[wimlib] / src / lzms-compress.c

/*
 * lzms-compress.c
 */

/*
 * Copyright (C) 2013 Eric Biggers
 *
 * This file is part of wimlib, a library for working with WIM files.
 *
 * wimlib 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.
 *
 * wimlib 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 wimlib; if not, see http://www.gnu.org/licenses/.
 */

/* This a compressor for the LZMS compression format.  More details about this
 * format can be found in lzms-decompress.c.
 *
 * This is currently an unsophisticated implementation that is fast but does not
 * attain the best compression ratios allowed by the format.
 */

#ifdef HAVE_CONFIG_H
#  include "config.h"
#endif

#include "wimlib.h"
#include "wimlib/compiler.h"
#include "wimlib/compressor_ops.h"
#include "wimlib/compress_common.h"
#include "wimlib/endianness.h"
#include "wimlib/error.h"
#include "wimlib/lz_hash.h"
#include "wimlib/lz_sarray.h"
#include "wimlib/lzms.h"
#include "wimlib/util.h"

#include <string.h>
#include <limits.h>
#include <pthread.h>

#define LZMS_OPTIM_ARRAY_SIZE   1024

struct lzms_compressor;
struct lzms_adaptive_state {
        struct lzms_lz_lru_queues lru;
        u8 main_state;
        u8 match_state;
        u8 lz_match_state;
        u8 lz_repeat_match_state[LZMS_NUM_RECENT_OFFSETS - 1];
};
#define LZ_ADAPTIVE_STATE struct lzms_adaptive_state
#define LZ_COMPRESSOR     struct lzms_compressor
#include "wimlib/lz_optimal.h"

/* Stucture used for writing raw bits to the end of the LZMS-compressed data as
 * a series of 16-bit little endian coding units.  */
struct lzms_output_bitstream {
        /* Buffer variable containing zero or more bits that have been logically
         * written to the bitstream but not yet written to memory.  This must be
         * at least as large as the coding unit size.  */
        u16 bitbuf;

        /* Number of bits in @bitbuf that are valid.  */
        unsigned num_free_bits;

        /* Pointer to one past the next position in the compressed data buffer
         * at which to output a 16-bit coding unit.  */
        le16 *out;

        /* Maximum number of 16-bit coding units that can still be output to
         * the compressed data buffer.  */
        size_t num_le16_remaining;

        /* Set to %true if not all coding units could be output due to
         * insufficient space.  */
        bool overrun;
};

/* Stucture used for range encoding (raw version).  */
struct lzms_range_encoder_raw {

        /* A 33-bit variable that holds the low boundary of the current range.
         * The 33rd bit is needed to catch carries.  */
        u64 low;

        /* Size of the current range.  */
        u32 range;

        /* Next 16-bit coding unit to output.  */
        u16 cache;

        /* Number of 16-bit coding units whose output has been delayed due to
         * possible carrying.  The first such coding unit is @cache; all
         * subsequent such coding units are 0xffff.  */
        u32 cache_size;

        /* Pointer to the next position in the compressed data buffer at which
         * to output a 16-bit coding unit.  */
        le16 *out;

        /* Maximum number of 16-bit coding units that can still be output to
         * the compressed data buffer.  */
        size_t num_le16_remaining;

        /* %true when the very first coding unit has not yet been output.  */
        bool first;

        /* Set to %true if not all coding units could be output due to
         * insufficient space.  */
        bool overrun;
};

/* Structure used for range encoding.  This wraps around `struct
 * lzms_range_encoder_raw' to use and maintain probability entries.  */
struct lzms_range_encoder {
        /* Pointer to the raw range encoder, which has no persistent knowledge
         * of probabilities.  Multiple lzms_range_encoder's share the same
         * lzms_range_encoder_raw.  */
        struct lzms_range_encoder_raw *rc;

        /* Bits recently encoded by this range encoder.  This are used as in
         * index into @prob_entries.  */
        u32 state;

        /* Bitmask for @state to prevent its value from exceeding the number of
         * probability entries.  */
        u32 mask;

        /* Probability entries being used for this range encoder.  */
        struct lzms_probability_entry prob_entries[LZMS_MAX_NUM_STATES];
};

/* Structure used for Huffman encoding.  */
struct lzms_huffman_encoder {

        /* Bitstream to write Huffman-encoded symbols and verbatim bits to.
         * Multiple lzms_huffman_encoder's share the same lzms_output_bitstream.
         */
        struct lzms_output_bitstream *os;

        /* Number of symbols that have been written using this code far.  Reset
         * to 0 whenever the code is rebuilt.  */
        u32 num_syms_written;

        /* When @num_syms_written reaches this number, the Huffman code must be
         * rebuilt.  */
        u32 rebuild_freq;

        /* Number of symbols in the represented Huffman code.  */
        unsigned num_syms;

        /* Running totals of symbol frequencies.  These are diluted slightly
         * whenever the code is rebuilt.  */
        u32 sym_freqs[LZMS_MAX_NUM_SYMS];

        /* The length, in bits, of each symbol in the Huffman code.  */
        u8 lens[LZMS_MAX_NUM_SYMS];

        /* The codeword of each symbol in the Huffman code.  */
        u16 codewords[LZMS_MAX_NUM_SYMS];
};

/* State of the LZMS compressor.  */
struct lzms_compressor {
        /* Pointer to a buffer holding the preprocessed data to compress.  */
        u8 *window;

        /* Current position in @buffer.  */
        u32 cur_window_pos;

        /* Size of the data in @buffer.  */
        u32 window_size;

#if 0
        /* Temporary array used by lz_analyze_block(); must be at least as long
         * as the window.  */
        u32 *prev_tab;
#endif

        /* Suffix array match-finder.  */
        struct lz_sarray lz_sarray;

        /* Temporary space to store found matches.  */
        struct raw_match *matches;

        /* Match-chooser.  */
        struct lz_match_chooser mc;

        /* Maximum block size this compressor instantiation allows.  This is the
         * allocated size of @window.  */
        u32 max_block_size;

        /* Raw range encoder which outputs to the beginning of the compressed
         * data buffer, proceeding forwards.  */
        struct lzms_range_encoder_raw rc;

        /* Bitstream which outputs to the end of the compressed data buffer,
         * proceeding backwards.  */
        struct lzms_output_bitstream os;

        /* Range encoders.  */
        struct lzms_range_encoder main_range_encoder;
        struct lzms_range_encoder match_range_encoder;
        struct lzms_range_encoder lz_match_range_encoder;
        struct lzms_range_encoder lz_repeat_match_range_encoders[LZMS_NUM_RECENT_OFFSETS - 1];
        struct lzms_range_encoder delta_match_range_encoder;
        struct lzms_range_encoder delta_repeat_match_range_encoders[LZMS_NUM_RECENT_OFFSETS - 1];

        /* Huffman encoders.  */
        struct lzms_huffman_encoder literal_encoder;
        struct lzms_huffman_encoder lz_offset_encoder;
        struct lzms_huffman_encoder length_encoder;
        struct lzms_huffman_encoder delta_power_encoder;
        struct lzms_huffman_encoder delta_offset_encoder;

        /* LRU (least-recently-used) queues for match information.  */
        struct lzms_lru_queues lru;

        /* Used for preprocessing.  */
        s32 last_target_usages[65536];
};

/* Initialize the output bitstream @os to write forwards to the specified
 * compressed data buffer @out that is @out_limit 16-bit integers long.  */
static void
lzms_output_bitstream_init(struct lzms_output_bitstream *os,
                           le16 *out, size_t out_limit)
{
        os->bitbuf = 0;
        os->num_free_bits = 16;
        os->out = out + out_limit;
        os->num_le16_remaining = out_limit;
        os->overrun = false;
}

/* Write @num_bits bits, contained in the low @num_bits bits of @bits (ordered
 * from high-order to low-order), to the output bitstream @os.  */
static void
lzms_output_bitstream_put_bits(struct lzms_output_bitstream *os,
                               u32 bits, unsigned num_bits)
{
        bits &= (1U << num_bits) - 1;

        while (num_bits > os->num_free_bits) {

                if (unlikely(os->num_le16_remaining == 0)) {
                        os->overrun = true;
                        return;
                }

                unsigned num_fill_bits = os->num_free_bits;

                os->bitbuf <<= num_fill_bits;
                os->bitbuf |= bits >> (num_bits - num_fill_bits);

                *--os->out = cpu_to_le16(os->bitbuf);
                --os->num_le16_remaining;

                os->num_free_bits = 16;
                num_bits -= num_fill_bits;
                bits &= (1U << num_bits) - 1;
        }
        os->bitbuf <<= num_bits;
        os->bitbuf |= bits;
        os->num_free_bits -= num_bits;
}

/* Flush the output bitstream, ensuring that all bits written to it have been
 * written to memory.  Returns %true if all bits were output successfully, or
 * %false if an overrun occurred.  */
static bool
lzms_output_bitstream_flush(struct lzms_output_bitstream *os)
{
        if (os->num_free_bits != 16)
                lzms_output_bitstream_put_bits(os, 0, os->num_free_bits + 1);
        return !os->overrun;
}

/* Initialize the range encoder @rc to write forwards to the specified
 * compressed data buffer @out that is @out_limit 16-bit integers long.  */
static void
lzms_range_encoder_raw_init(struct lzms_range_encoder_raw *rc,
                            le16 *out, size_t out_limit)
{
        rc->low = 0;
        rc->range = 0xffffffff;
        rc->cache = 0;
        rc->cache_size = 1;
        rc->out = out;
        rc->num_le16_remaining = out_limit;
        rc->first = true;
        rc->overrun = false;
}

/*
 * Attempt to flush bits from the range encoder.
 *
 * Note: this is based on the public domain code for LZMA written by Igor
 * Pavlov.  The only differences in this function are that in LZMS the bits must
 * be output in 16-bit coding units instead of 8-bit coding units, and that in
 * LZMS the first coding unit is not ignored by the decompressor, so the encoder
 * cannot output a dummy value to that position.
 *
 * The basic idea is that we're writing bits from @rc->low to the output.
 * However, due to carrying, the writing of coding units with value 0xffff, as
 * well as one prior coding unit, must be delayed until it is determined whether
 * a carry is needed.
 */
static void
lzms_range_encoder_raw_shift_low(struct lzms_range_encoder_raw *rc)
{
        LZMS_DEBUG("low=%"PRIx64", cache=%"PRIx64", cache_size=%u",
                   rc->low, rc->cache, rc->cache_size);
        if ((u32)(rc->low) < 0xffff0000 ||
            (u32)(rc->low >> 32) != 0)
        {
                /* Carry not needed (rc->low < 0xffff0000), or carry occurred
                 * ((rc->low >> 32) != 0, a.k.a. the carry bit is 1).  */
                do {
                        if (!rc->first) {
                                if (rc->num_le16_remaining == 0) {
                                        rc->overrun = true;
                                        return;
                                }
                                *rc->out++ = cpu_to_le16(rc->cache +
                                                         (u16)(rc->low >> 32));
                                --rc->num_le16_remaining;
                        } else {
                                rc->first = false;
                        }

                        rc->cache = 0xffff;
                } while (--rc->cache_size != 0);

                rc->cache = (rc->low >> 16) & 0xffff;
        }
        ++rc->cache_size;
        rc->low = (rc->low & 0xffff) << 16;
}

static void
lzms_range_encoder_raw_normalize(struct lzms_range_encoder_raw *rc)
{
        if (rc->range <= 0xffff) {
                rc->range <<= 16;
                lzms_range_encoder_raw_shift_low(rc);
        }
}

static bool
lzms_range_encoder_raw_flush(struct lzms_range_encoder_raw *rc)
{
        for (unsigned i = 0; i < 4; i++)
                lzms_range_encoder_raw_shift_low(rc);
        return !rc->overrun;
}

/* Encode the next bit using the range encoder (raw version).
 *
 * @prob is the chance out of LZMS_PROBABILITY_MAX that the next bit is 0.  */
static void
lzms_range_encoder_raw_encode_bit(struct lzms_range_encoder_raw *rc, int bit,
                                  u32 prob)
{
        lzms_range_encoder_raw_normalize(rc);

        u32 bound = (rc->range >> LZMS_PROBABILITY_BITS) * prob;
        if (bit == 0) {
                rc->range = bound;
        } else {
                rc->low += bound;
                rc->range -= bound;
        }
}

/* Encode a bit using the specified range encoder. This wraps around
 * lzms_range_encoder_raw_encode_bit() to handle using and updating the
 * appropriate probability table.  */
static void
lzms_range_encode_bit(struct lzms_range_encoder *enc, int bit)
{
        struct lzms_probability_entry *prob_entry;
        u32 prob;

        /* Load the probability entry corresponding to the current state.  */
        prob_entry = &enc->prob_entries[enc->state];

        /* Treat the number of zero bits in the most recently encoded
         * LZMS_PROBABILITY_MAX bits with this probability entry as the chance,
         * out of LZMS_PROBABILITY_MAX, that the next bit will be a 0.  However,
         * don't allow 0% or 100% probabilities.  */
        prob = prob_entry->num_recent_zero_bits;
        if (prob == 0)
                prob = 1;
        else if (prob == LZMS_PROBABILITY_MAX)
                prob = LZMS_PROBABILITY_MAX - 1;

        /* Encode the next bit.  */
        lzms_range_encoder_raw_encode_bit(enc->rc, bit, prob);

        /* Update the state based on the newly encoded bit.  */
        enc->state = ((enc->state << 1) | bit) & enc->mask;

        /* Update the recent bits, including the cached count of 0's.  */
        BUILD_BUG_ON(LZMS_PROBABILITY_MAX > sizeof(prob_entry->recent_bits) * 8);
        if (bit == 0) {
                if (prob_entry->recent_bits & (1ULL << (LZMS_PROBABILITY_MAX - 1))) {
                        /* Replacing 1 bit with 0 bit; increment the zero count.
                         */
                        prob_entry->num_recent_zero_bits++;
                }
        } else {
                if (!(prob_entry->recent_bits & (1ULL << (LZMS_PROBABILITY_MAX - 1)))) {
                        /* Replacing 0 bit with 1 bit; decrement the zero count.
                         */
                        prob_entry->num_recent_zero_bits--;
                }
        }
        prob_entry->recent_bits = (prob_entry->recent_bits << 1) | bit;
}

/* Encode a symbol using the specified Huffman encoder.  */
static void
lzms_huffman_encode_symbol(struct lzms_huffman_encoder *enc, u32 sym)
{
        LZMS_ASSERT(sym < enc->num_syms);
        lzms_output_bitstream_put_bits(enc->os,
                                       enc->codewords[sym],
                                       enc->lens[sym]);
        ++enc->sym_freqs[sym];
        if (++enc->num_syms_written == enc->rebuild_freq) {
                /* Adaptive code needs to be rebuilt.  */
                LZMS_DEBUG("Rebuilding code (num_syms=%u)", enc->num_syms);
                make_canonical_huffman_code(enc->num_syms,
                                            LZMS_MAX_CODEWORD_LEN,
                                            enc->sym_freqs,
                                            enc->lens,
                                            enc->codewords);

                /* Dilute the frequencies.  */
                for (unsigned i = 0; i < enc->num_syms; i++) {
                        enc->sym_freqs[i] >>= 1;
                        enc->sym_freqs[i] += 1;
                }
                enc->num_syms_written = 0;
        }
}

static void
lzms_encode_length(struct lzms_huffman_encoder *enc, u32 length)
{
        unsigned slot;
        unsigned num_extra_bits;
        u32 extra_bits;

        slot = lzms_get_length_slot(length);

        num_extra_bits = lzms_extra_length_bits[slot];

        extra_bits = length - lzms_length_slot_base[slot];

        lzms_huffman_encode_symbol(enc, slot);
        lzms_output_bitstream_put_bits(enc->os, extra_bits, num_extra_bits);
}

static void
lzms_encode_offset(struct lzms_huffman_encoder *enc, u32 offset)
{
        unsigned slot;
        unsigned num_extra_bits;
        u32 extra_bits;

        slot = lzms_get_position_slot(offset);

        num_extra_bits = lzms_extra_position_bits[slot];

        extra_bits = offset - lzms_position_slot_base[slot];

        lzms_huffman_encode_symbol(enc, slot);
        lzms_output_bitstream_put_bits(enc->os, extra_bits, num_extra_bits);
}

static void
lzms_begin_encode_item(struct lzms_compressor *ctx)
{
        ctx->lru.lz.upcoming_offset = 0;
        ctx->lru.delta.upcoming_offset = 0;
        ctx->lru.delta.upcoming_power = 0;
}

static void
lzms_end_encode_item(struct lzms_compressor *ctx, u32 length)
{
        LZMS_ASSERT(ctx->window_size - ctx->cur_window_pos >= length);
        ctx->cur_window_pos += length;
        lzms_update_lru_queues(&ctx->lru);
}

/* Encode a literal byte.  */
static void
lzms_encode_literal(struct lzms_compressor *ctx, u8 literal)
{
        LZMS_DEBUG("Position %u: Encoding literal 0x%02x ('%c')",
                   ctx->cur_window_pos, literal, literal);

        lzms_begin_encode_item(ctx);

        /* Main bit: 0 = a literal, not a match.  */
        lzms_range_encode_bit(&ctx->main_range_encoder, 0);

        /* Encode the literal using the current literal Huffman code.  */
        lzms_huffman_encode_symbol(&ctx->literal_encoder, literal);

        lzms_end_encode_item(ctx, 1);
}

/* Encode a (length, offset) pair (LZ match).  */
static void
lzms_encode_lz_match(struct lzms_compressor *ctx, u32 length, u32 offset)
{
        int recent_offset_idx;

        LZMS_DEBUG("Position %u: Encoding LZ match {length=%u, offset=%u}",
                   ctx->cur_window_pos, length, offset);

        LZMS_ASSERT(length <= ctx->window_size - ctx->cur_window_pos);
        LZMS_ASSERT(offset <= ctx->cur_window_pos);
        LZMS_ASSERT(!memcmp(&ctx->window[ctx->cur_window_pos],
                            &ctx->window[ctx->cur_window_pos - offset],
                            length));

        lzms_begin_encode_item(ctx);

        /* Main bit: 1 = a match, not a literal.  */
        lzms_range_encode_bit(&ctx->main_range_encoder, 1);

        /* Match bit: 0 = a LZ match, not a delta match.  */
        lzms_range_encode_bit(&ctx->match_range_encoder, 0);

        /* Determine if the offset can be represented as a recent offset.  */
        for (recent_offset_idx = 0;
             recent_offset_idx < LZMS_NUM_RECENT_OFFSETS;
             recent_offset_idx++)
                if (offset == ctx->lru.lz.recent_offsets[recent_offset_idx])
                        break;

        if (recent_offset_idx == LZMS_NUM_RECENT_OFFSETS) {
                /* Explicit offset.  */

                /* LZ match bit: 0 = explicit offset, not a recent offset.  */
                lzms_range_encode_bit(&ctx->lz_match_range_encoder, 0);

                /* Encode the match offset.  */
                lzms_encode_offset(&ctx->lz_offset_encoder, offset);
        } else {
                int i;

                /* Recent offset.  */

                /* LZ match bit: 1 = recent offset, not an explicit offset.  */
                lzms_range_encode_bit(&ctx->lz_match_range_encoder, 1);

                /* Encode the recent offset index.  A 1 bit is encoded for each
                 * index passed up.  This sequence of 1 bits is terminated by a
                 * 0 bit, or automatically when (LZMS_NUM_RECENT_OFFSETS - 1) 1
                 * bits have been encoded.  */
                for (i = 0; i < recent_offset_idx; i++)
                        lzms_range_encode_bit(&ctx->lz_repeat_match_range_encoders[i], 1);

                if (i < LZMS_NUM_RECENT_OFFSETS - 1)
                        lzms_range_encode_bit(&ctx->lz_repeat_match_range_encoders[i], 0);

                /* Initial update of the LZ match offset LRU queue.  */
                for (; i < LZMS_NUM_RECENT_OFFSETS; i++)
                        ctx->lru.lz.recent_offsets[i] = ctx->lru.lz.recent_offsets[i + 1];
        }

        /* Encode the match length.  */
        lzms_encode_length(&ctx->length_encoder, length);

        /* Save the match offset for later insertion at the front of the LZ
         * match offset LRU queue.  */
        ctx->lru.lz.upcoming_offset = offset;

        lzms_end_encode_item(ctx, length);
}

#if 0
static void
lzms_record_literal(u8 literal, void *_ctx)
{
        struct lzms_compressor *ctx = _ctx;

        lzms_encode_literal(ctx, literal);
}

static void
lzms_record_match(unsigned length, unsigned offset, void *_ctx)
{
        struct lzms_compressor *ctx = _ctx;

        lzms_encode_lz_match(ctx, length, offset);
}

static void
lzms_fast_encode(struct lzms_compressor *ctx)
{
        static const struct lz_params lzms_lz_params = {
                .min_match      = 3,
                .max_match      = UINT_MAX,
                .max_offset     = UINT_MAX,
                .nice_match     = 64,
                .good_match     = 32,
                .max_chain_len  = 64,
                .max_lazy_match = 258,
                .too_far        = 4096,
        };

        lz_analyze_block(ctx->window,
                         ctx->window_size,
                         lzms_record_match,
                         lzms_record_literal,
                         ctx,
                         &lzms_lz_params,
                         ctx->prev_tab);

}
#endif

/* Fast heuristic cost evaluation to use in the inner loop of the match-finder.
 * Unlike lzms_get_lz_match_cost(), which does a true cost evaluation, this
 * simply prioritize matches based on their offset.  */
static input_idx_t
lzms_lz_match_cost_fast(input_idx_t length, input_idx_t offset, const void *_lru)
{
        const struct lzms_lz_lru_queues *lru = _lru;

        for (input_idx_t i = 0; i < LZMS_NUM_RECENT_OFFSETS; i++)
                if (offset == lru->recent_offsets[i])
                        return i;

        return offset;
}

#define LZMS_COST_SHIFT 5

/*#define LZMS_RC_COSTS_USE_FLOATING_POINT*/

static u32
lzms_rc_costs[LZMS_PROBABILITY_MAX + 1];

#ifdef LZMS_RC_COSTS_USE_FLOATING_POINT
#  include <math.h>
#endif

static void
lzms_do_init_rc_costs(void)
{
        /* Fill in a table that maps range coding probabilities needed to code a
         * bit X (0 or 1) to the number of bits (scaled by a constant factor, to
         * handle fractional costs) needed to code that bit X.
         *
         * Consider the range of the range decoder.  To eliminate exactly half
         * the range (logical probability of 0.5), we need exactly 1 bit.  For
         * lower probabilities we need more bits and for higher probabilities we
         * need fewer bits.  In general, a logical probability of N will
         * eliminate the proportion 1 - N of the range; this information takes
         * log2(1 / N) bits to encode.
         *
         * The below loop is simply calculating this number of bits for each
         * possible probability allowed by the LZMS compression format, but
         * without using real numbers.  To handle fractional probabilities, each
         * cost is multiplied by (1 << LZMS_COST_SHIFT).  These techniques are
         * based on those used by LZMA.
         *
         * Note that in LZMS, a probability x really means x / 64, and 0 / 64 is
         * really interpreted as 1 / 64 and 64 / 64 is really interpreted as
         * 63 / 64.
         */
        for (u32 i = 0; i <= LZMS_PROBABILITY_MAX; i++) {
                u32 prob = i;

                if (prob == 0)
                        prob = 1;
                else if (prob == LZMS_PROBABILITY_MAX)
                        prob = LZMS_PROBABILITY_MAX - 1;

        #ifdef LZMS_RC_COSTS_USE_FLOATING_POINT
                lzms_rc_costs[i] = log2((double)LZMS_PROBABILITY_MAX / prob) *
                                        (1 << LZMS_COST_SHIFT);
        #else
                u32 w = prob;
                u32 bit_count = 0;
                for (u32 j = 0; j < LZMS_COST_SHIFT; j++) {
                        w *= w;
                        bit_count <<= 1;
                        while (w >= (1U << 16)) {
                                w >>= 1;
                                ++bit_count;
                        }
                }
                lzms_rc_costs[i] = (LZMS_PROBABILITY_BITS << LZMS_COST_SHIFT) -
                                   (15 + bit_count);
        #endif
        }
}

static void
lzms_init_rc_costs(void)
{
        static bool done = false;
        static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;

        if (unlikely(!done)) {
                pthread_mutex_lock(&mutex);
                if (!done) {
                        lzms_do_init_rc_costs();
                        done = true;
                }
                pthread_mutex_unlock(&mutex);
        }
}

/*
 * Return the cost to range-encode the specified bit when in the specified
 * state.
 *
 * @enc         The range encoder to use.
 * @cur_state   Current state, which indicates the probability entry to choose.
 *              Updated by this function.
 * @bit         The bit to encode (0 or 1).
 */
static u32
lzms_rc_bit_cost(const struct lzms_range_encoder *enc, u8 *cur_state, int bit)
{
        u32 prob_zero;
        u32 prob_correct;

        prob_zero = enc->prob_entries[*cur_state & enc->mask].num_recent_zero_bits;

        *cur_state = (*cur_state << 1) | bit;

        if (bit == 0)
                prob_correct = prob_zero;
        else
                prob_correct = LZMS_PROBABILITY_MAX - prob_zero;

        return lzms_rc_costs[prob_correct];
}

static u32
lzms_huffman_symbol_cost(const struct lzms_huffman_encoder *enc, u32 sym)
{
        return enc->lens[sym] << LZMS_COST_SHIFT;
}

static u32
lzms_offset_cost(const struct lzms_huffman_encoder *enc, u32 offset)
{
        u32 slot;
        u32 num_extra_bits;
        u32 cost = 0;

        slot = lzms_get_position_slot(offset);

        cost += lzms_huffman_symbol_cost(enc, slot);

        num_extra_bits = lzms_extra_position_bits[slot];

        cost += num_extra_bits << LZMS_COST_SHIFT;

        return cost;
}

static u32
lzms_length_cost(const struct lzms_huffman_encoder *enc, u32 length)
{
        u32 slot;
        u32 num_extra_bits;
        u32 cost = 0;

        slot = lzms_get_length_slot(length);

        cost += lzms_huffman_symbol_cost(enc, slot);

        num_extra_bits = lzms_extra_length_bits[slot];

        cost += num_extra_bits << LZMS_COST_SHIFT;

        return cost;
}

static u32
lzms_get_matches(struct lzms_compressor *ctx,
                 const struct lzms_adaptive_state *state,
                 struct raw_match **matches_ret)
{
        *matches_ret = ctx->matches;
        return lz_sarray_get_matches(&ctx->lz_sarray,
                                     ctx->matches,
                                     lzms_lz_match_cost_fast,
                                     &state->lru);
}

static void
lzms_skip_bytes(struct lzms_compressor *ctx, input_idx_t n)
{
        while (n--)
                lz_sarray_skip_position(&ctx->lz_sarray);
}

static u32
lzms_get_prev_literal_cost(struct lzms_compressor *ctx,
                           struct lzms_adaptive_state *state)
{
        u8 literal = ctx->window[lz_sarray_get_pos(&ctx->lz_sarray) - 1];
        u32 cost = 0;

        state->lru.upcoming_offset = 0;
        lzms_update_lz_lru_queues(&state->lru);

        cost += lzms_rc_bit_cost(&ctx->main_range_encoder,
                                 &state->main_state, 0);

        cost += lzms_huffman_symbol_cost(&ctx->literal_encoder, literal);

        return cost;
}

static u32
lzms_get_lz_match_cost(struct lzms_compressor *ctx,
                       struct lzms_adaptive_state *state,
                       input_idx_t length, input_idx_t offset)
{
        u32 cost = 0;
        int recent_offset_idx;

        cost += lzms_rc_bit_cost(&ctx->main_range_encoder,
                                 &state->main_state, 1);
        cost += lzms_rc_bit_cost(&ctx->match_range_encoder,
                                 &state->match_state, 0);

        for (recent_offset_idx = 0;
             recent_offset_idx < LZMS_NUM_RECENT_OFFSETS;
             recent_offset_idx++)
                if (offset == state->lru.recent_offsets[recent_offset_idx])
                        break;

        if (recent_offset_idx == LZMS_NUM_RECENT_OFFSETS) {
                /* Explicit offset.  */
                cost += lzms_rc_bit_cost(&ctx->lz_match_range_encoder,
                                         &state->lz_match_state, 0);

                cost += lzms_offset_cost(&ctx->lz_offset_encoder, offset);
        } else {
                int i;

                /* Recent offset.  */
                cost += lzms_rc_bit_cost(&ctx->lz_match_range_encoder,
                                         &state->lz_match_state, 1);

                for (i = 0; i < recent_offset_idx; i++)
                        cost += lzms_rc_bit_cost(&ctx->lz_repeat_match_range_encoders[i],
                                                 &state->lz_repeat_match_state[i], 0);

                if (i < LZMS_NUM_RECENT_OFFSETS - 1)
                        cost += lzms_rc_bit_cost(&ctx->lz_repeat_match_range_encoders[i],
                                                 &state->lz_repeat_match_state[i], 1);


                /* Initial update of the LZ match offset LRU queue.  */
                for (; i < LZMS_NUM_RECENT_OFFSETS; i++)
                        state->lru.recent_offsets[i] = state->lru.recent_offsets[i + 1];
        }

        cost += lzms_length_cost(&ctx->length_encoder, length);

        state->lru.upcoming_offset = offset;
        lzms_update_lz_lru_queues(&state->lru);

        return cost;
}

static struct raw_match
lzms_get_near_optimal_match(struct lzms_compressor *ctx)
{
        struct lzms_adaptive_state initial_state;

        initial_state.lru = ctx->lru.lz;
        initial_state.main_state = ctx->main_range_encoder.state;
        initial_state.match_state = ctx->match_range_encoder.state;
        initial_state.lz_match_state = ctx->lz_match_range_encoder.state;
        for (int i = 0; i < LZMS_NUM_RECENT_OFFSETS - 1; i++)
                initial_state.lz_repeat_match_state[i] =
                        ctx->lz_repeat_match_range_encoders[i].state;
        return lz_get_near_optimal_match(&ctx->mc,
                                         lzms_get_matches,
                                         lzms_skip_bytes,
                                         lzms_get_prev_literal_cost,
                                         lzms_get_lz_match_cost,
                                         ctx,
                                         &initial_state);
}

/*
 * The main loop for the LZMS compressor.
 *
 * Notes:
 *
 * - This uses near-optimal LZ parsing backed by a suffix-array match-finder.
 *   More details can be found in the corresponding files (lz_optimal.h,
 *   lz_sarray.{h,c}).
 *
 * - This does not output any delta matches.  It would take a specialized
 *   algorithm to find them, then more code in lz_optimal.h and here to handle
 *   evaluating and outputting them.
 *
 * - The costs of literals and matches are estimated using the range encoder
 *   states and the semi-adaptive Huffman codes.  Except for range encoding
 *   states, costs are assumed to be constant throughout a single run of the
 *   parsing algorithm, which can parse up to LZMS_OPTIM_ARRAY_SIZE bytes of
 *   data.  This introduces a source of inaccuracy because the probabilities and
 *   Huffman codes can change over this part of the data.
 */
static void
lzms_normal_encode(struct lzms_compressor *ctx)
{
        struct raw_match match;

        /* Load window into suffix array match-finder.  */
        lz_sarray_load_window(&ctx->lz_sarray, ctx->window, ctx->window_size);

        /* Reset the match-chooser.  */
        lz_match_chooser_begin(&ctx->mc);

        while (ctx->cur_window_pos != ctx->window_size) {
                match = lzms_get_near_optimal_match(ctx);
                if (match.len <= 1)
                        lzms_encode_literal(ctx, ctx->window[ctx->cur_window_pos]);
                else
                        lzms_encode_lz_match(ctx, match.len, match.offset);
        }
}

static void
lzms_init_range_encoder(struct lzms_range_encoder *enc,
                        struct lzms_range_encoder_raw *rc, u32 num_states)
{
        enc->rc = rc;
        enc->state = 0;
        enc->mask = num_states - 1;
        for (u32 i = 0; i < num_states; i++) {
                enc->prob_entries[i].num_recent_zero_bits = LZMS_INITIAL_PROBABILITY;
                enc->prob_entries[i].recent_bits = LZMS_INITIAL_RECENT_BITS;
        }
}

static void
lzms_init_huffman_encoder(struct lzms_huffman_encoder *enc,
                          struct lzms_output_bitstream *os,
                          unsigned num_syms,
                          unsigned rebuild_freq)
{
        enc->os = os;
        enc->num_syms_written = 0;
        enc->rebuild_freq = rebuild_freq;
        enc->num_syms = num_syms;
        for (unsigned i = 0; i < num_syms; i++)
                enc->sym_freqs[i] = 1;

        make_canonical_huffman_code(enc->num_syms,
                                    LZMS_MAX_CODEWORD_LEN,
                                    enc->sym_freqs,
                                    enc->lens,
                                    enc->codewords);
}

/* Initialize the LZMS compressor.  */
static void
lzms_init_compressor(struct lzms_compressor *ctx, const u8 *udata, u32 ulen,
                     le16 *cdata, u32 clen16)
{
        unsigned num_position_slots;

        /* Copy the uncompressed data into the @ctx->window buffer.  */
        memcpy(ctx->window, udata, ulen);
        memset(&ctx->window[ulen], 0, 8);
        ctx->cur_window_pos = 0;
        ctx->window_size = ulen;

        /* Initialize the raw range encoder (writing forwards).  */
        lzms_range_encoder_raw_init(&ctx->rc, cdata, clen16);

        /* Initialize the output bitstream for Huffman symbols and verbatim bits
         * (writing backwards).  */
        lzms_output_bitstream_init(&ctx->os, cdata, clen16);

        /* Calculate the number of position slots needed for this compressed
         * block.  */
        num_position_slots = lzms_get_position_slot(ulen - 1) + 1;

        LZMS_DEBUG("Using %u position slots", num_position_slots);

        /* Initialize Huffman encoders for each alphabet used in the compressed
         * representation.  */
        lzms_init_huffman_encoder(&ctx->literal_encoder, &ctx->os,
                                  LZMS_NUM_LITERAL_SYMS,
                                  LZMS_LITERAL_CODE_REBUILD_FREQ);

        lzms_init_huffman_encoder(&ctx->lz_offset_encoder, &ctx->os,
                                  num_position_slots,
                                  LZMS_LZ_OFFSET_CODE_REBUILD_FREQ);

        lzms_init_huffman_encoder(&ctx->length_encoder, &ctx->os,
                                  LZMS_NUM_LEN_SYMS,
                                  LZMS_LENGTH_CODE_REBUILD_FREQ);

        lzms_init_huffman_encoder(&ctx->delta_offset_encoder, &ctx->os,
                                  num_position_slots,
                                  LZMS_DELTA_OFFSET_CODE_REBUILD_FREQ);

        lzms_init_huffman_encoder(&ctx->delta_power_encoder, &ctx->os,
                                  LZMS_NUM_DELTA_POWER_SYMS,
                                  LZMS_DELTA_POWER_CODE_REBUILD_FREQ);

        /* Initialize range encoders, all of which wrap around the same
         * lzms_range_encoder_raw.  */
        lzms_init_range_encoder(&ctx->main_range_encoder,
                                &ctx->rc, LZMS_NUM_MAIN_STATES);

        lzms_init_range_encoder(&ctx->match_range_encoder,
                                &ctx->rc, LZMS_NUM_MATCH_STATES);

        lzms_init_range_encoder(&ctx->lz_match_range_encoder,
                                &ctx->rc, LZMS_NUM_LZ_MATCH_STATES);

        for (size_t i = 0; i < ARRAY_LEN(ctx->lz_repeat_match_range_encoders); i++)
                lzms_init_range_encoder(&ctx->lz_repeat_match_range_encoders[i],
                                        &ctx->rc, LZMS_NUM_LZ_REPEAT_MATCH_STATES);

        lzms_init_range_encoder(&ctx->delta_match_range_encoder,
                                &ctx->rc, LZMS_NUM_DELTA_MATCH_STATES);

        for (size_t i = 0; i < ARRAY_LEN(ctx->delta_repeat_match_range_encoders); i++)
                lzms_init_range_encoder(&ctx->delta_repeat_match_range_encoders[i],
                                        &ctx->rc, LZMS_NUM_DELTA_REPEAT_MATCH_STATES);

        /* Initialize LRU match information.  */
        lzms_init_lru_queues(&ctx->lru);
}

/* Flush the output streams, prepare the final compressed data, and return its
 * size in bytes.
 *
 * A return value of 0 indicates that the data could not be compressed to fit in
 * the available space.  */
static size_t
lzms_finalize(struct lzms_compressor *ctx, u8 *cdata, size_t csize_avail)
{
        size_t num_forwards_bytes;
        size_t num_backwards_bytes;
        size_t compressed_size;

        /* Flush both the forwards and backwards streams, and make sure they
         * didn't cross each other and start overwriting each other's data.  */
        if (!lzms_output_bitstream_flush(&ctx->os)) {
                LZMS_DEBUG("Backwards bitstream overrun.");
                return 0;
        }

        if (!lzms_range_encoder_raw_flush(&ctx->rc)) {
                LZMS_DEBUG("Forwards bitstream overrun.");
                return 0;
        }

        if (ctx->rc.out > ctx->os.out) {
                LZMS_DEBUG("Two bitstreams crossed.");
                return 0;
        }

        /* Now the compressed buffer contains the data output by the forwards
         * bitstream, then empty space, then data output by the backwards
         * bitstream.  Move the data output by the backwards bitstream to be
         * adjacent to the data output by the forward bitstream, and calculate
         * the compressed size that this results in.  */
        num_forwards_bytes = (u8*)ctx->rc.out - (u8*)cdata;
        num_backwards_bytes = ((u8*)cdata + csize_avail) - (u8*)ctx->os.out;

        memmove(cdata + num_forwards_bytes, ctx->os.out, num_backwards_bytes);

        compressed_size = num_forwards_bytes + num_backwards_bytes;
        LZMS_DEBUG("num_forwards_bytes=%zu, num_backwards_bytes=%zu, "
                   "compressed_size=%zu",
                   num_forwards_bytes, num_backwards_bytes, compressed_size);
        LZMS_ASSERT(compressed_size % 2 == 0);
        return compressed_size;
}

static size_t
lzms_compress(const void *uncompressed_data, size_t uncompressed_size,
              void *compressed_data, size_t compressed_size_avail, void *_ctx)
{
        struct lzms_compressor *ctx = _ctx;
        size_t compressed_size;

        LZMS_DEBUG("uncompressed_size=%zu, compressed_size_avail=%zu",
                   uncompressed_size, compressed_size_avail);

        /* Make sure the uncompressed size is compatible with this compressor.
         */
        if (uncompressed_size > ctx->max_block_size) {
                LZMS_DEBUG("Can't compress %zu bytes: LZMS context "
                           "only supports %u bytes",
                           uncompressed_size, ctx->max_block_size);
                return 0;
        }

        /* Don't bother compressing extremely small inputs.  */
        if (uncompressed_size < 4) {
                LZMS_DEBUG("Input too small to bother compressing.");
                return 0;
        }

        /* Cap the available compressed size to a 32-bit integer and round it
         * down to the nearest multiple of 2.  */
        if (compressed_size_avail > UINT32_MAX)
                compressed_size_avail = UINT32_MAX;
        if (compressed_size_avail & 1)
                compressed_size_avail--;

        /* Initialize the compressor structures.  */
        lzms_init_compressor(ctx, uncompressed_data, uncompressed_size,
                             compressed_data, compressed_size_avail / 2);

        /* Preprocess the uncompressed data.  */
        lzms_x86_filter(ctx->window, ctx->window_size,
                        ctx->last_target_usages, false);

        /* Compute and encode a literal/match sequence that decompresses to the
         * preprocessed data.  */
#if 1
        lzms_normal_encode(ctx);
#else
        lzms_fast_encode(ctx);
#endif

        /* Get and return the compressed data size.  */
        compressed_size = lzms_finalize(ctx, compressed_data,
                                        compressed_size_avail);

        if (compressed_size == 0) {
                LZMS_DEBUG("Data did not compress to requested size or less.");
                return 0;
        }

        LZMS_DEBUG("Compressed %zu => %zu bytes",
                   uncompressed_size, compressed_size);

#if defined(ENABLE_VERIFY_COMPRESSION) || defined(ENABLE_LZMS_DEBUG)
        /* Verify that we really get the same thing back when decompressing.  */
        {
                struct wimlib_decompressor *decompressor;

                LZMS_DEBUG("Verifying LZMS compression.");

                if (0 == wimlib_create_decompressor(WIMLIB_COMPRESSION_TYPE_LZMS,
                                                    ctx->max_block_size,
                                                    NULL,
                                                    &decompressor))
                {
                        int ret;
                        ret = wimlib_decompress(compressed_data,
                                                compressed_size,
                                                ctx->window,
                                                uncompressed_size,
                                                decompressor);
                        wimlib_free_decompressor(decompressor);

                        if (ret) {
                                ERROR("Failed to decompress data we "
                                      "compressed using LZMS algorithm");
                                wimlib_assert(0);
                                return 0;
                        }
                        if (memcmp(uncompressed_data, ctx->window,
                                   uncompressed_size))
                        {
                                ERROR("Data we compressed using LZMS algorithm "
                                      "didn't decompress to original");
                                wimlib_assert(0);
                                return 0;
                        }
                } else {
                        WARNING("Failed to create decompressor for "
                                "data verification!");
                }
        }
#endif /* ENABLE_LZMS_DEBUG || ENABLE_VERIFY_COMPRESSION  */

        return compressed_size;
}

static void
lzms_free_compressor(void *_ctx)
{
        struct lzms_compressor *ctx = _ctx;

        if (ctx) {
                FREE(ctx->window);
#if 0
                FREE(ctx->prev_tab);
#endif
                FREE(ctx->matches);
                lz_sarray_destroy(&ctx->lz_sarray);
                lz_match_chooser_destroy(&ctx->mc);
                FREE(ctx);
        }
}

static const struct wimlib_lzms_compressor_params lzms_default = {
        .hdr = sizeof(struct wimlib_lzms_compressor_params),
        .min_match_length = 2,
        .max_match_length = UINT32_MAX,
        .nice_match_length = 32,
        .max_search_depth = 50,
        .max_matches_per_pos = 3,
        .optim_array_length = 1024,
};

static const struct wimlib_lzms_compressor_params *
lzms_get_params(const struct wimlib_compressor_params_header *_params)
{
        const struct wimlib_lzms_compressor_params *params =
                (const struct wimlib_lzms_compressor_params*)_params;

        if (params == NULL)
                params = &lzms_default;

        return params;
}

static int
lzms_create_compressor(size_t max_block_size,
                       const struct wimlib_compressor_params_header *_params,
                       void **ctx_ret)
{
        struct lzms_compressor *ctx;
        const struct wimlib_lzms_compressor_params *params = lzms_get_params(_params);

        if (max_block_size == 0 || max_block_size >= INT32_MAX) {
                LZMS_DEBUG("Invalid max_block_size (%u)", max_block_size);
                return WIMLIB_ERR_INVALID_PARAM;
        }

        ctx = CALLOC(1, sizeof(struct lzms_compressor));
        if (ctx == NULL)
                goto oom;

        ctx->window = MALLOC(max_block_size);
        if (ctx->window == NULL)
                goto oom;

#if 0
        ctx->prev_tab = MALLOC(max_block_size * sizeof(ctx->prev_tab[0]));
        if (ctx->prev_tab == NULL)
                goto oom;
#endif

        ctx->matches = MALLOC(min(params->max_match_length -
                                        params->min_match_length + 1,
                                  params->max_matches_per_pos) *
                                sizeof(ctx->matches[0]));
        if (ctx->matches == NULL)
                goto oom;

        if (!lz_sarray_init(&ctx->lz_sarray, max_block_size,
                            params->min_match_length,
                            params->max_match_length,
                            params->max_search_depth,
                            params->max_matches_per_pos))
                goto oom;

        if (!lz_match_chooser_init(&ctx->mc,
                                   params->optim_array_length,
                                   params->nice_match_length,
                                   params->max_match_length))
                goto oom;

        /* Initialize position and length slot data if not done already.  */
        lzms_init_slots();

        /* Initialize range encoding cost table if not done already.  */
        lzms_init_rc_costs();

        ctx->max_block_size = max_block_size;

        *ctx_ret = ctx;
        return 0;

oom:
        lzms_free_compressor(ctx);
        return WIMLIB_ERR_NOMEM;
}

static u64
lzms_get_needed_memory(size_t max_block_size,
                       const struct wimlib_compressor_params_header *_params)
{
        const struct wimlib_lzms_compressor_params *params = lzms_get_params(_params);

        u64 size = 0;

        size += max_block_size;
        size += sizeof(struct lzms_compressor);
        size += lz_sarray_get_needed_memory(max_block_size);
        size += lz_match_chooser_get_needed_memory(params->optim_array_length,
                                                   params->nice_match_length,
                                                   params->max_match_length);
        size += min(params->max_match_length -
                    params->min_match_length + 1,
                    params->max_matches_per_pos) *
                sizeof(((struct lzms_compressor*)0)->matches[0]);
        return size;
}

static bool
lzms_params_valid(const struct wimlib_compressor_params_header *_params)
{
        const struct wimlib_lzms_compressor_params *params =
                (const struct wimlib_lzms_compressor_params*)_params;

        if (params->hdr.size != sizeof(*params) ||
            params->max_match_length < params->min_match_length ||
            params->min_match_length < 2 ||
            params->optim_array_length == 0 ||
            min(params->max_match_length, params->nice_match_length) > 65536)
                return false;

        return true;
}

const struct compressor_ops lzms_compressor_ops = {
        .params_valid       = lzms_params_valid,
        .get_needed_memory  = lzms_get_needed_memory,
        .create_compressor  = lzms_create_compressor,
        .compress           = lzms_compress,
        .free_compressor    = lzms_free_compressor,
};