/*
 * Copyright (c) 2019 Tom Marshall <tdm.code@gmail.com>
 *
 * 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 2
 * 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, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
 * 02110-1301, USA. 
 */

#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/bio.h>
#include <linux/device-mapper.h>
#include <linux/workqueue.h>
#include <linux/mutex.h>

#include <linux/lz4.h>

#include <linux/dm-compress.h>

/**************************************
 * Core memory management.
 **************************************/

struct page*
cbd_alloc_page(void)
{
    return alloc_page(GFP_KERNEL);
}

void
cbd_free_page(struct page* page)
{
    __free_page(page);
}

struct page*
cbd_alloc_pages(size_t len)
{
    return alloc_pages(GFP_KERNEL, get_order(len * PAGE_SIZE));
}

void
cbd_free_pages(struct page* pages, size_t len)
{
    __free_pages(pages, get_order(len * PAGE_SIZE));
}

bool
cbd_alloc_pagev(struct page** pagev, size_t len)
{
    size_t n;

    for (n = 0; n < len; ++n) {
        pagev[n] = cbd_alloc_page();
        if (!pagev[n]) {
            goto err;
        }
    }

    return true;

err:
    while (n--) {
        cbd_free_page(pagev[n]);
        pagev[n] = NULL;
    }
    return false;
}

void
cbd_free_pagev(struct page** pagev, size_t len)
{
    size_t n;

    for (n = 0; n < len; ++n) {
        cbd_free_page(pagev[n]);
        pagev[n] = NULL;
    }
}

/**************************************
 * Core low-level I/O.
 *
 * pblk count are in units of physical blocks (4096 bytes), NOT sectors.
 * data is a page address (obtained via __get_free_pages and friends).
 **************************************/

static struct bio*
pblk_io_prepare(struct cbd_params* params, unsigned int op,
        u64 pblk, u32 count, struct page** pagev)
{
    struct bio* bio;
    u32 n;

    bio = bio_alloc(GFP_KERNEL, count);
    if (!bio) {
        printk(KERN_ERR "%s: out of memory\n", __func__);
        return NULL;
    }
    bio_set_dev(bio, (struct block_device*)params->priv);
    bio->bi_opf = op;

    bio->bi_iter.bi_sector = (pblk << (PBLK_SHIFT - SECTOR_SHIFT));
    for (n = 0; n < count; ++n) {
        if (bio_add_page(bio, pagev[n], PAGE_SIZE, 0) != PAGE_SIZE) {
            BUG();
        }
    }

    return bio;
}

int
pblk_read_wait(struct cbd_params* params,
        u64 pblk, u32 count, struct page** pagev)
{
    int ret;
    struct bio* bio;

    bio = pblk_io_prepare(params, REQ_OP_READ, pblk, count, pagev);
    if (!bio) {
        printk(KERN_ERR "%s: out of memory\n", __func__);
        return -ENOMEM;
    }
    ret = submit_bio_wait(bio);
    if (ret) {
        printk(KERN_ERR "%s: submit_bio_wait failed: %d\n", __func__, ret);
    }
    bio_put(bio);

    return ret;
}

int
pblk_read(struct cbd_params* params,
        u64 pblk, u32 count, struct page** pagev,
        pblk_endio_t endio, void* endio_priv)
{
    int ret;
    struct bio* bio;

    bio = pblk_io_prepare(params, REQ_OP_READ, pblk, count, pagev);
    if (!bio) {
        printk(KERN_ERR "%s: out of memory\n", __func__);
        return -ENOMEM;
    }
    bio->bi_end_io = endio;
    bio->bi_private = endio_priv;

    ret = submit_bio(bio);
    if (ret != 0) {
        printk(KERN_ERR "%s: submit_bio_wait failed: %d\n", __func__, ret);
    }
    bio_put(bio);

    return ret;
}

void
pblk_write(struct cbd_params* params,
        u64 pblk, u32 count, struct page** pagev,
        pblk_endio_t endio, void* endio_priv)
{
    struct bio* bio;

    bio = pblk_io_prepare(params, REQ_OP_WRITE, pblk, count, pagev);
    if (!bio) {
        printk(KERN_ERR "%s: out of memory\n", __func__);
        return;
    }
    bio->bi_end_io = endio;
    bio->bi_private = endio_priv;

    if (pblk < CBD_HEADER_BLOCKS) {
        printk(KERN_ERR "%s: *** Attempt to write header\n", __func__);
        dump_stack();
        bio->bi_status = BLK_STS_IOERR;
        endio(bio);
        return;
    }

    submit_bio(bio);
}

int
pblk_endio(struct bio* bio)
{
    int ret;

    ret = blk_status_to_errno(bio->bi_status);
    bio_put(bio);
    return ret;
}

static inline u32
cshift (u32 x, uint n)
{
    return (x << n) | (x >> (32 - n));
}

#define A ctx->counter[0]
#define B ctx->counter[1]
#define C ctx->counter[2]
#define D ctx->counter[3]
#define X data

void
md5_init(struct md5* ctx)
{
    ctx->sz[0] = 0;
    ctx->sz[1] = 0;
    D = 0x10325476;
    C = 0x98badcfe;
    B = 0xefcdab89;
    A = 0x67452301;
}

#define F(x,y,z) ((x & y) | (~x & z))
#define G(x,y,z) ((x & z) | (y & ~z))
#define H(x,y,z) (x ^ y ^ z)
#define I(x,y,z) (y ^ (x | ~z))

#define DOIT(a,b,c,d,k,s,i,OP) \
a = b + cshift(a + OP(b,c,d) + X[k] + (i), s)

#define DO1(a,b,c,d,k,s,i) DOIT(a,b,c,d,k,s,i,F)
#define DO2(a,b,c,d,k,s,i) DOIT(a,b,c,d,k,s,i,G)
#define DO3(a,b,c,d,k,s,i) DOIT(a,b,c,d,k,s,i,H)
#define DO4(a,b,c,d,k,s,i) DOIT(a,b,c,d,k,s,i,I)

static inline void
calc(struct md5* ctx, u32* data)
{
    u32 AA, BB, CC, DD;

    AA = A;
    BB = B;
    CC = C;
    DD = D;

    /* Round 1 */

    DO1(A,B,C,D,0,7,0xd76aa478);
    DO1(D,A,B,C,1,12,0xe8c7b756);
    DO1(C,D,A,B,2,17,0x242070db);
    DO1(B,C,D,A,3,22,0xc1bdceee);

    DO1(A,B,C,D,4,7,0xf57c0faf);
    DO1(D,A,B,C,5,12,0x4787c62a);
    DO1(C,D,A,B,6,17,0xa8304613);
    DO1(B,C,D,A,7,22,0xfd469501);

    DO1(A,B,C,D,8,7,0x698098d8);
    DO1(D,A,B,C,9,12,0x8b44f7af);
    DO1(C,D,A,B,10,17,0xffff5bb1);
    DO1(B,C,D,A,11,22,0x895cd7be);

    DO1(A,B,C,D,12,7,0x6b901122);
    DO1(D,A,B,C,13,12,0xfd987193);
    DO1(C,D,A,B,14,17,0xa679438e);
    DO1(B,C,D,A,15,22,0x49b40821);

    /* Round 2 */

    DO2(A,B,C,D,1,5,0xf61e2562);
    DO2(D,A,B,C,6,9,0xc040b340);
    DO2(C,D,A,B,11,14,0x265e5a51);
    DO2(B,C,D,A,0,20,0xe9b6c7aa);

    DO2(A,B,C,D,5,5,0xd62f105d);
    DO2(D,A,B,C,10,9,0x2441453);
    DO2(C,D,A,B,15,14,0xd8a1e681);
    DO2(B,C,D,A,4,20,0xe7d3fbc8);

    DO2(A,B,C,D,9,5,0x21e1cde6);
    DO2(D,A,B,C,14,9,0xc33707d6);
    DO2(C,D,A,B,3,14,0xf4d50d87);
    DO2(B,C,D,A,8,20,0x455a14ed);

    DO2(A,B,C,D,13,5,0xa9e3e905);
    DO2(D,A,B,C,2,9,0xfcefa3f8);
    DO2(C,D,A,B,7,14,0x676f02d9);
    DO2(B,C,D,A,12,20,0x8d2a4c8a);

    /* Round 3 */

    DO3(A,B,C,D,5,4,0xfffa3942);
    DO3(D,A,B,C,8,11,0x8771f681);
    DO3(C,D,A,B,11,16,0x6d9d6122);
    DO3(B,C,D,A,14,23,0xfde5380c);

    DO3(A,B,C,D,1,4,0xa4beea44);
    DO3(D,A,B,C,4,11,0x4bdecfa9);
    DO3(C,D,A,B,7,16,0xf6bb4b60);
    DO3(B,C,D,A,10,23,0xbebfbc70);

    DO3(A,B,C,D,13,4,0x289b7ec6);
    DO3(D,A,B,C,0,11,0xeaa127fa);
    DO3(C,D,A,B,3,16,0xd4ef3085);
    DO3(B,C,D,A,6,23,0x4881d05);

    DO3(A,B,C,D,9,4,0xd9d4d039);
    DO3(D,A,B,C,12,11,0xe6db99e5);
    DO3(C,D,A,B,15,16,0x1fa27cf8);
    DO3(B,C,D,A,2,23,0xc4ac5665);

    /* Round 4 */

    DO4(A,B,C,D,0,6,0xf4292244);
    DO4(D,A,B,C,7,10,0x432aff97);
    DO4(C,D,A,B,14,15,0xab9423a7);
    DO4(B,C,D,A,5,21,0xfc93a039);

    DO4(A,B,C,D,12,6,0x655b59c3);
    DO4(D,A,B,C,3,10,0x8f0ccc92);
    DO4(C,D,A,B,10,15,0xffeff47d);
    DO4(B,C,D,A,1,21,0x85845dd1);

    DO4(A,B,C,D,8,6,0x6fa87e4f);
    DO4(D,A,B,C,15,10,0xfe2ce6e0);
    DO4(C,D,A,B,6,15,0xa3014314);
    DO4(B,C,D,A,13,21,0x4e0811a1);

    DO4(A,B,C,D,4,6,0xf7537e82);
    DO4(D,A,B,C,11,10,0xbd3af235);
    DO4(C,D,A,B,2,15,0x2ad7d2bb);
    DO4(B,C,D,A,9,21,0xeb86d391);

    A += AA;
    B += BB;
    C += CC;
    D += DD;
}

/*
 * From `Performance analysis of MD5' by Joseph D. Touch <touch@isi.edu>
 */
#if !defined(__BYTE_ORDER__) || !defined (__ORDER_BIG_ENDIAN__)
#error __BYTE_ORDER macros not defined
#endif

#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
static inline u32
swap_u32(u32 t)
{
    u32 temp1, temp2;

    temp1   = cshift(t, 16);
    temp2   = temp1 >> 8;
    temp1  &= 0x00ff00ff;
    temp2  &= 0x00ff00ff;
    temp1 <<= 8;
    return temp1 | temp2;
}
#endif

struct x32 {
    uint a:32;
    uint b:32;
};

void
md5_update(struct md5* ctx, const void* data, size_t len)
{
    const byte* p = data;
    size_t old_sz = ctx->sz[0];
    size_t offset;

    ctx->sz[0] += len * 8;
    if (ctx->sz[0] < old_sz)
        ++ctx->sz[1];
    offset = (old_sz / 8)  % 64;
    while(len > 0){
        size_t l = min(len, 64 - offset);
        memcpy(ctx->save + offset, p, l);
        offset += l;
        p += l;
        len -= l;
        if (offset == 64) {
#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
            int i;
            u32 current[16];
            struct x32 *u = (struct x32*)ctx->save;
            for (i = 0; i < 8; i++) {
                current[2*i+0] = swap_u32(u[i].a);
                current[2*i+1] = swap_u32(u[i].b);
            }
            calc(ctx, current);
#else
            calc(ctx, (u32*)ctx->save);
#endif
            offset = 0;
        }
    }
}

void
md5_final(struct md5* ctx, byte* buf)
{
    byte zeros[72];
    uint offset = (ctx->sz[0] / 8) % 64;
    uint dstart = (120 - offset - 1) % 64 + 1;

    *zeros = 0x80;
    memset (zeros + 1, 0, sizeof(zeros) - 1);
    zeros[dstart+0] = (ctx->sz[0] >> 0) & 0xff;
    zeros[dstart+1] = (ctx->sz[0] >> 8) & 0xff;
    zeros[dstart+2] = (ctx->sz[0] >> 16) & 0xff;
    zeros[dstart+3] = (ctx->sz[0] >> 24) & 0xff;
    zeros[dstart+4] = (ctx->sz[1] >> 0) & 0xff;
    zeros[dstart+5] = (ctx->sz[1] >> 8) & 0xff;
    zeros[dstart+6] = (ctx->sz[1] >> 16) & 0xff;
    zeros[dstart+7] = (ctx->sz[1] >> 24) & 0xff;
    md5_update(ctx, zeros, dstart + 8);
    {
        int i;
        byte *r = buf;

        for (i = 0; i < 4; ++i) {
            r[4*i]   = ctx->counter[i] & 0xFF;
            r[4*i+1] = (ctx->counter[i] >> 8) & 0xFF;
            r[4*i+2] = (ctx->counter[i] >> 16) & 0xFF;
            r[4*i+3] = (ctx->counter[i] >> 24) & 0xFF;
        }
    }
}