/*
* This file is part of fastboot 3DS
* Copyright (C) 2017 derrek, profi200
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program 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 for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "mem_map.h"
#include "fb_assert.h"
#include "types.h"
#include "arm9/hardware/crypto.h"
#include "arm9/hardware/cfg9.h"
#include "arm9/hardware/interrupt.h"
#include "arm9/hardware/ndma.h"
#include "arm.h"
#include "mmio.h"
//////////////////////////////////
// AES //
//////////////////////////////////
#define AES_REGS_BASE (IO_MEM_ARM9_ONLY + 0x9000)
#define REG_AESCNT *((vu32*)(AES_REGS_BASE + 0x000))
#define REG_AESBLKCNT *((vu32*)(AES_REGS_BASE + 0x004))
#define REG_AES_BLKCNT_LOW *((vu16*)(AES_REGS_BASE + 0x004))
#define REG_AES_BLKCNT_HIGH *((vu16*)(AES_REGS_BASE + 0x006))
#define REG_AESWRFIFO ( AES_REGS_BASE + 0x008)
#define REG_AESRDFIFO ( AES_REGS_BASE + 0x00C)
#define REG_AESKEYSEL *((vu8* )(AES_REGS_BASE + 0x010))
#define REG_AESKEYCNT *((vu8* )(AES_REGS_BASE + 0x011))
#define REG_AESCTR ((vu32*)(AES_REGS_BASE + 0x020))
#define REG_AESMAC ((vu32*)(AES_REGS_BASE + 0x030))
#define REG_AESKEY0 ((vu32*)(AES_REGS_BASE + 0x040))
#define REG_AESKEYX0 ((vu32*)(AES_REGS_BASE + 0x050))
#define REG_AESKEYY0 ((vu32*)(AES_REGS_BASE + 0x060))
#define REG_AESKEY1 ((vu32*)(AES_REGS_BASE + 0x070))
#define REG_AESKEYX1 ((vu32*)(AES_REGS_BASE + 0x080))
#define REG_AESKEYY1 ((vu32*)(AES_REGS_BASE + 0x090))
#define REG_AESKEY2 ((vu32*)(AES_REGS_BASE + 0x0A0))
#define REG_AESKEYX2 ((vu32*)(AES_REGS_BASE + 0x0B0))
#define REG_AESKEYY2 ((vu32*)(AES_REGS_BASE + 0x0C0))
#define REG_AESKEY3 ((vu32*)(AES_REGS_BASE + 0x0D0))
#define REG_AESKEYX3 ((vu32*)(AES_REGS_BASE + 0x0E0))
#define REG_AESKEYY3 ((vu32*)(AES_REGS_BASE + 0x0F0))
#define REG_AESKEYFIFO ((vu32*)(AES_REGS_BASE + 0x100))
#define REG_AESKEYXFIFO ((vu32*)(AES_REGS_BASE + 0x104))
#define REG_AESKEYYFIFO ((vu32*)(AES_REGS_BASE + 0x108))
static void setupKeys(void)
{
// Setup TWL unit info and console ID
const bool isDevUnit = REG_CFG9_UNITINFO != 0;
const u64 twlConsoleId = (isDevUnit ? (*((vu64*)0x10012000)) :
((*((vu64*)0x01FFB808) ^ 0x8C267B7B358A6AFULL) | 0x80000000ULL));
*((vu8*)0x10010014) = REG_CFG9_UNITINFO;
*((vu64*)0x10012100) = twlConsoleId;
// TWL key init
REG_AESCNT = 0;
REG_AESKEYX1[2] = (u32)(twlConsoleId>>32);
REG_AESKEYX1[3] = (u32)twlConsoleId;
AES_setKey(0x02, AES_KEY_X, AES_INPUT_LITTLE | AES_INPUT_REVERSED, false, (u32*)0x01FFD398);
u32 key3X[4] = {(u32)twlConsoleId, 0, 0, (u32)(twlConsoleId>>32)};
if(isDevUnit)
{
key3X[1] = 0xEE7A4B1E;
key3X[2] = 0xAF42C08B;
AES_setKey(0x03, AES_KEY_X, AES_INPUT_LITTLE | AES_INPUT_REVERSED, false, key3X);
alignas(4) static const u8 key2YDev[16] = {
0x3B, 0x06, 0x86, 0x57, 0x33, 0x04, 0x88, 0x11, 0x49, 0x04, 0x6B, 0x33, 0x12, 0x02, 0xAC, 0xF3};
alignas(4) static const u8 key3YDev[16] = {
0xAA, 0xBF, 0x76, 0xF1, 0x7A, 0xB8, 0xE8, 0x66, 0x97, 0x64, 0x6A, 0x26, 0x05, 0x00, 0xA0, 0xE1};
AES_setKey(0x02, AES_KEY_Y, AES_INPUT_LITTLE | AES_INPUT_REVERSED, false, (const u32*)key2YDev);
AES_setKey(0x03, AES_KEY_Y, AES_INPUT_LITTLE | AES_INPUT_REVERSED, false, (const u32*)key3YDev);
}
else
{
key3X[1] = 0x544E494E; // "NINT"
key3X[2] = 0x4F444E45; // "ENDO"
AES_setKey(0x03, AES_KEY_X, AES_INPUT_LITTLE | AES_INPUT_REVERSED, false, key3X);
AES_setKey(0x02, AES_KEY_Y, AES_INPUT_LITTLE | AES_INPUT_REVERSED, false, (u32*)0x01FFD220);
u32 key3YRetail[4];
key3YRetail[0] = ((u32*)0x01FFD3C8)[0];
key3YRetail[1] = ((u32*)0x01FFD3C8)[1];
key3YRetail[2] = ((u32*)0x01FFD3C8)[2];
key3YRetail[3] = 0xE1A00005;
AES_setKey(0x03, AES_KEY_Y, AES_INPUT_LITTLE | AES_INPUT_REVERSED, false, key3YRetail);
}
// 3DS key init
if(REG_CFG9_SOCINFO & 2) // New 3DS
{
alignas(4) static const u8 keyY0x05[16] = {
0x4D, 0x80, 0x4F, 0x4E, 0x99, 0x90, 0x19, 0x46, 0x13, 0xA2, 0x04, 0xAC, 0x58, 0x44, 0x60, 0xBE};
AES_setKey(0x05, AES_KEY_Y, AES_INPUT_BIG | AES_INPUT_NORMAL, false, (const u32*)keyY0x05);
}
alignas(4) static const u8 keyY0x24[16] = {
0x74, 0xCA, 0x07, 0x48, 0x84, 0xF4, 0x22, 0x8D, 0xEB, 0x2A, 0x1C, 0xA7, 0x2D, 0x28, 0x77, 0x62};
AES_setKey(0x24, AES_KEY_Y, AES_INPUT_BIG | AES_INPUT_NORMAL, false, (const u32*)keyY0x24);
alignas(4) static const u8 keyX0x25s[2][16] = {
{0xCE, 0xE7, 0xD8, 0xAB, 0x30, 0xC0, 0x0D, 0xAE, 0x85, 0x0E, 0xF5, 0xE3, 0x82, 0xAC, 0x5A, 0xF3},
{0x81, 0x90, 0x7A, 0x4B, 0x6F, 0x1B, 0x47, 0x32, 0x3A, 0x67, 0x79, 0x74, 0xCE, 0x4A, 0xD7, 0x1B}};
AES_setKey(0x25, AES_KEY_X, AES_INPUT_BIG | AES_INPUT_NORMAL, false, (const u32*)keyX0x25s[isDevUnit]);
alignas(4) static const u8 keyY0x2Fs[2][16] = {
{0xC3, 0x69, 0xBA, 0xA2, 0x1E, 0x18, 0x8A, 0x88, 0xA9, 0xAA, 0x94, 0xE5, 0x50, 0x6A, 0x9F, 0x16},
{0x73, 0x25, 0xC4, 0xEB, 0x14, 0x3A, 0x0D, 0x5F, 0x5D, 0xB6, 0xE5, 0xC5, 0x7A, 0x21, 0x95, 0xAC}};
AES_setKey(0x2F, AES_KEY_Y, AES_INPUT_BIG | AES_INPUT_NORMAL, false, (const u32*)keyY0x2Fs[isDevUnit]);
// Set 0x11 keyslot
alignas(4) static const u8 key1s[2][16] = {
{0x07, 0x29, 0x44, 0x38, 0xF8, 0xC9, 0x75, 0x93, 0xAA, 0x0E, 0x4A, 0xB4, 0xAE, 0x84, 0xC1, 0xD8},
{0xA2, 0xF4, 0x00, 0x3C, 0x7A, 0x95, 0x10, 0x25, 0xDF, 0x4E, 0x9E, 0x74, 0xE3, 0x0C, 0x92, 0x99}};
alignas(4) static const u8 key2s[2][16] = {
{0x42, 0x3F, 0x81, 0x7A, 0x23, 0x52, 0x58, 0x31, 0x6E, 0x75, 0x8E, 0x3A, 0x39, 0x43, 0x2E, 0xD0},
{0xFF, 0x77, 0xA0, 0x9A, 0x99, 0x81, 0xE9, 0x48, 0xEC, 0x51, 0xC9, 0x32, 0x5D, 0x14, 0xEC, 0x25}};
alignas(4) u8 keyBlocks[2][16] = {
{0xA4, 0x8D, 0xE4, 0xF1, 0x0B, 0x36, 0x44, 0xAA, 0x90, 0x31, 0x28, 0xFF, 0x4D, 0xCA, 0x76, 0xDF},
{0xDD, 0xDA, 0xA4, 0xC6, 0x2C, 0xC4, 0x50, 0xE9, 0xDA, 0xB6, 0x9B, 0x0D, 0x9D, 0x2A, 0x21, 0x98}};
u32 decKey[4];
AES_ctx ctx;
AES_setCryptParams(&ctx, AES_INPUT_BIG | AES_INPUT_NORMAL, AES_OUTPUT_BIG | AES_OUTPUT_NORMAL);
// key 0x18
AES_setKey(0x11, AES_KEY_NORMAL, AES_INPUT_BIG | AES_INPUT_NORMAL, false, (const u32*)key1s[isDevUnit]);
AES_selectKeyslot(0x11);
AES_ecb(&ctx, (const u32*)keyBlocks[0], decKey, 1, false, false);
AES_setKey(0x18, AES_KEY_X, AES_INPUT_BIG | AES_INPUT_NORMAL, false, decKey);
AES_setKey(0x11, AES_KEY_NORMAL, AES_INPUT_BIG | AES_INPUT_NORMAL, false, (const u32*)key2s[isDevUnit]);
AES_selectKeyslot(0x11);
for(u8 slot = 0x19; slot < 0x20; slot++, keyBlocks[1][0xF]++)
{
AES_ecb(&ctx, (const u32*)keyBlocks[1], decKey, 1, false, false);
AES_setKey(slot, AES_KEY_X, AES_INPUT_BIG | AES_INPUT_NORMAL, false, decKey);
}
}
void AES_init(void)
{
REG_AESCNT = AES_MAC_SIZE(4) | AES_FLUSH_WRITE_FIFO | AES_FLUSH_READ_FIFO;
*((vu8*)0x10000008) = 0; // ??
REG_NDMA0_DST_ADDR = REG_AESWRFIFO;
REG_NDMA0_INT_CNT = NDMA_INT_SYS_FREQ;
REG_NDMA1_SRC_ADDR = REG_AESRDFIFO;
REG_NDMA1_INT_CNT = NDMA_INT_SYS_FREQ;
IRQ_registerHandler(IRQ_AES, NULL);
setupKeys();
}
void AES_setKey(u8 keyslot, AesKeyType type, u8 orderEndianess, bool twlScrambler, const u32 key[4])
{
fb_assert(keyslot < 0x40);
fb_assert(key != NULL);
REG_AESCNT = (u32)orderEndianess<<23;
if(keyslot > 3)
{
REG_AESKEYCNT = 1u<<7 | (type > AES_KEY_NORMAL && twlScrambler ? 1u : 0u)<<6 | keyslot;
for(u32 i = 0; i < 4; i++) REG_AESKEYFIFO[type] = key[i];
}
else
{
vu32 *twlKeyNReg = ®_AESKEY0[12u * keyslot + type * 4u];
if(orderEndianess & AES_INPUT_NORMAL)
{
for(u32 i = 0; i < 4; i++) twlKeyNReg[i] = key[3u - i];
}
else
{
for(u32 i = 0; i < 4; i++) twlKeyNReg[i] = key[i];
}
}
}
void AES_selectKeyslot(u8 keyslot)
{
fb_assert(keyslot < 0x40);
REG_AESKEYSEL = keyslot;
REG_AESCNT |= AES_UPDATE_KEYSLOT;
}
void AES_setNonce(AES_ctx *const ctx, u8 orderEndianess, const u32 nonce[3])
{
fb_assert(ctx != NULL);
fb_assert(nonce != NULL);
ctx->ctrIvNonceParams = (u32)orderEndianess<<23;
u32 *const ctxNonce = ctx->ctrIvNonce;
if(orderEndianess & AES_INPUT_NORMAL)
{
for(u32 i = 0; i < 3; i++) ctxNonce[i] = nonce[2u - i];
}
else
{
for(u32 i = 0; i < 3; i++) ctxNonce[i] = nonce[i];
}
}
void AES_setCtrIv(AES_ctx *const ctx, u8 orderEndianess, const u32 ctrIv[4])
{
fb_assert(ctx != NULL);
fb_assert(ctrIv != NULL);
ctx->ctrIvNonceParams = (u32)orderEndianess<<23;
u32 *const ctxCtrIv = ctx->ctrIvNonce;
if(orderEndianess & AES_INPUT_NORMAL)
{
for(u32 i = 0; i < 4; i++) ctxCtrIv[i] = ctrIv[3u - i];
}
else
{
for(u32 i = 0; i < 4; i++) ctxCtrIv[i] = ctrIv[i];
}
}
NAKED void AES_addCounter(u32 ctr[4], u32 val)
{
__asm__
(
"str lr, [sp, #-4]!\n\t"
"ldm r0, {r2, r3, r12, lr}\n\t"
"adds r2, r1, lsr #4\n\t"
"addcss r3, r3, #1\n\t"
"addcss r12, r12, #1\n\t"
"addcs lr, lr, #1\n\t"
"stm r0, {r2, r3, r12, lr}\n\t"
"ldr pc, [sp], #4\n\t"
: : "r" (ctr), "r" (val) :
);
}
void AES_setCryptParams(AES_ctx *const ctx, u8 inEndianessOrder, u8 outEndianessOrder)
{
fb_assert(ctx != NULL);
ctx->aesParams = (u32)inEndianessOrder<<23 | (u32)outEndianessOrder<<22;
}
static void aesProcessBlocksCpu(const u32 *in, u32 *out, u32 blocks)
{
REG_AES_BLKCNT_HIGH = blocks;
REG_AESCNT |= AES_ENABLE | 3u<<12 | AES_FLUSH_READ_FIFO | AES_FLUSH_WRITE_FIFO;
for(u32 i = 0; i < blocks * 4; i += 4)
{
_u128 tmp = *((const _u128*)&in[i]);
*((vu32*)REG_AESWRFIFO) = tmp.data[0];
*((vu32*)REG_AESWRFIFO) = tmp.data[1];
*((vu32*)REG_AESWRFIFO) = tmp.data[2];
*((vu32*)REG_AESWRFIFO) = tmp.data[3];
while(AES_READ_FIFO_COUNT == 0);
tmp.data[0] = *((vu32*)REG_AESRDFIFO);
tmp.data[1] = *((vu32*)REG_AESRDFIFO);
tmp.data[2] = *((vu32*)REG_AESRDFIFO);
tmp.data[3] = *((vu32*)REG_AESRDFIFO);
*((_u128*)&out[i]) = tmp;
}
}
// AES_init() must be called before this works
static void aesProcessBlocksDma(const u32 *in, u32 *out, u32 blocks)
{
// DMA can't reach TCMs
fb_assert(((u32)in >= ITCM_BOOT9_MIRROR + ITCM_SIZE) && (((u32)in < DTCM_BASE) || ((u32)in >= DTCM_BASE + DTCM_SIZE)));
fb_assert(((u32)out >= ITCM_BOOT9_MIRROR + ITCM_SIZE) && (((u32)out < DTCM_BASE) || ((u32)out >= DTCM_BASE + DTCM_SIZE)));
// Check block alignment
const u8 aesFifoSize = (blocks & 1u ? 0u : 1u); // 1 = 32 bytes, 0 = 16 bytes
REG_NDMA0_SRC_ADDR = (u32)in;
REG_NDMA0_TOTAL_CNT = blocks<<2;
REG_NDMA0_LOG_BLK_CNT = aesFifoSize * 4 + 4;
REG_NDMA0_CNT = NDMA_ENABLE | NDMA_TOTAL_CNT_MODE | NDMA_STARTUP_AES_IN |
NDMA_BURST_WORDS(4) | NDMA_SRC_UPDATE_INC | NDMA_DST_UPDATE_FIXED;
REG_NDMA1_DST_ADDR = (u32)out;
REG_NDMA1_TOTAL_CNT = blocks<<2;
REG_NDMA1_LOG_BLK_CNT = aesFifoSize * 4 + 4;
REG_NDMA1_CNT = NDMA_ENABLE | NDMA_TOTAL_CNT_MODE | NDMA_STARTUP_AES_OUT |
NDMA_BURST_WORDS(4) | NDMA_SRC_UPDATE_FIXED | NDMA_DST_UPDATE_INC;
REG_AES_BLKCNT_HIGH = blocks;
REG_AESCNT |= AES_ENABLE | AES_IRQ_ENABLE | aesFifoSize<<14 | (3 - aesFifoSize)<<12 |
AES_FLUSH_READ_FIFO | AES_FLUSH_WRITE_FIFO;
do
{
__wfi();
} while(REG_AESCNT & AES_ENABLE);
}
void AES_ctr(AES_ctx *const ctx, const u32 *in, u32 *out, u32 blocks, bool dma)
{
fb_assert(ctx != NULL);
fb_assert(in != NULL);
fb_assert(out != NULL);
const u32 ctrParams = ctx->ctrIvNonceParams;
u32 *const ctr = ctx->ctrIvNonce;
const u32 aesParams = AES_MODE_CTR | ctx->aesParams;
while(blocks)
{
REG_AESCNT = ctrParams;
for(u32 i = 0; i < 4; i++) REG_AESCTR[i] = ctr[i];
REG_AESCNT = aesParams;
u32 blockNum = ((blocks > AES_MAX_BLOCKS) ? AES_MAX_BLOCKS : blocks);
if(dma) aesProcessBlocksDma(in, out, blockNum);
else aesProcessBlocksCpu(in, out, blockNum);
AES_addCounter(ctr, blockNum<<4);
in += blockNum<<2;
out += blockNum<<2;
blocks -= blockNum;
}
}
/*void AES_cbc(AES_ctx *const ctx, const u32 *in, u32 *out, u32 blocks, bool enc, bool dma)
{
fb_assert(ctx != NULL);
fb_assert(in != NULL);
fb_assert(out != NULL);
const u32 ivParams = ctx->ctrIvNonceParams;
u32 *const iv = ctx->ctrIvNonce;
const u32 aesParams = (enc ? AES_MODE_CBC_ENCRYPT : AES_MODE_CBC_DECRYPT) | ctx->aesParams;
while(blocks)
{
REG_AESCNT = ivParams;
REG_AESCTR[0] = iv[0];
REG_AESCTR[1] = iv[1];
REG_AESCTR[2] = iv[2];
REG_AESCTR[3] = iv[3];
u32 blockNum = ((blocks > AES_MAX_BLOCKS) ? AES_MAX_BLOCKS : blocks);
if(!enc)
{
// Save last 16 bytes of the input blocks as next IV
const u32 *const nextIv = in + (blockNum<<2) - 4;
if(aesParams>>23 & AES_INPUT_NORMAL)
{
iv[0] = nextIv[3];
iv[1] = nextIv[2];
iv[2] = nextIv[1];
iv[3] = nextIv[0];
}
else
{
iv[0] = nextIv[0];
iv[1] = nextIv[1];
iv[2] = nextIv[2];
iv[3] = nextIv[3];
}
}
REG_AESCNT = aesParams;
if(dma) aesProcessBlocksDma(in, out, blockNum);
else aesProcessBlocksCpu(in, out, blockNum);
if(enc)
{
// Save last 16 bytes of the output blocks as next IV
const u32 *const nextIv = out + (blockNum<<2) - 4;
if(aesParams>>23 & AES_INPUT_NORMAL)
{
iv[0] = nextIv[3];
iv[1] = nextIv[2];
iv[2] = nextIv[1];
iv[3] = nextIv[0];
}
else
{
iv[0] = nextIv[0];
iv[1] = nextIv[1];
iv[2] = nextIv[2];
iv[3] = nextIv[3];
}
}
in += blockNum<<2;
out += blockNum<<2;
blocks -= blockNum;
}
}*/
void AES_ecb(AES_ctx *const ctx, const u32 *in, u32 *out, u32 blocks, bool enc, bool dma)
{
fb_assert(ctx != NULL);
fb_assert(in != NULL);
fb_assert(out != NULL);
const u32 aesParams = (enc ? AES_MODE_ECB_ENCRYPT : AES_MODE_ECB_DECRYPT) | ctx->aesParams;
while(blocks)
{
REG_AESCNT = aesParams;
u32 blockNum = ((blocks > AES_MAX_BLOCKS) ? AES_MAX_BLOCKS : blocks);
if(dma) aesProcessBlocksDma(in, out, blockNum);
else aesProcessBlocksCpu(in, out, blockNum);
in += blockNum<<2;
out += blockNum<<2;
blocks -= blockNum;
}
}
bool AES_ccm(const AES_ctx *const ctx, const u32 *const in, u32 *const out, u32 macSize,
u32 mac[4], u16 blocks, bool enc)
{
fb_assert(ctx != NULL);
fb_assert(in != NULL);
fb_assert(out != NULL);
fb_assert(macSize != 0);
fb_assert(mac != NULL);
fb_assert(blocks != 0);
REG_AESCNT = ctx->ctrIvNonceParams;
for(u32 i = 0; i < 3; i++) REG_AESCTR[i] = ctx->ctrIvNonce[i];
REG_AES_BLKCNT_LOW = 0;
REG_AESCNT = (enc ? AES_MODE_CCM_ENCRYPT : AES_MODE_CCM_DECRYPT) |
AES_MAC_SIZE(macSize) | ctx->aesParams;
aesProcessBlocksCpu(in, out, blocks);
// This is broken right now with DMA due to a (AES engine?) bug.
if(!enc)
{
for(u32 i = 0; i < 4; i++) *((vu32*)REG_AESWRFIFO) = mac[i];
while(REG_AESCNT & AES_ENABLE);
}
else
{
for(u32 i = 0; i < 4; i++) mac[i] = *((vu32*)REG_AESRDFIFO);
}
if(enc) return true;
else return AES_IS_MAC_VALID;
}
//////////////////////////////////
// SHA //
//////////////////////////////////
#define SHA_REGS_BASE (IO_MEM_ARM9_ONLY + 0xA000)
#define REG_SHA_CNT *((vu32*)(SHA_REGS_BASE + 0x00))
#define REG_SHA_BLKCNT *((vu32*)(SHA_REGS_BASE + 0x04))
#define REGs_SHA_HASH ((vu32*)(SHA_REGS_BASE + 0x40))
#define REGs_SHA_INFIFO ((vu32*)(SHA_REGS_BASE + 0x80))
void SHA_start(u8 params)
{
REG_SHA_CNT = params | SHA_IN_DMA_ENABLE | SHA_ENABLE;
}
void SHA_update(const u32 *data, u32 size)
{
while(size >= 64)
{
*((volatile _u512*)REGs_SHA_INFIFO) = *((const _u512*)data);
data += 64 / 4;
size -= 64;
while(REG_SHA_CNT & SHA_ENABLE);
}
if(size) iomemcpy(REGs_SHA_INFIFO, data, size);
}
void SHA_finish(u32 *const hash, u8 endianess)
{
REG_SHA_CNT = (REG_SHA_CNT & SHA_MODE_MASK) | endianess | SHA_FINAL_ROUND;
while(REG_SHA_CNT & SHA_ENABLE);
SHA_getState(hash);
}
void SHA_getState(u32 *const out)
{
u32 size;
switch(REG_SHA_CNT & SHA_MODE_MASK)
{
case SHA_MODE_256:
size = 32;
break;
case SHA_MODE_224:
size = 28;
break;
case SHA_MODE_1:
default: // 2 and 3 are both SHA1
size = 20;
}
iomemcpy(out, REGs_SHA_HASH, size);
}
void sha(const u32 *data, u32 size, u32 *const hash, u8 params, u8 hashEndianess)
{
SHA_start(params);
SHA_update(data, size);
SHA_finish(hash, hashEndianess);
}
/*void sha_dma(const u32 *data, u32 size, u32 *const hash, u8 params, u8 hashEndianess)
{
REG_NDMA2_SRC_ADDR = (u32)data;
REG_NDMA2_DST_ADDR = REG_SHA_INFIFO;
REG_NDMA2_TOTAL_CNT = size / 4;
REG_NDMA2_LOG_BLK_CNT = 64 / 4;
REG_NDMA2_INT_CNT = NDMA_INT_SYS_FREQ;
REG_NDMA2_CNT = NDMA_DST_UPDATE_INC | NDMA_DST_ADDR_RELOAD | NDMA_SRC_UPDATE_INC |
NDMA_BURST_WORDS(64 / 4) | NDMA_TOTAL_CNT_MODE | NDMA_STARTUP_SHA | NDMA_ENABLE;
SHA_start(params);
while(REG_NDMA2_CNT & NDMA_ENABLE);
while(REG_SHA_CNT & SHA_ENABLE);
SHA_finish(hash, hashEndianess);
}*/
//////////////////////////////////
// RSA //
//////////////////////////////////
#define RSA_REGS_BASE (IO_MEM_ARM9_ONLY + 0xB000)
#define REG_RSA_CNT *((vu32*)(RSA_REGS_BASE + 0x000))
#define REG_RSA_UNK_F0 *((vu32*)(RSA_REGS_BASE + 0x0F0))
#define REGs_RSA_SLOT0 ((vu32*)(RSA_REGS_BASE + 0x100))
#define REGs_RSA_SLOT1 ((vu32*)(RSA_REGS_BASE + 0x110))
#define REGs_RSA_SLOT2 ((vu32*)(RSA_REGS_BASE + 0x120))
#define REGs_RSA_SLOT3 ((vu32*)(RSA_REGS_BASE + 0x130))
#define REGs_RSA_EXP ((vu32*)(RSA_REGS_BASE + 0x200))
#define REGs_RSA_MOD ((vu32*)(RSA_REGS_BASE + 0x400))
#define REGs_RSA_TXT ((vu32*)(RSA_REGS_BASE + 0x800))
typedef struct
{
vu32 REG_RSA_SLOTCNT;
vu32 REG_RSA_SLOTSIZE;
vu32 REG_RSA_SLOTUNK_0x8;
vu32 REG_RSA_SLOTUNK_0xC;
} RsaSlot;
static inline void rsaWaitBusyIrq(void)
{
do
{
__wfi();
} while(REG_RSA_CNT & RSA_CNT_ENABLE);
}
void RSA_init(void)
{
REG_RSA_UNK_F0 = 0;
IRQ_registerHandler(IRQ_RSA, NULL);
}
void RSA_selectKeyslot(u8 keyslot)
{
fb_assert(keyslot < 4);
REG_RSA_CNT = keyslot<<RSA_CNT_KEYSLOT_SHIFT;
}
bool RSA_setKey2048(u8 keyslot, const u32 *const mod, u32 exp)
{
fb_assert(keyslot < 4);
fb_assert(mod != NULL);
RsaSlot *const slot = &((RsaSlot*)REGs_RSA_SLOT0)[keyslot];
if(slot->REG_RSA_SLOTCNT & RSA_SLOTCNT_WR_PROT) return false;
if(!(slot->REG_RSA_SLOTCNT & RSA_SLOTCNT_BIT31)) slot->REG_RSA_SLOTCNT &= ~RSA_SLOTCNT_SET;
REG_RSA_CNT = RSA_CNT_INPUT_NORMAL | RSA_CNT_INPUT_BIG | keyslot<<RSA_CNT_KEYSLOT_SHIFT;
iomemset(REGs_RSA_EXP, 0, 0x100 - 4);
REGs_RSA_EXP[(0x100>>2) - 1] = exp;
if(slot->REG_RSA_SLOTSIZE != RSA_SLOTSIZE_2048) return false;
iomemcpy(REGs_RSA_MOD, mod, 0x100);
return true;
}
bool RSA_decrypt2048(u32 *const decSig, const u32 *const encSig)
{
fb_assert(decSig != NULL);
fb_assert(encSig != NULL);
const u8 keyslot = (REG_RSA_CNT & RSA_CNT_KEYSLOT_MASK)>>RSA_CNT_KEYSLOT_SHIFT;
RsaSlot *const slot = &((RsaSlot*)REGs_RSA_SLOT0)[keyslot];
if(!(slot->REG_RSA_SLOTCNT & RSA_SLOTCNT_SET)) return false;
REG_RSA_CNT = (REG_RSA_CNT & ~RSA_CNT_INPUT_MASK) | RSA_CNT_INPUT_NORMAL | RSA_CNT_INPUT_BIG;
iomemcpy(REGs_RSA_TXT, encSig, 0x100);
REG_RSA_CNT |= RSA_CNT_IRQ_ENABLE | RSA_CNT_ENABLE;
rsaWaitBusyIrq();
iomemcpy(decSig, REGs_RSA_TXT, 0x100);
return true;
}
bool RSA_verify2048(const u32 *const encSig, const u32 *const data, u32 size)
{
fb_assert(encSig != NULL);
fb_assert(data != NULL);
alignas(4) u8 decSig[0x100];
if(!RSA_decrypt2048((u32*)decSig, encSig)) return false;
if(*((u16*)decSig) != 0x0100u) return false;
u32 read = 2;
while(decSig[read] == 0xFF && ++read < 0x100);
if(read != 0xCC || decSig[read] != 0x00) return false;
// ASN.1 is a clusterfuck so we skip parsing the remaining headers
// and hardcode the hash location.
u32 calcHash[8];
sha(data, size, calcHash, SHA_INPUT_BIG | SHA_MODE_256, SHA_OUTPUT_BIG);
// Compare hash
u32 res = 0;
for(u32 i = 0; i < 8; i++) res |= ((u32*)(decSig + 0xE0))[i] ^ calcHash[i];
return res == 0;
}
