From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from [87.239.111.99] (localhost [127.0.0.1]) by dev.tarantool.org (Postfix) with ESMTP id 935EB5C3F1C; Tue, 15 Aug 2023 12:43:53 +0300 (MSK) DKIM-Filter: OpenDKIM Filter v2.11.0 dev.tarantool.org 935EB5C3F1C DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=tarantool.org; s=dev; t=1692092633; bh=g8w0Tz/0Gu88BqHmc1ZUyvfz25XPAH5Wxstw8NI4ek4=; h=To:Date:In-Reply-To:References:Subject:List-Id:List-Unsubscribe: List-Archive:List-Post:List-Help:List-Subscribe:From:Reply-To:Cc: From; b=lsnZZtvj1US55hDJQDqORtM0yBLB+ZgjKxkhlbOZhZbLtCsO2cyI9BZBAmbET7a/N QgFXjBNj1lcy/bTK0UXlNzGuycQh3tNNZ5/dP1SVfjiWhJmMgtY0vhwAEupM5TzjgN vnLcD5LiK28NG2IejuqXU2tXMRqSnU7Eu2skMTfQ= Received: from smtp30.i.mail.ru (smtp30.i.mail.ru [95.163.41.71]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by dev.tarantool.org (Postfix) with ESMTPS id 4EBE85C3F0F for ; Tue, 15 Aug 2023 12:41:29 +0300 (MSK) DKIM-Filter: OpenDKIM Filter v2.11.0 dev.tarantool.org 4EBE85C3F0F Received: by smtp30.i.mail.ru with esmtpa (envelope-from ) id 1qVqYJ-009Mi1-1s; Tue, 15 Aug 2023 12:41:28 +0300 To: Maxim Kokryashkin , Sergey Bronnikov Date: Tue, 15 Aug 2023 12:36:31 +0300 Message-ID: <3d4ed7eb3db111ca1571688b281c560bd7a0f674.1692089299.git.skaplun@tarantool.org> X-Mailer: git-send-email 2.41.0 In-Reply-To: References: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Mailru-Src: smtp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eAau8CL7WIMRKs4sN3D3tLDjz0dLbV79QFUyzQ2Ujvy7cMT6pYYqY16iZVKkSc3dCLJ7zSJH7+u4VD18S7Vl4ZUrpaVfd2+vE6kuoey4m4VkSEu530nj6fImhcD4MUrOEAnl0W826KZ9Q+tr5ycPtXkTV4k65bRjmOUUP8cvGozZ33TWg5HZplvhhXbhDGzqmQDTd6OAevLeAnq3Ra9uf7zvY2zzsIhlcp/Y7m53TZgf2aB4JOg4gkr2biojJ1ceUZTkownNfVMDlhb41Q== X-Mailru-Sender: 11C2EC085EDE56FAC07928AF2646A769FE4FC501550FAAC973301DC292E5090A2F2E01304D987C7FDEDBA653FF35249392D99EB8CC7091A70E183A470755BFD208F19895AA18418972D6B4FCE48DF648AE208404248635DF X-Mras: Ok Subject: [Tarantool-patches] [PATCH luajit 5/5] Revert to trival pow() optimizations to prevent inaccuracies. X-BeenThere: tarantool-patches@dev.tarantool.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: Tarantool development patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , From: Sergey Kaplun via Tarantool-patches Reply-To: Sergey Kaplun Cc: tarantool-patches@dev.tarantool.org Errors-To: tarantool-patches-bounces@dev.tarantool.org Sender: "Tarantool-patches" From: Mike Pall (cherry-picked from commit 96d6d5032098ea9f0002165394a8774dcaa0c0ce) This patch fixes different misbehaviour between JIT-compiled code and the interpreter for power operator with the following ways: * Drop folding optimizations for base ^ n => base * base ..., as far as pow(base, n) isn't interchangeable with just multiplicity of numbers and depends on the implementation. * Since the internal power function is inaccurate for very big or small powers, it is dropped, and `pow()` from the standard library is used instead. To save consistency between JIT behaviour and the VM narrowing optimization is dropped, and only trivial folding optimizations are used. Also, `math_extern2` version with two parameters is dropped, since it's no more used. Also, this fixes failures of the [220/502] lib/string/format/num.lua test [1] from LuaJIT-test suite. [1]: https://www.exploringbinary.com/incorrect-floating-point-to-decimal-conversions/ Sergey Kaplun: * added the description and the test for the problem Part of tarantool/tarantool#8825 --- src/lj_asm.c | 3 +- src/lj_dispatch.h | 2 +- src/lj_ffrecord.c | 4 +- src/lj_ircall.h | 3 +- src/lj_iropt.h | 1 - src/lj_opt_fold.c | 37 ++++------------ src/lj_opt_narrow.c | 24 ---------- src/lj_opt_split.c | 2 +- src/lj_record.c | 2 +- src/lj_vm.h | 3 -- src/lj_vmmath.c | 44 +------------------ src/vm_arm.dasc | 13 +++--- src/vm_arm64.dasc | 11 ++--- src/vm_mips.dasc | 11 ++--- src/vm_mips64.dasc | 11 ++--- src/vm_ppc.dasc | 11 ++--- src/vm_x64.dasc | 9 ++-- src/vm_x86.dasc | 11 ++--- .../lj-684-pow-inconsistencies.test.lua | 21 ++++++++- 19 files changed, 64 insertions(+), 159 deletions(-) diff --git a/src/lj_asm.c b/src/lj_asm.c index 65261d50..3a1909d5 100644 --- a/src/lj_asm.c +++ b/src/lj_asm.c @@ -1660,8 +1660,7 @@ static void asm_pow(ASMState *as, IRIns *ir) IRCALL_lj_carith_powu64); else #endif - asm_callid(as, ir, irt_isnum(IR(ir->op2)->t) ? IRCALL_lj_vm_pow : - IRCALL_lj_vm_powi); + asm_callid(as, ir, IRCALL_pow); } static void asm_div(ASMState *as, IRIns *ir) diff --git a/src/lj_dispatch.h b/src/lj_dispatch.h index af870a75..b8bc2594 100644 --- a/src/lj_dispatch.h +++ b/src/lj_dispatch.h @@ -44,7 +44,7 @@ extern double __divdf3(double a, double b); #define GOTDEF(_) \ _(floor) _(ceil) _(trunc) _(log) _(log10) _(exp) _(sin) _(cos) _(tan) \ _(asin) _(acos) _(atan) _(sinh) _(cosh) _(tanh) _(frexp) _(modf) _(atan2) \ - _(lj_vm_pow) _(fmod) _(ldexp) _(lj_vm_modi) \ + _(pow) _(fmod) _(ldexp) _(lj_vm_modi) \ _(lj_dispatch_call) _(lj_dispatch_ins) _(lj_dispatch_stitch) \ _(lj_dispatch_profile) _(lj_err_throw) \ _(lj_ffh_coroutine_wrap_err) _(lj_func_closeuv) _(lj_func_newL_gc) \ diff --git a/src/lj_ffrecord.c b/src/lj_ffrecord.c index 0746ec64..99a6b918 100644 --- a/src/lj_ffrecord.c +++ b/src/lj_ffrecord.c @@ -590,8 +590,8 @@ static void LJ_FASTCALL recff_math_call(jit_State *J, RecordFFData *rd) static void LJ_FASTCALL recff_math_pow(jit_State *J, RecordFFData *rd) { - J->base[0] = lj_opt_narrow_pow(J, J->base[0], J->base[1], - &rd->argv[0], &rd->argv[1]); + J->base[0] = lj_opt_narrow_arith(J, J->base[0], J->base[1], + &rd->argv[0], &rd->argv[1], IR_POW); UNUSED(rd); } diff --git a/src/lj_ircall.h b/src/lj_ircall.h index ac0888a0..9c195918 100644 --- a/src/lj_ircall.h +++ b/src/lj_ircall.h @@ -194,8 +194,7 @@ typedef struct CCallInfo { _(FPMATH, sqrt, 1, N, NUM, XA_FP) \ _(ANY, log, 1, N, NUM, XA_FP) \ _(ANY, lj_vm_log2, 1, N, NUM, XA_FP) \ - _(ANY, lj_vm_powi, 2, N, NUM, XA_FP) \ - _(ANY, lj_vm_pow, 2, N, NUM, XA2_FP) \ + _(ANY, pow, 2, N, NUM, XA2_FP) \ _(ANY, atan2, 2, N, NUM, XA2_FP) \ _(ANY, ldexp, 2, N, NUM, XA_FP) \ _(SOFTFP, lj_vm_tobit, 1, N, INT, XA_FP32) \ diff --git a/src/lj_iropt.h b/src/lj_iropt.h index a59ba3f4..7ee1ea86 100644 --- a/src/lj_iropt.h +++ b/src/lj_iropt.h @@ -144,7 +144,6 @@ LJ_FUNC TRef lj_opt_narrow_arith(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc, IROp op); LJ_FUNC TRef lj_opt_narrow_unm(jit_State *J, TRef rc, TValue *vc); LJ_FUNC TRef lj_opt_narrow_mod(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc); -LJ_FUNC TRef lj_opt_narrow_pow(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc); LJ_FUNC IRType lj_opt_narrow_forl(jit_State *J, cTValue *forbase); /* Optimization passes. */ diff --git a/src/lj_opt_fold.c b/src/lj_opt_fold.c index 7d7cc9d1..09e6c87b 100644 --- a/src/lj_opt_fold.c +++ b/src/lj_opt_fold.c @@ -236,14 +236,10 @@ LJFOLDF(kfold_fpcall2) return NEXTFOLD; } -LJFOLD(POW KNUM KINT) LJFOLD(POW KNUM KNUM) LJFOLDF(kfold_numpow) { - lua_Number a = knumleft; - lua_Number b = fright->o == IR_KINT ? (lua_Number)fright->i : knumright; - lua_Number y = lj_vm_foldarith(a, b, IR_POW - IR_ADD); - return lj_ir_knum(J, y); + return lj_ir_knum(J, lj_vm_foldarith(knumleft, knumright, IR_POW - IR_ADD)); } /* Must not use kfold_kref for numbers (could be NaN). */ @@ -1084,34 +1080,17 @@ LJFOLDF(simplify_nummuldiv_negneg) return RETRYFOLD; } -LJFOLD(POW any KINT) -LJFOLDF(simplify_numpow_xkint) +LJFOLD(POW any KNUM) +LJFOLDF(simplify_numpow_k) { - int32_t k = fright->i; - TRef ref = fins->op1; - if (k == 0) /* x ^ 0 ==> 1 */ + if (knumright == 0) /* x ^ 0 ==> 1 */ return lj_ir_knum_one(J); /* Result must be a number, not an int. */ - if (k == 1) /* x ^ 1 ==> x */ + else if (knumright == 1) /* x ^ 1 ==> x */ return LEFTFOLD; - if ((uint32_t)(k+65536) > 2*65536u) /* Limit code explosion. */ + else if (knumright == 2) /* x ^ 2 ==> x * x */ + return emitir(IRTN(IR_MUL), fins->op1, fins->op1); + else return NEXTFOLD; - if (k < 0) { /* x ^ (-k) ==> (1/x) ^ k. */ - ref = emitir(IRTN(IR_DIV), lj_ir_knum_one(J), ref); - k = -k; - } - /* Unroll x^k for 1 <= k <= 65536. */ - for (; (k & 1) == 0; k >>= 1) /* Handle leading zeros. */ - ref = emitir(IRTN(IR_MUL), ref, ref); - if ((k >>= 1) != 0) { /* Handle trailing bits. */ - TRef tmp = emitir(IRTN(IR_MUL), ref, ref); - for (; k != 1; k >>= 1) { - if (k & 1) - ref = emitir(IRTN(IR_MUL), ref, tmp); - tmp = emitir(IRTN(IR_MUL), tmp, tmp); - } - ref = emitir(IRTN(IR_MUL), ref, tmp); - } - return ref; } /* -- Simplify conversions ------------------------------------------------ */ diff --git a/src/lj_opt_narrow.c b/src/lj_opt_narrow.c index d6601f4c..db0da10f 100644 --- a/src/lj_opt_narrow.c +++ b/src/lj_opt_narrow.c @@ -584,30 +584,6 @@ TRef lj_opt_narrow_mod(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc) return emitir(IRTN(IR_SUB), rb, tmp); } -/* Narrowing of power operator or math.pow. */ -TRef lj_opt_narrow_pow(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc) -{ - rb = conv_str_tonum(J, rb, vb); - rb = lj_ir_tonum(J, rb); /* Left arg is always treated as an FP number. */ - rc = conv_str_tonum(J, rc, vc); - if (tvisint(vc) || numisint(numV(vc))) { - int32_t k = numberVint(vc); - if (!(k >= -65536 && k <= 65536)) goto force_pow_num; - if (!tref_isinteger(rc)) { - /* Guarded conversion to integer! */ - rc = emitir(IRTGI(IR_CONV), rc, IRCONV_INT_NUM|IRCONV_CHECK); - } - if (!tref_isk(rc)) { /* Range guard: -65536 <= i <= 65536 */ - TRef tmp = emitir(IRTI(IR_ADD), rc, lj_ir_kint(J, 65536)); - emitir(IRTGI(IR_ULE), tmp, lj_ir_kint(J, 2*65536)); - } - } else { -force_pow_num: - rc = lj_ir_tonum(J, rc); /* Want POW(num, num), not POW(num, int). */ - } - return emitir(IRTN(IR_POW), rb, rc); -} - /* -- Predictive narrowing of induction variables ------------------------- */ /* Narrow a single runtime value. */ diff --git a/src/lj_opt_split.c b/src/lj_opt_split.c index a619d852..0dc6394f 100644 --- a/src/lj_opt_split.c +++ b/src/lj_opt_split.c @@ -400,7 +400,7 @@ static void split_ir(jit_State *J) hi = split_call_ll(J, hisubst, oir, ir, IRCALL_softfp_div); break; case IR_POW: - hi = split_call_li(J, hisubst, oir, ir, IRCALL_lj_vm_powi); + hi = split_call_li(J, hisubst, oir, ir, IRCALL_pow); break; case IR_FPMATH: hi = split_call_l(J, hisubst, oir, ir, IRCALL_lj_vm_floor + ir->op2); diff --git a/src/lj_record.c b/src/lj_record.c index d1332bfc..34d1210a 100644 --- a/src/lj_record.c +++ b/src/lj_record.c @@ -2268,7 +2268,7 @@ void lj_record_ins(jit_State *J) case BC_POW: if (tref_isnumber_str(rb) && tref_isnumber_str(rc)) - rc = lj_opt_narrow_pow(J, rb, rc, rbv, rcv); + rc = lj_opt_narrow_arith(J, rb, rc, rbv, rcv, IR_POW); else rc = rec_mm_arith(J, &ix, MM_pow); break; diff --git a/src/lj_vm.h b/src/lj_vm.h index f6f28a08..79166e5e 100644 --- a/src/lj_vm.h +++ b/src/lj_vm.h @@ -96,9 +96,6 @@ LJ_ASMF int lj_vm_errno(void); #endif #endif -LJ_ASMF double lj_vm_powi(double, int32_t); -LJ_ASMF double lj_vm_pow(double, double); - /* Continuations for metamethods. */ LJ_ASMF void lj_cont_cat(void); /* Continue with concatenation. */ LJ_ASMF void lj_cont_ra(void); /* Store result in RA from instruction. */ diff --git a/src/lj_vmmath.c b/src/lj_vmmath.c index 539f955b..506867f8 100644 --- a/src/lj_vmmath.c +++ b/src/lj_vmmath.c @@ -30,52 +30,12 @@ LJ_FUNCA double lj_wrap_sinh(double x) { return sinh(x); } LJ_FUNCA double lj_wrap_cosh(double x) { return cosh(x); } LJ_FUNCA double lj_wrap_tanh(double x) { return tanh(x); } LJ_FUNCA double lj_wrap_atan2(double x, double y) { return atan2(x, y); } +LJ_FUNCA double lj_wrap_pow(double x, double y) { return pow(x, y); } LJ_FUNCA double lj_wrap_fmod(double x, double y) { return fmod(x, y); } #endif /* -- Helper functions ---------------------------------------------------- */ -/* Unsigned x^k. */ -static double lj_vm_powui(double x, uint32_t k) -{ - double y; - lj_assertX(k != 0, "pow with zero exponent"); - for (; (k & 1) == 0; k >>= 1) x *= x; - y = x; - if ((k >>= 1) != 0) { - for (;;) { - x *= x; - if (k == 1) break; - if (k & 1) y *= x; - k >>= 1; - } - y *= x; - } - return y; -} - -/* Signed x^k. */ -double lj_vm_powi(double x, int32_t k) -{ - if (k > 1) - return lj_vm_powui(x, (uint32_t)k); - else if (k == 1) - return x; - else if (k == 0) - return 1.0; - else - return 1.0 / lj_vm_powui(x, (uint32_t)-k); -} - -double lj_vm_pow(double x, double y) -{ - int32_t k = lj_num2int(y); - if ((k >= -65536 && k <= 65536) && y == (double)k) - return lj_vm_powi(x, k); - else - return pow(x, y); -} - double lj_vm_foldarith(double x, double y, int op) { switch (op) { @@ -84,7 +44,7 @@ double lj_vm_foldarith(double x, double y, int op) case IR_MUL - IR_ADD: return x*y; break; case IR_DIV - IR_ADD: return x/y; break; case IR_MOD - IR_ADD: return x-lj_vm_floor(x/y)*y; break; - case IR_POW - IR_ADD: return lj_vm_pow(x, y); break; + case IR_POW - IR_ADD: return pow(x, y); break; case IR_NEG - IR_ADD: return -x; break; case IR_ABS - IR_ADD: return fabs(x); break; #if LJ_HASJIT diff --git a/src/vm_arm.dasc b/src/vm_arm.dasc index 792f0363..767d31f9 100644 --- a/src/vm_arm.dasc +++ b/src/vm_arm.dasc @@ -1485,11 +1485,11 @@ static void build_subroutines(BuildCtx *ctx) |.endif |.endmacro | - |.macro math_extern2, name, func + |.macro math_extern2, func |.if HFABI - | .ffunc_dd math_ .. name + | .ffunc_dd math_ .. func |.else - | .ffunc_nn math_ .. name + | .ffunc_nn math_ .. func |.endif | .IOS mov RA, BASE | bl extern func @@ -1500,9 +1500,6 @@ static void build_subroutines(BuildCtx *ctx) | b ->fff_restv |.endif |.endmacro - |.macro math_extern2, func - | math_extern2 func, func - |.endmacro | |.if FPU | .ffunc_d math_sqrt @@ -1548,7 +1545,7 @@ static void build_subroutines(BuildCtx *ctx) | math_extern sinh | math_extern cosh | math_extern tanh - | math_extern2 pow, lj_vm_pow + | math_extern2 pow | math_extern2 atan2 | math_extern2 fmod | @@ -3156,7 +3153,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop) break; case BC_POW: | // NYI: (partial) integer arithmetic. - | ins_arithfp extern, extern lj_vm_pow + | ins_arithfp extern, extern pow break; case BC_CAT: diff --git a/src/vm_arm64.dasc b/src/vm_arm64.dasc index fb267a76..de33bde4 100644 --- a/src/vm_arm64.dasc +++ b/src/vm_arm64.dasc @@ -1391,14 +1391,11 @@ static void build_subroutines(BuildCtx *ctx) | b ->fff_resn |.endmacro | - |.macro math_extern2, name, func - | .ffunc_nn math_ .. name + |.macro math_extern2, func + | .ffunc_nn math_ .. func | bl extern func | b ->fff_resn |.endmacro - |.macro math_extern2, func - | math_extern2 func, func - |.endmacro | |.ffunc_n math_sqrt | fsqrt d0, d0 @@ -1427,7 +1424,7 @@ static void build_subroutines(BuildCtx *ctx) | math_extern sinh | math_extern cosh | math_extern tanh - | math_extern2 pow, lj_vm_pow + | math_extern2 pow | math_extern2 atan2 | math_extern2 fmod | @@ -2624,7 +2621,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop) | ins_arithload FARG1, FARG2 | ins_arithfallback ins_arithcheck_num |.if "fpins" == "fpow" - | bl extern lj_vm_pow + | bl extern pow |.else | fpins FARG1, FARG1, FARG2 |.endif diff --git a/src/vm_mips.dasc b/src/vm_mips.dasc index 5664f503..32caabf7 100644 --- a/src/vm_mips.dasc +++ b/src/vm_mips.dasc @@ -1631,17 +1631,14 @@ static void build_subroutines(BuildCtx *ctx) |. nop |.endmacro | - |.macro math_extern2, name, func - | .ffunc_nn math_ .. name + |.macro math_extern2, func + | .ffunc_nn math_ .. func |. load_got func | call_extern |. nop | b ->fff_resn |. nop |.endmacro - |.macro math_extern2, func - | math_extern2 func, func - |.endmacro | |// TODO: Return integer type if result is integer (own sf implementation). |.macro math_round, func @@ -1695,7 +1692,7 @@ static void build_subroutines(BuildCtx *ctx) | math_extern sinh | math_extern cosh | math_extern tanh - | math_extern2 pow, lj_vm_pow + | math_extern2 pow | math_extern2 atan2 | math_extern2 fmod | @@ -3588,7 +3585,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop) | sltiu AT, SFARG1HI, LJ_TISNUM | sltiu TMP0, SFARG2HI, LJ_TISNUM | and AT, AT, TMP0 - | load_got lj_vm_pow + | load_got pow | beqz AT, ->vmeta_arith |. addu RA, BASE, RA |.if FPU diff --git a/src/vm_mips64.dasc b/src/vm_mips64.dasc index 249605d4..44fba36c 100644 --- a/src/vm_mips64.dasc +++ b/src/vm_mips64.dasc @@ -1669,17 +1669,14 @@ static void build_subroutines(BuildCtx *ctx) |. nop |.endmacro | - |.macro math_extern2, name, func - | .ffunc_nn math_ .. name + |.macro math_extern2, func + | .ffunc_nn math_ .. func |. load_got func | call_extern |. nop | b ->fff_resn |. nop |.endmacro - |.macro math_extern2, func - | math_extern2 func, func - |.endmacro | |// TODO: Return integer type if result is integer (own sf implementation). |.macro math_round, func @@ -1733,7 +1730,7 @@ static void build_subroutines(BuildCtx *ctx) | math_extern sinh | math_extern cosh | math_extern tanh - | math_extern2 pow, lj_vm_pow + | math_extern2 pow | math_extern2 atan2 | math_extern2 fmod | @@ -3826,7 +3823,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop) | sltiu TMP0, TMP0, LJ_TISNUM | sltiu TMP1, TMP1, LJ_TISNUM | and AT, TMP0, TMP1 - | load_got lj_vm_pow + | load_got pow | beqz AT, ->vmeta_arith |. daddu RA, BASE, RA |.if FPU diff --git a/src/vm_ppc.dasc b/src/vm_ppc.dasc index 94af63e6..980ad897 100644 --- a/src/vm_ppc.dasc +++ b/src/vm_ppc.dasc @@ -2032,14 +2032,11 @@ static void build_subroutines(BuildCtx *ctx) | b ->fff_resn |.endmacro | - |.macro math_extern2, name, func - | .ffunc_nn math_ .. name + |.macro math_extern2, func + | .ffunc_nn math_ .. func | blex func | b ->fff_resn |.endmacro - |.macro math_extern2, func - | math_extern2 func, func - |.endmacro | |.macro math_round, func | .ffunc_1 math_ .. func @@ -2164,7 +2161,7 @@ static void build_subroutines(BuildCtx *ctx) | math_extern sinh | math_extern cosh | math_extern tanh - | math_extern2 pow, lj_vm_pow + | math_extern2 pow | math_extern2 atan2 | math_extern2 fmod | @@ -4157,7 +4154,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop) | checknum cr1, CARG3 | crand 4*cr0+lt, 4*cr0+lt, 4*cr1+lt | bge ->vmeta_arith_vv - | blex lj_vm_pow + | blex pow | ins_next1 |.if FPU | stfdx FARG1, BASE, RA diff --git a/src/vm_x64.dasc b/src/vm_x64.dasc index acbe8dc2..09bf67e5 100644 --- a/src/vm_x64.dasc +++ b/src/vm_x64.dasc @@ -1825,16 +1825,13 @@ static void build_subroutines(BuildCtx *ctx) | jmp ->fff_resxmm0 |.endmacro | - |.macro math_extern2, name, func - | .ffunc_nn math_ .. name + |.macro math_extern2, func + | .ffunc_nn math_ .. func | mov RB, BASE | call extern func | mov BASE, RB | jmp ->fff_resxmm0 |.endmacro - |.macro math_extern2, func - | math_extern2 func, func - |.endmacro | | math_extern log10 | math_extern exp @@ -1847,7 +1844,7 @@ static void build_subroutines(BuildCtx *ctx) | math_extern sinh | math_extern cosh | math_extern tanh - | math_extern2 pow, lj_vm_pow + | math_extern2 pow | math_extern2 atan2 | math_extern2 fmod | diff --git a/src/vm_x86.dasc b/src/vm_x86.dasc index bf30cce6..f16ade1a 100644 --- a/src/vm_x86.dasc +++ b/src/vm_x86.dasc @@ -2240,8 +2240,8 @@ static void build_subroutines(BuildCtx *ctx) | jmp ->fff_resfp |.endmacro | - |.macro math_extern2, name, func - | .ffunc_nnsse math_ .. name + |.macro math_extern2, func + | .ffunc_nnsse math_ .. func |.if not X64 | movsd FPARG1, xmm0 | movsd FPARG3, xmm1 @@ -2251,9 +2251,6 @@ static void build_subroutines(BuildCtx *ctx) | mov BASE, RB | jmp ->fff_resfp |.endmacro - |.macro math_extern2, func - | math_extern2 func, func - |.endmacro | | math_extern log10 | math_extern exp @@ -2266,7 +2263,7 @@ static void build_subroutines(BuildCtx *ctx) | math_extern sinh | math_extern cosh | math_extern tanh - | math_extern2 pow, lj_vm_pow + | math_extern2 pow | math_extern2 atan2 | math_extern2 fmod | @@ -3944,7 +3941,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop) | movsd FPARG1, xmm0 | movsd FPARG3, xmm1 |.endif - | call extern lj_vm_pow + | call extern pow | movzx RA, PC_RA | mov BASE, RB |.if X64 diff --git a/test/tarantool-tests/lj-684-pow-inconsistencies.test.lua b/test/tarantool-tests/lj-684-pow-inconsistencies.test.lua index 5129fc45..ab9db3df 100644 --- a/test/tarantool-tests/lj-684-pow-inconsistencies.test.lua +++ b/test/tarantool-tests/lj-684-pow-inconsistencies.test.lua @@ -2,14 +2,15 @@ local tap = require('tap') -- Test to demonstrate the incorrect JIT behaviour for different -- power operation optimizations. -- See also: --- https://github.com/LuaJIT/LuaJIT/issues/684. +-- https://github.com/LuaJIT/LuaJIT/issues/684, +-- https://github.com/LuaJIT/LuaJIT/issues/817. local test = tap.test('lj-684-pow-inconsistencies'):skipcond({ ['Test requires JIT enabled'] = not jit.status(), }) local tostring = tostring -test:plan(4) +test:plan(5) jit.opt.start('hotloop=1') @@ -64,6 +65,22 @@ jit.flush() test:samevalues(res, ('consistent results for folding 2921 ^ 0.5')) +-- -948388 ^ 3 = -0x1.7ad0e8ad7439dp+59. +res = {} +-- XXX: use local variable to prevent folding via parser. +-- XXX: use stack slot out of trace to prevent constant folding. +local corner_case_3 = -948388 +jit.on() +for i = 1, 4 do + res[i] = corner_case_3 ^ 3 +end + +-- XXX: Prevent hotcount side effects. +jit.off() +jit.flush() + +test:samevalues(res, ('consistent results for int pow (-948388) ^ 3')) + -- Narrowing for non-constant base of power operation. local function pow(base, power) return base ^ power -- 2.41.0