ | bgez CARG3, ->fff_newstr
+ |. addiu CARG3, CARG3, 1 // len++
+ |->fff_emptystr: // Return empty string.
+ | addiu STR:SFARG1LO, DISPATCH, DISPATCH_GL(strempty)
+ | b ->fff_restv
+ |. li SFARG1HI, LJ_TSTR
+ |
+ |.macro ffstring_op, name
+ | .ffunc string_ .. name
+ | ffgccheck
+ |. nop
+ | lw CARG3, HI(BASE)
+ | lw STR:CARG2, LO(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ |. li AT, LJ_TSTR
+ | bne CARG3, AT, ->fff_fallback
+ |. addiu SBUF:CARG1, DISPATCH, DISPATCH_GL(tmpbuf)
+ | load_got lj_buf_putstr_ .. name
+ | lw TMP0, SBUF:CARG1->b
+ | sw L, SBUF:CARG1->L
+ | sw BASE, L->base
+ | sw TMP0, SBUF:CARG1->w
+ | call_intern extern lj_buf_putstr_ .. name
+ |. sw PC, SAVE_PC
+ | load_got lj_buf_tostr
+ | call_intern lj_buf_tostr
+ |. move SBUF:CARG1, SBUF:CRET1
+ | b ->fff_resstr
+ |. lw BASE, L->base
+ |.endmacro
+ |
+ |ffstring_op reverse
+ |ffstring_op lower
+ |ffstring_op upper
+ |
+ |//-- Bit library --------------------------------------------------------
+ |
+ |->vm_tobit_fb:
+ | beqz TMP1, ->fff_fallback
+ |.if FPU
+ |. ldc1 FARG1, 0(BASE)
+ | add.d FARG1, FARG1, TOBIT
+ | jr ra
+ |. mfc1 CRET1, FARG1
+ |.else
+ |// FP number to bit conversion for soft-float.
+ |->vm_tobit:
+ | sll TMP0, SFARG1HI, 1
+ | lui AT, 0x0020
+ | addu TMP0, TMP0, AT
+ | slt AT, TMP0, r0
+ | movz SFARG1LO, r0, AT
+ | beqz AT, >2
+ |. li TMP1, 0x3e0
+ | not TMP1, TMP1
+ | sra TMP0, TMP0, 21
+ | subu TMP0, TMP1, TMP0
+ | slt AT, TMP0, r0
+ | bnez AT, >1
+ |. sll TMP1, SFARG1HI, 11
+ | lui AT, 0x8000
+ | or TMP1, TMP1, AT
+ | srl AT, SFARG1LO, 21
+ | or TMP1, TMP1, AT
+ | slt AT, SFARG1HI, r0
+ | beqz AT, >2
+ |. srlv SFARG1LO, TMP1, TMP0
+ | subu SFARG1LO, r0, SFARG1LO
+ |2:
+ | jr ra
+ |. move CRET1, SFARG1LO
+ |1:
+ | addiu TMP0, TMP0, 21
+ | srlv TMP1, SFARG1LO, TMP0
+ | li AT, 20
+ | subu TMP0, AT, TMP0
+ | sll SFARG1LO, SFARG1HI, 12
+ | sllv AT, SFARG1LO, TMP0
+ | or SFARG1LO, TMP1, AT
+ | slt AT, SFARG1HI, r0
+ | beqz AT, <2
+ |. nop
+ | jr ra
+ |. subu CRET1, r0, SFARG1LO
+ |.endif
+ |
+ |.macro .ffunc_bit, name
+ | .ffunc_1 bit_..name
+ | beq SFARG1HI, TISNUM, >6
+ |. move CRET1, SFARG1LO
+ | bal ->vm_tobit_fb
+ |. sltu TMP1, SFARG1HI, TISNUM
+ |6:
+ |.endmacro
+ |
+ |.macro .ffunc_bit_op, name, ins
+ | .ffunc_bit name
+ | addiu TMP2, BASE, 8
+ | addu TMP3, BASE, NARGS8:RC
+ |1:
+ | lw SFARG1HI, HI(TMP2)
+ | beq TMP2, TMP3, ->fff_resi
+ |. lw SFARG1LO, LO(TMP2)
+ |.if FPU
+ | bne SFARG1HI, TISNUM, >2
+ |. addiu TMP2, TMP2, 8
+ | b <1
+ |. ins CRET1, CRET1, SFARG1LO
+ |2:
+ | ldc1 FARG1, -8(TMP2)
+ | sltu TMP1, SFARG1HI, TISNUM
+ | beqz TMP1, ->fff_fallback
+ |. add.d FARG1, FARG1, TOBIT
+ | mfc1 SFARG1LO, FARG1
+ | b <1
+ |. ins CRET1, CRET1, SFARG1LO
+ |.else
+ | beq SFARG1HI, TISNUM, >2
+ |. move CRET2, CRET1
+ | bal ->vm_tobit_fb
+ |. sltu TMP1, SFARG1HI, TISNUM
+ | move SFARG1LO, CRET2
+ |2:
+ | ins CRET1, CRET1, SFARG1LO
+ | b <1
+ |. addiu TMP2, TMP2, 8
+ |.endif
+ |.endmacro
+ |
+ |.ffunc_bit_op band, and
+ |.ffunc_bit_op bor, or
+ |.ffunc_bit_op bxor, xor
+ |
+ |.ffunc_bit bswap
+ | srl TMP0, CRET1, 24
+ | srl TMP2, CRET1, 8
+ | sll TMP1, CRET1, 24
+ | andi TMP2, TMP2, 0xff00
+ | or TMP0, TMP0, TMP1
+ | andi CRET1, CRET1, 0xff00
+ | or TMP0, TMP0, TMP2
+ | sll CRET1, CRET1, 8
+ | b ->fff_resi
+ |. or CRET1, TMP0, CRET1
+ |
+ |.ffunc_bit bnot
+ | b ->fff_resi
+ |. not CRET1, CRET1
+ |
+ |.macro .ffunc_bit_sh, name, ins, shmod
+ | .ffunc_2 bit_..name
+ | beq SFARG1HI, TISNUM, >1
+ |. nop
+ | bal ->vm_tobit_fb
+ |. sltu TMP1, SFARG1HI, TISNUM
+ | move SFARG1LO, CRET1
+ |1:
+ | bne SFARG2HI, TISNUM, ->fff_fallback
+ |. nop
+ |.if shmod == 1
+ | li AT, 32
+ | subu TMP0, AT, SFARG2LO
+ | sllv SFARG2LO, SFARG1LO, SFARG2LO
+ | srlv SFARG1LO, SFARG1LO, TMP0
+ |.elif shmod == 2
+ | li AT, 32
+ | subu TMP0, AT, SFARG2LO
+ | srlv SFARG2LO, SFARG1LO, SFARG2LO
+ | sllv SFARG1LO, SFARG1LO, TMP0
+ |.endif
+ | b ->fff_resi
+ |. ins CRET1, SFARG1LO, SFARG2LO
+ |.endmacro
+ |
+ |.ffunc_bit_sh lshift, sllv, 0
+ |.ffunc_bit_sh rshift, srlv, 0
+ |.ffunc_bit_sh arshift, srav, 0
+ |// Can't use rotrv, since it's only in MIPS32R2.
+ |.ffunc_bit_sh rol, or, 1
+ |.ffunc_bit_sh ror, or, 2
+ |
+ |.ffunc_bit tobit
+ |->fff_resi:
+ | lw PC, FRAME_PC(BASE)
+ | addiu RA, BASE, -8
+ | sw TISNUM, -8+HI(BASE)
+ | b ->fff_res1
+ |. sw CRET1, -8+LO(BASE)
+ |
+ |//-----------------------------------------------------------------------
+ |
+ |->fff_fallback: // Call fast function fallback handler.
+ | // BASE = new base, RB = CFUNC, RC = nargs*8
+ | lw TMP3, CFUNC:RB->f
+ | addu TMP1, BASE, NARGS8:RC
+ | lw PC, FRAME_PC(BASE) // Fallback may overwrite PC.
+ | addiu TMP0, TMP1, 8*LUA_MINSTACK
+ | lw TMP2, L->maxstack
+ | sw PC, SAVE_PC // Redundant (but a defined value).
+ | sltu AT, TMP2, TMP0
+ | sw BASE, L->base
+ | sw TMP1, L->top
+ | bnez AT, >5 // Need to grow stack.
+ |. move CFUNCADDR, TMP3
+ | jalr TMP3 // (lua_State *L)
+ |. move CARG1, L
+ | // Either throws an error, or recovers and returns -1, 0 or nresults+1.
+ | lw BASE, L->base
+ | sll RD, CRET1, 3
+ | bgtz CRET1, ->fff_res // Returned nresults+1?
+ |. addiu RA, BASE, -8
+ |1: // Returned 0 or -1: retry fast path.
+ | lw TMP0, L->top
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | bnez CRET1, ->vm_call_tail // Returned -1?
+ |. subu NARGS8:RC, TMP0, BASE
+ | ins_callt // Returned 0: retry fast path.
+ |
+ |// Reconstruct previous base for vmeta_call during tailcall.
+ |->vm_call_tail:
+ | andi TMP0, PC, FRAME_TYPE
+ | li AT, -4
+ | bnez TMP0, >3
+ |. and TMP1, PC, AT
+ | lbu TMP1, OFS_RA(PC)
+ | sll TMP1, TMP1, 3
+ | addiu TMP1, TMP1, 8
+ |3:
+ | b ->vm_call_dispatch // Resolve again for tailcall.
+ |. subu TMP2, BASE, TMP1
+ |
+ |5: // Grow stack for fallback handler.
+ | load_got lj_state_growstack
+ | li CARG2, LUA_MINSTACK
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | lw BASE, L->base
+ | b <1
+ |. li CRET1, 0 // Force retry.
+ |
+ |->fff_gcstep: // Call GC step function.
+ | // BASE = new base, RC = nargs*8
+ | move MULTRES, ra
+ | load_got lj_gc_step
+ | sw BASE, L->base
+ | addu TMP0, BASE, NARGS8:RC
+ | sw PC, SAVE_PC // Redundant (but a defined value).
+ | sw TMP0, L->top
+ | call_intern lj_gc_step // (lua_State *L)
+ |. move CARG1, L
+ | lw BASE, L->base
+ | move ra, MULTRES
+ | lw TMP0, L->top
+ | lw CFUNC:RB, FRAME_FUNC(BASE)
+ | jr ra
+ |. subu NARGS8:RC, TMP0, BASE
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Special dispatch targets -------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |->vm_record: // Dispatch target for recording phase.
+ |.if JIT
+ | lbu TMP3, DISPATCH_GL(hookmask)(DISPATCH)
+ | andi AT, TMP3, HOOK_VMEVENT // No recording while in vmevent.
+ | bnez AT, >5
+ | // Decrement the hookcount for consistency, but always do the call.
+ |. lw TMP2, DISPATCH_GL(hookcount)(DISPATCH)
+ | andi AT, TMP3, HOOK_ACTIVE
+ | bnez AT, >1
+ |. addiu TMP2, TMP2, -1
+ | andi AT, TMP3, LUA_MASKLINE|LUA_MASKCOUNT
+ | beqz AT, >1
+ |. nop
+ | b >1
+ |. sw TMP2, DISPATCH_GL(hookcount)(DISPATCH)
+ |.endif
+ |
+ |->vm_rethook: // Dispatch target for return hooks.
+ | lbu TMP3, DISPATCH_GL(hookmask)(DISPATCH)
+ | andi AT, TMP3, HOOK_ACTIVE // Hook already active?
+ | beqz AT, >1
+ |5: // Re-dispatch to static ins.
+ |. lw AT, GG_DISP2STATIC(TMP0) // Assumes TMP0 holds DISPATCH+OP*4.
+ | jr AT
+ |. nop
+ |
+ |->vm_inshook: // Dispatch target for instr/line hooks.
+ | lbu TMP3, DISPATCH_GL(hookmask)(DISPATCH)
+ | lw TMP2, DISPATCH_GL(hookcount)(DISPATCH)
+ | andi AT, TMP3, HOOK_ACTIVE // Hook already active?
+ | bnez AT, <5
+ |. andi AT, TMP3, LUA_MASKLINE|LUA_MASKCOUNT
+ | beqz AT, <5
+ |. addiu TMP2, TMP2, -1
+ | beqz TMP2, >1
+ |. sw TMP2, DISPATCH_GL(hookcount)(DISPATCH)
+ | andi AT, TMP3, LUA_MASKLINE
+ | beqz AT, <5
+ |1:
+ |. load_got lj_dispatch_ins
+ | sw MULTRES, SAVE_MULTRES
+ | move CARG2, PC
+ | sw BASE, L->base
+ | // SAVE_PC must hold the _previous_ PC. The callee updates it with PC.
+ | call_intern lj_dispatch_ins // (lua_State *L, const BCIns *pc)
+ |. move CARG1, L
+ |3:
+ | lw BASE, L->base
+ |4: // Re-dispatch to static ins.
+ | lw INS, -4(PC)
+ | decode_OP4a TMP1, INS
+ | decode_OP4b TMP1
+ | addu TMP0, DISPATCH, TMP1
+ | decode_RD8a RD, INS
+ | lw AT, GG_DISP2STATIC(TMP0)
+ | decode_RA8a RA, INS
+ | decode_RD8b RD
+ | jr AT
+ | decode_RA8b RA
+ |
+ |->cont_hook: // Continue from hook yield.
+ | addiu PC, PC, 4
+ | b <4
+ |. lw MULTRES, -24+LO(RB) // Restore MULTRES for *M ins.
+ |
+ |->vm_hotloop: // Hot loop counter underflow.
+ |.if JIT
+ | lw LFUNC:TMP1, FRAME_FUNC(BASE)
+ | addiu CARG1, DISPATCH, GG_DISP2J
+ | sw PC, SAVE_PC
+ | lw TMP1, LFUNC:TMP1->pc
+ | move CARG2, PC
+ | sw L, DISPATCH_J(L)(DISPATCH)
+ | lbu TMP1, PC2PROTO(framesize)(TMP1)
+ | load_got lj_trace_hot
+ | sw BASE, L->base
+ | sll TMP1, TMP1, 3
+ | addu TMP1, BASE, TMP1
+ | call_intern lj_trace_hot // (jit_State *J, const BCIns *pc)
+ |. sw TMP1, L->top
+ | b <3
+ |. nop
+ |.endif
+ |
+ |->vm_callhook: // Dispatch target for call hooks.
+ |.if JIT
+ | b >1
+ |.endif
+ |. move CARG2, PC
+ |
+ |->vm_hotcall: // Hot call counter underflow.
+ |.if JIT
+ | ori CARG2, PC, 1
+ |1:
+ |.endif
+ | load_got lj_dispatch_call
+ | addu TMP0, BASE, RC
+ | sw PC, SAVE_PC
+ | sw BASE, L->base
+ | subu RA, RA, BASE
+ | sw TMP0, L->top
+ | call_intern lj_dispatch_call // (lua_State *L, const BCIns *pc)
+ |. move CARG1, L
+ | // Returns ASMFunction.
+ | lw BASE, L->base
+ | lw TMP0, L->top
+ | sw r0, SAVE_PC // Invalidate for subsequent line hook.
+ | subu NARGS8:RC, TMP0, BASE
+ | addu RA, BASE, RA
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | jr CRET1
+ |. lw INS, -4(PC)
+ |
+ |->cont_stitch: // Trace stitching.
+ |.if JIT
+ | // RA = resultptr, RB = meta base
+ | lw INS, -4(PC)
+ | lw TMP2, -24+LO(RB) // Save previous trace.
+ | decode_RA8a RC, INS
+ | addiu AT, MULTRES, -8
+ | decode_RA8b RC
+ | beqz AT, >2
+ |. addu RC, BASE, RC // Call base.
+ |1: // Move results down.
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ | addiu AT, AT, -8
+ | addiu RA, RA, 8
+ | sw SFRETHI, HI(RC)
+ | sw SFRETLO, LO(RC)
+ | bnez AT, <1
+ |. addiu RC, RC, 8
+ |2:
+ | decode_RA8a RA, INS
+ | decode_RB8a RB, INS
+ | decode_RA8b RA
+ | decode_RB8b RB
+ | addu RA, RA, RB
+ | addu RA, BASE, RA
+ |3:
+ | sltu AT, RC, RA
+ | bnez AT, >9 // More results wanted?
+ |. nop
+ |
+ | lhu TMP3, TRACE:TMP2->traceno
+ | lhu RD, TRACE:TMP2->link
+ | beq RD, TMP3, ->cont_nop // Blacklisted.
+ |. load_got lj_dispatch_stitch
+ | bnez RD, =>BC_JLOOP // Jump to stitched trace.
+ |. sll RD, RD, 3
+ |
+ | // Stitch a new trace to the previous trace.
+ | sw TMP3, DISPATCH_J(exitno)(DISPATCH)
+ | sw L, DISPATCH_J(L)(DISPATCH)
+ | sw BASE, L->base
+ | addiu CARG1, DISPATCH, GG_DISP2J
+ | call_intern lj_dispatch_stitch // (jit_State *J, const BCIns *pc)
+ |. move CARG2, PC
+ | b ->cont_nop
+ |. lw BASE, L->base
+ |
+ |9:
+ | sw TISNIL, HI(RC)
+ | b <3
+ |. addiu RC, RC, 8
+ |.endif
+ |
+ |->vm_profhook: // Dispatch target for profiler hook.
+#if LJ_HASPROFILE
+ | load_got lj_dispatch_profile
+ | sw MULTRES, SAVE_MULTRES
+ | move CARG2, PC
+ | sw BASE, L->base
+ | call_intern lj_dispatch_profile // (lua_State *L, const BCIns *pc)
+ |. move CARG1, L
+ | // HOOK_PROFILE is off again, so re-dispatch to dynamic instruction.
+ | addiu PC, PC, -4
+ | b ->cont_nop
+ |. lw BASE, L->base
+#endif
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Trace exit handler -------------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |.macro savex_, a, b
+ |.if FPU
+ | sdc1 f..a, 16+a*8(sp)
+ | sw r..a, 16+32*8+a*4(sp)
+ | sw r..b, 16+32*8+b*4(sp)
+ |.else
+ | sw r..a, 16+a*4(sp)
+ | sw r..b, 16+b*4(sp)
+ |.endif
+ |.endmacro
+ |
+ |->vm_exit_handler:
+ |.if JIT
+ |.if FPU
+ | addiu sp, sp, -(16+32*8+32*4)
+ |.else
+ | addiu sp, sp, -(16+32*4)
+ |.endif
+ | savex_ 0, 1
+ | savex_ 2, 3
+ | savex_ 4, 5
+ | savex_ 6, 7
+ | savex_ 8, 9
+ | savex_ 10, 11
+ | savex_ 12, 13
+ | savex_ 14, 15
+ | savex_ 16, 17
+ | savex_ 18, 19
+ | savex_ 20, 21
+ | savex_ 22, 23
+ | savex_ 24, 25
+ | savex_ 26, 27
+ |.if FPU
+ | sdc1 f28, 16+28*8(sp)
+ | sdc1 f30, 16+30*8(sp)
+ | sw r28, 16+32*8+28*4(sp)
+ | sw r30, 16+32*8+30*4(sp)
+ | sw r0, 16+32*8+31*4(sp) // Clear RID_TMP.
+ | addiu TMP2, sp, 16+32*8+32*4 // Recompute original value of sp.
+ | sw TMP2, 16+32*8+29*4(sp) // Store sp in RID_SP
+ |.else
+ | sw r28, 16+28*4(sp)
+ | sw r30, 16+30*4(sp)
+ | sw r0, 16+31*4(sp) // Clear RID_TMP.
+ | addiu TMP2, sp, 16+32*4 // Recompute original value of sp.
+ | sw TMP2, 16+29*4(sp) // Store sp in RID_SP
+ |.endif
+ | li_vmstate EXIT
+ | addiu DISPATCH, JGL, -GG_DISP2G-32768
+ | lw TMP1, 0(TMP2) // Load exit number.
+ | st_vmstate
+ | lw L, DISPATCH_GL(cur_L)(DISPATCH)
+ | lw BASE, DISPATCH_GL(jit_base)(DISPATCH)
+ | load_got lj_trace_exit
+ | sw L, DISPATCH_J(L)(DISPATCH)
+ | sw ra, DISPATCH_J(parent)(DISPATCH) // Store trace number.
+ | sw BASE, L->base
+ | sw TMP1, DISPATCH_J(exitno)(DISPATCH) // Store exit number.
+ | addiu CARG1, DISPATCH, GG_DISP2J
+ | sw r0, DISPATCH_GL(jit_base)(DISPATCH)
+ | call_intern lj_trace_exit // (jit_State *J, ExitState *ex)
+ |. addiu CARG2, sp, 16
+ | // Returns MULTRES (unscaled) or negated error code.
+ | lw TMP1, L->cframe
+ | li AT, -4
+ | lw BASE, L->base
+ | and sp, TMP1, AT
+ | lw PC, SAVE_PC // Get SAVE_PC.
+ | b >1
+ |. sw L, SAVE_L // Set SAVE_L (on-trace resume/yield).
+ |.endif
+ |->vm_exit_interp:
+ |.if JIT
+ | // CRET1 = MULTRES or negated error code, BASE, PC and JGL set.
+ | lw L, SAVE_L
+ | addiu DISPATCH, JGL, -GG_DISP2G-32768
+ | sw BASE, L->base
+ |1:
+ | bltz CRET1, >9 // Check for error from exit.
+ |. lw LFUNC:RB, FRAME_FUNC(BASE)
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | sll MULTRES, CRET1, 3
+ | li TISNIL, LJ_TNIL
+ | li TISNUM, LJ_TISNUM // Setup type comparison constants.
+ | sw MULTRES, SAVE_MULTRES
+ | .FPU mtc1 TMP3, TOBIT
+ | lw TMP1, LFUNC:RB->pc
+ | sw r0, DISPATCH_GL(jit_base)(DISPATCH)
+ | lw KBASE, PC2PROTO(k)(TMP1)
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | // Modified copy of ins_next which handles function header dispatch, too.
+ | lw INS, 0(PC)
+ | addiu PC, PC, 4
+ | // Assumes TISNIL == ~LJ_VMST_INTERP == -1
+ | sw TISNIL, DISPATCH_GL(vmstate)(DISPATCH)
+ | decode_OP4a TMP1, INS
+ | decode_OP4b TMP1
+ | sltiu TMP2, TMP1, BC_FUNCF*4
+ | addu TMP0, DISPATCH, TMP1
+ | decode_RD8a RD, INS
+ | lw AT, 0(TMP0)
+ | decode_RA8a RA, INS
+ | beqz TMP2, >2
+ |. decode_RA8b RA
+ | jr AT
+ |. decode_RD8b RD
+ |2:
+ | sltiu TMP2, TMP1, (BC_FUNCC+2)*4 // Fast function?
+ | bnez TMP2, >3
+ |. lw TMP1, FRAME_PC(BASE)
+ | // Check frame below fast function.
+ | andi TMP0, TMP1, FRAME_TYPE
+ | bnez TMP0, >3 // Trace stitching continuation?
+ |. nop
+ | // Otherwise set KBASE for Lua function below fast function.
+ | lw TMP2, -4(TMP1)
+ | decode_RA8a TMP0, TMP2
+ | decode_RA8b TMP0
+ | subu TMP1, BASE, TMP0
+ | lw LFUNC:TMP2, -8+FRAME_FUNC(TMP1)
+ | lw TMP1, LFUNC:TMP2->pc
+ | lw KBASE, PC2PROTO(k)(TMP1)
+ |3:
+ | addiu RC, MULTRES, -8
+ | jr AT
+ |. addu RA, RA, BASE
+ |
+ |9: // Rethrow error from the right C frame.
+ | load_got lj_err_trace
+ | sub CARG2, r0, CRET1
+ | call_intern lj_err_trace // (lua_State *L, int errcode)
+ |. move CARG1, L
+ |.endif
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Math helper functions ----------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |// Hard-float round to integer.
+ |// Modifies AT, TMP0, FRET1, FRET2, f4. Keeps all others incl. FARG1.
+ |.macro vm_round_hf, func
+ | lui TMP0, 0x4330 // Hiword of 2^52 (double).
+ | mtc1 r0, f4
+ | mtc1 TMP0, f5
+ | abs.d FRET2, FARG1 // |x|
+ | mfc1 AT, f13
+ | c.olt.d 0, FRET2, f4
+ | add.d FRET1, FRET2, f4 // (|x| + 2^52) - 2^52
+ | bc1f 0, >1 // Truncate only if |x| < 2^52.
+ |. sub.d FRET1, FRET1, f4
+ | slt AT, AT, r0
+ |.if "func" == "ceil"
+ | lui TMP0, 0xbff0 // Hiword of -1 (double). Preserves -0.
+ |.else
+ | lui TMP0, 0x3ff0 // Hiword of +1 (double).
+ |.endif
+ |.if "func" == "trunc"
+ | mtc1 TMP0, f5
+ | c.olt.d 0, FRET2, FRET1 // |x| < result?
+ | sub.d FRET2, FRET1, f4
+ | movt.d FRET1, FRET2, 0 // If yes, subtract +1.
+ | neg.d FRET2, FRET1
+ | jr ra
+ |. movn.d FRET1, FRET2, AT // Merge sign bit back in.
+ |.else
+ | neg.d FRET2, FRET1
+ | mtc1 TMP0, f5
+ | movn.d FRET1, FRET2, AT // Merge sign bit back in.
+ |.if "func" == "ceil"
+ | c.olt.d 0, FRET1, FARG1 // x > result?
+ |.else
+ | c.olt.d 0, FARG1, FRET1 // x < result?
+ |.endif
+ | sub.d FRET2, FRET1, f4 // If yes, subtract +-1.
+ | jr ra
+ |. movt.d FRET1, FRET2, 0
+ |.endif
+ |1:
+ | jr ra
+ |. mov.d FRET1, FARG1
+ |.endmacro
+ |
+ |.macro vm_round, func
+ |.if FPU
+ | vm_round_hf, func
+ |.endif
+ |.endmacro
+ |
+ |->vm_floor:
+ | vm_round floor
+ |->vm_ceil:
+ | vm_round ceil
+ |->vm_trunc:
+ |.if JIT
+ | vm_round trunc
+ |.endif
+ |
+ |// Soft-float integer to number conversion.
+ |.macro sfi2d, AHI, ALO
+ |.if not FPU
+ | beqz ALO, >9 // Handle zero first.
+ |. sra TMP0, ALO, 31
+ | xor TMP1, ALO, TMP0
+ | subu TMP1, TMP1, TMP0 // Absolute value in TMP1.
+ | clz AHI, TMP1
+ | andi TMP0, TMP0, 0x800 // Mask sign bit.
+ | li AT, 0x3ff+31-1
+ | sllv TMP1, TMP1, AHI // Align mantissa left with leading 1.
+ | subu AHI, AT, AHI // Exponent - 1 in AHI.
+ | sll ALO, TMP1, 21
+ | or AHI, AHI, TMP0 // Sign | Exponent.
+ | srl TMP1, TMP1, 11
+ | sll AHI, AHI, 20 // Align left.
+ | jr ra
+ |. addu AHI, AHI, TMP1 // Add mantissa, increment exponent.
+ |9:
+ | jr ra
+ |. li AHI, 0
+ |.endif
+ |.endmacro
+ |
+ |// Input SFARG1LO. Output: SFARG1*. Temporaries: AT, TMP0, TMP1.
+ |->vm_sfi2d_1:
+ | sfi2d SFARG1HI, SFARG1LO
+ |
+ |// Input SFARG2LO. Output: SFARG2*. Temporaries: AT, TMP0, TMP1.
+ |->vm_sfi2d_2:
+ | sfi2d SFARG2HI, SFARG2LO
+ |
+ |// Soft-float comparison. Equivalent to c.eq.d.
+ |// Input: SFARG*. Output: CRET1. Temporaries: AT, TMP0, TMP1.
+ |->vm_sfcmpeq:
+ |.if not FPU
+ | sll AT, SFARG1HI, 1
+ | sll TMP0, SFARG2HI, 1
+ | or CRET1, SFARG1LO, SFARG2LO
+ | or TMP1, AT, TMP0
+ | or TMP1, TMP1, CRET1
+ | beqz TMP1, >8 // Both args +-0: return 1.
+ |. sltu CRET1, r0, SFARG1LO
+ | lui TMP1, 0xffe0
+ | addu AT, AT, CRET1
+ | sltu CRET1, r0, SFARG2LO
+ | 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;
+ |. xor TMP0, SFARG1HI, SFARG2HI
+ | xor TMP1, SFARG1LO, SFARG2LO
+ | or AT, TMP0, TMP1
+ | jr ra
+ |. sltiu CRET1, AT, 1 // Same values: return 1.
+ |8:
+ | jr ra
+ |. li CRET1, 1
+ |9:
+ | jr ra
+ |. li CRET1, 0
+ |.endif
+ |
+ |// Soft-float comparison. Equivalent to c.ult.d and c.olt.d.
+ |// Input: SFARG*. Output: CRET1. Temporaries: AT, TMP0, TMP1, CRET2.
+ |->vm_sfcmpult:
+ |.if not FPU
+ | b >1
+ |. li CRET2, 1
+ |.endif
+ |
+ |->vm_sfcmpolt:
+ |.if not FPU
+ | li CRET2, 0
+ |1:
+ | sll AT, SFARG1HI, 1
+ | sll TMP0, SFARG2HI, 1
+ | or CRET1, SFARG1LO, SFARG2LO
+ | or TMP1, AT, TMP0
+ | or TMP1, TMP1, CRET1
+ | beqz TMP1, >8 // Both args +-0: return 0.
+ |. sltu CRET1, r0, SFARG1LO
+ | lui TMP1, 0xffe0
+ | addu AT, AT, CRET1
+ | sltu CRET1, r0, SFARG2LO
+ | 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, SFARG1HI, SFARG2HI
+ | bltz AT, >5 // Both args negative?
+ |. nop
+ | beq SFARG1HI, SFARG2HI, >8
+ |. sltu CRET1, SFARG1LO, SFARG2LO
+ | jr ra
+ |. slt CRET1, SFARG1HI, SFARG2HI
+ |5: // Swap conditions if both operands are negative.
+ | beq SFARG1HI, SFARG2HI, >8
+ |. sltu CRET1, SFARG2LO, SFARG1LO
+ | jr ra
+ |. slt CRET1, SFARG2HI, SFARG1HI
+ |8:
+ | jr ra
+ |. nop
+ |9:
+ | jr ra
+ |. 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:
+ |.if not FPU
+ | sll AT, SFARG1HI, 1
+ | sll TMP0, SFARG2HI, 1
+ | or CRET1, SFARG1LO, SFARG2LO
+ | or TMP1, AT, TMP0
+ | or TMP1, TMP1, CRET1
+ | beqz TMP1, >8 // Both args +-0: return 1.
+ |. sltu CRET1, r0, SFARG1LO
+ | lui TMP1, 0xffe0
+ | addu AT, AT, CRET1
+ | sltu CRET1, r0, SFARG2LO
+ | 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;
+ |. and AT, SFARG1HI, SFARG2HI
+ | xor AT, AT, TMP3
+ | bltz AT, >5 // Both args negative?
+ |. nop
+ | beq SFARG1HI, SFARG2HI, >6
+ |. sltu CRET1, SFARG2LO, SFARG1LO
+ | jr ra
+ |. slt CRET1, SFARG2HI, SFARG1HI
+ |5: // Swap conditions if both operands are negative.
+ | beq SFARG1HI, SFARG2HI, >6
+ |. sltu CRET1, SFARG1LO, SFARG2LO
+ | slt CRET1, SFARG1HI, SFARG2HI
+ |6:
+ | jr ra
+ |. nop
+ |8:
+ | jr ra
+ |. li CRET1, 1
+ |9:
+ | jr ra
+ |. li CRET1, 0
+ |.endif
+ |
+ |.macro sfmin_max, name, fpcall
+ |->vm_sf .. name:
+ |.if JIT and not FPU
+ | move TMP2, ra
+ | bal ->fpcall
+ |. nop
+ | move TMP0, CRET1
+ | move SFRETHI, SFARG1HI
+ | move SFRETLO, SFARG1LO
+ | move ra, TMP2
+ | movz SFRETHI, SFARG2HI, TMP0
+ | jr ra
+ |. movz SFRETLO, SFARG2LO, TMP0
+ |.endif
+ |.endmacro
+ |
+ | sfmin_max min, vm_sfcmpolt
+ | sfmin_max max, vm_sfcmpogt
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Miscellaneous functions --------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |.define NEXT_TAB, TAB:CARG1
+ |.define NEXT_IDX, CARG2
+ |.define NEXT_ASIZE, CARG3
+ |.define NEXT_NIL, CARG4
+ |.define NEXT_TMP0, r12
+ |.define NEXT_TMP1, r13
+ |.define NEXT_TMP2, r14
+ |.define NEXT_RES_VK, CRET1
+ |.define NEXT_RES_IDX, CRET2
+ |.define NEXT_RES_PTR, sp
+ |.define NEXT_RES_VAL_I, 0(sp)
+ |.define NEXT_RES_VAL_IT, 4(sp)
+ |.define NEXT_RES_KEY_I, 8(sp)
+ |.define NEXT_RES_KEY_IT, 12(sp)
+ |
+ |// TValue *lj_vm_next(GCtab *t, uint32_t idx)
+ |// Next idx returned in CRET2.
+ |->vm_next:
+ |.if JIT and ENDIAN_LE
+ | lw NEXT_ASIZE, NEXT_TAB->asize
+ | lw NEXT_TMP0, NEXT_TAB->array
+ | li NEXT_NIL, LJ_TNIL
+ |1: // Traverse array part.
+ | sltu AT, NEXT_IDX, NEXT_ASIZE
+ | sll NEXT_TMP1, NEXT_IDX, 3
+ | beqz AT, >5
+ |. addu NEXT_TMP1, NEXT_TMP0, NEXT_TMP1
+ | lw NEXT_TMP2, 4(NEXT_TMP1)
+ | sw NEXT_IDX, NEXT_RES_KEY_I
+ | beq NEXT_TMP2, NEXT_NIL, <1
+ |. addiu NEXT_IDX, NEXT_IDX, 1
+ | lw NEXT_TMP0, 0(NEXT_TMP1)
+ | li AT, LJ_TISNUM
+ | sw NEXT_TMP2, NEXT_RES_VAL_IT
+ | sw AT, NEXT_RES_KEY_IT
+ | sw NEXT_TMP0, NEXT_RES_VAL_I
+ | move NEXT_RES_VK, NEXT_RES_PTR
+ | jr ra
+ |. move NEXT_RES_IDX, NEXT_IDX
+ |
+ |5: // Traverse hash part.
+ | subu NEXT_RES_IDX, NEXT_IDX, NEXT_ASIZE
+ | lw NODE:NEXT_RES_VK, NEXT_TAB->node
+ | sll NEXT_TMP2, NEXT_RES_IDX, 5
+ | lw NEXT_TMP0, NEXT_TAB->hmask
+ | sll AT, NEXT_RES_IDX, 3
+ | subu AT, NEXT_TMP2, AT
+ | addu NODE:NEXT_RES_VK, NODE:NEXT_RES_VK, AT
+ |6:
+ | sltu AT, NEXT_TMP0, NEXT_RES_IDX
+ | bnez AT, >8
+ |. nop
+ | lw NEXT_TMP2, NODE:NEXT_RES_VK->val.it
+ | bne NEXT_TMP2, NEXT_NIL, >9
+ |. addiu NEXT_RES_IDX, NEXT_RES_IDX, 1
+ | // Skip holes in hash part.
+ | b <6
+ |. addiu NODE:NEXT_RES_VK, NODE:NEXT_RES_VK, sizeof(Node)
+ |
+ |8: // End of iteration. Set the key to nil (not the value).
+ | sw NEXT_NIL, NEXT_RES_KEY_IT
+ | move NEXT_RES_VK, NEXT_RES_PTR
+ |9:
+ | jr ra
+ |. addu NEXT_RES_IDX, NEXT_RES_IDX, NEXT_ASIZE
+ |.endif
+ |
+ |//-----------------------------------------------------------------------
+ |//-- FFI helper functions -----------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |// Handler for callback functions. Callback slot number in r1, g in r2.
+ |->vm_ffi_callback:
+ |.if FFI
+ |.type CTSTATE, CTState, PC
+ | saveregs
+ | lw CTSTATE, GL:r2->ctype_state
+ | addiu DISPATCH, r2, GG_G2DISP
+ | load_got lj_ccallback_enter
+ | sw r1, CTSTATE->cb.slot
+ | sw CARG1, CTSTATE->cb.gpr[0]
+ | sw CARG2, CTSTATE->cb.gpr[1]
+ | .FPU sdc1 FARG1, CTSTATE->cb.fpr[0]
+ | sw CARG3, CTSTATE->cb.gpr[2]
+ | sw CARG4, CTSTATE->cb.gpr[3]
+ | .FPU sdc1 FARG2, CTSTATE->cb.fpr[1]
+ | addiu TMP0, sp, CFRAME_SPACE+16
+ | sw TMP0, CTSTATE->cb.stack
+ | sw r0, SAVE_PC // Any value outside of bytecode is ok.
+ | move CARG2, sp
+ | call_intern lj_ccallback_enter // (CTState *cts, void *cf)
+ |. move CARG1, CTSTATE
+ | // Returns lua_State *.
+ | lw BASE, L:CRET1->base
+ | lw RC, L:CRET1->top
+ | li TISNUM, LJ_TISNUM // Setup type comparison constants.
+ | move L, CRET1
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | .FPU mtc1 TMP3, TOBIT
+ | li_vmstate INTERP
+ | li TISNIL, LJ_TNIL
+ | subu RC, RC, BASE
+ | st_vmstate
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | ins_callt
+ |.endif
+ |
+ |->cont_ffi_callback: // Return from FFI callback.
+ |.if FFI
+ | load_got lj_ccallback_leave
+ | lw CTSTATE, DISPATCH_GL(ctype_state)(DISPATCH)
+ | sw BASE, L->base
+ | sw RB, L->top
+ | sw L, CTSTATE->L
+ | move CARG2, RA
+ | call_intern lj_ccallback_leave // (CTState *cts, TValue *o)
+ |. move CARG1, CTSTATE
+ | .FPU ldc1 FRET1, CTSTATE->cb.fpr[0]
+ | lw CRET1, CTSTATE->cb.gpr[0]
+ | .FPU ldc1 FRET2, CTSTATE->cb.fpr[1]
+ | b ->vm_leave_unw
+ |. lw CRET2, CTSTATE->cb.gpr[1]
+ |.endif
+ |
+ |->vm_ffi_call: // Call C function via FFI.
+ | // Caveat: needs special frame unwinding, see below.
+ |.if FFI
+ | .type CCSTATE, CCallState, CARG1
+ | lw TMP1, CCSTATE->spadj
+ | lbu CARG2, CCSTATE->nsp
+ | move TMP2, sp
+ | subu sp, sp, TMP1
+ | sw ra, -4(TMP2)
+ | sll CARG2, CARG2, 2
+ | sw r16, -8(TMP2)
+ | sw CCSTATE, -12(TMP2)
+ | move r16, TMP2
+ | addiu TMP1, CCSTATE, offsetof(CCallState, stack)
+ | addiu TMP2, sp, 16
+ | beqz CARG2, >2
+ |. addu TMP3, TMP1, CARG2
+ |1:
+ | lw TMP0, 0(TMP1)
+ | addiu TMP1, TMP1, 4
+ | sltu AT, TMP1, TMP3
+ | sw TMP0, 0(TMP2)
+ | bnez AT, <1
+ |. addiu TMP2, TMP2, 4
+ |2:
+ | lw CFUNCADDR, CCSTATE->func
+ | lw CARG2, CCSTATE->gpr[1]
+ | lw CARG3, CCSTATE->gpr[2]
+ | lw CARG4, CCSTATE->gpr[3]
+ | .FPU ldc1 FARG1, CCSTATE->fpr[0]
+ | .FPU ldc1 FARG2, CCSTATE->fpr[1]
+ | jalr CFUNCADDR
+ |. lw CARG1, CCSTATE->gpr[0] // Do this last, since CCSTATE is CARG1.
+ | lw CCSTATE:TMP1, -12(r16)
+ | lw TMP2, -8(r16)
+ | lw ra, -4(r16)
+ | sw CRET1, CCSTATE:TMP1->gpr[0]
+ | sw CRET2, CCSTATE:TMP1->gpr[1]
+ |.if FPU
+ | sdc1 FRET1, CCSTATE:TMP1->fpr[0]
+ | sdc1 FRET2, CCSTATE:TMP1->fpr[1]
+ |.else
+ | sw CARG1, CCSTATE:TMP1->gpr[2] // Soft-float: complex double .im part.
+ | sw CARG2, CCSTATE:TMP1->gpr[3]
+ |.endif
+ | move sp, r16
+ | jr ra
+ |. move r16, TMP2
+ |.endif
+ |// Note: vm_ffi_call must be the last function in this object file!
+ |
+ |//-----------------------------------------------------------------------
+}
+
+/* Generate the code for a single instruction. */
+static void build_ins(BuildCtx *ctx, BCOp op, int defop)
+{
+ int vk = 0;
+ |=>defop:
+
+ switch (op) {
+
+ /* -- Comparison ops ---------------------------------------------------- */
+
+ /* Remember: all ops branch for a true comparison, fall through otherwise. */
+
+ case BC_ISLT: case BC_ISGE: case BC_ISLE: case BC_ISGT:
+ | // RA = src1*8, RD = src2*8, JMP with RD = target
+ |.macro bc_comp, FRA, FRD, RAHI, RALO, RDHI, RDLO, movop, fmovop, fcomp, sfcomp
+ | addu RA, BASE, RA
+ | addu RD, BASE, RD
+ | lw RAHI, HI(RA)
+ | lw RDHI, HI(RD)
+ | lhu TMP2, OFS_RD(PC)
+ | addiu PC, PC, 4
+ | bne RAHI, TISNUM, >2
+ |. lw RALO, LO(RA)
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | lw RDLO, LO(RD)
+ | bne RDHI, TISNUM, >5
+ |. decode_RD4b TMP2
+ | slt AT, SFARG1LO, SFARG2LO
+ | addu TMP2, TMP2, TMP3
+ | movop TMP2, r0, AT
+ |1:
+ | addu PC, PC, TMP2
+ | ins_next
+ |
+ |2: // RA is not an integer.
+ | sltiu AT, RAHI, LJ_TISNUM
+ | beqz AT, ->vmeta_comp
+ |. lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | sltiu AT, RDHI, LJ_TISNUM
+ |.if FPU
+ | ldc1 FRA, 0(RA)
+ | ldc1 FRD, 0(RD)
+ |.else
+ | lw RDLO, LO(RD)
+ |.endif
+ | beqz AT, >4
+ |. decode_RD4b TMP2
+ |3: // RA and RD are both numbers.
+ |.if FPU
+ | fcomp f20, f22
+ | addu TMP2, TMP2, TMP3
+ | b <1
+ |. fmovop TMP2, r0
+ |.else
+ | bal sfcomp
+ |. addu TMP2, TMP2, TMP3
+ | b <1
+ |. movop TMP2, r0, CRET1
+ |.endif
+ |
+ |4: // RA is a number, RD is not a number.
+ | bne RDHI, TISNUM, ->vmeta_comp
+ | // RA is a number, RD is an integer. Convert RD to a number.
+ |.if FPU
+ |. lwc1 FRD, LO(RD)
+ | b <3
+ |. cvt.d.w FRD, FRD
+ |.else
+ |. nop
+ |.if "RDHI" == "SFARG1HI"
+ | bal ->vm_sfi2d_1
+ |.else
+ | bal ->vm_sfi2d_2
+ |.endif
+ |. nop
+ | b <3
+ |. nop
+ |.endif
+ |
+ |5: // RA is an integer, RD is not an integer
+ | sltiu AT, RDHI, LJ_TISNUM
+ | beqz AT, ->vmeta_comp
+ | // RA is an integer, RD is a number. Convert RA to a number.
+ |.if FPU
+ |. mtc1 RALO, FRA
+ | ldc1 FRD, 0(RD)
+ | b <3
+ | cvt.d.w FRA, FRA
+ |.else
+ |. nop
+ |.if "RAHI" == "SFARG1HI"
+ | bal ->vm_sfi2d_1
+ |.else
+ | bal ->vm_sfi2d_2
+ |.endif
+ |. nop
+ | b <3
+ |. nop
+ |.endif
+ |.endmacro
+ |
+ if (op == BC_ISLT) {
+ | bc_comp f20, f22, SFARG1HI, SFARG1LO, SFARG2HI, SFARG2LO, movz, movf, c.olt.d, ->vm_sfcmpolt
+ } else if (op == BC_ISGE) {
+ | bc_comp f20, f22, SFARG1HI, SFARG1LO, SFARG2HI, SFARG2LO, movn, movt, c.olt.d, ->vm_sfcmpolt
+ } else if (op == BC_ISLE) {
+ | bc_comp f22, f20, SFARG2HI, SFARG2LO, SFARG1HI, SFARG1LO, movn, movt, c.ult.d, ->vm_sfcmpult
+ } else {
+ | bc_comp f22, f20, SFARG2HI, SFARG2LO, SFARG1HI, SFARG1LO, movz, movf, c.ult.d, ->vm_sfcmpult
+ }
+ break;
+
+ case BC_ISEQV: case BC_ISNEV:
+ vk = op == BC_ISEQV;
+ | // RA = src1*8, RD = src2*8, JMP with RD = target
+ | addu RA, BASE, RA
+ | addiu PC, PC, 4
+ | addu RD, BASE, RD
+ | lw SFARG1HI, HI(RA)
+ | lhu TMP2, -4+OFS_RD(PC)
+ | lw SFARG2HI, HI(RD)
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | sltu AT, TISNUM, SFARG1HI
+ | sltu TMP0, TISNUM, SFARG2HI
+ | or AT, AT, TMP0
+ if (vk) {
+ | beqz AT, ->BC_ISEQN_Z
+ } else {
+ | beqz AT, ->BC_ISNEN_Z
+ }
+ |. decode_RD4b TMP2
+ | // Either or both types are not numbers.
+ | lw SFARG1LO, LO(RA)
+ | lw SFARG2LO, LO(RD)
+ | addu TMP2, TMP2, TMP3
+ |.if FFI
+ | li TMP3, LJ_TCDATA
+ | beq SFARG1HI, TMP3, ->vmeta_equal_cd
+ |.endif
+ |. sltiu AT, SFARG1HI, LJ_TISPRI // Not a primitive?
+ |.if FFI
+ | beq SFARG2HI, TMP3, ->vmeta_equal_cd
+ |.endif
+ |. xor TMP3, SFARG1LO, SFARG2LO // Same tv?
+ | xor SFARG2HI, SFARG2HI, SFARG1HI // Same type?
+ | sltiu TMP0, SFARG1HI, LJ_TISTABUD+1 // Table or userdata?
+ | movz TMP3, r0, AT // Ignore tv if primitive.
+ | movn TMP0, r0, SFARG2HI // Tab/ud and same type?
+ | or AT, SFARG2HI, TMP3 // Same type && (pri||same tv).
+ | movz TMP0, r0, AT
+ | beqz TMP0, >1 // Done if not tab/ud or not same type or same tv.
+ if (vk) {
+ |. movn TMP2, r0, AT
+ } else {
+ |. movz TMP2, r0, AT
+ }
+ | // Different tables or userdatas. Need to check __eq metamethod.
+ | // Field metatable must be at same offset for GCtab and GCudata!
+ | lw TAB:TMP1, TAB:SFARG1LO->metatable
+ | beqz TAB:TMP1, >1 // No metatable?
+ |. nop
+ | lbu TMP1, TAB:TMP1->nomm
+ | andi TMP1, TMP1, 1<<MM_eq
+ | bnez TMP1, >1 // Or 'no __eq' flag set?
+ |. nop
+ | b ->vmeta_equal // Handle __eq metamethod.
+ |. li TMP0, 1-vk // ne = 0 or 1.
+ |1:
+ | addu PC, PC, TMP2
+ | ins_next
+ break;
+
+ case BC_ISEQS: case BC_ISNES:
+ vk = op == BC_ISEQS;
+ | // RA = src*8, RD = str_const*8 (~), JMP with RD = target
+ | addu RA, BASE, RA
+ | addiu PC, PC, 4
+ | lw TMP0, HI(RA)
+ | srl RD, RD, 1
+ | lw STR:TMP3, LO(RA)
+ | subu RD, KBASE, RD
+ | lhu TMP2, -4+OFS_RD(PC)
+ |.if FFI
+ | li AT, LJ_TCDATA
+ | beq TMP0, AT, ->vmeta_equal_cd
+ |.endif
+ |. lw STR:TMP1, -4(RD) // KBASE-4-str_const*4
+ | addiu TMP0, TMP0, -LJ_TSTR
+ | decode_RD4b TMP2
+ | xor TMP1, STR:TMP1, STR:TMP3
+ | or TMP0, TMP0, TMP1
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | addu TMP2, TMP2, TMP3
+ if (vk) {
+ | movn TMP2, r0, TMP0
+ } else {
+ | movz TMP2, r0, TMP0
+ }
+ | addu PC, PC, TMP2
+ | ins_next
+ break;
+
+ case BC_ISEQN: case BC_ISNEN:
+ vk = op == BC_ISEQN;
+ | // RA = src*8, RD = num_const*8, JMP with RD = target
+ | addu RA, BASE, RA
+ | addu RD, KBASE, RD
+ | lw SFARG1HI, HI(RA)
+ | lw SFARG2HI, HI(RD)
+ | lhu TMP2, OFS_RD(PC)
+ | addiu PC, PC, 4
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | decode_RD4b TMP2
+ if (vk) {
+ |->BC_ISEQN_Z:
+ } else {
+ |->BC_ISNEN_Z:
+ }
+ | bne SFARG1HI, TISNUM, >3
+ |. lw SFARG1LO, LO(RA)
+ | lw SFARG2LO, LO(RD)
+ | addu TMP2, TMP2, TMP3
+ | bne SFARG2HI, TISNUM, >6
+ |. xor AT, SFARG1LO, SFARG2LO
+ if (vk) {
+ | movn TMP2, r0, AT
+ |1:
+ | addu PC, PC, TMP2
+ |2:
+ } else {
+ | movz TMP2, r0, AT
+ |1:
+ |2:
+ | addu PC, PC, TMP2
+ }
+ | ins_next
+ |
+ |3: // RA is not an integer.
+ | sltiu AT, SFARG1HI, LJ_TISNUM
+ |.if FFI
+ | beqz AT, >8
+ |.else
+ | beqz AT, <2
+ |.endif
+ |. addu TMP2, TMP2, TMP3
+ | sltiu AT, SFARG2HI, LJ_TISNUM
+ |.if FPU
+ | ldc1 f20, 0(RA)
+ | ldc1 f22, 0(RD)
+ |.endif
+ | beqz AT, >5
+ |. lw SFARG2LO, LO(RD)
+ |4: // RA and RD are both numbers.
+ |.if FPU
+ | c.eq.d f20, f22
+ | b <1
+ if (vk) {
+ |. movf TMP2, r0
+ } else {
+ |. movt TMP2, r0
+ }
+ |.else
+ | bal ->vm_sfcmpeq
+ |. nop
+ | b <1
+ if (vk) {
+ |. movz TMP2, r0, CRET1
+ } else {
+ |. movn TMP2, r0, CRET1
+ }
+ |.endif
+ |
+ |5: // RA is a number, RD is not a number.
+ |.if FFI
+ | bne SFARG2HI, TISNUM, >9
+ |.else
+ | bne SFARG2HI, TISNUM, <2
+ |.endif
+ | // RA is a number, RD is an integer. Convert RD to a number.
+ |.if FPU
+ |. lwc1 f22, LO(RD)
+ | b <4
+ |. cvt.d.w f22, f22
+ |.else
+ |. nop
+ | bal ->vm_sfi2d_2
+ |. nop
+ | b <4
+ |. nop
+ |.endif
+ |
+ |6: // RA is an integer, RD is not an integer
+ | sltiu AT, SFARG2HI, LJ_TISNUM
+ |.if FFI
+ | beqz AT, >9
+ |.else
+ | beqz AT, <2
+ |.endif
+ | // RA is an integer, RD is a number. Convert RA to a number.
+ |.if FPU
+ |. mtc1 SFARG1LO, f20
+ | ldc1 f22, 0(RD)
+ | b <4
+ | cvt.d.w f20, f20
+ |.else
+ |. nop
+ | bal ->vm_sfi2d_1
+ |. nop
+ | b <4
+ |. nop
+ |.endif
+ |
+ |.if FFI
+ |8:
+ | li AT, LJ_TCDATA
+ | bne SFARG1HI, AT, <2
+ |. nop
+ | b ->vmeta_equal_cd
+ |. nop
+ |9:
+ | li AT, LJ_TCDATA
+ | bne SFARG2HI, AT, <2
+ |. nop
+ | b ->vmeta_equal_cd
+ |. nop
+ |.endif
+ break;
+
+ case BC_ISEQP: case BC_ISNEP:
+ vk = op == BC_ISEQP;
+ | // RA = src*8, RD = primitive_type*8 (~), JMP with RD = target
+ | addu RA, BASE, RA
+ | srl TMP1, RD, 3
+ | lw TMP0, HI(RA)
+ | lhu TMP2, OFS_RD(PC)
+ | not TMP1, TMP1
+ | addiu PC, PC, 4
+ |.if FFI
+ | li AT, LJ_TCDATA
+ | beq TMP0, AT, ->vmeta_equal_cd
+ |.endif
+ |. xor TMP0, TMP0, TMP1
+ | decode_RD4b TMP2
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | addu TMP2, TMP2, TMP3
+ if (vk) {
+ | movn TMP2, r0, TMP0
+ } else {
+ | movz TMP2, r0, TMP0
+ }
+ | addu PC, PC, TMP2
+ | ins_next
+ break;
+
+ /* -- Unary test and copy ops ------------------------------------------- */
+
+ case BC_ISTC: case BC_ISFC: case BC_IST: case BC_ISF:
+ | // RA = dst*8 or unused, RD = src*8, JMP with RD = target
+ | addu RD, BASE, RD
+ | lhu TMP2, OFS_RD(PC)
+ | lw TMP0, HI(RD)
+ | addiu PC, PC, 4
+ if (op == BC_IST || op == BC_ISF) {
+ | sltiu TMP0, TMP0, LJ_TISTRUECOND
+ | decode_RD4b TMP2
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | addu TMP2, TMP2, TMP3
+ if (op == BC_IST) {
+ | movz TMP2, r0, TMP0
+ } else {
+ | movn TMP2, r0, TMP0
+ }
+ | addu PC, PC, TMP2
+ } else {
+ | sltiu TMP0, TMP0, LJ_TISTRUECOND
+ | lw SFRETHI, HI(RD)
+ | lw SFRETLO, LO(RD)
+ if (op == BC_ISTC) {
+ | beqz TMP0, >1
+ } else {
+ | bnez TMP0, >1
+ }
+ |. addu RA, BASE, RA
+ | decode_RD4b TMP2
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | addu TMP2, TMP2, TMP3
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | addu PC, PC, TMP2
+ |1:
+ }
+ | ins_next
+ break;
+
+ case BC_ISTYPE:
+ | // RA = src*8, RD = -type*8
+ | addu TMP2, BASE, RA
+ | srl TMP1, RD, 3
+ | lw TMP0, HI(TMP2)
+ | ins_next1
+ | addu AT, TMP0, TMP1
+ | bnez AT, ->vmeta_istype
+ |. ins_next2
+ break;
+ case BC_ISNUM:
+ | // RA = src*8, RD = -(TISNUM-1)*8
+ | addu TMP2, BASE, RA
+ | lw TMP0, HI(TMP2)
+ | ins_next1
+ | sltiu AT, TMP0, LJ_TISNUM
+ | beqz AT, ->vmeta_istype
+ |. ins_next2
+ break;
+
+ /* -- Unary ops --------------------------------------------------------- */
+
+ case BC_MOV:
+ | // RA = dst*8, RD = src*8
+ | addu RD, BASE, RD
+ | addu RA, BASE, RA
+ | lw SFRETHI, HI(RD)
+ | lw SFRETLO, LO(RD)
+ | ins_next1
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | ins_next2
+ break;
+ case BC_NOT:
+ | // RA = dst*8, RD = src*8
+ | addu RD, BASE, RD
+ | addu RA, BASE, RA
+ | lw TMP0, HI(RD)
+ | li TMP1, LJ_TFALSE
+ | sltiu TMP0, TMP0, LJ_TISTRUECOND
+ | addiu TMP1, TMP0, LJ_TTRUE
+ | ins_next1
+ | sw TMP1, HI(RA)
+ | ins_next2
+ break;
+ case BC_UNM:
+ | // RA = dst*8, RD = src*8
+ | addu RB, BASE, RD
+ | lw SFARG1HI, HI(RB)
+ | addu RA, BASE, RA
+ | bne SFARG1HI, TISNUM, >2
+ |. lw SFARG1LO, LO(RB)
+ | lui TMP1, 0x8000
+ | beq SFARG1LO, TMP1, ->vmeta_unm // Meta handler deals with -2^31.
+ |. negu SFARG1LO, SFARG1LO
+ |1:
+ | ins_next1
+ | sw SFARG1HI, HI(RA)
+ | sw SFARG1LO, LO(RA)
+ | ins_next2
+ |2:
+ | sltiu AT, SFARG1HI, LJ_TISNUM
+ | beqz AT, ->vmeta_unm
+ |. lui TMP1, 0x8000
+ | b <1
+ |. xor SFARG1HI, SFARG1HI, TMP1
+ break;
+ case BC_LEN:
+ | // RA = dst*8, RD = src*8
+ | addu CARG2, BASE, RD
+ | addu RA, BASE, RA
+ | lw TMP0, HI(CARG2)
+ | lw CARG1, LO(CARG2)
+ | li AT, LJ_TSTR
+ | bne TMP0, AT, >2
+ |. li AT, LJ_TTAB
+ | lw CRET1, STR:CARG1->len
+ |1:
+ | ins_next1
+ | sw TISNUM, HI(RA)
+ | sw CRET1, LO(RA)
+ | ins_next2
+ |2:
+ | bne TMP0, AT, ->vmeta_len
+ |. nop
+#if LJ_52
+ | lw TAB:TMP2, TAB:CARG1->metatable
+ | bnez TAB:TMP2, >9
+ |. nop
+ |3:
+#endif
+ |->BC_LEN_Z:
+ | load_got lj_tab_len
+ | call_intern lj_tab_len // (GCtab *t)
+ |. nop
+ | // Returns uint32_t (but less than 2^31).
+ | b <1
+ |. nop
+#if LJ_52
+ |9:
+ | lbu TMP0, TAB:TMP2->nomm
+ | andi TMP0, TMP0, 1<<MM_len
+ | bnez TMP0, <3 // 'no __len' flag set: done.
+ |. nop
+ | b ->vmeta_len
+ |. nop
+#endif
+ break;
+
+ /* -- Binary ops -------------------------------------------------------- */
+
+ |.macro fpmod, a, b, c
+ | bal ->vm_floor // floor(b/c)
+ |. div.d FARG1, b, c
+ | mul.d a, FRET1, c
+ | sub.d a, b, a // b - floor(b/c)*c
+ |.endmacro
+
+ |.macro sfpmod
+ | addiu sp, sp, -16
+ |
+ | load_got __divdf3
+ | sw SFARG1HI, HI(sp)
+ | sw SFARG1LO, LO(sp)
+ | sw SFARG2HI, 8+HI(sp)
+ | call_extern
+ |. sw SFARG2LO, 8+LO(sp)
+ |
+ | load_got floor
+ | move SFARG1HI, SFRETHI
+ | call_extern
+ |. move SFARG1LO, SFRETLO
+ |
+ | load_got __muldf3
+ | move SFARG1HI, SFRETHI
+ | move SFARG1LO, SFRETLO
+ | lw SFARG2HI, 8+HI(sp)
+ | call_extern
+ |. lw SFARG2LO, 8+LO(sp)
+ |
+ | load_got __subdf3
+ | lw SFARG1HI, HI(sp)
+ | lw SFARG1LO, LO(sp)
+ | move SFARG2HI, SFRETHI
+ | call_extern
+ |. move SFARG2LO, SFRETLO
+ |
+ | addiu sp, sp, 16
+ |.endmacro
+
+ |.macro ins_arithpre, label
+ ||vk = ((int)op - BC_ADDVN) / (BC_ADDNV-BC_ADDVN);
+ | // RA = dst*8, RB = src1*8, RC = src2*8 | num_const*8
+ ||switch (vk) {
+ ||case 0:
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | // RA = dst*8, RB = src1*8, RC = num_const*8
+ | addu RB, BASE, RB
+ |.if "label" ~= "none"
+ | b label
+ |.endif
+ |. addu RC, KBASE, RC
+ || break;
+ ||case 1:
+ | decode_RB8a RC, INS
+ | decode_RB8b RC
+ | decode_RDtoRC8 RB, RD
+ | // RA = dst*8, RB = num_const*8, RC = src1*8
+ | addu RC, BASE, RC
+ |.if "label" ~= "none"
+ | b label
+ |.endif
+ |. addu RB, KBASE, RB
+ || break;
+ ||default:
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | // RA = dst*8, RB = src1*8, RC = src2*8
+ | addu RB, BASE, RB
+ |.if "label" ~= "none"
+ | b label
+ |.endif
+ |. addu RC, BASE, RC
+ || break;
+ ||}
+ |.endmacro
+ |
+ |.macro ins_arith, intins, fpins, fpcall, label
+ | ins_arithpre none
+ |
+ |.if "label" ~= "none"
+ |label:
+ |.endif
+ |
+ | lw SFARG1HI, HI(RB)
+ | lw SFARG2HI, HI(RC)
+ |
+ |.if "intins" ~= "div"
+ |
+ | // Check for two integers.
+ | lw SFARG1LO, LO(RB)
+ | bne SFARG1HI, TISNUM, >5
+ |. lw SFARG2LO, LO(RC)
+ | bne SFARG2HI, TISNUM, >5
+ |
+ |.if "intins" == "addu"
+ |. intins CRET1, SFARG1LO, SFARG2LO
+ | xor TMP1, CRET1, SFARG1LO // ((y^a) & (y^b)) < 0: overflow.
+ | xor TMP2, CRET1, SFARG2LO
+ | and TMP1, TMP1, TMP2
+ | bltz TMP1, ->vmeta_arith
+ |. addu RA, BASE, RA
+ |.elif "intins" == "subu"
+ |. intins CRET1, SFARG1LO, SFARG2LO
+ | xor TMP1, CRET1, SFARG1LO // ((y^a) & (a^b)) < 0: overflow.
+ | xor TMP2, SFARG1LO, SFARG2LO
+ | and TMP1, TMP1, TMP2
+ | bltz TMP1, ->vmeta_arith
+ |. addu RA, BASE, RA
+ |.elif "intins" == "mult"
+ |. intins SFARG1LO, SFARG2LO
+ | mflo CRET1
+ | mfhi TMP2
+ | sra TMP1, CRET1, 31
+ | bne TMP1, TMP2, ->vmeta_arith
+ |. addu RA, BASE, RA
+ |.else
+ |. load_got lj_vm_modi
+ | beqz SFARG2LO, ->vmeta_arith
+ |. addu RA, BASE, RA
+ |.if ENDIAN_BE
+ | move CARG1, SFARG1LO
+ |.endif
+ | call_extern
+ |. move CARG2, SFARG2LO
+ |.endif
+ |
+ | ins_next1
+ | sw TISNUM, HI(RA)
+ | sw CRET1, LO(RA)
+ |3:
+ | ins_next2
+ |
+ |.elif not FPU
+ |
+ | lw SFARG1LO, LO(RB)
+ | lw SFARG2LO, LO(RC)
+ |
+ |.endif
+ |
+ |5: // Check for two numbers.
+ | .FPU ldc1 f20, 0(RB)
+ | sltiu AT, SFARG1HI, LJ_TISNUM
+ | sltiu TMP0, SFARG2HI, LJ_TISNUM
+ | .FPU ldc1 f22, 0(RC)
+ | and AT, AT, TMP0
+ | beqz AT, ->vmeta_arith
+ |. addu RA, BASE, RA
+ |
+ |.if FPU
+ | fpins FRET1, f20, f22
+ |.elif "fpcall" == "sfpmod"
+ | sfpmod
+ |.else
+ | load_got fpcall
+ | call_extern
+ |. nop
+ |.endif
+ |
+ | ins_next1
+ |.if not FPU
+ | sw SFRETHI, HI(RA)
+ |.endif
+ |.if "intins" ~= "div"
+ | b <3
+ |.endif
+ |.if FPU
+ |. sdc1 FRET1, 0(RA)
+ |.else
+ |. sw SFRETLO, LO(RA)
+ |.endif
+ |.if "intins" == "div"
+ | ins_next2
+ |.endif
+ |
+ |.endmacro
+
+ case BC_ADDVN: case BC_ADDNV: case BC_ADDVV:
+ | ins_arith addu, add.d, __adddf3, none
+ break;
+ case BC_SUBVN: case BC_SUBNV: case BC_SUBVV:
+ | ins_arith subu, sub.d, __subdf3, none
+ break;
+ case BC_MULVN: case BC_MULNV: case BC_MULVV:
+ | ins_arith mult, mul.d, __muldf3, none
+ break;
+ case BC_DIVVN:
+ | ins_arith div, div.d, __divdf3, ->BC_DIVVN_Z
+ break;
+ case BC_DIVNV: case BC_DIVVV:
+ | ins_arithpre ->BC_DIVVN_Z
+ break;
+ case BC_MODVN:
+ | ins_arith modi, fpmod, sfpmod, ->BC_MODVN_Z
+ break;
+ case BC_MODNV: case BC_MODVV:
+ | ins_arithpre ->BC_MODVN_Z
+ break;
+ case BC_POW:
+ | ins_arithpre none
+ | lw SFARG1HI, HI(RB)
+ | lw SFARG2HI, HI(RC)
+ | sltiu AT, SFARG1HI, LJ_TISNUM
+ | sltiu TMP0, SFARG2HI, LJ_TISNUM
+ | and AT, AT, TMP0
+ | load_got pow
+ | beqz AT, ->vmeta_arith
+ |. addu RA, BASE, RA
+ |.if FPU
+ | ldc1 FARG1, 0(RB)
+ | ldc1 FARG2, 0(RC)
+ |.else
+ | lw SFARG1LO, LO(RB)
+ | lw SFARG2LO, LO(RC)
+ |.endif
+ | call_extern
+ |. nop
+ | ins_next1
+ |.if FPU
+ | sdc1 FRET1, 0(RA)
+ |.else
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ |.endif
+ | ins_next2
+ break;
+
+ case BC_CAT:
+ | // RA = dst*8, RB = src_start*8, RC = src_end*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | subu CARG3, RC, RB
+ | sw BASE, L->base
+ | addu CARG2, BASE, RC
+ | move MULTRES, RB
+ |->BC_CAT_Z:
+ | load_got lj_meta_cat
+ | srl CARG3, CARG3, 3
+ | sw PC, SAVE_PC
+ | call_intern lj_meta_cat // (lua_State *L, TValue *top, int left)
+ |. move CARG1, L
+ | // Returns NULL (finished) or TValue * (metamethod).
+ | bnez CRET1, ->vmeta_binop
+ |. lw BASE, L->base
+ | addu RB, BASE, MULTRES
+ | lw SFRETHI, HI(RB)
+ | lw SFRETLO, LO(RB)
+ | addu RA, BASE, RA
+ | ins_next1
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | ins_next2
+ break;
+
+ /* -- Constant ops ------------------------------------------------------ */
+
+ case BC_KSTR:
+ | // RA = dst*8, RD = str_const*8 (~)
+ | srl TMP1, RD, 1
+ | subu TMP1, KBASE, TMP1
+ | ins_next1
+ | lw TMP0, -4(TMP1) // KBASE-4-str_const*4
+ | addu RA, BASE, RA
+ | li TMP2, LJ_TSTR
+ | sw TMP0, LO(RA)
+ | sw TMP2, HI(RA)
+ | ins_next2
+ break;
+ case BC_KCDATA:
+ |.if FFI
+ | // RA = dst*8, RD = cdata_const*8 (~)
+ | srl TMP1, RD, 1
+ | subu TMP1, KBASE, TMP1
+ | ins_next1
+ | lw TMP0, -4(TMP1) // KBASE-4-cdata_const*4
+ | addu RA, BASE, RA
+ | li TMP2, LJ_TCDATA
+ | sw TMP0, LO(RA)
+ | sw TMP2, HI(RA)
+ | ins_next2
+ |.endif
+ break;
+ case BC_KSHORT:
+ | // RA = dst*8, RD = int16_literal*8
+ | sra RD, INS, 16
+ | addu RA, BASE, RA
+ | ins_next1
+ | sw TISNUM, HI(RA)
+ | sw RD, LO(RA)
+ | ins_next2
+ break;
+ case BC_KNUM:
+ | // RA = dst*8, RD = num_const*8
+ | addu RD, KBASE, RD
+ | addu RA, BASE, RA
+ | lw SFRETHI, HI(RD)
+ | lw SFRETLO, LO(RD)
+ | ins_next1
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | ins_next2
+ break;
+ case BC_KPRI:
+ | // RA = dst*8, RD = primitive_type*8 (~)
+ | srl TMP1, RD, 3
+ | addu RA, BASE, RA
+ | not TMP0, TMP1
+ | ins_next1
+ | sw TMP0, HI(RA)
+ | ins_next2
+ break;
+ case BC_KNIL:
+ | // RA = base*8, RD = end*8
+ | addu RA, BASE, RA
+ | sw TISNIL, HI(RA)
+ | addiu RA, RA, 8
+ | addu RD, BASE, RD
+ |1:
+ | sw TISNIL, HI(RA)
+ | slt AT, RA, RD
+ | bnez AT, <1
+ |. addiu RA, RA, 8
+ | ins_next_
+ break;
+
+ /* -- Upvalue and function ops ------------------------------------------ */
+
+ case BC_UGET:
+ | // RA = dst*8, RD = uvnum*8
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | srl RD, RD, 1
+ | addu RD, RD, LFUNC:RB
+ | lw UPVAL:RB, LFUNC:RD->uvptr
+ | ins_next1
+ | lw TMP1, UPVAL:RB->v
+ | lw SFRETHI, HI(TMP1)
+ | lw SFRETLO, LO(TMP1)
+ | addu RA, BASE, RA
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | ins_next2
+ break;
+ case BC_USETV:
+ | // RA = uvnum*8, RD = src*8
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | srl RA, RA, 1
+ | addu RD, BASE, RD
+ | addu RA, RA, LFUNC:RB
+ | lw UPVAL:RB, LFUNC:RA->uvptr
+ | lw SFRETHI, HI(RD)
+ | lw SFRETLO, LO(RD)
+ | lbu TMP3, UPVAL:RB->marked
+ | lw CARG2, UPVAL:RB->v
+ | andi TMP3, TMP3, LJ_GC_BLACK // isblack(uv)
+ | lbu TMP0, UPVAL:RB->closed
+ | sw SFRETHI, HI(CARG2)
+ | sw SFRETLO, LO(CARG2)
+ | li AT, LJ_GC_BLACK|1
+ | or TMP3, TMP3, TMP0
+ | beq TMP3, AT, >2 // Upvalue is closed and black?
+ |. addiu TMP2, SFRETHI, -(LJ_TNUMX+1)
+ |1:
+ | ins_next
+ |
+ |2: // Check if new value is collectable.
+ | sltiu AT, TMP2, LJ_TISGCV - (LJ_TNUMX+1)
+ | beqz AT, <1 // tvisgcv(v)
+ |. nop
+ | lbu TMP3, GCOBJ:SFRETLO->gch.marked
+ | andi TMP3, TMP3, LJ_GC_WHITES // iswhite(v)
+ | beqz TMP3, <1
+ |. load_got lj_gc_barrieruv
+ | // Crossed a write barrier. Move the barrier forward.
+ | call_intern lj_gc_barrieruv // (global_State *g, TValue *tv)
+ |. addiu CARG1, DISPATCH, GG_DISP2G
+ | b <1
+ |. nop
+ break;
+ case BC_USETS:
+ | // RA = uvnum*8, RD = str_const*8 (~)
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | srl RA, RA, 1
+ | srl TMP1, RD, 1
+ | addu RA, RA, LFUNC:RB
+ | subu TMP1, KBASE, TMP1
+ | lw UPVAL:RB, LFUNC:RA->uvptr
+ | lw STR:TMP1, -4(TMP1) // KBASE-4-str_const*4
+ | lbu TMP2, UPVAL:RB->marked
+ | lw CARG2, UPVAL:RB->v
+ | lbu TMP3, STR:TMP1->marked
+ | andi AT, TMP2, LJ_GC_BLACK // isblack(uv)
+ | lbu TMP2, UPVAL:RB->closed
+ | li TMP0, LJ_TSTR
+ | sw STR:TMP1, LO(CARG2)
+ | bnez AT, >2
+ |. sw TMP0, HI(CARG2)
+ |1:
+ | ins_next
+ |
+ |2: // Check if string is white and ensure upvalue is closed.
+ | beqz TMP2, <1
+ |. andi AT, TMP3, LJ_GC_WHITES // iswhite(str)
+ | beqz AT, <1
+ |. load_got lj_gc_barrieruv
+ | // Crossed a write barrier. Move the barrier forward.
+ | call_intern lj_gc_barrieruv // (global_State *g, TValue *tv)
+ |. addiu CARG1, DISPATCH, GG_DISP2G
+ | b <1
+ |. nop
+ break;
+ case BC_USETN:
+ | // RA = uvnum*8, RD = num_const*8
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | srl RA, RA, 1
+ | addu RD, KBASE, RD
+ | addu RA, RA, LFUNC:RB
+ | lw UPVAL:RB, LFUNC:RA->uvptr
+ | lw SFRETHI, HI(RD)
+ | lw SFRETLO, LO(RD)
+ | lw TMP1, UPVAL:RB->v
+ | ins_next1
+ | sw SFRETHI, HI(TMP1)
+ | sw SFRETLO, LO(TMP1)
+ | ins_next2
+ break;
+ case BC_USETP:
+ | // RA = uvnum*8, RD = primitive_type*8 (~)
+ | lw LFUNC:RB, FRAME_FUNC(BASE)
+ | srl RA, RA, 1
+ | srl TMP0, RD, 3
+ | addu RA, RA, LFUNC:RB
+ | not TMP0, TMP0
+ | lw UPVAL:RB, LFUNC:RA->uvptr
+ | ins_next1
+ | lw TMP1, UPVAL:RB->v
+ | sw TMP0, HI(TMP1)
+ | ins_next2
+ break;
+
+ case BC_UCLO:
+ | // RA = level*8, RD = target
+ | lw TMP2, L->openupval
+ | branch_RD // Do this first since RD is not saved.
+ | load_got lj_func_closeuv
+ | sw BASE, L->base
+ | beqz TMP2, >1
+ |. move CARG1, L
+ | call_intern lj_func_closeuv // (lua_State *L, TValue *level)
+ |. addu CARG2, BASE, RA
+ | lw BASE, L->base
+ |1:
+ | ins_next
+ break;
+
+ case BC_FNEW:
+ | // RA = dst*8, RD = proto_const*8 (~) (holding function prototype)
+ | srl TMP1, RD, 1
+ | load_got lj_func_newL_gc
+ | subu TMP1, KBASE, TMP1
+ | lw CARG3, FRAME_FUNC(BASE)
+ | lw CARG2, -4(TMP1) // KBASE-4-tab_const*4
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | // (lua_State *L, GCproto *pt, GCfuncL *parent)
+ | call_intern lj_func_newL_gc
+ |. move CARG1, L
+ | // Returns GCfuncL *.
+ | lw BASE, L->base
+ | li TMP0, LJ_TFUNC
+ | ins_next1
+ | addu RA, BASE, RA
+ | sw LFUNC:CRET1, LO(RA)
+ | sw TMP0, HI(RA)
+ | ins_next2
+ break;
+
+ /* -- Table ops --------------------------------------------------------- */
+
+ case BC_TNEW:
+ case BC_TDUP:
+ | // RA = dst*8, RD = (hbits|asize)*8 | tab_const*8 (~)
+ | lw TMP0, DISPATCH_GL(gc.total)(DISPATCH)
+ | lw TMP1, DISPATCH_GL(gc.threshold)(DISPATCH)
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | sltu AT, TMP0, TMP1
+ | beqz AT, >5
+ |1:
+ if (op == BC_TNEW) {
+ | load_got lj_tab_new
+ | srl CARG2, RD, 3
+ | andi CARG2, CARG2, 0x7ff
+ | li TMP0, 0x801
+ | addiu AT, CARG2, -0x7ff
+ | srl CARG3, RD, 14
+ | movz CARG2, TMP0, AT
+ | // (lua_State *L, int32_t asize, uint32_t hbits)
+ | call_intern lj_tab_new
+ |. move CARG1, L
+ | // Returns Table *.
+ } else {
+ | load_got lj_tab_dup
+ | srl TMP1, RD, 1
+ | subu TMP1, KBASE, TMP1
+ | move CARG1, L
+ | call_intern lj_tab_dup // (lua_State *L, Table *kt)
+ |. lw CARG2, -4(TMP1) // KBASE-4-str_const*4
+ | // Returns Table *.
+ }
+ | lw BASE, L->base
+ | ins_next1
+ | addu RA, BASE, RA
+ | li TMP0, LJ_TTAB
+ | sw TAB:CRET1, LO(RA)
+ | sw TMP0, HI(RA)
+ | ins_next2
+ |5:
+ | load_got lj_gc_step_fixtop
+ | move MULTRES, RD
+ | call_intern lj_gc_step_fixtop // (lua_State *L)
+ |. move CARG1, L
+ | b <1
+ |. move RD, MULTRES
+ break;
+
+ case BC_GGET:
+ | // RA = dst*8, RD = str_const*8 (~)
+ case BC_GSET:
+ | // RA = src*8, RD = str_const*8 (~)
+ | lw LFUNC:TMP2, FRAME_FUNC(BASE)
+ | srl TMP1, RD, 1
+ | subu TMP1, KBASE, TMP1
+ | lw TAB:RB, LFUNC:TMP2->env
+ | lw STR:RC, -4(TMP1) // KBASE-4-str_const*4
+ if (op == BC_GGET) {
+ | b ->BC_TGETS_Z
+ } else {
+ | b ->BC_TSETS_Z
+ }
+ |. addu RA, BASE, RA
+ break;
+
+ case BC_TGETV:
+ | // RA = dst*8, RB = table*8, RC = key*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | addu CARG2, BASE, RB
+ | addu CARG3, BASE, RC
+ | lw TMP1, HI(CARG2)
+ | lw TMP2, HI(CARG3)
+ | lw TAB:RB, LO(CARG2)
+ | li AT, LJ_TTAB
+ | bne TMP1, AT, ->vmeta_tgetv
+ |. addu RA, BASE, RA
+ | bne TMP2, TISNUM, >5
+ |. lw RC, LO(CARG3)
+ | lw TMP0, TAB:RB->asize
+ | lw TMP1, TAB:RB->array
+ | sltu AT, RC, TMP0
+ | sll TMP2, RC, 3
+ | beqz AT, ->vmeta_tgetv // Integer key and in array part?
+ |. addu TMP2, TMP1, TMP2
+ | lw SFRETHI, HI(TMP2)
+ | beq SFRETHI, TISNIL, >2
+ |. lw SFRETLO, LO(TMP2)
+ |1:
+ | ins_next1
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | ins_next2
+ |
+ |2: // Check for __index if table value is nil.
+ | lw TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, <1 // No metatable: done.
+ |. nop
+ | lbu TMP0, TAB:TMP2->nomm
+ | andi TMP0, TMP0, 1<<MM_index
+ | bnez TMP0, <1 // 'no __index' flag set: done.
+ |. nop
+ | b ->vmeta_tgetv
+ |. nop
+ |
+ |5:
+ | li AT, LJ_TSTR
+ | bne TMP2, AT, ->vmeta_tgetv
+ |. nop
+ | b ->BC_TGETS_Z // String key?
+ |. nop
+ break;
+ case BC_TGETS:
+ | // RA = dst*8, RB = table*8, RC = str_const*4 (~)
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | addu CARG2, BASE, RB
+ | decode_RC4a RC, INS
+ | lw TMP0, HI(CARG2)
+ | decode_RC4b RC
+ | li AT, LJ_TTAB
+ | lw TAB:RB, LO(CARG2)
+ | subu CARG3, KBASE, RC
+ | lw STR:RC, -4(CARG3) // KBASE-4-str_const*4
+ | bne TMP0, AT, ->vmeta_tgets1
+ |. addu RA, BASE, RA
+ |->BC_TGETS_Z:
+ | // TAB:RB = GCtab *, STR:RC = GCstr *, RA = dst*8
+ | lw TMP0, TAB:RB->hmask
+ | lw TMP1, STR:RC->sid
+ | lw NODE:TMP2, TAB:RB->node
+ | and TMP1, TMP1, TMP0 // idx = str->sid & tab->hmask
+ | sll TMP0, TMP1, 5
+ | sll TMP1, TMP1, 3
+ | subu TMP1, TMP0, TMP1
+ | addu NODE:TMP2, NODE:TMP2, TMP1 // node = tab->node + (idx*32-idx*8)
+ |1:
+ | lw CARG1, offsetof(Node, key)+HI(NODE:TMP2)
+ | lw TMP0, offsetof(Node, key)+LO(NODE:TMP2)
+ | lw NODE:TMP1, NODE:TMP2->next
+ | lw SFRETHI, offsetof(Node, val)+HI(NODE:TMP2)
+ | addiu CARG1, CARG1, -LJ_TSTR
+ | xor TMP0, TMP0, STR:RC
+ | or AT, CARG1, TMP0
+ | bnez AT, >4
+ |. lw TAB:TMP3, TAB:RB->metatable
+ | beq SFRETHI, TISNIL, >5 // Key found, but nil value?
+ |. lw SFRETLO, offsetof(Node, val)+LO(NODE:TMP2)
+ |3:
+ | ins_next1
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | ins_next2
+ |
+ |4: // Follow hash chain.
+ | bnez NODE:TMP1, <1
+ |. move NODE:TMP2, NODE:TMP1
+ | // End of hash chain: key not found, nil result.
+ |
+ |5: // Check for __index if table value is nil.
+ | beqz TAB:TMP3, <3 // No metatable: done.
+ |. li SFRETHI, LJ_TNIL
+ | lbu TMP0, TAB:TMP3->nomm
+ | andi TMP0, TMP0, 1<<MM_index
+ | bnez TMP0, <3 // 'no __index' flag set: done.
+ |. nop
+ | b ->vmeta_tgets
+ |. nop
+ break;
+ case BC_TGETB:
+ | // RA = dst*8, RB = table*8, RC = index*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | addu CARG2, BASE, RB
+ | decode_RDtoRC8 RC, RD
+ | lw CARG1, HI(CARG2)
+ | li AT, LJ_TTAB
+ | lw TAB:RB, LO(CARG2)
+ | addu RA, BASE, RA
+ | bne CARG1, AT, ->vmeta_tgetb
+ |. srl TMP0, RC, 3
+ | lw TMP1, TAB:RB->asize
+ | lw TMP2, TAB:RB->array
+ | sltu AT, TMP0, TMP1
+ | beqz AT, ->vmeta_tgetb
+ |. addu RC, TMP2, RC
+ | lw SFRETHI, HI(RC)
+ | beq SFRETHI, TISNIL, >5
+ |. lw SFRETLO, LO(RC)
+ |1:
+ | ins_next1
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | ins_next2
+ |
+ |5: // Check for __index if table value is nil.
+ | lw TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, <1 // No metatable: done.
+ |. nop
+ | lbu TMP1, TAB:TMP2->nomm
+ | andi TMP1, TMP1, 1<<MM_index
+ | bnez TMP1, <1 // 'no __index' flag set: done.
+ |. nop
+ | b ->vmeta_tgetb // Caveat: preserve TMP0 and CARG2!
+ |. nop
+ break;
+ case BC_TGETR:
+ | // RA = dst*8, RB = table*8, RC = key*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | addu RB, BASE, RB
+ | addu RC, BASE, RC
+ | lw TAB:CARG1, LO(RB)
+ | lw CARG2, LO(RC)
+ | addu RA, BASE, RA
+ | lw TMP0, TAB:CARG1->asize
+ | lw TMP1, TAB:CARG1->array
+ | sltu AT, CARG2, TMP0
+ | sll TMP2, CARG2, 3
+ | beqz AT, ->vmeta_tgetr // In array part?
+ |. addu CRET1, TMP1, TMP2
+ | lw SFARG2HI, HI(CRET1)
+ | lw SFARG2LO, LO(CRET1)
+ |->BC_TGETR_Z:
+ | ins_next1
+ | sw SFARG2HI, HI(RA)
+ | sw SFARG2LO, LO(RA)
+ | ins_next2
+ break;
+
+ case BC_TSETV:
+ | // RA = src*8, RB = table*8, RC = key*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | addu CARG2, BASE, RB
+ | addu CARG3, BASE, RC
+ | lw TMP1, HI(CARG2)
+ | lw TMP2, HI(CARG3)
+ | lw TAB:RB, LO(CARG2)
+ | li AT, LJ_TTAB
+ | bne TMP1, AT, ->vmeta_tsetv
+ |. addu RA, BASE, RA
+ | bne TMP2, TISNUM, >5
+ |. lw RC, LO(CARG3)
+ | lw TMP0, TAB:RB->asize
+ | lw TMP1, TAB:RB->array
+ | sltu AT, RC, TMP0
+ | sll TMP2, RC, 3
+ | beqz AT, ->vmeta_tsetv // Integer key and in array part?
+ |. addu TMP1, TMP1, TMP2
+ | lw TMP0, HI(TMP1)
+ | lbu TMP3, TAB:RB->marked
+ | lw SFRETHI, HI(RA)
+ | beq TMP0, TISNIL, >3
+ |. lw SFRETLO, LO(RA)
+ |1:
+ | andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ | sw SFRETHI, HI(TMP1)
+ | bnez AT, >7
+ |. sw SFRETLO, LO(TMP1)
+ |2:
+ | ins_next
+ |
+ |3: // Check for __newindex if previous value is nil.
+ | lw TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, <1 // No metatable: done.
+ |. nop
+ | lbu TMP2, TAB:TMP2->nomm
+ | andi TMP2, TMP2, 1<<MM_newindex
+ | bnez TMP2, <1 // 'no __newindex' flag set: done.
+ |. nop
+ | b ->vmeta_tsetv
+ |. nop
+ |
+ |5:
+ | li AT, LJ_TSTR
+ | bne TMP2, AT, ->vmeta_tsetv
+ |. nop
+ | b ->BC_TSETS_Z // String key?
+ |. nop
+ |
+ |7: // Possible table write barrier for the value. Skip valiswhite check.
+ | barrierback TAB:RB, TMP3, TMP0, <2
+ break;
+ case BC_TSETS:
+ | // RA = src*8, RB = table*8, RC = str_const*8 (~)
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | addu CARG2, BASE, RB
+ | decode_RC4a RC, INS
+ | lw TMP0, HI(CARG2)
+ | decode_RC4b RC
+ | li AT, LJ_TTAB
+ | subu CARG3, KBASE, RC
+ | lw TAB:RB, LO(CARG2)
+ | lw STR:RC, -4(CARG3) // KBASE-4-str_const*4
+ | bne TMP0, AT, ->vmeta_tsets1
+ |. addu RA, BASE, RA
+ |->BC_TSETS_Z:
+ | // TAB:RB = GCtab *, STR:RC = GCstr *, RA = BASE+src*8
+ | lw TMP0, TAB:RB->hmask
+ | lw TMP1, STR:RC->sid
+ | lw NODE:TMP2, TAB:RB->node
+ | sb r0, TAB:RB->nomm // Clear metamethod cache.
+ | and TMP1, TMP1, TMP0 // idx = str->sid & tab->hmask
+ | sll TMP0, TMP1, 5
+ | sll TMP1, TMP1, 3
+ | subu TMP1, TMP0, TMP1
+ | addu NODE:TMP2, NODE:TMP2, TMP1 // node = tab->node + (idx*32-idx*8)
+ |.if FPU
+ | ldc1 f20, 0(RA)
+ |.else
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ |.endif
+ |1:
+ | lw CARG1, offsetof(Node, key)+HI(NODE:TMP2)
+ | lw TMP0, offsetof(Node, key)+LO(NODE:TMP2)
+ | li AT, LJ_TSTR
+ | lw NODE:TMP1, NODE:TMP2->next
+ | bne CARG1, AT, >5
+ |. lw CARG2, offsetof(Node, val)+HI(NODE:TMP2)
+ | bne TMP0, STR:RC, >5
+ |. lbu TMP3, TAB:RB->marked
+ | beq CARG2, TISNIL, >4 // Key found, but nil value?
+ |. lw TAB:TMP0, TAB:RB->metatable
+ |2:
+ | andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ |.if FPU
+ | bnez AT, >7
+ |. sdc1 f20, NODE:TMP2->val
+ |.else
+ | sw SFRETHI, NODE:TMP2->val.u32.hi
+ | bnez AT, >7
+ |. sw SFRETLO, NODE:TMP2->val.u32.lo
+ |.endif
+ |3:
+ | ins_next
+ |
+ |4: // Check for __newindex if previous value is nil.
+ | beqz TAB:TMP0, <2 // No metatable: done.
+ |. nop
+ | lbu TMP0, TAB:TMP0->nomm
+ | andi TMP0, TMP0, 1<<MM_newindex
+ | bnez TMP0, <2 // 'no __newindex' flag set: done.
+ |. nop
+ | b ->vmeta_tsets
+ |. nop
+ |
+ |5: // Follow hash chain.
+ | bnez NODE:TMP1, <1
+ |. move NODE:TMP2, NODE:TMP1
+ | // End of hash chain: key not found, add a new one
+ |
+ | // But check for __newindex first.
+ | lw TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, >6 // No metatable: continue.
+ |. addiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | lbu TMP0, TAB:TMP2->nomm
+ | andi TMP0, TMP0, 1<<MM_newindex
+ | beqz TMP0, ->vmeta_tsets // 'no __newindex' flag NOT set: check.
+ |. li AT, LJ_TSTR
+ |6:
+ | load_got lj_tab_newkey
+ | sw STR:RC, LO(CARG3)
+ | sw AT, HI(CARG3)
+ | sw BASE, L->base
+ | move CARG2, TAB:RB
+ | sw PC, SAVE_PC
+ | call_intern lj_tab_newkey // (lua_State *L, GCtab *t, TValue *k
+ |. move CARG1, L
+ | // Returns TValue *.
+ | lw BASE, L->base
+ |.if FPU
+ | b <3 // No 2nd write barrier needed.
+ |. sdc1 f20, 0(CRET1)
+ |.else
+ | lw SFARG1HI, HI(RA)
+ | lw SFARG1LO, LO(RA)
+ | sw SFARG1HI, HI(CRET1)
+ | b <3 // No 2nd write barrier needed.
+ |. sw SFARG1LO, LO(CRET1)
+ |.endif
+ |
+ |7: // Possible table write barrier for the value. Skip valiswhite check.
+ | barrierback TAB:RB, TMP3, TMP0, <3
+ break;
+ case BC_TSETB:
+ | // RA = src*8, RB = table*8, RC = index*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | addu CARG2, BASE, RB
+ | decode_RDtoRC8 RC, RD
+ | lw CARG1, HI(CARG2)
+ | li AT, LJ_TTAB
+ | lw TAB:RB, LO(CARG2)
+ | addu RA, BASE, RA
+ | bne CARG1, AT, ->vmeta_tsetb
+ |. srl TMP0, RC, 3
+ | lw TMP1, TAB:RB->asize
+ | lw TMP2, TAB:RB->array
+ | sltu AT, TMP0, TMP1
+ | beqz AT, ->vmeta_tsetb
+ |. addu RC, TMP2, RC
+ | lw TMP1, HI(RC)
+ | lbu TMP3, TAB:RB->marked
+ | beq TMP1, TISNIL, >5
+ |1:
+ |. lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ | andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ | sw SFRETHI, HI(RC)
+ | bnez AT, >7
+ |. sw SFRETLO, LO(RC)
+ |2:
+ | ins_next
+ |
+ |5: // Check for __newindex if previous value is nil.
+ | lw TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, <1 // No metatable: done.
+ |. nop
+ | lbu TMP1, TAB:TMP2->nomm
+ | andi TMP1, TMP1, 1<<MM_newindex
+ | bnez TMP1, <1 // 'no __newindex' flag set: done.
+ |. nop
+ | b ->vmeta_tsetb // Caveat: preserve TMP0 and CARG2!
+ |. nop
+ |
+ |7: // Possible table write barrier for the value. Skip valiswhite check.
+ | barrierback TAB:RB, TMP3, TMP0, <2
+ break;
+ case BC_TSETR:
+ | // RA = dst*8, RB = table*8, RC = key*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | addu CARG1, BASE, RB
+ | addu CARG3, BASE, RC
+ | lw TAB:CARG2, LO(CARG1)
+ | lw CARG3, LO(CARG3)
+ | lbu TMP3, TAB:CARG2->marked
+ | lw TMP0, TAB:CARG2->asize
+ | lw TMP1, TAB:CARG2->array
+ | andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ | bnez AT, >7
+ |. addu RA, BASE, RA
+ |2:
+ | sltu AT, CARG3, TMP0
+ | sll TMP2, CARG3, 3
+ | beqz AT, ->vmeta_tsetr // In array part?
+ |. addu CRET1, TMP1, TMP2
+ |->BC_TSETR_Z:
+ | lw SFARG1HI, HI(RA)
+ | lw SFARG1LO, LO(RA)
+ | ins_next1
+ | sw SFARG1HI, HI(CRET1)
+ | sw SFARG1LO, LO(CRET1)
+ | ins_next2
+ |
+ |7: // Possible table write barrier for the value. Skip valiswhite check.
+ | barrierback TAB:CARG2, TMP3, CRET1, <2
+ break;
+
+ case BC_TSETM:
+ | // RA = base*8 (table at base-1), RD = num_const*8 (start index)
+ | addu RA, BASE, RA
+ |1:
+ | addu TMP3, KBASE, RD
+ | lw TAB:CARG2, -8+LO(RA) // Guaranteed to be a table.
+ | addiu TMP0, MULTRES, -8
+ | lw TMP3, LO(TMP3) // Integer constant is in lo-word.
+ | beqz TMP0, >4 // Nothing to copy?
+ |. srl CARG3, TMP0, 3
+ | addu CARG3, CARG3, TMP3
+ | lw TMP2, TAB:CARG2->asize
+ | sll TMP1, TMP3, 3
+ | lbu TMP3, TAB:CARG2->marked
+ | lw CARG1, TAB:CARG2->array
+ | sltu AT, TMP2, CARG3
+ | bnez AT, >5
+ |. addu TMP2, RA, TMP0
+ | addu TMP1, TMP1, CARG1
+ | andi TMP0, TMP3, LJ_GC_BLACK // isblack(table)
+ |3: // Copy result slots to table.
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ | addiu RA, RA, 8
+ | sltu AT, RA, TMP2
+ | sw SFRETHI, HI(TMP1)
+ | sw SFRETLO, LO(TMP1)
+ | bnez AT, <3
+ |. addiu TMP1, TMP1, 8
+ | bnez TMP0, >7
+ |. nop
+ |4:
+ | ins_next
+ |
+ |5: // Need to resize array part.
+ | load_got lj_tab_reasize
+ | sw BASE, L->base
+ | sw PC, SAVE_PC
+ | move BASE, RD
+ | call_intern lj_tab_reasize // (lua_State *L, GCtab *t, int nasize)
+ |. move CARG1, L
+ | // Must not reallocate the stack.
+ | move RD, BASE
+ | b <1
+ |. lw BASE, L->base // Reload BASE for lack of a saved register.
+ |
+ |7: // Possible table write barrier for any value. Skip valiswhite check.
+ | barrierback TAB:CARG2, TMP3, TMP0, <4
+ break;
+
+ /* -- Calls and vararg handling ----------------------------------------- */
+
+ case BC_CALLM:
+ | // RA = base*8, (RB = (nresults+1)*8,) RC = extra_nargs*8
+ | decode_RDtoRC8 NARGS8:RC, RD
+ | b ->BC_CALL_Z
+ |. addu NARGS8:RC, NARGS8:RC, MULTRES
+ break;
+ case BC_CALL:
+ | // RA = base*8, (RB = (nresults+1)*8,) RC = (nargs+1)*8
+ | decode_RDtoRC8 NARGS8:RC, RD
+ |->BC_CALL_Z:
+ | move TMP2, BASE
+ | addu BASE, BASE, RA
+ | li AT, LJ_TFUNC
+ | lw TMP0, HI(BASE)
+ | lw LFUNC:RB, LO(BASE)
+ | addiu BASE, BASE, 8
+ | bne TMP0, AT, ->vmeta_call
+ |. addiu NARGS8:RC, NARGS8:RC, -8
+ | ins_call
+ break;
+
+ case BC_CALLMT:
+ | // RA = base*8, (RB = 0,) RC = extra_nargs*8
+ | addu NARGS8:RD, NARGS8:RD, MULTRES // BC_CALLT gets RC from RD.
+ | // Fall through. Assumes BC_CALLT follows.
+ break;
+ case BC_CALLT:
+ | // RA = base*8, (RB = 0,) RC = (nargs+1)*8
+ | addu RA, BASE, RA
+ | li AT, LJ_TFUNC
+ | lw TMP0, HI(RA)
+ | lw LFUNC:RB, LO(RA)
+ | move NARGS8:RC, RD
+ | lw TMP1, FRAME_PC(BASE)
+ | addiu RA, RA, 8
+ | bne TMP0, AT, ->vmeta_callt
+ |. addiu NARGS8:RC, NARGS8:RC, -8
+ |->BC_CALLT_Z:
+ | andi TMP0, TMP1, FRAME_TYPE // Caveat: preserve TMP0 until the 'or'.
+ | lbu TMP3, LFUNC:RB->ffid
+ | bnez TMP0, >7
+ |. xori TMP2, TMP1, FRAME_VARG
+ |1:
+ | sw LFUNC:RB, FRAME_FUNC(BASE) // Copy function down, but keep PC.
+ | sltiu AT, TMP3, 2 // (> FF_C) Calling a fast function?
+ | move TMP2, BASE
+ | beqz NARGS8:RC, >3
+ |. move TMP3, NARGS8:RC
+ |2:
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ | addiu RA, RA, 8
+ | addiu TMP3, TMP3, -8
+ | sw SFRETHI, HI(TMP2)
+ | sw SFRETLO, LO(TMP2)
+ | bnez TMP3, <2
+ |. addiu TMP2, TMP2, 8
+ |3:
+ | or TMP0, TMP0, AT
+ | beqz TMP0, >5
+ |. nop
+ |4:
+ | ins_callt
+ |
+ |5: // Tailcall to a fast function with a Lua frame below.
+ | lw INS, -4(TMP1)
+ | decode_RA8a RA, INS
+ | decode_RA8b RA
+ | subu TMP1, BASE, RA
+ | lw LFUNC:TMP1, -8+FRAME_FUNC(TMP1)
+ | lw TMP1, LFUNC:TMP1->pc
+ | b <4
+ |. lw KBASE, PC2PROTO(k)(TMP1) // Need to prepare KBASE.
+ |
+ |7: // Tailcall from a vararg function.
+ | andi AT, TMP2, FRAME_TYPEP
+ | bnez AT, <1 // Vararg frame below?
+ |. subu TMP2, BASE, TMP2 // Relocate BASE down.
+ | move BASE, TMP2
+ | lw TMP1, FRAME_PC(TMP2)
+ | b <1
+ |. andi TMP0, TMP1, FRAME_TYPE
+ break;
+
+ case BC_ITERC:
+ | // RA = base*8, (RB = (nresults+1)*8, RC = (nargs+1)*8 ((2+1)*8))
+ | move TMP2, BASE
+ | addu BASE, BASE, RA
+ | li AT, LJ_TFUNC
+ | lw TMP1, -24+HI(BASE)
+ | lw LFUNC:RB, -24+LO(BASE)
+ | lw SFARG1HI, -16+HI(BASE)
+ | lw SFARG1LO, -16+LO(BASE)
+ | lw SFARG2HI, -8+HI(BASE)
+ | lw SFARG2LO, -8+LO(BASE)
+ | sw TMP1, HI(BASE) // Copy callable.
+ | sw LFUNC:RB, LO(BASE)
+ | sw SFARG1HI, 8+HI(BASE) // Copy state.
+ | sw SFARG1LO, 8+LO(BASE)
+ | sw SFARG2HI, 16+HI(BASE) // Copy control var.
+ | sw SFARG2LO, 16+LO(BASE)
+ | addiu BASE, BASE, 8
+ | bne TMP1, AT, ->vmeta_call
+ |. li NARGS8:RC, 16 // Iterators get 2 arguments.
+ | ins_call
+ break;
+
+ case BC_ITERN:
+ |.if JIT and ENDIAN_LE
+ | hotloop
+ |.endif
+ |->vm_IITERN:
+ | // RA = base*8, (RB = (nresults+1)*8, RC = (nargs+1)*8 (2+1)*8)
+ | addu RA, BASE, RA
+ | lw TAB:RB, -16+LO(RA)
+ | lw RC, -8+LO(RA) // Get index from control var.
+ | lw TMP0, TAB:RB->asize
+ | lw TMP1, TAB:RB->array
+ | addiu PC, PC, 4
+ |1: // Traverse array part.
+ | sltu AT, RC, TMP0
+ | beqz AT, >5 // Index points after array part?
+ |. sll TMP3, RC, 3
+ | addu TMP3, TMP1, TMP3
+ | lw SFARG1HI, HI(TMP3)
+ | lw SFARG1LO, LO(TMP3)
+ | lhu RD, -4+OFS_RD(PC)
+ | sw TISNUM, HI(RA)
+ | sw RC, LO(RA)
+ | beq SFARG1HI, TISNIL, <1 // Skip holes in array part.
+ |. addiu RC, RC, 1
+ | sw SFARG1HI, 8+HI(RA)
+ | sw SFARG1LO, 8+LO(RA)
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | decode_RD4b RD
+ | addu RD, RD, TMP3
+ | sw RC, -8+LO(RA) // Update control var.
+ | addu PC, PC, RD
+ |3:
+ | ins_next
+ |
+ |5: // Traverse hash part.
+ | lw TMP1, TAB:RB->hmask
+ | subu RC, RC, TMP0
+ | lw TMP2, TAB:RB->node
+ |6:
+ | sltu AT, TMP1, RC // End of iteration? Branch to ITERL+1.
+ | bnez AT, <3
+ |. sll TMP3, RC, 5
+ | sll RB, RC, 3
+ | subu TMP3, TMP3, RB
+ | addu NODE:TMP3, TMP3, TMP2
+ | lw SFARG1HI, NODE:TMP3->val.u32.hi
+ | lw SFARG1LO, NODE:TMP3->val.u32.lo
+ | lhu RD, -4+OFS_RD(PC)
+ | beq SFARG1HI, TISNIL, <6 // Skip holes in hash part.
+ |. addiu RC, RC, 1
+ | lw SFARG2HI, NODE:TMP3->key.u32.hi
+ | lw SFARG2LO, NODE:TMP3->key.u32.lo
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | sw SFARG1HI, 8+HI(RA)
+ | sw SFARG1LO, 8+LO(RA)
+ | addu RC, RC, TMP0
+ | decode_RD4b RD
+ | addu RD, RD, TMP3
+ | sw SFARG2HI, HI(RA)
+ | sw SFARG2LO, LO(RA)
+ | addu PC, PC, RD
+ | b <3
+ |. sw RC, -8+LO(RA) // Update control var.
+ break;
+
+ case BC_ISNEXT:
+ | // RA = base*8, RD = target (points to ITERN)
+ | addu RA, BASE, RA
+ | srl TMP0, RD, 1
+ | lw CARG1, -24+HI(RA)
+ | lw CFUNC:CARG2, -24+LO(RA)
+ | addu TMP0, PC, TMP0
+ | lw CARG3, -16+HI(RA)
+ | lw CARG4, -8+HI(RA)
+ | li AT, LJ_TFUNC
+ | bne CARG1, AT, >5
+ |. lui TMP2, (-(BCBIAS_J*4 >> 16) & 65535)
+ | lbu CARG2, CFUNC:CARG2->ffid
+ | addiu CARG3, CARG3, -LJ_TTAB
+ | addiu CARG4, CARG4, -LJ_TNIL
+ | or CARG3, CARG3, CARG4
+ | addiu CARG2, CARG2, -FF_next_N
+ | or CARG2, CARG2, CARG3
+ | bnez CARG2, >5
+ |. lui TMP1, (LJ_KEYINDEX >> 16)
+ | addu PC, TMP0, TMP2
+ | ori TMP1, TMP1, (LJ_KEYINDEX & 0xffff)
+ | sw r0, -8+LO(RA) // Initialize control var.
+ | sw TMP1, -8+HI(RA)
+ |1:
+ | ins_next
+ |5: // Despecialize bytecode if any of the checks fail.
+ | li TMP3, BC_JMP
+ | li TMP1, BC_ITERC
+ | sb TMP3, -4+OFS_OP(PC)
+ | addu PC, TMP0, TMP2
+ |.if JIT
+ | lb TMP0, OFS_OP(PC)
+ | li AT, BC_ITERN
+ | bne TMP0, AT, >6
+ |. lhu TMP2, OFS_RD(PC)
+ |.endif
+ | b <1
+ |. sb TMP1, OFS_OP(PC)
+ |.if JIT
+ |6: // Unpatch JLOOP.
+ | lw TMP0, DISPATCH_J(trace)(DISPATCH)
+ | sll TMP2, TMP2, 2
+ | addu TMP0, TMP0, TMP2
+ | lw TRACE:TMP2, 0(TMP0)
+ | lw TMP0, TRACE:TMP2->startins
+ | li AT, -256
+ | and TMP0, TMP0, AT
+ | or TMP0, TMP0, TMP1
+ | b <1
+ |. sw TMP0, 0(PC)
+ |.endif
+ break;
+
+ case BC_VARG:
+ | // RA = base*8, RB = (nresults+1)*8, RC = numparams*8
+ | lw TMP0, FRAME_PC(BASE)
+ | decode_RDtoRC8 RC, RD
+ | decode_RB8a RB, INS
+ | addu RC, BASE, RC
+ | decode_RB8b RB
+ | addu RA, BASE, RA
+ | addiu RC, RC, FRAME_VARG
+ | addu TMP2, RA, RB
+ | addiu TMP3, BASE, -8 // TMP3 = vtop
+ | subu RC, RC, TMP0 // RC = vbase
+ | // Note: RC may now be even _above_ BASE if nargs was < numparams.
+ | beqz RB, >5 // Copy all varargs?
+ |. subu TMP1, TMP3, RC
+ | addiu TMP2, TMP2, -16
+ |1: // Copy vararg slots to destination slots.
+ | lw CARG1, HI(RC)
+ | sltu AT, RC, TMP3
+ | lw CARG2, LO(RC)
+ | addiu RC, RC, 8
+ | movz CARG1, TISNIL, AT
+ | sw CARG1, HI(RA)
+ | sw CARG2, LO(RA)
+ | sltu AT, RA, TMP2
+ | bnez AT, <1
+ |. addiu RA, RA, 8
+ |3:
+ | ins_next
+ |
+ |5: // Copy all varargs.
+ | lw TMP0, L->maxstack
+ | blez TMP1, <3 // No vararg slots?
+ |. li MULTRES, 8 // MULTRES = (0+1)*8
+ | addu TMP2, RA, TMP1
+ | sltu AT, TMP0, TMP2
+ | bnez AT, >7
+ |. addiu MULTRES, TMP1, 8
+ |6:
+ | lw SFRETHI, HI(RC)
+ | lw SFRETLO, LO(RC)
+ | addiu RC, RC, 8
+ | sw SFRETHI, HI(RA)
+ | sw SFRETLO, LO(RA)
+ | sltu AT, RC, TMP3
+ | bnez AT, <6 // More vararg slots?
+ |. addiu RA, RA, 8
+ | b <3
+ |. nop
+ |
+ |7: // Grow stack for varargs.
+ | load_got lj_state_growstack
+ | sw RA, L->top
+ | subu RA, RA, BASE
+ | sw BASE, L->base
+ | subu BASE, RC, BASE // Need delta, because BASE may change.
+ | sw PC, SAVE_PC
+ | srl CARG2, TMP1, 3
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | move RC, BASE
+ | lw BASE, L->base
+ | addu RA, BASE, RA
+ | addu RC, BASE, RC
+ | b <6
+ |. addiu TMP3, BASE, -8
+ break;
+
+ /* -- Returns ----------------------------------------------------------- */
+
+ case BC_RETM:
+ | // RA = results*8, RD = extra_nresults*8
+ | addu RD, RD, MULTRES // MULTRES >= 8, so RD >= 8.
+ | // Fall through. Assumes BC_RET follows.
+ break;
+
+ case BC_RET:
+ | // RA = results*8, RD = (nresults+1)*8
+ | lw PC, FRAME_PC(BASE)
+ | addu RA, BASE, RA
+ | move MULTRES, RD
+ |1:
+ | andi TMP0, PC, FRAME_TYPE
+ | bnez TMP0, ->BC_RETV_Z
+ |. xori TMP1, PC, FRAME_VARG
+ |
+ |->BC_RET_Z:
+ | // BASE = base, RA = resultptr, RD = (nresults+1)*8, PC = return
+ | lw INS, -4(PC)
+ | addiu TMP2, BASE, -8
+ | addiu RC, RD, -8
+ | decode_RA8a TMP0, INS
+ | decode_RB8a RB, INS
+ | decode_RA8b TMP0
+ | decode_RB8b RB
+ | addu TMP3, TMP2, RB
+ | beqz RC, >3
+ |. subu BASE, TMP2, TMP0
+ |2:
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ | addiu RA, RA, 8
+ | addiu RC, RC, -8
+ | sw SFRETHI, HI(TMP2)
+ | sw SFRETLO, LO(TMP2)
+ | bnez RC, <2
+ |. addiu TMP2, TMP2, 8
+ |3:
+ | addiu TMP3, TMP3, -8
+ |5:
+ | sltu AT, TMP2, TMP3
+ | bnez AT, >6
+ |. lw LFUNC:TMP1, FRAME_FUNC(BASE)
+ | ins_next1
+ | lw TMP1, LFUNC:TMP1->pc
+ | lw KBASE, PC2PROTO(k)(TMP1)
+ | ins_next2
+ |
+ |6: // Fill up results with nil.
+ | sw TISNIL, HI(TMP2)
+ | b <5
+ |. addiu TMP2, TMP2, 8
+ |
+ |->BC_RETV_Z: // Non-standard return case.
+ | andi TMP2, TMP1, FRAME_TYPEP
+ | bnez TMP2, ->vm_return
+ |. nop
+ | // Return from vararg function: relocate BASE down.
+ | subu BASE, BASE, TMP1
+ | b <1
+ |. lw PC, FRAME_PC(BASE)
+ break;
+
+ case BC_RET0: case BC_RET1:
+ | // RA = results*8, RD = (nresults+1)*8
+ | lw PC, FRAME_PC(BASE)
+ | addu RA, BASE, RA
+ | move MULTRES, RD
+ | andi TMP0, PC, FRAME_TYPE
+ | bnez TMP0, ->BC_RETV_Z
+ |. xori TMP1, PC, FRAME_VARG
+ |
+ | lw INS, -4(PC)
+ | addiu TMP2, BASE, -8
+ if (op == BC_RET1) {
+ | lw SFRETHI, HI(RA)
+ | lw SFRETLO, LO(RA)
+ }
+ | decode_RB8a RB, INS
+ | decode_RA8a RA, INS
+ | decode_RB8b RB
+ | decode_RA8b RA
+ if (op == BC_RET1) {
+ | sw SFRETHI, HI(TMP2)
+ | sw SFRETLO, LO(TMP2)
+ }
+ | subu BASE, TMP2, RA
+ |5:
+ | sltu AT, RD, RB
+ | bnez AT, >6
+ |. lw LFUNC:TMP1, FRAME_FUNC(BASE)
+ | ins_next1
+ | lw TMP1, LFUNC:TMP1->pc
+ | lw KBASE, PC2PROTO(k)(TMP1)
+ | ins_next2
+ |
+ |6: // Fill up results with nil.
+ | addiu TMP2, TMP2, 8
+ | addiu RD, RD, 8
+ | b <5
+ if (op == BC_RET1) {
+ |. sw TISNIL, HI(TMP2)
+ } else {
+ |. sw TISNIL, -8+HI(TMP2)
+ }
+ break;
+
+ /* -- Loops and branches ------------------------------------------------ */
+
+ case BC_FORL:
+ |.if JIT
+ | hotloop
+ |.endif
+ | // Fall through. Assumes BC_IFORL follows.
+ break;
+
+ case BC_JFORI:
+ case BC_JFORL:
+#if !LJ_HASJIT
+ break;
+#endif
+ case BC_FORI:
+ case BC_IFORL:
+ | // RA = base*8, RD = target (after end of loop or start of loop)
+ vk = (op == BC_IFORL || op == BC_JFORL);
+ | addu RA, BASE, RA
+ | lw SFARG1HI, FORL_IDX*8+HI(RA)
+ | lw SFARG1LO, FORL_IDX*8+LO(RA)
+ if (op != BC_JFORL) {
+ | srl RD, RD, 1
+ | lui TMP2, (-(BCBIAS_J*4 >> 16) & 65535)
+ | addu TMP2, RD, TMP2
+ }
+ if (!vk) {
+ | lw SFARG2HI, FORL_STOP*8+HI(RA)
+ | lw SFARG2LO, FORL_STOP*8+LO(RA)
+ | bne SFARG1HI, TISNUM, >5
+ |. lw SFRETHI, FORL_STEP*8+HI(RA)
+ | xor AT, SFARG2HI, TISNUM
+ | lw SFRETLO, FORL_STEP*8+LO(RA)
+ | xor TMP0, SFRETHI, TISNUM
+ | or AT, AT, TMP0
+ | bnez AT, ->vmeta_for
+ |. slt AT, SFRETLO, r0
+ | slt CRET1, SFARG2LO, SFARG1LO
+ | slt TMP1, SFARG1LO, SFARG2LO
+ | movn CRET1, TMP1, AT
+ } else {
+ | bne SFARG1HI, TISNUM, >5
+ |. lw SFARG2LO, FORL_STEP*8+LO(RA)
+ | lw SFRETLO, FORL_STOP*8+LO(RA)
+ | move TMP3, SFARG1LO
+ | addu SFARG1LO, SFARG1LO, SFARG2LO
+ | xor TMP0, SFARG1LO, TMP3
+ | xor TMP1, SFARG1LO, SFARG2LO
+ | and TMP0, TMP0, TMP1
+ | slt TMP1, SFARG1LO, SFRETLO
+ | slt CRET1, SFRETLO, SFARG1LO
+ | slt AT, SFARG2LO, r0
+ | slt TMP0, TMP0, r0 // ((y^a) & (y^b)) < 0: overflow.
+ | movn CRET1, TMP1, AT
+ | or CRET1, CRET1, TMP0
+ }
+ |1:
+ if (op == BC_FORI) {
+ | movz TMP2, r0, CRET1
+ | addu PC, PC, TMP2
+ } else if (op == BC_JFORI) {
+ | addu PC, PC, TMP2
+ | lhu RD, -4+OFS_RD(PC)
+ } else if (op == BC_IFORL) {
+ | movn TMP2, r0, CRET1
+ | addu PC, PC, TMP2
+ }
+ if (vk) {
+ | sw SFARG1HI, FORL_IDX*8+HI(RA)
+ | sw SFARG1LO, FORL_IDX*8+LO(RA)
+ }
+ | ins_next1
+ | sw SFARG1HI, FORL_EXT*8+HI(RA)
+ | sw SFARG1LO, FORL_EXT*8+LO(RA)
+ |2:
+ if (op == BC_JFORI) {
+ | beqz CRET1, =>BC_JLOOP
+ |. decode_RD8b RD
+ } else if (op == BC_JFORL) {
+ | beqz CRET1, =>BC_JLOOP
+ }
+ | ins_next2
+ |
+ |5: // FP loop.
+ |.if FPU
+ if (!vk) {
+ | ldc1 f0, FORL_IDX*8(RA)
+ | ldc1 f2, FORL_STOP*8(RA)
+ | sltiu TMP0, SFARG1HI, LJ_TISNUM
+ | sltiu TMP1, SFARG2HI, LJ_TISNUM
+ | sltiu AT, SFRETHI, LJ_TISNUM
+ | and TMP0, TMP0, TMP1
+ | and AT, AT, TMP0
+ | beqz AT, ->vmeta_for
+ |. slt TMP3, SFRETHI, r0
+ | c.ole.d 0, f0, f2
+ | c.ole.d 1, f2, f0
+ | li CRET1, 1
+ | movt CRET1, r0, 0
+ | movt AT, r0, 1
+ | b <1
+ |. movn CRET1, AT, TMP3
+ } else {
+ | ldc1 f0, FORL_IDX*8(RA)
+ | ldc1 f4, FORL_STEP*8(RA)
+ | ldc1 f2, FORL_STOP*8(RA)
+ | lw SFARG2HI, FORL_STEP*8+HI(RA)
+ | add.d f0, f0, f4
+ | c.ole.d 0, f0, f2
+ | c.ole.d 1, f2, f0
+ | slt TMP3, SFARG2HI, r0
+ | li CRET1, 1
+ | li AT, 1
+ | movt CRET1, r0, 0
+ | movt AT, r0, 1
+ | movn CRET1, AT, TMP3
+ if (op == BC_IFORL) {
+ | movn TMP2, r0, CRET1
+ | addu PC, PC, TMP2
+ }
+ | sdc1 f0, FORL_IDX*8(RA)
+ | ins_next1
+ | b <2
+ |. sdc1 f0, FORL_EXT*8(RA)
+ }
+ |.else
+ if (!vk) {
+ | sltiu TMP0, SFARG1HI, LJ_TISNUM
+ | sltiu TMP1, SFARG2HI, LJ_TISNUM
+ | sltiu AT, SFRETHI, LJ_TISNUM
+ | and TMP0, TMP0, TMP1
+ | and AT, AT, TMP0
+ | beqz AT, ->vmeta_for
+ |. nop
+ | bal ->vm_sfcmpolex
+ |. move TMP3, SFRETHI
+ | b <1
+ |. nop
+ } else {
+ | lw SFARG2HI, FORL_STEP*8+HI(RA)
+ | load_got __adddf3
+ | call_extern
+ |. sw TMP2, ARG5
+ | lw SFARG2HI, FORL_STOP*8+HI(RA)
+ | lw SFARG2LO, FORL_STOP*8+LO(RA)
+ | move SFARG1HI, SFRETHI
+ | move SFARG1LO, SFRETLO
+ | bal ->vm_sfcmpolex
+ |. lw TMP3, FORL_STEP*8+HI(RA)
+ if ( op == BC_JFORL ) {
+ | lhu RD, -4+OFS_RD(PC)
+ | lw TMP2, ARG5
+ | b <1
+ |. decode_RD8b RD
+ } else {
+ | b <1
+ |. lw TMP2, ARG5
+ }
+ }
+ |.endif
+ break;
+
+ case BC_ITERL:
+ |.if JIT
+ | hotloop
+ |.endif
+ | // Fall through. Assumes BC_IITERL follows.
+ break;
+
+ case BC_JITERL:
+#if !LJ_HASJIT
+ break;
+#endif
+ case BC_IITERL:
+ | // RA = base*8, RD = target
+ | addu RA, BASE, RA
+ | lw TMP1, HI(RA)
+ | beq TMP1, TISNIL, >1 // Stop if iterator returned nil.
+ |. lw TMP2, LO(RA)
+ if (op == BC_JITERL) {
+ | sw TMP1, -8+HI(RA)
+ | b =>BC_JLOOP
+ |. sw TMP2, -8+LO(RA)
+ } else {
+ | branch_RD // Otherwise save control var + branch.
+ | sw TMP1, -8+HI(RA)
+ | sw TMP2, -8+LO(RA)
+ }
+ |1:
+ | ins_next
+ break;
+
+ case BC_LOOP:
+ | // RA = base*8, RD = target (loop extent)
+ | // Note: RA/RD is only used by trace recorder to determine scope/extent
+ | // This opcode does NOT jump, it's only purpose is to detect a hot loop.
+ |.if JIT
+ | hotloop
+ |.endif
+ | // Fall through. Assumes BC_ILOOP follows.
+ break;
+
+ case BC_ILOOP:
+ | // RA = base*8, RD = target (loop extent)
+ | ins_next
+ break;
+
+ case BC_JLOOP:
+ |.if JIT
+ | // RA = base*8 (ignored), RD = traceno*8
+ | lw TMP1, DISPATCH_J(trace)(DISPATCH)
+ | srl RD, RD, 1
+ | li AT, 0
+ | addu TMP1, TMP1, RD
+ | // Traces on MIPS don't store the trace number, so use 0.
+ | sw AT, DISPATCH_GL(vmstate)(DISPATCH)
+ | lw TRACE:TMP2, 0(TMP1)
+ | sw BASE, DISPATCH_GL(jit_base)(DISPATCH)
+ | lw TMP2, TRACE:TMP2->mcode
+ | sw L, DISPATCH_GL(tmpbuf.L)(DISPATCH)
+ | jr TMP2
+ |. addiu JGL, DISPATCH, GG_DISP2G+32768
+ |.endif
+ break;
+
+ case BC_JMP:
+ | // RA = base*8 (only used by trace recorder), RD = target
+ | branch_RD
+ | ins_next
+ break;
+
+ /* -- Function headers -------------------------------------------------- */
+
+ case BC_FUNCF:
+ |.if JIT
+ | hotcall
+ |.endif
+ case BC_FUNCV: /* NYI: compiled vararg functions. */
+ | // Fall through. Assumes BC_IFUNCF/BC_IFUNCV follow.
+ break;
+
+ case BC_JFUNCF:
+#if !LJ_HASJIT
+ break;
+#endif
+ case BC_IFUNCF:
+ | // BASE = new base, RA = BASE+framesize*8, RB = LFUNC, RC = nargs*8
+ | lw TMP2, L->maxstack
+ | lbu TMP1, -4+PC2PROTO(numparams)(PC)
+ | lw KBASE, -4+PC2PROTO(k)(PC)
+ | sltu AT, TMP2, RA
+ | bnez AT, ->vm_growstack_l
+ |. sll TMP1, TMP1, 3
+ if (op != BC_JFUNCF) {
+ | ins_next1
+ }
+ |2:
+ | sltu AT, NARGS8:RC, TMP1 // Check for missing parameters.
+ | bnez AT, >3
+ |. addu AT, BASE, NARGS8:RC
+ if (op == BC_JFUNCF) {
+ | decode_RD8a RD, INS
+ | b =>BC_JLOOP
+ |. decode_RD8b RD
+ } else {
+ | ins_next2
+ }
+ |
+ |3: // Clear missing parameters.
+ | sw TISNIL, HI(AT)
+ | b <2
+ |. addiu NARGS8:RC, NARGS8:RC, 8
+ break;
+
+ case BC_JFUNCV:
+#if !LJ_HASJIT
+ break;
+#endif
+ | NYI // NYI: compiled vararg functions
+ break; /* NYI: compiled vararg functions. */
+
+ case BC_IFUNCV:
+ | // BASE = new base, RA = BASE+framesize*8, RB = LFUNC, RC = nargs*8
+ | addu TMP1, BASE, RC
+ | lw TMP2, L->maxstack
+ | addu TMP0, RA, RC
+ | sw LFUNC:RB, LO(TMP1) // Store copy of LFUNC.
+ | addiu TMP3, RC, 8+FRAME_VARG
+ | sltu AT, TMP0, TMP2
+ | lw KBASE, -4+PC2PROTO(k)(PC)
+ | beqz AT, ->vm_growstack_l
+ |. sw TMP3, HI(TMP1) // Store delta + FRAME_VARG.
+ | lbu TMP2, -4+PC2PROTO(numparams)(PC)
+ | move RA, BASE
+ | move RC, TMP1
+ | ins_next1
+ | beqz TMP2, >3
+ |. addiu BASE, TMP1, 8
+ |1:
+ | lw TMP0, HI(RA)
+ | lw TMP3, LO(RA)
+ | sltu AT, RA, RC // Less args than parameters?
+ | move CARG1, TMP0
+ | movz TMP0, TISNIL, AT // Clear missing parameters.
+ | movn CARG1, TISNIL, AT // Clear old fixarg slot (help the GC).
+ | sw TMP3, 8+LO(TMP1)
+ | addiu TMP2, TMP2, -1
+ | sw TMP0, 8+HI(TMP1)
+ | addiu TMP1, TMP1, 8
+ | sw CARG1, HI(RA)
+ | bnez TMP2, <1
+ |. addiu RA, RA, 8
+ |3:
+ | ins_next2
+ break;
+
+ case BC_FUNCC:
+ case BC_FUNCCW:
+ | // BASE = new base, RA = BASE+framesize*8, RB = CFUNC, RC = nargs*8
+ if (op == BC_FUNCC) {
+ | lw CFUNCADDR, CFUNC:RB->f
+ } else {
+ | lw CFUNCADDR, DISPATCH_GL(wrapf)(DISPATCH)
+ }
+ | addu TMP1, RA, NARGS8:RC
+ | lw TMP2, L->maxstack
+ | addu RC, BASE, NARGS8:RC
+ | sw BASE, L->base
+ | sltu AT, TMP2, TMP1
+ | sw RC, L->top
+ | li_vmstate C
+ if (op == BC_FUNCCW) {
+ | lw CARG2, CFUNC:RB->f
+ }
+ | bnez AT, ->vm_growstack_c // Need to grow stack.
+ |. move CARG1, L
+ | jalr CFUNCADDR // (lua_State *L [, lua_CFunction f])
+ |. st_vmstate
+ | // Returns nresults.
+ | lw BASE, L->base
+ | sll RD, CRET1, 3
+ | lw TMP1, L->top
+ | li_vmstate INTERP
+ | lw PC, FRAME_PC(BASE) // Fetch PC of caller.
+ | subu RA, TMP1, RD // RA = L->top - nresults*8
+ | sw L, DISPATCH_GL(cur_L)(DISPATCH)
+ | b ->vm_returnc
+ |. st_vmstate
+ break;
+
+ /* ---------------------------------------------------------------------- */
+
+ default:
+ fprintf(stderr, "Error: undefined opcode BC_%s\n", bc_names[op]);
+ exit(2);
+ break;
+ }
+}
+
+static int build_backend(BuildCtx *ctx)
+{
+ int op;
+
+ dasm_growpc(Dst, BC__MAX);
+
+ build_subroutines(ctx);
+
+ |.code_op
+ for (op = 0; op < BC__MAX; op++)
+ build_ins(ctx, (BCOp)op, op);
+
+ return BC__MAX;
+}
+
+/* Emit pseudo frame-info for all assembler functions. */
+static void emit_asm_debug(BuildCtx *ctx)
+{
+ int fcofs = (int)((uint8_t *)ctx->glob[GLOB_vm_ffi_call] - ctx->code);
+ int i;
+ switch (ctx->mode) {
+ case BUILD_elfasm:
+ fprintf(ctx->fp, "\t.section .debug_frame,\"\",@progbits\n");
+ fprintf(ctx->fp,
+ ".Lframe0:\n"
+ "\t.4byte .LECIE0-.LSCIE0\n"
+ ".LSCIE0:\n"
+ "\t.4byte 0xffffffff\n"
+ "\t.byte 0x1\n"
+ "\t.string \"\"\n"
+ "\t.uleb128 0x1\n"
+ "\t.sleb128 -4\n"
+ "\t.byte 31\n"
+ "\t.byte 0xc\n\t.uleb128 29\n\t.uleb128 0\n"
+ "\t.align 2\n"
+ ".LECIE0:\n\n");
+ fprintf(ctx->fp,
+ ".LSFDE0:\n"
+ "\t.4byte .LEFDE0-.LASFDE0\n"
+ ".LASFDE0:\n"
+ "\t.4byte .Lframe0\n"
+ "\t.4byte .Lbegin\n"
+ "\t.4byte %d\n"
+ "\t.byte 0xe\n\t.uleb128 %d\n"
+ "\t.byte 0x9f\n\t.sleb128 1\n"
+ "\t.byte 0x9e\n\t.sleb128 2\n",
+ fcofs, CFRAME_SIZE);
+ for (i = 23; i >= 16; i--)
+ fprintf(ctx->fp, "\t.byte %d\n\t.uleb128 %d\n", 0x80+i, 26-i);
+#if !LJ_SOFTFP
+ for (i = 30; i >= 20; i -= 2)
+ fprintf(ctx->fp, "\t.byte %d\n\t.uleb128 %d\n", 0x80+32+i, 42-i);
+#endif
+ fprintf(ctx->fp,
+ "\t.align 2\n"
+ ".LEFDE0:\n\n");
+#if LJ_HASFFI
+ fprintf(ctx->fp,
+ ".LSFDE1:\n"
+ "\t.4byte .LEFDE1-.LASFDE1\n"
+ ".LASFDE1:\n"
+ "\t.4byte .Lframe0\n"
+ "\t.4byte lj_vm_ffi_call\n"
+ "\t.4byte %d\n"
+ "\t.byte 0x9f\n\t.uleb128 1\n"
+ "\t.byte 0x90\n\t.uleb128 2\n"
+ "\t.byte 0xd\n\t.uleb128 0x10\n"
+ "\t.align 2\n"
+ ".LEFDE1:\n\n", (int)ctx->codesz - fcofs);
+#endif
+#if !LJ_NO_UNWIND
+ fprintf(ctx->fp, "\t.section .eh_frame,\"aw\",@progbits\n");
+ fprintf(ctx->fp,
+ "\t.globl lj_err_unwind_dwarf\n"
+ ".Lframe1:\n"
+ "\t.4byte .LECIE1-.LSCIE1\n"
+ ".LSCIE1:\n"
+ "\t.4byte 0\n"
+ "\t.byte 0x1\n"
+ "\t.string \"zPR\"\n"
+ "\t.uleb128 0x1\n"
+ "\t.sleb128 -4\n"
+ "\t.byte 31\n"
+ "\t.uleb128 6\n" /* augmentation length */
+ "\t.byte 0\n"
+ "\t.4byte lj_err_unwind_dwarf\n"
+ "\t.byte 0\n"
+ "\t.byte 0xc\n\t.uleb128 29\n\t.uleb128 0\n"
+ "\t.align 2\n"
+ ".LECIE1:\n\n");
+ fprintf(ctx->fp,
+ ".LSFDE2:\n"
+ "\t.4byte .LEFDE2-.LASFDE2\n"
+ ".LASFDE2:\n"
+ "\t.4byte .LASFDE2-.Lframe1\n"
+ "\t.4byte .Lbegin\n"
+ "\t.4byte %d\n"
+ "\t.uleb128 0\n" /* augmentation length */
+ "\t.byte 0xe\n\t.uleb128 %d\n"
+ "\t.byte 0x9f\n\t.sleb128 1\n"
+ "\t.byte 0x9e\n\t.sleb128 2\n",
+ fcofs, CFRAME_SIZE);
+ for (i = 23; i >= 16; i--)
+ fprintf(ctx->fp, "\t.byte %d\n\t.uleb128 %d\n", 0x80+i, 26-i);
+#if !LJ_SOFTFP
+ for (i = 30; i >= 20; i -= 2)
+ fprintf(ctx->fp, "\t.byte %d\n\t.uleb128 %d\n", 0x80+32+i, 42-i);
+#endif
+ fprintf(ctx->fp,
+ "\t.align 2\n"
+ ".LEFDE2:\n\n");
+#if LJ_HASFFI
+ fprintf(ctx->fp,
+ ".Lframe2:\n"
+ "\t.4byte .LECIE2-.LSCIE2\n"
+ ".LSCIE2:\n"
+ "\t.4byte 0\n"
+ "\t.byte 0x1\n"
+ "\t.string \"zR\"\n"
+ "\t.uleb128 0x1\n"
+ "\t.sleb128 -4\n"
+ "\t.byte 31\n"
+ "\t.uleb128 1\n" /* augmentation length */
+ "\t.byte 0\n"
+ "\t.byte 0xc\n\t.uleb128 29\n\t.uleb128 0\n"
+ "\t.align 2\n"
+ ".LECIE2:\n\n");
+ fprintf(ctx->fp,
+ ".LSFDE3:\n"
+ "\t.4byte .LEFDE3-.LASFDE3\n"
+ ".LASFDE3:\n"
+ "\t.4byte .LASFDE3-.Lframe2\n"
+ "\t.4byte lj_vm_ffi_call\n"
+ "\t.4byte %d\n"
+ "\t.uleb128 0\n" /* augmentation length */
+ "\t.byte 0x9f\n\t.uleb128 1\n"
+ "\t.byte 0x90\n\t.uleb128 2\n"
+ "\t.byte 0xd\n\t.uleb128 0x10\n"
+ "\t.align 2\n"
+ ".LEFDE3:\n\n", (int)ctx->codesz - fcofs);
+#endif
+#endif
+ break;
+ default:
+ break;
+ }
+}
+