X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Flzx-compress.c;h=d71c28dd9068c787b7e1f2ca196ee39f90e91e17;hb=fade5de00b08590d830a1775b90e3490e78cdca5;hp=360ee9c4e49a528da6b90fe6b26a4249f2ef3998;hpb=cf7090dcba4736f4ea56f272b5e0152a8a0a70b5;p=wimlib

diff --git a/src/lzx-compress.c b/src/lzx-compress.c
index 360ee9c4..d71c28dd 100644
--- a/src/lzx-compress.c
+++ b/src/lzx-compress.c
@@ -179,9 +179,9 @@ typedef u32 block_cost_t;
 
 /* Codewords for the LZX main, length, and aligned offset Huffman codes  */
 struct lzx_codewords {
-	u16 main[LZX_MAINCODE_MAX_NUM_SYMBOLS];
-	u16 len[LZX_LENCODE_NUM_SYMBOLS];
-	u16 aligned[LZX_ALIGNEDCODE_NUM_SYMBOLS];
+	u32 main[LZX_MAINCODE_MAX_NUM_SYMBOLS];
+	u32 len[LZX_LENCODE_NUM_SYMBOLS];
+	u32 aligned[LZX_ALIGNEDCODE_NUM_SYMBOLS];
 };
 
 /* Codeword lengths (in bits) for the LZX main, length, and aligned offset
@@ -519,7 +519,7 @@ lzx_build_precode(const u8 lens[restrict],
 		  input_idx_t precode_freqs[restrict LZX_PRECODE_NUM_SYMBOLS],
 		  u8 output_syms[restrict num_syms],
 		  u8 precode_lens[restrict LZX_PRECODE_NUM_SYMBOLS],
-		  u16 precode_codewords[restrict LZX_PRECODE_NUM_SYMBOLS],
+		  u32 precode_codewords[restrict LZX_PRECODE_NUM_SYMBOLS],
 		  unsigned *num_additional_bits_ret)
 {
 	memset(precode_freqs, 0,
@@ -688,7 +688,7 @@ lzx_write_compressed_code(struct output_bitstream *out,
 	input_idx_t precode_freqs[LZX_PRECODE_NUM_SYMBOLS];
 	u8 output_syms[num_syms];
 	u8 precode_lens[LZX_PRECODE_NUM_SYMBOLS];
-	u16 precode_codewords[LZX_PRECODE_NUM_SYMBOLS];
+	u32 precode_codewords[LZX_PRECODE_NUM_SYMBOLS];
 	unsigned i;
 	unsigned num_output_syms;
 	u8 precode_sym;
@@ -1303,6 +1303,7 @@ lzx_optimize_block(struct lzx_compressor *ctx, struct lzx_block_spec *spec,
 
 	unsigned orig_window_pos = spec->window_pos;
 	unsigned orig_cached_pos = ctx->cached_matches_pos;
+	unsigned num_passes_remaining = num_passes;
 
 	LZX_ASSERT(ctx->match_window_pos == spec->window_pos);
 
@@ -1314,7 +1315,8 @@ lzx_optimize_block(struct lzx_compressor *ctx, struct lzx_block_spec *spec,
 	/* The first optimal parsing pass is done using the cost model already
 	 * set in ctx->costs.  Each later pass is done using a cost model
 	 * computed from the previous pass.  */
-	for (unsigned pass = 0; pass < num_passes; pass++) {
+	do {
+		--num_passes_remaining;
 
 		ctx->match_window_pos = orig_window_pos;
 		ctx->cached_matches_pos = orig_cached_pos;
@@ -1322,7 +1324,12 @@ lzx_optimize_block(struct lzx_compressor *ctx, struct lzx_block_spec *spec,
 		spec->num_chosen_matches = 0;
 		memset(&freqs, 0, sizeof(freqs));
 
-		for (unsigned i = spec->window_pos; i < spec->window_pos + spec->block_size; ) {
+		const u8 *window_ptr = &ctx->window[spec->window_pos];
+		const u8 *window_end = &window_ptr[spec->block_size];
+		struct lzx_match *next_chosen_match =
+			&ctx->chosen_matches[spec->chosen_matches_start_pos];
+
+		while (window_ptr != window_end) {
 			struct raw_match raw_match;
 			struct lzx_match lzx_match;
 
@@ -1350,36 +1357,33 @@ lzx_optimize_block(struct lzx_compressor *ctx, struct lzx_block_spec *spec,
 					 * null value and therefore cannot
 					 * generate such matches.  */
 					BUILD_BUG_ON(LZX_MIN_MATCH_LEN != 2);
-					lzx_match.data = lzx_tally_literal(ctx->window[i],
+					lzx_match.data = lzx_tally_literal(*window_ptr++,
 									   &freqs);
-					i += 1;
-					ctx->chosen_matches[spec->chosen_matches_start_pos +
-							    spec->num_chosen_matches++]
-							    = lzx_match;
-					lzx_match.data = lzx_tally_literal(ctx->window[i],
+					*next_chosen_match++ = lzx_match;
+					lzx_match.data = lzx_tally_literal(*window_ptr++,
 									   &freqs);
-					i += 1;
 				} else {
 					lzx_match.data = lzx_tally_match(raw_match.len,
 									 raw_match.offset,
 									 &freqs,
 									 &ctx->queue);
-					i += raw_match.len;
+					window_ptr += raw_match.len;
 				}
 			} else {
-				lzx_match.data = lzx_tally_literal(ctx->window[i], &freqs);
-				i += 1;
+				lzx_match.data = lzx_tally_literal(*window_ptr++, &freqs);
 			}
-			ctx->chosen_matches[spec->chosen_matches_start_pos +
-					    spec->num_chosen_matches++] = lzx_match;
+			*next_chosen_match++ = lzx_match;
 		}
+		spec->num_chosen_matches = next_chosen_match -
+					   &ctx->chosen_matches[spec->chosen_matches_start_pos];
 
 		lzx_make_huffman_codes(&freqs, &spec->codes,
 				       ctx->num_main_syms);
-		if (pass < num_passes - 1)
+		if (num_passes_remaining)
 			lzx_set_costs(ctx, &spec->codes.lens);
 		ctx->matches_cached = true;
-	}
+	} while (num_passes_remaining);
+
 	spec->block_type = lzx_choose_verbatim_or_aligned(&freqs, &spec->codes);
 	ctx->matches_cached = false;
 }