[FFmpeg-devel] Patch for Cast128 algorithm implementation
supraja reddy
supraja0493 at gmail.com
Wed Oct 15 21:20:52 CEST 2014
I have updated the changes as requested . Please look into the attached
patch and let me know if there is anything else you wish me to change.
Thanks ,
Supraja
On Sun, Oct 12, 2014 at 8:38 PM, Giorgio Vazzana <mywing81 at gmail.com> wrote:
> 2014-10-12 13:59 GMT+02:00 supraja reddy <supraja0493 at gmail.com>:
> > Hello ,
> >
> > I , Supraja , am interested to work with ffmpeg as a part of OPW
> internship
> > . As a part of the qualification task , I have taken up the
> implementation
> > of Cast 128 algorithm implementation . I have written the code for the
> same
> > and have attached the patch of the same with this mail . It is currently
> > written for ecb mode . It takes keysizes of 40 , 80 and 128 bits for now
> .
> >
> > Please let me know if there are any further changes or suggestions .
>
> Hello,
>
> thanks for your patch! Please make sure to run tools/patcheck as
> Michael suggested and fix the reported problems.
> Some preliminary comments below, hopefully I'll have more time to
> test/review the patch in the following days.
>
> > From b7aab16b09db2a3ca16144221f1516f7fec8a7a2 Mon Sep 17 00:00:00 2001
> > From: Myra <supraja0493 at gmail.com>
>
> Make sure your name and email address are configured correctly in git.
>
> > Date: Sun, 12 Oct 2014 11:39:24 +0530
> > Subject: [PATCH] Added cast128 algorithm for ecb mode
> >
> > Signed-off-by: Myra <supraja0493 at gmail.com>
> > ---
> > libavutil/Makefile | 3 +
> > libavutil/cast5.c | 521
> +++++++++++++++++++++++++++++++++++++++++++++++++++++
> > libavutil/cast5.h | 37 ++++
> > 3 files changed, 561 insertions(+)
> > create mode 100644 libavutil/cast5.c
> > create mode 100644 libavutil/cast5.h
> >
> > diff --git a/libavutil/Makefile b/libavutil/Makefile
> > index 65c6b53..22d345e 100644
> > --- a/libavutil/Makefile
> > +++ b/libavutil/Makefile
> > @@ -15,6 +15,7 @@ HEADERS = adler32.h
> \
> > bprint.h
> \
> > bswap.h
> \
> > buffer.h
> \
> > + cast5.c
> \
> > channel_layout.h
> \
> > common.h
> \
> > cpu.h
> \
> > @@ -82,6 +83,7 @@ OBJS = adler32.o
> \
> > blowfish.o
> \
> > bprint.o
> \
> > buffer.o
> \
>
> > + cast5.o \
>
> You need to add one more space after cast5.o to align the '\'.
>
> > channel_layout.o
> \
> > cpu.o
> \
> > crc.o
> \
> > @@ -151,6 +153,7 @@ TESTPROGS = adler32
> \
> > base64
> \
> > blowfish
> \
> > bprint
> \
>
> > + cast5 \
>
> Same here, one more space is needed.
>
> > cpu
> \
> > crc
> \
> > des
> \
> > diff --git a/libavutil/cast5.c b/libavutil/cast5.c
> > new file mode 100644
> > index 0000000..ca7200c
> > --- /dev/null
> > +++ b/libavutil/cast5.c
> > @@ -0,0 +1,521 @@
> > +#include "cast5.h"
> > +#include "common.h"
> > +#include "intreadwrite.h"
>
> > +#include "timer.h"
>
> I don't think we need timer.h
>
> > +/*
> > + * An implementation of the CAST128 algorithm
>
> Maybe add a note mentioning the RFC2144.
>
> > + * Copyright (c) 2014 Supraja Meedinti
> > + *
> > + * This file is part of FFmpeg.
> > + *
> > + * FFmpeg is free software; you can redistribute it and/or
> > + * modify it under the terms of the GNU Lesser General Public
> > + * License as published by the Free Software Foundation; either
> > + * version 2.1 of the License, or (at your option) any later version.
> > + *
> > + * FFmpeg 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
> > + * Lesser General Public License for more details.
> > + *
> > + * You should have received a copy of the GNU Lesser General Public
> > + * License along with FFmpeg; if not, write to the Free Software
> > + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
> 02110-1301 USA
> > + */
> > +
> > +#define IA(x) ((x>>24) & 0xff)
> > +#define IB(x) ((x>>16) & 0xff)
> > +#define IC(x) ((x>>8) & 0xff)
> > +#define ID(x) (x & 0xff)
>
> I'd also suggest to add a macro for circular left shift (left
> rotation) and use it throughout the code:
>
> #define LR(x, c) ((x) << (c) | (x) >> (32 - (c)))
>
> > +
> > +typedef struct AVCAST5 {
> > + uint32_t Km[17];
> > + uint32_t Kr[17];
> > + int rounds;
> > +} AVCAST5;
> > +
>
> > +
> > +
> > +static uint32_t func(uint32_t D, uint32_t Km, uint32_t Kr, int rnum)
> > +{
> > + uint32_t I,t,val;
> > + switch(rnum%3){
> > + case 0:
> > + t=(Km-D);
> > + I=(t<<Kr)|(t>>(32-Kr));
> > + val=((S1[IA(I)]+S2[IB(I)])^S3[IC(I)])-S4[ID(I)];
> > + break;
>
> You can probably eliminate t, something like:
>
> I = LR(Km - D, Kr);
>
> > + case 1:
> > + t=Km+D;
> > + I=(t<<Kr)|(t>>(32-Kr));
> > + val=((S1[IA(I)]^S2[IB(I)])-S3[IC(I)])+S4[ID(I)];
> > + break;
> > + case 2:
> > + t=(Km^D);
> > + I=(t<<Kr)|(t>>(32-Kr));
> > + val=((S1[IA(I)]-S2[IB(I)])+S3[IC(I)])^S4[ID(I)];
> > + break;
> > + }
> > + return val;
> > +}
> > +
> > +static void GenerateRoundKeys(int rnds,uint32_t* K, uint32_t* q,
> uint32_t* p)
>
> We could call q->x and p->z to use the same names used in the RFC, and
> define:
>
> static void generate_round_keys(int rounds, uint32_t *K, uint32_t
> x[4], uint32_t z[4])
>
> > +{
>
> > +
> p[0]=q[0]^S5[IB(q[3])]^S6[ID(q[3])]^S7[IA(q[3])]^S8[IC(q[3])]^S7[IA(q[2])];
> > +
> p[1]=q[2]^S5[IA(p[0])]^S6[IC(p[0])]^S7[IB(p[0])]^S8[ID(p[0])]^S8[IC(q[2])];
> > +
> p[2]=q[3]^S5[ID(p[1])]^S6[IC(p[1])]^S7[IB(p[1])]^S8[IA(p[1])]^S5[IB(q[2])];
> > +
> p[3]=q[1]^S5[IC(p[2])]^S6[IB(p[2])]^S7[ID(p[2])]^S8[IA(p[2])]^S6[ID(q[2])];
>
> The block above computes p[] from (p[],q[]) and will be repeated
> later, it could be defined as a macro.
>
> > +
> > +
> K[1]=S5[IA(p[2])]^S6[IB(p[2])]^S7[ID(p[1])]^S8[IC(p[1])]^S5[IC(p[0])];
> > +
> K[2]=S5[IC(p[2])]^S6[ID(p[2])]^S7[IB(p[1])]^S8[IA(p[1])]^S6[IC(p[1])];
> > +
> K[3]=S5[IA(p[3])]^S6[IB(p[3])]^S7[ID(p[0])]^S8[IC(p[0])]^S7[IB(p[2])];
> > +
> K[4]=S5[IC(p[3])]^S6[ID(p[3])]^S7[IB(p[0])]^S8[IA(p[0])]^S8[IA(p[3])];
> > +
>
> > +
> q[0]=p[2]^S5[IB(p[1])]^S6[ID(p[1])]^S7[IA(p[1])]^S8[IC(p[1])]^S7[IA(p[0])];
> > +
> q[1]=p[0]^S5[IA(q[0])]^S6[IC(q[0])]^S7[IB(q[0])]^S8[ID(q[0])]^S8[IC(p[0])];
> > +
> q[2]=p[1]^S5[ID(q[1])]^S6[IC(q[1])]^S7[IB(q[1])]^S8[IA(q[1])]^S5[IB(p[0])];
> > +
> q[3]=p[3]^S5[IC(q[2])]^S6[IB(q[2])]^S7[ID(q[2])]^S8[IA(q[2])]^S6[ID(p[0])];
>
> Same here, the block above computes q[] from (p[],q[]) and will be
> repeated later, you could define a macro.
>
> > +
> > +
> K[5]=S5[ID(q[0])]^S6[IC(q[0])]^S7[IA(q[3])]^S8[IB(q[3])]^S5[IA(q[2])];
> > +
> K[6]=S5[IB(q[0])]^S6[IA(q[0])]^S7[IC(q[3])]^S8[ID(q[3])]^S6[IB(q[3])];
> > +
> K[7]=S5[ID(q[1])]^S6[IC(q[1])]^S7[IA(q[2])]^S8[IB(q[2])]^S7[ID(q[0])];
> > +
> K[8]=S5[IB(q[1])]^S6[IA(q[1])]^S7[IC(q[2])]^S8[ID(q[2])]^S8[ID(q[1])];
> > +
> > +
> p[0]=q[0]^S5[IB(q[3])]^S6[ID(q[3])]^S7[IA(q[3])]^S8[IC(q[3])]^S7[IA(q[2])];
> > +
> p[1]=q[2]^S5[IA(p[0])]^S6[IC(p[0])]^S7[IB(p[0])]^S8[ID(p[0])]^S8[IC(q[2])];
> > +
> p[2]=q[3]^S5[ID(p[1])]^S6[IC(p[1])]^S7[IB(p[1])]^S8[IA(p[1])]^S5[IB(q[2])];
> > +
> p[3]=q[1]^S5[IC(p[2])]^S6[IB(p[2])]^S7[ID(p[2])]^S8[IA(p[2])]^S6[ID(q[2])];
> > +
> > +
> K[9]=S5[ID(p[0])]^S6[IC(p[0])]^S7[IA(p[3])]^S8[IB(p[3])]^S5[IB(p[2])];
> > +
> K[10]=S5[IB(p[0])]^S6[IA(p[0])]^S7[IC(p[3])]^S8[ID(p[3])]^S6[IA(p[3])];
> > +
> K[11]=S5[ID(p[1])]^S6[IC(p[1])]^S7[IA(p[2])]^S8[IB(p[2])]^S7[IC(p[0])];
> > +
> K[12]=S5[IB(p[1])]^S6[IA(p[1])]^S7[IC(p[2])]^S8[ID(p[2])]^S8[IC(p[1])];
> > +
> > +
> q[0]=p[2]^S5[IB(p[1])]^S6[ID(p[1])]^S7[IA(p[1])]^S8[IC(p[1])]^S7[IA(p[0])];
> > +
> q[1]=p[0]^S5[IA(q[0])]^S6[IC(q[0])]^S7[IB(q[0])]^S8[ID(q[0])]^S8[IC(p[0])];
> > +
> q[2]=p[1]^S5[ID(q[1])]^S6[IC(q[1])]^S7[IB(q[1])]^S8[IA(q[1])]^S5[IB(p[0])];
> > +
> q[3]=p[3]^S5[IC(q[2])]^S6[IB(q[2])]^S7[ID(q[2])]^S8[IA(q[2])]^S6[ID(p[0])];
> > +
> > + if(rnds==16) {
> > +
> K[13]=S5[IA(q[2])]^S6[IB(q[2])]^S7[ID(q[1])]^S8[IC(q[1])]^S5[ID(q[0])];
> > +
> K[14]=S5[IC(q[2])]^S6[ID(q[2])]^S7[IB(q[1])]^S8[IA(q[1])]^S6[ID(q[1])];
> > +
> K[15]=S5[IA(q[3])]^S6[IB(q[3])]^S7[ID(q[0])]^S8[IC(q[0])]^S7[IA(q[2])];
> > +
> K[16]=S5[IC(q[3])]^S6[ID(q[3])]^S7[IB(q[0])]^S8[IA(q[0])]^S8[IB(q[3])];
> > + }
> > +}
> > +
> > +int av_cast5_init(AVCAST5* cs, const uint8_t *key, int key_bits)
> > +{
> > + uint8_t newKey[16];
> > + int i;
> > + uint32_t p[4],q[4];
>
> We need:
>
> memset(newKey, 0, sizeof(newKey));
>
> > + if(key_bits%8!=0||key_bits<40||key_bits>128)
>
> if (key_bits % 8 || key_bits < 40 || key_bits > 128)
>
> > + return -1;
> > + memcpy(newKey,key,16*sizeof(key[0]));
>
> memcpy(newKey, key, key_bits / 8);
>
> > + if (key_bits!=128) {
> > + for(i=15;i>=key_bits/8;i--)
> > + newKey[i]=0x00;
> > + }
>
> This if() is not needed anymore, since we used memset() on newKey.
>
> > + if(key_bits==128){
> > + cs->rounds=16;
> > + } else {
> > + cs->rounds=12;
> > + }
>
> cs->rounds = key_bits <= 80 ? 12 : 16;
>
> > + for(i=0;i<4;i++)
> > + q[i]=AV_RB32(newKey+(4*i));
> > + GenerateRoundKeys(cs->rounds,cs->Km,q,p);
> > + GenerateRoundKeys(cs->rounds,cs->Kr,q,p);
> > + for(i=0;i<cs->rounds;i++)
> > + cs->Kr[i]=cs->Kr[i]&0x1f;
> > + return 0;
> > +}
> > +
> > +static void encipher(AVCAST5* cs,uint8_t* dst,const uint8_t* src)
> > +{
> > + int i;
> > + uint32_t temp,sdst[2];
> > + sdst[0]=AV_RB32(src);
> > + sdst[1]=AV_RB32(src+4);
> > + for(i=1;i<=cs->rounds;i++){
> > + temp=sdst[1];
> > + sdst[1]=sdst[0]^(func(sdst[1],cs->Km[i],cs->Kr[i],i));
> > + sdst[0]=temp;
> > + }
> > + AV_WB32(dst,sdst[1]);
> > + AV_WB32(dst+4,sdst[0]);
> > +}
> > +
> > +static void decipher(AVCAST5* cs,uint8_t* dst,const uint8_t* src)
> > +{
> > + int i;
> > + uint32_t temp,sdst[2];
> > + sdst[0]=AV_RB32(src);
> > + sdst[1]=AV_RB32(src+4);
> > + for(i=cs->rounds;i>0;i--){
> > + temp=sdst[1];
> > + sdst[1]=sdst[0]^(func(sdst[1],cs->Km[i],cs->Kr[i],i));
> > + sdst[0]=temp;
> > + }
> > + AV_WB32(dst,sdst[1]);
> > + AV_WB32(dst+4,sdst[0]);
> > +}
> > +
> > +void av_cast5_crypt(AVCAST5* cs, uint8_t* dst, const uint8_t* src, int
> count, int decrypt)
> > +{
> > + while(count--)
> > + {
> > + if(decrypt){
> > + decipher(cs,dst,src);
> > + } else {
> > + encipher(cs,dst,src);
> > + }
> > + src=src+8;
> > + dst=dst+8;
> > + }
> > +}
> > +
> > +#include<stdio.h>
> > +#include<stdlib.h>
> > +#include"log.h"
>
> > +#include"lfg.h"
>
> Not sure we need lfg.h?
>
> > +
> > +#ifdef TEST
> > +int main(int argc, char** argv)
> > +{
> > + AVCAST5 cs;
> > + uint8_t
> Key[3][16]={{0x01,0x23,0x45,0x67,0x12,0x34,0x56,0x78,0x23,0x45,0x67,0x89,0x34,0x56,0x78,0x9a},
> > + {0x01,0x23,0x45,0x67,0x12,0x34,0x56,0x78,0x23,0x45},
> > + {0x01,0x23,0x45,0x67,0x12}};
> > + uint8_t rpt[8]={0x01,0x23,0x45,0x67,0x89,0xab,0xcd,0xef};
> > + uint8_t rct[3][8]={{0x23,0x8b,0x4f,0xe5,0x84,0x7e,0x44,0xb2},
> > + {0xeb,0x6a,0x71,0x1a,0x2c,0x02,0x27,0x1b},
> > + {0x7a,0xc8,0x16,0xd1,0x6e,0x9b,0x30,0x2e}};
> > + int i,err=0;
> > + uint8_t temp[8];
> > + av_log_set_level(AV_LOG_DEBUG);
> > + av_cast5_init(&cs,Key[0],128);
> > + av_cast5_crypt(&cs,temp,rpt,1,0);
> > + for(i=0;i<8;i++){
> > + if(rct[0][i]!=temp[i]){
> > + av_log(NULL,AV_LOG_ERROR,"%d %02x
> %02x\n",i,rct[0][i],temp[i]);
> > + err=1;
> > + }
> > + }
> > + av_cast5_crypt(&cs,temp,rct[0],1,1);
> > + for(i=0;i<8;i++){
> > + if(rpt[i]!=temp[i]){
> > + av_log(NULL,AV_LOG_ERROR,"%d %02x %02x\n",i,rpt[i],temp[i]);
> > + err=1;
> > + }
> > + }
> > + av_cast5_init(&cs,Key[1],80);
> > + av_cast5_crypt(&cs,temp,rpt,1,0);
> > + for(i=0;i<8;i++){
> > + if(rct[1][i]!=temp[i]){
> > + av_log(NULL,AV_LOG_ERROR,"%d %02x
> %02x\n",i,rct[1][i],temp[i]);
> > + err=1;
> > + }
> > + }
> > + av_cast5_crypt(&cs,temp,rct[1],1,1);
> > + for(i=0;i<8;i++){
> > + if(rpt[i]!=temp[i]){
> > + av_log(NULL,AV_LOG_ERROR,"%d %02x %02x\n",i,rpt[i],temp[i]);
> > + err=1;
> > + }
> > + }
> > + av_cast5_init(&cs,Key[2],40);
> > + av_cast5_crypt(&cs,temp,rpt,1,0);
> > + for(i=0;i<8;i++){
> > + if(rct[2][i]!=temp[i]){
> > + av_log(NULL,AV_LOG_ERROR,"%d %02x
> %02x\n",i,rct[2][i],temp[i]);
> > + err=1;
> > + }
> > + }
> > + av_cast5_crypt(&cs,temp,rct[2],1,1);
> > + for(i=0;i<8;i++){
> > + if(rpt[i]!=temp[i]){
> > + av_log(NULL,AV_LOG_ERROR,"%d %02x %02x\n",i,rpt[i],temp[i]);
> > + err=1;
> > + }
> > + }
> > + return err;
> > +}
> > +
> > +#endif
> > +
> > +
> > +
> > +
> > +
> > +
> > diff --git a/libavutil/cast5.h b/libavutil/cast5.h
> > new file mode 100644
> > index 0000000..d2196b9
> > --- /dev/null
> > +++ b/libavutil/cast5.h
> > @@ -0,0 +1,37 @@
> > +/*
> > + * An implementation of the CAST128 algorithm
> > + * Copyright (c) 2014 Supraja Meedinti
> > + *
> > + * This file is part of FFmpeg.
> > + *
> > + * FFmpeg is free software; you can redistribute it and/or
> > + * modify it under the terms of the GNU Lesser General Public
> > + * License as published by the Free Software Foundation; either
> > + * version 2.1 of the License, or (at your option) any later version.
> > + *
> > + * FFmpeg 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
> > + * Lesser General Public License for more details.
> > + *
> > + * You should have received a copy of the GNU Lesser General Public
> > + * License along with FFmpeg; if not, write to the Free Software
> > + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
> 02110-1301 USA
> > + */
> > +
> > +#ifndef AVUTIL_CAST5_H
> > +#define AVUTIL_CAST5_H
> > +
> > +#include <stdint.h>
> > +
> > +#include "attributes.h"
> > +#include "version.h"
> > +
> > +struct AVCAST5;
> > +
> > +int av_cast5_init(struct AVCAST5 *a, const uint8_t *key, int key_bits);
> > +
> > +void av_cast5_crypt(struct AVCAST5 *a, uint8_t *dst, const uint8_t
> *src, int count,int decrypt);
> > +
> > +
> > +#endif
> > --
> > 1.8.3.2
> >
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Updated-cast5.patch
Type: text/x-patch
Size: 39761 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20141016/3792a521/attachment.bin>
More information about the ffmpeg-devel
mailing list