mono/linux-packaging-mono
0
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Imported Upstream version 4.8.0.309

Browse Source 
Former-commit-id: 5f9c6ae75f295e057a7d2971f3a6df4656fa8850
This commit is contained in:
Xamarin Public Jenkins (auto-signing)
2016-11-10 13:04:39 +00:00
parent ee1447783b
commit 94b2861243
4912 changed files with 390737 additions and 49310 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
external/boringssl/crypto/rc4/CMakeLists.txt vendored Normal file
View File

@@ -0,0 +1,29 @@
include_directories(../../include)
if (${ARCH} STREQUAL "x86_64")
set(
RC4_ARCH_SOURCES
rc4-x86_64.${ASM_EXT}
)
endif()
if (${ARCH} STREQUAL "x86")
set(
RC4_ARCH_SOURCES
rc4-586.${ASM_EXT}
)
endif()
add_library(
rc4
OBJECT
rc4.c
${RC4_ARCH_SOURCES}
)
perlasm(rc4-x86_64.${ASM_EXT} asm/rc4-x86_64.pl)
perlasm(rc4-586.${ASM_EXT} asm/rc4-586.pl)

388
external/boringssl/crypto/rc4/asm/rc4-586.pl vendored Normal file
View File

@@ -0,0 +1,388 @@
#!/usr/bin/env perl
# ====================================================================
# [Re]written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
# project. The module is, however, dual licensed under OpenSSL and
# CRYPTOGAMS licenses depending on where you obtain it. For further
# details see http://www.openssl.org/~appro/cryptogams/.
# ====================================================================
# At some point it became apparent that the original SSLeay RC4
# assembler implementation performs suboptimally on latest IA-32
# microarchitectures. After re-tuning performance has changed as
# following:
#
# Pentium -10%
# Pentium III +12%
# AMD +50%(*)
# P4 +250%(**)
#
# (*) This number is actually a trade-off:-) It's possible to
# achieve +72%, but at the cost of -48% off PIII performance.
# In other words code performing further 13% faster on AMD
# would perform almost 2 times slower on Intel PIII...
# For reference! This code delivers ~80% of rc4-amd64.pl
# performance on the same Opteron machine.
# (**) This number requires compressed key schedule set up by
# RC4_set_key [see commentary below for further details].
#
# <appro@fy.chalmers.se>
# May 2011
#
# Optimize for Core2 and Westmere [and incidentally Opteron]. Current
# performance in cycles per processed byte (less is better) and
# improvement relative to previous version of this module is:
#
# Pentium 10.2 # original numbers
# Pentium III 7.8(*)
# Intel P4 7.5
#
# Opteron 6.1/+20% # new MMX numbers
# Core2 5.3/+67%(**)
# Westmere 5.1/+94%(**)
# Sandy Bridge 5.0/+8%
# Atom 12.6/+6%
#
# (*) PIII can actually deliver 6.6 cycles per byte with MMX code,
# but this specific code performs poorly on Core2. And vice
# versa, below MMX/SSE code delivering 5.8/7.1 on Core2 performs
# poorly on PIII, at 8.0/14.5:-( As PIII is not a "hot" CPU
# [anymore], I chose to discard PIII-specific code path and opt
# for original IALU-only code, which is why MMX/SSE code path
# is guarded by SSE2 bit (see below), not MMX/SSE.
# (**) Performance vs. block size on Core2 and Westmere had a maximum
# at ... 64 bytes block size. And it was quite a maximum, 40-60%
# in comparison to largest 8KB block size. Above improvement
# coefficients are for the largest block size.
$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
push(@INC,"${dir}","${dir}../../perlasm");
require "x86asm.pl";
&asm_init($ARGV[0],"rc4-586.pl",$x86only = $ARGV[$#ARGV] eq "386");
$xx="eax";
$yy="ebx";
$tx="ecx";
$ty="edx";
$inp="esi";
$out="ebp";
$dat="edi";
sub RC4_loop {
my $i=shift;
my $func = ($i==0)?*mov:*or;
&add (&LB($yy),&LB($tx));
&mov ($ty,&DWP(0,$dat,$yy,4));
&mov (&DWP(0,$dat,$yy,4),$tx);
&mov (&DWP(0,$dat,$xx,4),$ty);
&add ($ty,$tx);
&inc (&LB($xx));
&and ($ty,0xff);
&ror ($out,8) if ($i!=0);
if ($i<3) {
&mov ($tx,&DWP(0,$dat,$xx,4));
} else {
&mov ($tx,&wparam(3)); # reload [re-biased] out
}
&$func ($out,&DWP(0,$dat,$ty,4));
}
if ($alt=0) {
# >20% faster on Atom and Sandy Bridge[!], 8% faster on Opteron,
# but ~40% slower on Core2 and Westmere... Attempt to add movz
# brings down Opteron by 25%, Atom and Sandy Bridge by 15%, yet
# on Core2 with movz it's almost 20% slower than below alternative
# code... Yes, it's a total mess...
my @XX=($xx,$out);
$RC4_loop_mmx = sub { # SSE actually...
my $i=shift;
my $j=$i<=0?0:$i>>1;
my $mm=$i<=0?"mm0":"mm".($i&1);
&add (&LB($yy),&LB($tx));
&lea (@XX[1],&DWP(1,@XX[0]));
&pxor ("mm2","mm0") if ($i==0);
&psllq ("mm1",8) if ($i==0);
&and (@XX[1],0xff);
&pxor ("mm0","mm0") if ($i<=0);
&mov ($ty,&DWP(0,$dat,$yy,4));
&mov (&DWP(0,$dat,$yy,4),$tx);
&pxor ("mm1","mm2") if ($i==0);
&mov (&DWP(0,$dat,$XX[0],4),$ty);
&add (&LB($ty),&LB($tx));
&movd (@XX[0],"mm7") if ($i==0);
&mov ($tx,&DWP(0,$dat,@XX[1],4));
&pxor ("mm1","mm1") if ($i==1);
&movq ("mm2",&QWP(0,$inp)) if ($i==1);
&movq (&QWP(-8,(@XX[0],$inp)),"mm1") if ($i==0);
&pinsrw ($mm,&DWP(0,$dat,$ty,4),$j);
push (@XX,shift(@XX)) if ($i>=0);
}
} else {
# Using pinsrw here improves performane on Intel CPUs by 2-3%, but
# brings down AMD by 7%...
$RC4_loop_mmx = sub {
my $i=shift;
&add (&LB($yy),&LB($tx));
&psllq ("mm1",8*(($i-1)&7)) if (abs($i)!=1);
&mov ($ty,&DWP(0,$dat,$yy,4));
&mov (&DWP(0,$dat,$yy,4),$tx);
&mov (&DWP(0,$dat,$xx,4),$ty);
&inc ($xx);
&add ($ty,$tx);
&movz ($xx,&LB($xx)); # (*)
&movz ($ty,&LB($ty)); # (*)
&pxor ("mm2",$i==1?"mm0":"mm1") if ($i>=0);
&movq ("mm0",&QWP(0,$inp)) if ($i<=0);
&movq (&QWP(-8,($out,$inp)),"mm2") if ($i==0);
&mov ($tx,&DWP(0,$dat,$xx,4));
&movd ($i>0?"mm1":"mm2",&DWP(0,$dat,$ty,4));
# (*) This is the key to Core2 and Westmere performance.
# Whithout movz out-of-order execution logic confuses
# itself and fails to reorder loads and stores. Problem
# appears to be fixed in Sandy Bridge...
}
}
&external_label("OPENSSL_ia32cap_P");
# void asm_RC4(RC4_KEY *key,size_t len,const unsigned char *inp,unsigned char *out);
&function_begin("asm_RC4");
&mov ($dat,&wparam(0)); # load key schedule pointer
&mov ($ty, &wparam(1)); # load len
&mov ($inp,&wparam(2)); # load inp
&mov ($out,&wparam(3)); # load out
&xor ($xx,$xx); # avoid partial register stalls
&xor ($yy,$yy);
&cmp ($ty,0); # safety net
&je (&label("abort"));
&mov (&LB($xx),&BP(0,$dat)); # load key->x
&mov (&LB($yy),&BP(4,$dat)); # load key->y
&add ($dat,8);
&lea ($tx,&DWP(0,$inp,$ty));
&sub ($out,$inp); # re-bias out
&mov (&wparam(1),$tx); # save input+len
&inc (&LB($xx));
# detect compressed key schedule...
&cmp (&DWP(256,$dat),-1);
&je (&label("RC4_CHAR"));
&mov ($tx,&DWP(0,$dat,$xx,4));
&and ($ty,-4); # how many 4-byte chunks?
&jz (&label("loop1"));
&mov (&wparam(3),$out); # $out as accumulator in these loops
if ($x86only) {
&jmp (&label("go4loop4"));
} else {
&test ($ty,-8);
&jz (&label("go4loop4"));
&picmeup($out,"OPENSSL_ia32cap_P");
&bt (&DWP(0,$out),26); # check SSE2 bit [could have been MMX]
&jnc (&label("go4loop4"));
&mov ($out,&wparam(3)) if (!$alt);
&movd ("mm7",&wparam(3)) if ($alt);
&and ($ty,-8);
&lea ($ty,&DWP(-8,$inp,$ty));
&mov (&DWP(-4,$dat),$ty); # save input+(len/8)*8-8
&$RC4_loop_mmx(-1);
&jmp(&label("loop_mmx_enter"));
&set_label("loop_mmx",16);
&$RC4_loop_mmx(0);
&set_label("loop_mmx_enter");
for ($i=1;$i<8;$i++) { &$RC4_loop_mmx($i); }
&mov ($ty,$yy);
&xor ($yy,$yy); # this is second key to Core2
&mov (&LB($yy),&LB($ty)); # and Westmere performance...
&cmp ($inp,&DWP(-4,$dat));
&lea ($inp,&DWP(8,$inp));
&jb (&label("loop_mmx"));
if ($alt) {
&movd ($out,"mm7");
&pxor ("mm2","mm0");
&psllq ("mm1",8);
&pxor ("mm1","mm2");
&movq (&QWP(-8,$out,$inp),"mm1");
} else {
&psllq ("mm1",56);
&pxor ("mm2","mm1");
&movq (&QWP(-8,$out,$inp),"mm2");
}
&emms ();
&cmp ($inp,&wparam(1)); # compare to input+len
&je (&label("done"));
&jmp (&label("loop1"));
}
&set_label("go4loop4",16);
&lea ($ty,&DWP(-4,$inp,$ty));
&mov (&wparam(2),$ty); # save input+(len/4)*4-4
&set_label("loop4");
for ($i=0;$i<4;$i++) { RC4_loop($i); }
&ror ($out,8);
&xor ($out,&DWP(0,$inp));
&cmp ($inp,&wparam(2)); # compare to input+(len/4)*4-4
&mov (&DWP(0,$tx,$inp),$out);# $tx holds re-biased out here
&lea ($inp,&DWP(4,$inp));
&mov ($tx,&DWP(0,$dat,$xx,4));
&jb (&label("loop4"));
&cmp ($inp,&wparam(1)); # compare to input+len
&je (&label("done"));
&mov ($out,&wparam(3)); # restore $out
&set_label("loop1",16);
&add (&LB($yy),&LB($tx));
&mov ($ty,&DWP(0,$dat,$yy,4));
&mov (&DWP(0,$dat,$yy,4),$tx);
&mov (&DWP(0,$dat,$xx,4),$ty);
&add ($ty,$tx);
&inc (&LB($xx));
&and ($ty,0xff);
&mov ($ty,&DWP(0,$dat,$ty,4));
&xor (&LB($ty),&BP(0,$inp));
&lea ($inp,&DWP(1,$inp));
&mov ($tx,&DWP(0,$dat,$xx,4));
&cmp ($inp,&wparam(1)); # compare to input+len
&mov (&BP(-1,$out,$inp),&LB($ty));
&jb (&label("loop1"));
&jmp (&label("done"));
# this is essentially Intel P4 specific codepath...
&set_label("RC4_CHAR",16);
&movz ($tx,&BP(0,$dat,$xx));
# strangely enough unrolled loop performs over 20% slower...
&set_label("cloop1");
&add (&LB($yy),&LB($tx));
&movz ($ty,&BP(0,$dat,$yy));
&mov (&BP(0,$dat,$yy),&LB($tx));
&mov (&BP(0,$dat,$xx),&LB($ty));
&add (&LB($ty),&LB($tx));
&movz ($ty,&BP(0,$dat,$ty));
&add (&LB($xx),1);
&xor (&LB($ty),&BP(0,$inp));
&lea ($inp,&DWP(1,$inp));
&movz ($tx,&BP(0,$dat,$xx));
&cmp ($inp,&wparam(1));
&mov (&BP(-1,$out,$inp),&LB($ty));
&jb (&label("cloop1"));
&set_label("done");
&dec (&LB($xx));
&mov (&DWP(-4,$dat),$yy); # save key->y
&mov (&BP(-8,$dat),&LB($xx)); # save key->x
&set_label("abort");
&function_end("asm_RC4");
########################################################################
$inp="esi";
$out="edi";
$idi="ebp";
$ido="ecx";
$idx="edx";
# void asm_RC4_set_key(RC4_KEY *key,int len,const unsigned char *data);
&function_begin("asm_RC4_set_key");
&mov ($out,&wparam(0)); # load key
&mov ($idi,&wparam(1)); # load len
&mov ($inp,&wparam(2)); # load data
&picmeup($idx,"OPENSSL_ia32cap_P");
&lea ($out,&DWP(2*4,$out)); # &key->data
&lea ($inp,&DWP(0,$inp,$idi)); # $inp to point at the end
&neg ($idi);
&xor ("eax","eax");
&mov (&DWP(-4,$out),$idi); # borrow key->y
&bt (&DWP(0,$idx),20); # check for bit#20
&jc (&label("c1stloop"));
&set_label("w1stloop",16);
&mov (&DWP(0,$out,"eax",4),"eax"); # key->data[i]=i;
&add (&LB("eax"),1); # i++;
&jnc (&label("w1stloop"));
&xor ($ido,$ido);
&xor ($idx,$idx);
&set_label("w2ndloop",16);
&mov ("eax",&DWP(0,$out,$ido,4));
&add (&LB($idx),&BP(0,$inp,$idi));
&add (&LB($idx),&LB("eax"));
&add ($idi,1);
&mov ("ebx",&DWP(0,$out,$idx,4));
&jnz (&label("wnowrap"));
&mov ($idi,&DWP(-4,$out));
&set_label("wnowrap");
&mov (&DWP(0,$out,$idx,4),"eax");
&mov (&DWP(0,$out,$ido,4),"ebx");
&add (&LB($ido),1);
&jnc (&label("w2ndloop"));
&jmp (&label("exit"));
# Unlike all other x86 [and x86_64] implementations, Intel P4 core
# [including EM64T] was found to perform poorly with above "32-bit" key
# schedule, a.k.a. RC4_INT. Performance improvement for IA-32 hand-coded
# assembler turned out to be 3.5x if re-coded for compressed 8-bit one,
# a.k.a. RC4_CHAR! It's however inappropriate to just switch to 8-bit
# schedule for x86[_64], because non-P4 implementations suffer from
# significant performance losses then, e.g. PIII exhibits >2x
# deterioration, and so does Opteron. In order to assure optimal
# all-round performance, we detect P4 at run-time and set up compressed
# key schedule, which is recognized by RC4 procedure.
&set_label("c1stloop",16);
&mov (&BP(0,$out,"eax"),&LB("eax")); # key->data[i]=i;
&add (&LB("eax"),1); # i++;
&jnc (&label("c1stloop"));
&xor ($ido,$ido);
&xor ($idx,$idx);
&xor ("ebx","ebx");
&set_label("c2ndloop",16);
&mov (&LB("eax"),&BP(0,$out,$ido));
&add (&LB($idx),&BP(0,$inp,$idi));
&add (&LB($idx),&LB("eax"));
&add ($idi,1);
&mov (&LB("ebx"),&BP(0,$out,$idx));
&jnz (&label("cnowrap"));
&mov ($idi,&DWP(-4,$out));
&set_label("cnowrap");
&mov (&BP(0,$out,$idx),&LB("eax"));
&mov (&BP(0,$out,$ido),&LB("ebx"));
&add (&LB($ido),1);
&jnc (&label("c2ndloop"));
&mov (&DWP(256,$out),-1); # mark schedule as compressed
&set_label("exit");
&xor ("eax","eax");
&mov (&DWP(-8,$out),"eax"); # key->x=0;
&mov (&DWP(-4,$out),"eax"); # key->y=0;
&function_end("asm_RC4_set_key");
&asm_finish();

653
external/boringssl/crypto/rc4/asm/rc4-x86_64.pl vendored Normal file
View File

File diff suppressed because it is too large Load Diff

283
external/boringssl/crypto/rc4/rc4.c vendored Normal file
View File

@@ -0,0 +1,283 @@
/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
* All rights reserved.
*
* This package is an SSL implementation written
* by Eric Young (eay@cryptsoft.com).
* The implementation was written so as to conform with Netscapes SSL.
*
* This library is free for commercial and non-commercial use as long as
* the following conditions are aheared to. The following conditions
* apply to all code found in this distribution, be it the RC4, RSA,
* lhash, DES, etc., code; not just the SSL code. The SSL documentation
* included with this distribution is covered by the same copyright terms
* except that the holder is Tim Hudson (tjh@cryptsoft.com).
*
* Copyright remains Eric Young's, and as such any Copyright notices in
* the code are not to be removed.
* If this package is used in a product, Eric Young should be given attribution
* as the author of the parts of the library used.
* This can be in the form of a textual message at program startup or
* in documentation (online or textual) provided with the package.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* "This product includes cryptographic software written by
* Eric Young (eay@cryptsoft.com)"
* The word 'cryptographic' can be left out if the rouines from the library
* being used are not cryptographic related :-).
* 4. If you include any Windows specific code (or a derivative thereof) from
* the apps directory (application code) you must include an acknowledgement:
* "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
*
* THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* The licence and distribution terms for any publically available version or
* derivative of this code cannot be changed. i.e. this code cannot simply be
* copied and put under another distribution licence
* [including the GNU Public Licence.] */
#include <openssl/rc4.h>
#if defined(OPENSSL_NO_ASM) || \
(!defined(OPENSSL_X86_64) && !defined(OPENSSL_X86))
#if defined(OPENSSL_64_BIT)
#define RC4_CHUNK uint64_t
#elif defined(OPENSSL_32_BIT)
#define RC4_CHUNK uint32_t
#else
#error "Unknown word size"
#endif
/* RC4 as implemented from a posting from
* Newsgroups: sci.crypt
* From: sterndark@netcom.com (David Sterndark)
* Subject: RC4 Algorithm revealed.
* Message-ID: <sternCvKL4B.Hyy@netcom.com>
* Date: Wed, 14 Sep 1994 06:35:31 GMT */
void RC4(RC4_KEY *key, size_t len, const uint8_t *in, uint8_t *out) {
uint32_t *d;
uint32_t x, y, tx, ty;
size_t i;
x = key->x;
y = key->y;
d = key->data;
#define RC4_STEP \
(x = (x + 1) & 0xff, tx = d[x], y = (tx + y) & 0xff, ty = d[y], d[y] = tx, \
d[x] = ty, (RC4_CHUNK)d[(tx + ty) & 0xff])
if ((((size_t)in & (sizeof(RC4_CHUNK) - 1)) |
((size_t)out & (sizeof(RC4_CHUNK) - 1))) == 0) {
RC4_CHUNK ichunk, otp;
const union {
long one;
char little;
} is_endian = {1};
/* I reckon we can afford to implement both endian
* cases and to decide which way to take at run-time
* because the machine code appears to be very compact
* and redundant 1-2KB is perfectly tolerable (i.e.
* in case the compiler fails to eliminate it:-). By
* suggestion from Terrel Larson <terr@terralogic.net>
* who also stands for the is_endian union:-)
*
* Special notes.
*
* - is_endian is declared automatic as doing otherwise
* (declaring static) prevents gcc from eliminating
* the redundant code;
* - compilers (those I've tried) don't seem to have
* problems eliminating either the operators guarded
* by "if (sizeof(RC4_CHUNK)==8)" or the condition
* expressions themselves so I've got 'em to replace
* corresponding #ifdefs from the previous version;
* - I chose to let the redundant switch cases when
* sizeof(RC4_CHUNK)!=8 be (were also #ifdefed
* before);
* - in case you wonder "&(sizeof(RC4_CHUNK)*8-1)" in
* [LB]ESHFT guards against "shift is out of range"
* warnings when sizeof(RC4_CHUNK)!=8
*
* <appro@fy.chalmers.se> */
if (!is_endian.little) { /* BIG-ENDIAN CASE */
#define BESHFT(c) \
(((sizeof(RC4_CHUNK) - (c) - 1) * 8) & (sizeof(RC4_CHUNK) * 8 - 1))
for (; len & (0 - sizeof(RC4_CHUNK)); len -= sizeof(RC4_CHUNK)) {
ichunk = *(RC4_CHUNK *)in;
otp = RC4_STEP << BESHFT(0);
otp |= RC4_STEP << BESHFT(1);
otp |= RC4_STEP << BESHFT(2);
otp |= RC4_STEP << BESHFT(3);
#if defined(OPENSSL_64_BIT)
otp |= RC4_STEP << BESHFT(4);
otp |= RC4_STEP << BESHFT(5);
otp |= RC4_STEP << BESHFT(6);
otp |= RC4_STEP << BESHFT(7);
#endif
*(RC4_CHUNK *)out = otp ^ ichunk;
in += sizeof(RC4_CHUNK);
out += sizeof(RC4_CHUNK);
}
} else { /* LITTLE-ENDIAN CASE */
#define LESHFT(c) (((c) * 8) & (sizeof(RC4_CHUNK) * 8 - 1))
for (; len & (0 - sizeof(RC4_CHUNK)); len -= sizeof(RC4_CHUNK)) {
ichunk = *(RC4_CHUNK *)in;
otp = RC4_STEP;
otp |= RC4_STEP << 8;
otp |= RC4_STEP << 16;
otp |= RC4_STEP << 24;
#if defined(OPENSSL_64_BIT)
otp |= RC4_STEP << LESHFT(4);
otp |= RC4_STEP << LESHFT(5);
otp |= RC4_STEP << LESHFT(6);
otp |= RC4_STEP << LESHFT(7);
#endif
*(RC4_CHUNK *)out = otp ^ ichunk;
in += sizeof(RC4_CHUNK);
out += sizeof(RC4_CHUNK);
}
}
}
#define LOOP(in, out) \
x = ((x + 1) & 0xff); \
tx = d[x]; \
y = (tx + y) & 0xff; \
d[x] = ty = d[y]; \
d[y] = tx; \
(out) = d[(tx + ty) & 0xff] ^ (in);
#ifndef RC4_INDEX
#define RC4_LOOP(a, b, i) LOOP(*((a)++), *((b)++))
#else
#define RC4_LOOP(a, b, i) LOOP(a[i], b[i])
#endif
i = len >> 3;
if (i) {
for (;;) {
RC4_LOOP(in, out, 0);
RC4_LOOP(in, out, 1);
RC4_LOOP(in, out, 2);
RC4_LOOP(in, out, 3);
RC4_LOOP(in, out, 4);
RC4_LOOP(in, out, 5);
RC4_LOOP(in, out, 6);
RC4_LOOP(in, out, 7);
#ifdef RC4_INDEX
in += 8;
out += 8;
#endif
if (--i == 0) {
break;
}
}
}
i = len & 0x07;
if (i) {
for (;;) {
RC4_LOOP(in, out, 0);
if (--i == 0) {
break;
}
RC4_LOOP(in, out, 1);
if (--i == 0) {
break;
}
RC4_LOOP(in, out, 2);
if (--i == 0) {
break;
}
RC4_LOOP(in, out, 3);
if (--i == 0) {
break;
}
RC4_LOOP(in, out, 4);
if (--i == 0) {
break;
}
RC4_LOOP(in, out, 5);
if (--i == 0) {
break;
}
RC4_LOOP(in, out, 6);
if (--i == 0) {
break;
}
}
}
key->x = x;
key->y = y;
}
void RC4_set_key(RC4_KEY *rc4key, unsigned len, const uint8_t *key) {
uint32_t tmp;
unsigned i, id1, id2;
uint32_t *d;
d = &rc4key->data[0];
rc4key->x = 0;
rc4key->y = 0;
id1 = id2 = 0;
#define SK_LOOP(d, n) \
{ \
tmp = d[(n)]; \
id2 = (key[id1] + tmp + id2) & 0xff; \
if (++id1 == len) \
id1 = 0; \
d[(n)] = d[id2]; \
d[id2] = tmp; \
}
for (i = 0; i < 256; i++) {
d[i] = i;
}
for (i = 0; i < 256; i += 4) {
SK_LOOP(d, i + 0);
SK_LOOP(d, i + 1);
SK_LOOP(d, i + 2);
SK_LOOP(d, i + 3);
}
}
#else
/* In this case several functions are provided by asm code. However, one cannot
* control asm symbol visibility with command line flags and such so they are
* always hidden and wrapped by these C functions, which can be so
* controlled. */
void asm_RC4(RC4_KEY *key, size_t len, const uint8_t *in, uint8_t *out);
void RC4(RC4_KEY *key, size_t len, const uint8_t *in, uint8_t *out) {
asm_RC4(key, len, in, out);
}
void asm_RC4_set_key(RC4_KEY *rc4key, unsigned len, const uint8_t *key);
void RC4_set_key(RC4_KEY *rc4key, unsigned len, const uint8_t *key) {
asm_RC4_set_key(rc4key, len, key);
}
#endif /* OPENSSL_NO_ASM || (!OPENSSL_X86_64 && !OPENSSL_X86) */