Text file src/crypto/sha1/sha1block_arm.s

     1  // Copyright 2014 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  //
     5  // ARM version of md5block.go
     6  
     7  #include "textflag.h"
     8  
     9  // SHA-1 block routine. See sha1block.go for Go equivalent.
    10  //
    11  // There are 80 rounds of 4 types:
    12  //   - rounds 0-15 are type 1 and load data (ROUND1 macro).
    13  //   - rounds 16-19 are type 1 and do not load data (ROUND1x macro).
    14  //   - rounds 20-39 are type 2 and do not load data (ROUND2 macro).
    15  //   - rounds 40-59 are type 3 and do not load data (ROUND3 macro).
    16  //   - rounds 60-79 are type 4 and do not load data (ROUND4 macro).
    17  //
    18  // Each round loads or shuffles the data, then computes a per-round
    19  // function of b, c, d, and then mixes the result into and rotates the
    20  // five registers a, b, c, d, e holding the intermediate results.
    21  //
    22  // The register rotation is implemented by rotating the arguments to
    23  // the round macros instead of by explicit move instructions.
    24  
    25  // Register definitions
    26  #define Rdata	R0	// Pointer to incoming data
    27  #define Rconst	R1	// Current constant for SHA round
    28  #define Ra	R2		// SHA-1 accumulator
    29  #define Rb	R3		// SHA-1 accumulator
    30  #define Rc	R4		// SHA-1 accumulator
    31  #define Rd	R5		// SHA-1 accumulator
    32  #define Re	R6		// SHA-1 accumulator
    33  #define Rt0	R7		// Temporary
    34  #define Rt1	R8		// Temporary
    35  // r9, r10 are forbidden
    36  // r11 is OK provided you check the assembler that no synthetic instructions use it
    37  #define Rt2	R11		// Temporary
    38  #define Rctr	R12	// loop counter
    39  #define Rw	R14		// point to w buffer
    40  
    41  // func block(dig *digest, p []byte)
    42  // 0(FP) is *digest
    43  // 4(FP) is p.array (struct Slice)
    44  // 8(FP) is p.len
    45  //12(FP) is p.cap
    46  //
    47  // Stack frame
    48  #define p_end	end-4(SP)		// pointer to the end of data
    49  #define p_data	data-8(SP)	// current data pointer (unused?)
    50  #define w_buf	buf-(8+4*80)(SP)	//80 words temporary buffer w uint32[80]
    51  #define saved	abcde-(8+4*80+4*5)(SP)	// saved sha1 registers a,b,c,d,e - these must be last (unused?)
    52  // Total size +4 for saved LR is 352
    53  
    54  	// w[i] = p[j]<<24 | p[j+1]<<16 | p[j+2]<<8 | p[j+3]
    55  	// e += w[i]
    56  #define LOAD(Re) \
    57  	MOVBU	2(Rdata), Rt0 ; \
    58  	MOVBU	3(Rdata), Rt1 ; \
    59  	MOVBU	1(Rdata), Rt2 ; \
    60  	ORR	Rt0<<8, Rt1, Rt0	    ; \
    61  	MOVBU.P	4(Rdata), Rt1 ; \
    62  	ORR	Rt2<<16, Rt0, Rt0	    ; \
    63  	ORR	Rt1<<24, Rt0, Rt0	    ; \
    64  	MOVW.P	Rt0, 4(Rw)		    ; \
    65  	ADD	Rt0, Re, Re
    66  
    67  	// tmp := w[(i-3)&0xf] ^ w[(i-8)&0xf] ^ w[(i-14)&0xf] ^ w[(i)&0xf]
    68  	// w[i&0xf] = tmp<<1 | tmp>>(32-1)
    69  	// e += w[i&0xf]
    70  #define SHUFFLE(Re) \
    71  	MOVW	(-16*4)(Rw), Rt0 ; \
    72  	MOVW	(-14*4)(Rw), Rt1 ; \
    73  	MOVW	(-8*4)(Rw), Rt2  ; \
    74  	EOR	Rt0, Rt1, Rt0  ; \
    75  	MOVW	(-3*4)(Rw), Rt1  ; \
    76  	EOR	Rt2, Rt0, Rt0  ; \
    77  	EOR	Rt0, Rt1, Rt0  ; \
    78  	MOVW	Rt0@>(32-1), Rt0  ; \
    79  	MOVW.P	Rt0, 4(Rw)	  ; \
    80  	ADD	Rt0, Re, Re
    81  
    82  	// t1 = (b & c) | ((~b) & d)
    83  #define FUNC1(Ra, Rb, Rc, Rd, Re) \
    84  	MVN	Rb, Rt1	   ; \
    85  	AND	Rb, Rc, Rt0  ; \
    86  	AND	Rd, Rt1, Rt1 ; \
    87  	ORR	Rt0, Rt1, Rt1
    88  
    89  	// t1 = b ^ c ^ d
    90  #define FUNC2(Ra, Rb, Rc, Rd, Re) \
    91  	EOR	Rb, Rc, Rt1 ; \
    92  	EOR	Rd, Rt1, Rt1
    93  
    94  	// t1 = (b & c) | (b & d) | (c & d) =
    95  	// t1 = (b & c) | ((b | c) & d)
    96  #define FUNC3(Ra, Rb, Rc, Rd, Re) \
    97  	ORR	Rb, Rc, Rt0  ; \
    98  	AND	Rb, Rc, Rt1  ; \
    99  	AND	Rd, Rt0, Rt0 ; \
   100  	ORR	Rt0, Rt1, Rt1
   101  
   102  #define FUNC4 FUNC2
   103  
   104  	// a5 := a<<5 | a>>(32-5)
   105  	// b = b<<30 | b>>(32-30)
   106  	// e = a5 + t1 + e + const
   107  #define MIX(Ra, Rb, Rc, Rd, Re) \
   108  	ADD	Rt1, Re, Re	 ; \
   109  	MOVW	Rb@>(32-30), Rb	 ; \
   110  	ADD	Ra@>(32-5), Re, Re ; \
   111  	ADD	Rconst, Re, Re
   112  
   113  #define ROUND1(Ra, Rb, Rc, Rd, Re) \
   114  	LOAD(Re)		; \
   115  	FUNC1(Ra, Rb, Rc, Rd, Re)	; \
   116  	MIX(Ra, Rb, Rc, Rd, Re)
   117  
   118  #define ROUND1x(Ra, Rb, Rc, Rd, Re) \
   119  	SHUFFLE(Re)	; \
   120  	FUNC1(Ra, Rb, Rc, Rd, Re)	; \
   121  	MIX(Ra, Rb, Rc, Rd, Re)
   122  
   123  #define ROUND2(Ra, Rb, Rc, Rd, Re) \
   124  	SHUFFLE(Re)	; \
   125  	FUNC2(Ra, Rb, Rc, Rd, Re)	; \
   126  	MIX(Ra, Rb, Rc, Rd, Re)
   127  
   128  #define ROUND3(Ra, Rb, Rc, Rd, Re) \
   129  	SHUFFLE(Re)	; \
   130  	FUNC3(Ra, Rb, Rc, Rd, Re)	; \
   131  	MIX(Ra, Rb, Rc, Rd, Re)
   132  
   133  #define ROUND4(Ra, Rb, Rc, Rd, Re) \
   134  	SHUFFLE(Re)	; \
   135  	FUNC4(Ra, Rb, Rc, Rd, Re)	; \
   136  	MIX(Ra, Rb, Rc, Rd, Re)
   137  
   138  
   139  // func block(dig *digest, p []byte)
   140  TEXT	·block(SB), 0, $352-16
   141  	MOVW	p+4(FP), Rdata	// pointer to the data
   142  	MOVW	p_len+8(FP), Rt0	// number of bytes
   143  	ADD	Rdata, Rt0
   144  	MOVW	Rt0, p_end	// pointer to end of data
   145  
   146  	// Load up initial SHA-1 accumulator
   147  	MOVW	dig+0(FP), Rt0
   148  	MOVM.IA (Rt0), [Ra,Rb,Rc,Rd,Re]
   149  
   150  loop:
   151  	// Save registers at SP+4 onwards
   152  	MOVM.IB [Ra,Rb,Rc,Rd,Re], (R13)
   153  
   154  	MOVW	$w_buf, Rw
   155  	MOVW	$0x5A827999, Rconst
   156  	MOVW	$3, Rctr
   157  loop1:	ROUND1(Ra, Rb, Rc, Rd, Re)
   158  	ROUND1(Re, Ra, Rb, Rc, Rd)
   159  	ROUND1(Rd, Re, Ra, Rb, Rc)
   160  	ROUND1(Rc, Rd, Re, Ra, Rb)
   161  	ROUND1(Rb, Rc, Rd, Re, Ra)
   162  	SUB.S	$1, Rctr
   163  	BNE	loop1
   164  
   165  	ROUND1(Ra, Rb, Rc, Rd, Re)
   166  	ROUND1x(Re, Ra, Rb, Rc, Rd)
   167  	ROUND1x(Rd, Re, Ra, Rb, Rc)
   168  	ROUND1x(Rc, Rd, Re, Ra, Rb)
   169  	ROUND1x(Rb, Rc, Rd, Re, Ra)
   170  
   171  	MOVW	$0x6ED9EBA1, Rconst
   172  	MOVW	$4, Rctr
   173  loop2:	ROUND2(Ra, Rb, Rc, Rd, Re)
   174  	ROUND2(Re, Ra, Rb, Rc, Rd)
   175  	ROUND2(Rd, Re, Ra, Rb, Rc)
   176  	ROUND2(Rc, Rd, Re, Ra, Rb)
   177  	ROUND2(Rb, Rc, Rd, Re, Ra)
   178  	SUB.S	$1, Rctr
   179  	BNE	loop2
   180  
   181  	MOVW	$0x8F1BBCDC, Rconst
   182  	MOVW	$4, Rctr
   183  loop3:	ROUND3(Ra, Rb, Rc, Rd, Re)
   184  	ROUND3(Re, Ra, Rb, Rc, Rd)
   185  	ROUND3(Rd, Re, Ra, Rb, Rc)
   186  	ROUND3(Rc, Rd, Re, Ra, Rb)
   187  	ROUND3(Rb, Rc, Rd, Re, Ra)
   188  	SUB.S	$1, Rctr
   189  	BNE	loop3
   190  
   191  	MOVW	$0xCA62C1D6, Rconst
   192  	MOVW	$4, Rctr
   193  loop4:	ROUND4(Ra, Rb, Rc, Rd, Re)
   194  	ROUND4(Re, Ra, Rb, Rc, Rd)
   195  	ROUND4(Rd, Re, Ra, Rb, Rc)
   196  	ROUND4(Rc, Rd, Re, Ra, Rb)
   197  	ROUND4(Rb, Rc, Rd, Re, Ra)
   198  	SUB.S	$1, Rctr
   199  	BNE	loop4
   200  
   201  	// Accumulate - restoring registers from SP+4
   202  	MOVM.IB (R13), [Rt0,Rt1,Rt2,Rctr,Rw]
   203  	ADD	Rt0, Ra
   204  	ADD	Rt1, Rb
   205  	ADD	Rt2, Rc
   206  	ADD	Rctr, Rd
   207  	ADD	Rw, Re
   208  
   209  	MOVW	p_end, Rt0
   210  	CMP	Rt0, Rdata
   211  	BLO	loop
   212  
   213  	// Save final SHA-1 accumulator
   214  	MOVW	dig+0(FP), Rt0
   215  	MOVM.IA [Ra,Rb,Rc,Rd,Re], (Rt0)
   216  
   217  	RET
   218  

View as plain text