/* * Copyright (c) 2016, 2024, Intel Corporation. All rights reserved. * Intel Math Library (LIBM) Source Code * * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. * */ #include "macroAssembler_x86.hpp" #include "stubGenerator_x86_64.hpp" /******************************************************************************/ // ALGORITHM DESCRIPTION - SIN() // --------------------- // // 1. RANGE REDUCTION // // We perform an initial range reduction from X to r with // // X =~= N * pi/32 + r // // so that |r| <= pi/64 + epsilon. We restrict inputs to those // where |N| <= 932560. Beyond this, the range reduction is // insufficiently accurate. For extremely small inputs, // denormalization can occur internally, impacting performance. // This means that the main path is actually only taken for // 2^-252 <= |X| < 90112. // // To avoid branches, we perform the range reduction to full // accuracy each time. // // X - N * (P_1 + P_2 + P_3) // // where P_1 and P_2 are 32-bit numbers (so multiplication by N // is exact) and P_3 is a 53-bit number. Together, these // approximate pi well enough for all cases in the restricted // range. // // The main reduction sequence is: // // y = 32/pi * x // N = integer(y) // (computed by adding and subtracting off SHIFTER) // // m_1 = N * P_1 // m_2 = N * P_2 // r_1 = x - m_1 // r = r_1 - m_2 // (this r can be used for most of the calculation) // // c_1 = r_1 - r // m_3 = N * P_3 // c_2 = c_1 - m_2 // c = c_2 - m_3 // // 2. MAIN ALGORITHM // // The algorithm uses a table lookup based on B = M * pi / 32 // where M = N mod 64. The stored values are: // sigma closest power of 2 to cos(B) // C_hl 53-bit cos(B) - sigma // S_hi + S_lo 2 * 53-bit sin(B) // // The computation is organized as follows: // // sin(B + r + c) = [sin(B) + sigma * r] + // r * (cos(B) - sigma) + // sin(B) * [cos(r + c) - 1] + // cos(B) * [sin(r + c) - r] // // which is approximately: // // [S_hi + sigma * r] + // C_hl * r + // S_lo + S_hi * [(cos(r) - 1) - r * c] + // (C_hl + sigma) * [(sin(r) - r) + c] // // and this is what is actually computed. We separate this sum // into four parts: // // hi + med + pols + corr // // where // // hi = S_hi + sigma r // med = C_hl * r // pols = S_hi * (cos(r) - 1) + (C_hl + sigma) * (sin(r) - r) // corr = S_lo + c * ((C_hl + sigma) - S_hi * r) // // 3. POLYNOMIAL // // The polynomial S_hi * (cos(r) - 1) + (C_hl + sigma) * // (sin(r) - r) can be rearranged freely, since it is quite // small, so we exploit parallelism to the fullest. // // psc4 = SC_4 * r_1 // msc4 = psc4 * r // r2 = r * r // msc2 = SC_2 * r2 // r4 = r2 * r2 // psc3 = SC_3 + msc4 // psc1 = SC_1 + msc2 // msc3 = r4 * psc3 // sincospols = psc1 + msc3 // pols = sincospols * // // // 4. CORRECTION TERM // // This is where the "c" component of the range reduction is // taken into account; recall that just "r" is used for most of // the calculation. // // -c = m_3 - c_2 // -d = S_hi * r - (C_hl + sigma) // corr = -c * -d + S_lo // // 5. COMPENSATED SUMMATIONS // // The two successive compensated summations add up the high // and medium parts, leaving just the low parts to add up at // the end. // // rs = sigma * r // res_int = S_hi + rs // k_0 = S_hi - res_int // k_2 = k_0 + rs // med = C_hl * r // res_hi = res_int + med // k_1 = res_int - res_hi // k_3 = k_1 + med // // 6. FINAL SUMMATION // // We now add up all the small parts: // // res_lo = pols(hi) + pols(lo) + corr + k_1 + k_3 // // Now the overall result is just: // // res_hi + res_lo // // 7. SMALL ARGUMENTS // // If |x| < SNN (SNN meaning the smallest normal number), we // simply perform 0.1111111 cdots 1111 * x. For SNN <= |x|, we // do 2^-55 * (2^55 * x - x). // // Special cases: // sin(NaN) = quiet NaN, and raise invalid exception // sin(INF) = NaN and raise invalid exception // sin(+/-0) = +/-0 // /******************************************************************************/ // The 64 bit code is at most SSE2 compliant ATTRIBUTE_ALIGNED(8) static const juint _ALL_ONES[] = { 0xffffffffUL, 0x3fefffffUL }; #define __ _masm-> address StubGenerator::generate_libmSin() { StubGenStubId stub_id = StubGenStubId::dsin_id; StubCodeMark mark(this, stub_id); address start = __ pc(); Label L_2TAG_PACKET_0_0_1, L_2TAG_PACKET_1_0_1, L_2TAG_PACKET_2_0_1, L_2TAG_PACKET_3_0_1; Label L_2TAG_PACKET_4_0_1, L_2TAG_PACKET_5_0_1, L_2TAG_PACKET_6_0_1, L_2TAG_PACKET_7_0_1; Label L_2TAG_PACKET_8_0_1, L_2TAG_PACKET_9_0_1, L_2TAG_PACKET_10_0_1, L_2TAG_PACKET_11_0_1; Label L_2TAG_PACKET_13_0_1, L_2TAG_PACKET_14_0_1; Label L_2TAG_PACKET_12_0_1, B1_4; address ALL_ONES = (address)_ALL_ONES; __ enter(); // required for proper stackwalking of RuntimeStub frame #ifdef _WIN64 __ push(rsi); __ push(rdi); #endif __ push(rbx); __ subq(rsp, 16); __ movsd(Address(rsp, 8), xmm0); __ movl(rax, Address(rsp, 12)); __ movq(xmm1, ExternalAddress(PI32INV), r8 /*rscratch*/); //0x6dc9c883UL, 0x40245f30UL __ movq(xmm2, ExternalAddress(SHIFTER), r8 /*rscratch*/); //0x00000000UL, 0x43380000UL __ andl(rax, 2147418112); __ subl(rax, 808452096); __ cmpl(rax, 281346048); __ jcc(Assembler::above, L_2TAG_PACKET_0_0_1); __ mulsd(xmm1, xmm0); __ movdqu(xmm5, ExternalAddress(ONEHALF), r8 /*rscratch*/); //0x00000000UL, 0x3fe00000UL, 0x00000000UL, 0x3fe00000UL __ movq(xmm4, ExternalAddress(SIGN_MASK), r8 /*rscratch*/); //0x00000000UL, 0x80000000UL __ pand(xmm4, xmm0); __ por(xmm5, xmm4); __ addpd(xmm1, xmm5); __ cvttsd2sil(rdx, xmm1); __ cvtsi2sdl(xmm1, rdx); __ movdqu(xmm6, ExternalAddress(P_2), r8 /*rscratch*/); //0x1a600000UL, 0x3d90b461UL, 0x1a600000UL, 0x3d90b461UL __ mov64(r8, 0x3fb921fb54400000); __ movdq(xmm3, r8); __ movdqu(xmm5, ExternalAddress(SC_4), r8 /*rscratch*/); //0xa556c734UL, 0x3ec71de3UL, 0x1a01a01aUL, 0x3efa01a0UL __ pshufd(xmm4, xmm0, 68); __ mulsd(xmm3, xmm1); if (VM_Version::supports_sse3()) { __ movddup(xmm1, xmm1); } else { __ movlhps(xmm1, xmm1); } __ andl(rdx, 63); __ shll(rdx, 5); __ lea(rax, ExternalAddress(Ctable)); __ addq(rax, rdx); __ mulpd(xmm6, xmm1); __ mulsd(xmm1, ExternalAddress(P_3), r8 /*rscratch*/); //0x2e037073UL, 0x3b63198aUL __ subsd(xmm4, xmm3); __ movq(xmm7, Address(rax, 8)); __ subsd(xmm0, xmm3); if (VM_Version::supports_sse3()) { __ movddup(xmm3, xmm4); } else { __ movdqu(xmm3, xmm4); __ movlhps(xmm3, xmm3); } __ subsd(xmm4, xmm6); __ pshufd(xmm0, xmm0, 68); __ movdqu(xmm2, Address(rax, 0)); __ mulpd(xmm5, xmm0); __ subpd(xmm0, xmm6); __ mulsd(xmm7, xmm4); __ subsd(xmm3, xmm4); __ mulpd(xmm5, xmm0); __ mulpd(xmm0, xmm0); __ subsd(xmm3, xmm6); __ movdqu(xmm6, ExternalAddress(SC_2), r8 /*rscratch*/); //0x11111111UL, 0x3f811111UL, 0x55555555UL, 0x3fa55555UL __ subsd(xmm1, xmm3); __ movq(xmm3, Address(rax, 24)); __ addsd(xmm2, xmm3); __ subsd(xmm7, xmm2); __ mulsd(xmm2, xmm4); __ mulpd(xmm6, xmm0); __ mulsd(xmm3, xmm4); __ mulpd(xmm2, xmm0); __ mulpd(xmm0, xmm0); __ addpd(xmm5, ExternalAddress(SC_3), r8 /*rscratch*/); //0x1a01a01aUL, 0xbf2a01a0UL, 0x16c16c17UL, 0xbf56c16cUL __ mulsd(xmm4, Address(rax, 0)); __ addpd(xmm6, ExternalAddress(SC_1), r8 /*rscratch*/); //0x55555555UL, 0xbfc55555UL, 0x00000000UL, 0xbfe00000UL __ mulpd(xmm5, xmm0); __ movdqu(xmm0, xmm3); __ addsd(xmm3, Address(rax, 8)); __ mulpd(xmm1, xmm7); __ movdqu(xmm7, xmm4); __ addsd(xmm4, xmm3); __ addpd(xmm6, xmm5); __ movq(xmm5, Address(rax, 8)); __ subsd(xmm5, xmm3); __ subsd(xmm3, xmm4); __ addsd(xmm1, Address(rax, 16)); __ mulpd(xmm6, xmm2); __ addsd(xmm5, xmm0); __ addsd(xmm3, xmm7); __ addsd(xmm1, xmm5); __ addsd(xmm1, xmm3); __ addsd(xmm1, xmm6); __ unpckhpd(xmm6, xmm6); __ movdqu(xmm0, xmm4); __ addsd(xmm1, xmm6); __ addsd(xmm0, xmm1); __ jmp(B1_4); __ bind(L_2TAG_PACKET_0_0_1); __ jcc(Assembler::greater, L_2TAG_PACKET_1_0_1); __ shrl(rax, 20); __ cmpl(rax, 3325); __ jcc(Assembler::notEqual, L_2TAG_PACKET_2_0_1); __ mulsd(xmm0, ExternalAddress(ALL_ONES), r8 /*rscratch*/); //0xffffffffUL, 0x3fefffffUL __ jmp(B1_4); __ bind(L_2TAG_PACKET_2_0_1); __ movq(xmm3, ExternalAddress(TWO_POW_55), r8 /*rscratch*/); //0x00000000UL, 0x43600000UL __ mulsd(xmm3, xmm0); __ subsd(xmm3, xmm0); __ mulsd(xmm3, ExternalAddress(TWO_POW_M55), r8 /*rscratch*/); //0x00000000UL, 0x3c800000UL __ jmp(B1_4); __ bind(L_2TAG_PACKET_1_0_1); __ pextrw(rax, xmm0, 3); __ andl(rax, 32752); __ cmpl(rax, 32752); __ jcc(Assembler::equal, L_2TAG_PACKET_3_0_1); __ pextrw(rcx, xmm0, 3); __ andl(rcx, 32752); __ subl(rcx, 16224); __ shrl(rcx, 7); __ andl(rcx, 65532); __ lea(r11, ExternalAddress(PI_INV_TABLE)); __ addq(rcx, r11); __ movdq(rax, xmm0); __ movl(r10, Address(rcx, 20)); __ movl(r8, Address(rcx, 24)); __ movl(rdx, rax); __ shrq(rax, 21); __ orl(rax, INT_MIN); __ shrl(rax, 11); __ movl(r9, r10); __ imulq(r10, rdx); __ imulq(r9, rax); __ imulq(r8, rax); __ movl(rsi, Address(rcx, 16)); __ movl(rdi, Address(rcx, 12)); __ movl(r11, r10); __ shrq(r10, 32); __ addq(r9, r10); __ addq(r11, r8); __ movl(r8, r11); __ shrq(r11, 32); __ addq(r9, r11); __ movl(r10, rsi); __ imulq(rsi, rdx); __ imulq(r10, rax); __ movl(r11, rdi); __ imulq(rdi, rdx); __ movl(rbx, rsi); __ shrq(rsi, 32); __ addq(r9, rbx); __ movl(rbx, r9); __ shrq(r9, 32); __ addq(r10, rsi); __ addq(r10, r9); __ shlq(rbx, 32); __ orq(r8, rbx); __ imulq(r11, rax); __ movl(r9, Address(rcx, 8)); __ movl(rsi, Address(rcx, 4)); __ movl(rbx, rdi); __ shrq(rdi, 32); __ addq(r10, rbx); __ movl(rbx, r10); __ shrq(r10, 32); __ addq(r11, rdi); __ addq(r11, r10); __ movq(rdi, r9); __ imulq(r9, rdx); __ imulq(rdi, rax); __ movl(r10, r9); __ shrq(r9, 32); __ addq(r11, r10); __ movl(r10, r11); __ shrq(r11, 32); __ addq(rdi, r9); __ addq(rdi, r11); __ movq(r9, rsi); __ imulq(rsi, rdx); __ imulq(r9, rax); __ shlq(r10, 32); __ orq(r10, rbx); __ movl(rax, Address(rcx, 0)); __ movl(r11, rsi); __ shrq(rsi, 32); __ addq(rdi, r11); __ movl(r11, rdi); __ shrq(rdi, 32); __ addq(r9, rsi); __ addq(r9, rdi); __ imulq(rdx, rax); __ pextrw(rbx, xmm0, 3); __ lea(rdi, ExternalAddress(PI_INV_TABLE)); __ subq(rcx, rdi); __ addl(rcx, rcx); __ addl(rcx, rcx); __ addl(rcx, rcx); __ addl(rcx, 19); __ movl(rsi, 32768); __ andl(rsi, rbx); __ shrl(rbx, 4); __ andl(rbx, 2047); __ subl(rbx, 1023); __ subl(rcx, rbx); __ addq(r9, rdx); __ movl(rdx, rcx); __ addl(rdx, 32); __ cmpl(rcx, 1); __ jcc(Assembler::less, L_2TAG_PACKET_4_0_1); __ negl(rcx); __ addl(rcx, 29); __ shll(r9); __ movl(rdi, r9); __ andl(r9, 536870911); __ testl(r9, 268435456); __ jcc(Assembler::notEqual, L_2TAG_PACKET_5_0_1); __ shrl(r9); __ movl(rbx, 0); __ shlq(r9, 32); __ orq(r9, r11); __ bind(L_2TAG_PACKET_6_0_1); __ bind(L_2TAG_PACKET_7_0_1); __ cmpq(r9, 0); __ jcc(Assembler::equal, L_2TAG_PACKET_8_0_1); __ bind(L_2TAG_PACKET_9_0_1); __ bsrq(r11, r9); __ movl(rcx, 29); __ subl(rcx, r11); __ jcc(Assembler::lessEqual, L_2TAG_PACKET_10_0_1); __ shlq(r9); __ movq(rax, r10); __ shlq(r10); __ addl(rdx, rcx); __ negl(rcx); __ addl(rcx, 64); __ shrq(rax); __ shrq(r8); __ orq(r9, rax); __ orq(r10, r8); __ bind(L_2TAG_PACKET_11_0_1); __ cvtsi2sdq(xmm0, r9); __ shrq(r10, 1); __ cvtsi2sdq(xmm3, r10); __ xorpd(xmm4, xmm4); __ shll(rdx, 4); __ negl(rdx); __ addl(rdx, 16368); __ orl(rdx, rsi); __ xorl(rdx, rbx); __ pinsrw(xmm4, rdx, 3); __ movq(xmm2, ExternalAddress(PI_4), r8 /*rscratch*/); //0x40000000UL, 0x3fe921fbUL, 0x18469899UL, 0x3e64442dUL __ movq(xmm6, ExternalAddress(PI_4 + 8), r8 /*rscratch*/); //0x3fe921fbUL, 0x18469899UL, 0x3e64442dUL __ xorpd(xmm5, xmm5); __ subl(rdx, 1008); __ pinsrw(xmm5, rdx, 3); __ mulsd(xmm0, xmm4); __ shll(rsi, 16); __ sarl(rsi, 31); __ mulsd(xmm3, xmm5); __ movdqu(xmm1, xmm0); __ mulsd(xmm0, xmm2); __ shrl(rdi, 29); __ addsd(xmm1, xmm3); __ mulsd(xmm3, xmm2); __ addl(rdi, rsi); __ xorl(rdi, rsi); __ mulsd(xmm6, xmm1); __ movl(rax, rdi); __ addsd(xmm6, xmm3); __ movdqu(xmm2, xmm0); __ addsd(xmm0, xmm6); __ subsd(xmm2, xmm0); __ addsd(xmm6, xmm2); __ bind(L_2TAG_PACKET_12_0_1); __ movq(xmm1, ExternalAddress(PI32INV), r8 /*rscratch*/); //0x6dc9c883UL, 0x40245f30UL __ mulsd(xmm1, xmm0); __ movq(xmm5, ExternalAddress(ONEHALF), r8 /*rscratch*/); //0x00000000UL, 0x3fe00000UL, 0x00000000UL, 0x3fe00000UL __ movq(xmm4, ExternalAddress(SIGN_MASK), r8 /*rscratch*/); //0x00000000UL, 0x80000000UL __ pand(xmm4, xmm0); __ por(xmm5, xmm4); __ addpd(xmm1, xmm5); __ cvttsd2sil(rdx, xmm1); __ cvtsi2sdl(xmm1, rdx); __ movq(xmm3, ExternalAddress(P_1), r8 /*rscratch*/); //0x54400000UL, 0x3fb921fbUL __ movdqu(xmm2, ExternalAddress(P_2), r8 /*rscratch*/); //0x1a600000UL, 0x3d90b461UL, 0x1a600000UL, 0x3d90b461UL __ mulsd(xmm3, xmm1); __ unpcklpd(xmm1, xmm1); __ shll(rax, 3); __ addl(rdx, 1865216); __ movdqu(xmm4, xmm0); __ addl(rdx, rax); __ andl(rdx, 63); __ movdqu(xmm5, ExternalAddress(SC_4), r8 /*rscratch*/); //0x54400000UL, 0x3fb921fbUL __ lea(rax, ExternalAddress(Ctable)); __ shll(rdx, 5); __ addq(rax, rdx); __ mulpd(xmm2, xmm1); __ subsd(xmm0, xmm3); __ mulsd(xmm1, ExternalAddress(P_3), r8 /*rscratch*/); //0x2e037073UL, 0x3b63198aUL __ subsd(xmm4, xmm3); __ movq(xmm7, Address(rax, 8)); __ unpcklpd(xmm0, xmm0); __ movdqu(xmm3, xmm4); __ subsd(xmm4, xmm2); __ mulpd(xmm5, xmm0); __ subpd(xmm0, xmm2); __ mulsd(xmm7, xmm4); __ subsd(xmm3, xmm4); __ mulpd(xmm5, xmm0); __ mulpd(xmm0, xmm0); __ subsd(xmm3, xmm2); __ movdqu(xmm2, Address(rax, 0)); __ subsd(xmm1, xmm3); __ movq(xmm3, Address(rax, 24)); __ addsd(xmm2, xmm3); __ subsd(xmm7, xmm2); __ subsd(xmm1, xmm6); __ movdqu(xmm6, ExternalAddress(SC_2), r8 /*rscratch*/); //0x11111111UL, 0x3f811111UL, 0x55555555UL, 0x3fa55555UL __ mulsd(xmm2, xmm4); __ mulpd(xmm6, xmm0); __ mulsd(xmm3, xmm4); __ mulpd(xmm2, xmm0); __ mulpd(xmm0, xmm0); __ addpd(xmm5, ExternalAddress(SC_3), r8 /*rscratch*/); //0x1a01a01aUL, 0xbf2a01a0UL, 0x16c16c17UL, 0xbf56c16cUL __ mulsd(xmm4, Address(rax, 0)); __ addpd(xmm6, ExternalAddress(SC_1), r8 /*rscratch*/); //0x55555555UL, 0xbfc55555UL, 0x00000000UL, 0xbfe00000UL __ mulpd(xmm5, xmm0); __ movdqu(xmm0, xmm3); __ addsd(xmm3, Address(rax, 8)); __ mulpd(xmm1, xmm7); __ movdqu(xmm7, xmm4); __ addsd(xmm4, xmm3); __ addpd(xmm6, xmm5); __ movq(xmm5, Address(rax, 8)); __ subsd(xmm5, xmm3); __ subsd(xmm3, xmm4); __ addsd(xmm1, Address(rax, 16)); __ mulpd(xmm6, xmm2); __ addsd(xmm5, xmm0); __ addsd(xmm3, xmm7); __ addsd(xmm1, xmm5); __ addsd(xmm1, xmm3); __ addsd(xmm1, xmm6); __ unpckhpd(xmm6, xmm6); __ movdqu(xmm0, xmm4); __ addsd(xmm1, xmm6); __ addsd(xmm0, xmm1); __ jmp(B1_4); __ bind(L_2TAG_PACKET_8_0_1); __ addl(rdx, 64); __ movq(r9, r10); __ movq(r10, r8); __ movl(r8, 0); __ cmpq(r9, 0); __ jcc(Assembler::notEqual, L_2TAG_PACKET_9_0_1); __ addl(rdx, 64); __ movq(r9, r10); __ movq(r10, r8); __ cmpq(r9, 0); __ jcc(Assembler::notEqual, L_2TAG_PACKET_9_0_1); __ xorpd(xmm0, xmm0); __ xorpd(xmm6, xmm6); __ jmp(L_2TAG_PACKET_12_0_1); __ bind(L_2TAG_PACKET_10_0_1); __ jcc(Assembler::equal, L_2TAG_PACKET_11_0_1); __ negl(rcx); __ shrq(r10); __ movq(rax, r9); __ shrq(r9); __ subl(rdx, rcx); __ negl(rcx); __ addl(rcx, 64); __ shlq(rax); __ orq(r10, rax); __ jmp(L_2TAG_PACKET_11_0_1); __ bind(L_2TAG_PACKET_4_0_1); __ negl(rcx); __ shlq(r9, 32); __ orq(r9, r11); __ shlq(r9); __ movq(rdi, r9); __ testl(r9, INT_MIN); __ jcc(Assembler::notEqual, L_2TAG_PACKET_13_0_1); __ shrl(r9); __ movl(rbx, 0); __ shrq(rdi, 3); __ jmp(L_2TAG_PACKET_7_0_1); __ bind(L_2TAG_PACKET_5_0_1); __ shrl(r9); __ movl(rbx, 536870912); __ shrl(rbx); __ shlq(r9, 32); __ orq(r9, r11); __ shlq(rbx, 32); __ addl(rdi, 536870912); __ movl(rcx, 0); __ movl(r11, 0); __ subq(rcx, r8); __ sbbq(r11, r10); __ sbbq(rbx, r9); __ movq(r8, rcx); __ movq(r10, r11); __ movq(r9, rbx); __ movl(rbx, 32768); __ jmp(L_2TAG_PACKET_6_0_1); __ bind(L_2TAG_PACKET_13_0_1); __ shrl(r9); __ mov64(rbx, 0x100000000); __ shrq(rbx); __ movl(rcx, 0); __ movl(r11, 0); __ subq(rcx, r8); __ sbbq(r11, r10); __ sbbq(rbx, r9); __ movq(r8, rcx); __ movq(r10, r11); __ movq(r9, rbx); __ movl(rbx, 32768); __ shrq(rdi, 3); __ addl(rdi, 536870912); __ jmp(L_2TAG_PACKET_7_0_1); __ bind(L_2TAG_PACKET_3_0_1); __ movq(xmm0, Address(rsp, 8)); __ mulsd(xmm0, ExternalAddress(NEG_ZERO), r8 /*rscratch*/); //0x00000000UL, 0x80000000UL __ movq(Address(rsp, 0), xmm0); __ bind(L_2TAG_PACKET_14_0_1); __ bind(B1_4); __ addq(rsp, 16); __ pop(rbx); #ifdef _WIN64 __ pop(rdi); __ pop(rsi); #endif __ leave(); // required for proper stackwalking of RuntimeStub frame __ ret(0); return start; } #undef __