Skip to content

Commit c72f951

Browse files
YaSuenagYaSuenag
committed
8301853: C4819 warnings were reported in HotSpot on Windows
Reviewed-by: dholmes, ihse
1 parent 36478ee commit c72f951

File tree

2 files changed

+54
-54
lines changed

2 files changed

+54
-54
lines changed

‎src/hotspot/cpu/x86/stubGenerator_x86_64_poly.cpp‎

Lines changed: 53 additions & 53 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
/*
2-
* Copyright (c) 2022, Intel Corporation. All rights reserved.
2+
* Copyright (c) 2022, 2023, Intel Corporation. All rights reserved.
33
*
44
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
55
*
@@ -37,19 +37,19 @@
3737
// Explanation for the 'well known' modular arithmetic optimization, reduction by pseudo-Mersene prime 2^130-5:
3838
//
3939
// Reduction by 2^130-5 can be expressed as follows:
40-
// ( a×2^130 + b ) mod 2^130-5 //i.e. number split along the 130-bit boundary
41-
// = ( a×2^130 - 5×a + 5×a + b ) mod 2^130-5
42-
// = ( a×(2^130 - 5) + 5×a + b ) mod 2^130-5 // i.e. adding multiples of modulus is a noop
43-
// = ( 5×a + b ) mod 2^130-5
40+
// ( ax2^130 + b ) mod 2^130-5 //i.e. number split along the 130-bit boundary
41+
// = ( ax2^130 - 5xa + 5xa + b ) mod 2^130-5
42+
// = ( ax(2^130 - 5) + 5xa + b ) mod 2^130-5 // i.e. adding multiples of modulus is a noop
43+
// = ( 5xa + b ) mod 2^130-5
4444
// QED: shows mathematically the well known algorithm of 'split the number down the middle, multiply upper and add'
4545
// This is particularly useful to understand when combining with 'odd-sized' limbs that might cause misallignment
4646
//
4747

4848
// Pseudocode for this file (in general):
4949
// * used for poly1305_multiply_scalar
50-
// × used for poly1305_multiply8_avx512
51-
// lower-case variables are scalar numbers in 3×44-bit limbs (in gprs)
52-
// upper-case variables are 8-element vector numbers in 3×44-bit limbs (in zmm registers)
50+
// x used for poly1305_multiply8_avx512
51+
// lower-case variables are scalar numbers in 3x44-bit limbs (in gprs)
52+
// upper-case variables are 8-element vector numbers in 3x44-bit limbs (in zmm registers)
5353
// [ ] used to denote vector numbers (with their elements)
5454

5555
// Constant Pool:
@@ -84,41 +84,41 @@ static address poly1305_mask44() {
8484
}
8585

8686
// Compute product for 8 16-byte message blocks,
87-
// i.e. For each block, compute [a2 a1 a0] = [a2 a1 a0] × [r2 r1 r0]
87+
// i.e. For each block, compute [a2 a1 a0] = [a2 a1 a0] x [r2 r1 r0]
8888
//
8989
// Each block/number is represented by 3 44-bit limb digits, start with multiplication
9090
//
9191
// a2 a1 a0
92-
// × r2 r1 r0
92+
// x r2 r1 r0
9393
// ----------------------------------
94-
// a2×r0 a1×r0 a0×r0
95-
// + a1×r1 a0×r1 5×a2×r1' (r1' = r1<<2)
96-
// + a0×r2 5×a2×r2' 5×a1×r2' (r2' = r2<<2)
94+
// a2xr0 a1xr0 a0xr0
95+
// + a1xr1 a0xr1 5xa2xr1' (r1' = r1<<2)
96+
// + a0xr2 5xa2xr2' 5xa1xr2' (r2' = r2<<2)
9797
// ----------------------------------
9898
// p2 p1 p0
9999
//
100100
// Then, propagate the carry (bits after bit 44) from lower limbs into higher limbs.
101101
// Then, modular reduction from upper limb wrapped to lower limbs
102102
//
103103
// Math Note 1: 'carry propagation' from p2 to p0 involves multiplication by 5 (i.e. slightly modified modular reduction from above):
104-
// ( p2×2^88 ) mod 2^130-5
105-
// = ( p2'×2^88 + p2''×2^130) mod 2^130-5 // Split on 130-bit boudary
106-
// = ( p2'×2^88 + p2''×2^130 - 5×p2'' + 5×p2'') mod 2^130-5
107-
// = ( p2'×2^88 + p2''×(2^130 - 5) + 5×p2'') mod 2^130-5 // i.e. adding multiples of modulus is a noop
108-
// = ( p2'×2^88 + 5×p2'') mod 2^130-5
104+
// ( p2x2^88 ) mod 2^130-5
105+
// = ( p2'x2^88 + p2''x2^130) mod 2^130-5 // Split on 130-bit boudary
106+
// = ( p2'x2^88 + p2''x2^130 - 5xp2'' + 5xp2'') mod 2^130-5
107+
// = ( p2'x2^88 + p2''x(2^130 - 5) + 5xp2'') mod 2^130-5 // i.e. adding multiples of modulus is a noop
108+
// = ( p2'x2^88 + 5xp2'') mod 2^130-5
109109
//
110110
// Math Note 2: R1P = 4*5*R1 and R2P = 4*5*R2; This precomputation allows simultaneous reduction and multiplication.
111111
// This is not the standard 'multiply-upper-by-5', here is why the factor is 4*5 instead of 5.
112-
// For example, partial product (a2×r2):
113-
// (a2×2^88)×(r2×2^88) mod 2^130-5
114-
// = (a2×r2 × 2^176) mod 2^130-5
115-
// = (a2×r2 × 2^46×2^130) mod 2^130-5
116-
// = (a2×r2×2^46 × 2^130- 5×a2×r2×2^46 + 5×a2×r2×2^46) mod 2^130-5
117-
// = (a2×r2×2^46 × (2^130- 5) + 5×a2×r2×2^46) mod 2^130-5 // i.e. adding multiples of modulus is a noop
118-
// = (5×a2×r2×2^46) mod 2^130-5
119-
// = (a2×5×r2×2^2 × 2^44) mod 2^130-5 // Align to limb boudary
120-
// = (a2×[5×r2×4] × 2^44) mod 2^130-5
121-
// = (a2×R2P × 2^44) mod 2^130-5 // i.e. R2P = 4*5*R2
112+
// For example, partial product (a2xr2):
113+
// (a2x2^88)x(r2x2^88) mod 2^130-5
114+
// = (a2xr2 x 2^176) mod 2^130-5
115+
// = (a2xr2 x 2^46x2^130) mod 2^130-5
116+
// = (a2xr2x2^46 x 2^130- 5xa2xr2x2^46 + 5xa2xr2x2^46) mod 2^130-5
117+
// = (a2xr2x2^46 x (2^130- 5) + 5xa2xr2x2^46) mod 2^130-5 // i.e. adding multiples of modulus is a noop
118+
// = (5xa2xr2x2^46) mod 2^130-5
119+
// = (a2x5xr2x2^2 x 2^44) mod 2^130-5 // Align to limb boudary
120+
// = (a2x[5xr2x4] x 2^44) mod 2^130-5
121+
// = (a2xR2P x 2^44) mod 2^130-5 // i.e. R2P = 4*5*R2
122122
//
123123
void StubGenerator::poly1305_multiply8_avx512(
124124
const XMMRegister A0, const XMMRegister A1, const XMMRegister A2,
@@ -136,29 +136,29 @@ void StubGenerator::poly1305_multiply8_avx512(
136136
__ evpxorq(P2H, P2H, P2H, Assembler::AVX_512bit);
137137

138138
// Calculate partial products
139-
// p0 = a2×r1'
140-
// p1 = a2×r2'
141-
// p2 = a2×r0
139+
// p0 = a2xr1'
140+
// p1 = a2xr2'
141+
// p2 = a2xr0
142142
__ evpmadd52luq(P0L, A2, R1P, Assembler::AVX_512bit);
143143
__ evpmadd52huq(P0H, A2, R1P, Assembler::AVX_512bit);
144144
__ evpmadd52luq(P1L, A2, R2P, Assembler::AVX_512bit);
145145
__ evpmadd52huq(P1H, A2, R2P, Assembler::AVX_512bit);
146146
__ evpmadd52luq(P2L, A2, R0, Assembler::AVX_512bit);
147147
__ evpmadd52huq(P2H, A2, R0, Assembler::AVX_512bit);
148148

149-
// p0 += a0×r0
150-
// p1 += a0×r1
151-
// p2 += a0×r2
149+
// p0 += a0xr0
150+
// p1 += a0xr1
151+
// p2 += a0xr2
152152
__ evpmadd52luq(P1L, A0, R1, Assembler::AVX_512bit);
153153
__ evpmadd52huq(P1H, A0, R1, Assembler::AVX_512bit);
154154
__ evpmadd52luq(P2L, A0, R2, Assembler::AVX_512bit);
155155
__ evpmadd52huq(P2H, A0, R2, Assembler::AVX_512bit);
156156
__ evpmadd52luq(P0L, A0, R0, Assembler::AVX_512bit);
157157
__ evpmadd52huq(P0H, A0, R0, Assembler::AVX_512bit);
158158

159-
// p0 += a1×r2'
160-
// p1 += a1×r0
161-
// p2 += a1×r1
159+
// p0 += a1xr2'
160+
// p1 += a1xr0
161+
// p2 += a1xr1
162162
__ evpmadd52luq(P0L, A1, R2P, Assembler::AVX_512bit);
163163
__ evpmadd52huq(P0H, A1, R2P, Assembler::AVX_512bit);
164164
__ evpmadd52luq(P1L, A1, R0, Assembler::AVX_512bit);
@@ -168,10 +168,10 @@ void StubGenerator::poly1305_multiply8_avx512(
168168

169169
// Carry propagation:
170170
// (Not quite aligned) | More mathematically correct:
171-
// P2L P1L P0L | P2L×2^88 + P1L×2^44 + P0L×2^0
172-
// + P2H P1H P0H | + P2H×2^140 + P1H×2^96 + P0H×2^52
171+
// P2L P1L P0L | P2Lx2^88 + P1Lx2^44 + P0Lx2^0
172+
// + P2H P1H P0H | + P2Hx2^140 + P1Hx2^96 + P0Hx2^52
173173
// --------------------------- | -----------------------------------------------
174-
// = P2H A2 A1 A0 | = P2H×2^130 + A2×2^88 + A1×2^44 + A0×2^0
174+
// = P2H A2 A1 A0 | = P2Hx2^130 + A2x2^88 + A1x2^44 + A0x2^0
175175
//
176176
__ vpsrlq(TMP, P0L, 44, Assembler::AVX_512bit);
177177
__ evpandq(A0, P0L, ExternalAddress(poly1305_mask44()), Assembler::AVX_512bit, rscratch); // Clear top 20 bits
@@ -214,8 +214,8 @@ void StubGenerator::poly1305_multiply8_avx512(
214214
// a2 a1 a0
215215
// x r1 r0
216216
// -----------------------------
217-
// a2×r0 a1×r0 a0×r0
218-
// + a0×r1
217+
// a2xr0 a1xr0 a0xr0
218+
// + a0xr1
219219
// + 5xa2xr1 5xa1xr1
220220
// -----------------------------
221221
// [0|L2L] [L1H|L1L] [L0H|L0L]
@@ -347,7 +347,7 @@ void StubGenerator::poly1305_limbs_avx512(
347347
}
348348

349349
/**
350-
* Copy 5×26-bit (unreduced) limbs stored at Register limbs into a2:a1:a0 (3×64-bit limbs)
350+
* Copy 5x26-bit (unreduced) limbs stored at Register limbs into a2:a1:a0 (3x64-bit limbs)
351351
*
352352
* a2 is optional. When only128 is set, limbs are expected to fit into 128-bits (i.e. a1:a0 such as clamped R)
353353
*/
@@ -393,7 +393,7 @@ void StubGenerator::poly1305_limbs(
393393
}
394394

395395
/**
396-
* Break 3×64-bit a2:a1:a0 limbs into 5×26-bit limbs and store out into 5 quadwords at address `limbs`
396+
* Break 3x64-bit a2:a1:a0 limbs into 5x26-bit limbs and store out into 5 quadwords at address `limbs`
397397
*/
398398
void StubGenerator::poly1305_limbs_out(
399399
const Register a0, const Register a1, const Register a2,
@@ -474,9 +474,9 @@ void StubGenerator::poly1305_limbs_out(
474474
//
475475
// Pseudocode:
476476
// * used for poly1305_multiply_scalar
477-
// × used for poly1305_multiply8_avx512
478-
// lower-case variables are scalar numbers in 3×44-bit limbs (in gprs)
479-
// upper-case variables are 8&16-element vector numbers in 3×44-bit limbs (in zmm registers)
477+
// x used for poly1305_multiply8_avx512
478+
// lower-case variables are scalar numbers in 3x44-bit limbs (in gprs)
479+
// upper-case variables are 8&16-element vector numbers in 3x44-bit limbs (in zmm registers)
480480
//
481481
// CL = a // [0 0 0 0 0 0 0 a]
482482
// AL = poly1305_limbs_avx512(input)
@@ -496,24 +496,24 @@ void StubGenerator::poly1305_limbs_out(
496496
// B = limbs(T) // [r^4 0 r^3 0 r^2 0 r^1 0 ]
497497
// CL = B >> 1 // [ 0 r^4 0 r^3 0 r^2 0 r^1]
498498
// R = r^4 || r^4 || .. // [r^4 r^4 r^4 r^4 r^4 r^4 r^4 r^4]
499-
// B = B×R // [r^8 0 r^7 0 r^6 0 r^5 0 ]
499+
// B = BxR // [r^8 0 r^7 0 r^6 0 r^5 0 ]
500500
// B = B | CL // [r^8 r^4 r^7 r^3 r^6 r^2 r^5 r^1]
501501
// CL = B
502502
// R = r^8 || r^8 || .. // [r^8 r^8 r^8 r^8 r^8 r^8 r^8 r^8]
503-
// B = B × R // [r^16 r^12 r^15 r^11 r^14 r^10 r^13 r^9]
503+
// B = B x R // [r^16 r^12 r^15 r^11 r^14 r^10 r^13 r^9]
504504
// CH = B
505505
// R = r^16 || r^16 || .. // [r^16 r^16 r^16 r^16 r^16 r^16 r^16 r^16]
506506
//
507507
// for (;length>=16; input+=16, length-=16)
508508
// BL = poly1305_limbs_avx512(input)
509509
// BH = poly1305_limbs_avx512(input+8)
510-
// AL = AL × R
511-
// AH = AH × R
510+
// AL = AL x R
511+
// AH = AH x R
512512
// AL = AL + BL
513513
// AH = AH + BH
514514
//
515-
// AL = AL × CL
516-
// AH = AH × CH
515+
// AL = AL x CL
516+
// AH = AH x CH
517517
// A = AL + AH // 16->8 blocks
518518
// T = A >> 4 // 8 ->4 blocks
519519
// A = A + T

‎src/hotspot/share/utilities/elfFile.hpp‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -797,7 +797,7 @@ class DwarfFile : public ElfFile {
797797
uint32_t _line;
798798

799799
// A column number within a source line. Columns are numbered beginning at 1. The value 0 is reserved to indicate
800-
// that a statement begins at the “left edge” of the line.
800+
// that a statement begins at the "left edge" of the line.
801801
uint32_t _column;
802802

803803
// Indicates that the current instruction is a recommended breakpoint location.

0 commit comments

Comments
 (0)