Oh well, big messy commit again. Some highlights:

  - Updated to the latest, probably final file format version.
  - Command line tool reworked to not use threads anymore.
    Threading will probably go into liblzma anyway.
  - Memory usage limit is now about 30 % for uncompression
    and about 90 % for compression.
  - Progress indicator with --verbose
  - Simplified --help and full --long-help
  - Upgraded to the last LGPLv2.1+ getopt_long from gnulib.
  - Some bug fixes

20 files changed, 407 insertions, 358 deletions

diff --git a/src/liblzma/api/lzma/block.h b/src/liblzma/api/lzma/block.h
index eb3768e2..06c1633c 100644
--- a/src/liblzma/api/lzma/block.h
+++ b/src/liblzma/api/lzma/block.h
@@ -1,6 +1,6 @@
 /**
  * \file        lzma/block.h
- * \brief       .lzma Block handling
+ * \brief       .xz Block handling
  *
  * \author      Copyright (C) 1999-2006 Igor Pavlov
  * \author      Copyright (C) 2007 Lasse Collin
@@ -131,11 +131,10 @@ typedef struct {
 	 *
 	 * \note        Because of the array is terminated with
 	 *              .id = LZMA_VLI_UNKNOWN, the actual array must
-	 *              have LZMA_BLOCK_FILTERS_MAX + 1 members or the Block
+	 *              have LZMA_FILTERS_MAX + 1 members or the Block
 	 *              Header decoder will overflow the buffer.
 	 */
 	lzma_filter *filters;
-#	define LZMA_BLOCK_FILTERS_MAX 4
 
 } lzma_block;
 
@@ -148,6 +147,8 @@ typedef struct {
  * The size can be calculated from the first byte of a Block using this macro.
  * Note that if the first byte is 0x00, it indicates beginning of Index; use
  * this macro only when the byte is not 0x00.
+ *
+ * There is no encoding macro, because Block Header encoder is enough for that.
  */
 #define lzma_block_header_size_decode(b) (((uint32_t)(b) + 1) * 4)
 
@@ -211,38 +212,50 @@ extern lzma_ret lzma_block_header_decode(lzma_block *options,
 
 
 /**
- * \brief       Sets Compressed Size according to Total Size
+ * \brief       Sets Compressed Size according to Unpadded Size
  *
- * Block Header stores Compressed Size, but Index has Total Size. If the
+ * Block Header stores Compressed Size, but Index has Unpadded Size. If the
  * application has already parsed the Index and is now decoding Blocks,
- * it can calculate Compressed Size from Total Size. This function does
+ * it can calculate Compressed Size from Unpadded Size. This function does
  * exactly that with error checking, so application doesn't need to check,
  * for example, if the value in Index is too small to contain even the
- * Block Header. Note that you need to call this function after decoding
+ * Block Header. Note that you need to call this function _after_ decoding
  * the Block Header field.
  *
  * \return      - LZMA_OK: options->compressed_size was set successfully.
- *              - LZMA_DATA_ERROR: total_size is too small compared to
+ *              - LZMA_DATA_ERROR: unpadded_size is too small compared to
  *                options->header_size and lzma_check_sizes[options->check].
  *              - LZMA_PROG_ERROR: Some values are invalid. For example,
- *                total_size and options->header_size must be multiples
- *                of four, total_size must be at least 12, and
+ *                options->header_size must be a multiple of four, and
  *                options->header_size between 8 and 1024 inclusive.
  */
-extern lzma_ret lzma_block_total_size_set(
-		lzma_block *options, lzma_vli total_size)
+extern lzma_ret lzma_block_compressed_size(
+		lzma_block *options, lzma_vli unpadded_size)
 		lzma_attr_warn_unused_result;
 
 
 /**
- * \brief       Calculates Total Size
+ * \brief       Calculates Unpadded Size
  *
- * This function can be useful after decoding a Block to get Total Size
+ * This function can be useful after decoding a Block to get Unpadded Size
  * that is stored in Index.
  *
- * \return      Total Size on success, or zero on error.
+ * \return      Unpadded Size on success, or zero on error.
+ */
+extern lzma_vli lzma_block_unpadded_size(const lzma_block *options)
+		lzma_attr_pure;
+
+
+/**
+ * \brief       Calculates the total encoded size of a Block
+ *
+ * This is equivalent to lzma_block_unpadded_size() except that the returned
+ * value includes the size of the Block Padding field.
+ *
+ * \return      On success, total encoded size of the Block. On error,
+ *              zero is returned.
  */
-extern lzma_vli lzma_block_total_size_get(const lzma_block *options)
+extern lzma_vli lzma_block_total_size(const lzma_block *options)
 		lzma_attr_pure;
 
 
@@ -255,8 +268,6 @@ extern lzma_vli lzma_block_total_size_get(const lzma_block *options)
  * \return      - LZMA_OK: All good, continue with lzma_code().
  *              - LZMA_MEM_ERROR
  *              - LZMA_OPTIONS_ERROR
- *              - LZMA_DATA_ERROR: Limits (total_limit and uncompressed_limit)
- *                have been reached already.
  *              - LZMA_UNSUPPORTED_CHECK: options->check specfies a Check
  *                that is not supported by this buid of liblzma. Initializing
  *                the encoder failed.
diff --git a/src/liblzma/api/lzma/filter.h b/src/liblzma/api/lzma/filter.h
index 53e5737e..b4fb02a7 100644
--- a/src/liblzma/api/lzma/filter.h
+++ b/src/liblzma/api/lzma/filter.h
@@ -55,6 +55,14 @@ typedef struct {
 
 
 /**
+ * \brief       Maximum number of filters in a chain
+ *
+ * FIXME desc
+ */
+#define LZMA_FILTERS_MAX 4
+
+
+/**
  * \brief       Test if the given Filter ID is supported for encoding
  *
  * Returns true if the give Filter ID  is supported for encoding by this
diff --git a/src/liblzma/api/lzma/index.h b/src/liblzma/api/lzma/index.h
index 522969d4..d6072614 100644
--- a/src/liblzma/api/lzma/index.h
+++ b/src/liblzma/api/lzma/index.h
@@ -32,12 +32,24 @@ typedef struct lzma_index_s lzma_index;
  */
 typedef struct {
 	/**
-	 * Total Size of a Block.
+	 * \brief       Total encoded size of a Block including Block Padding
+	 *
+	 * This value is useful if you need to know the actual size of the
+	 * Block that the Block decoder will read.
 	 */
 	lzma_vli total_size;
 
 	/**
-	 * Uncompressed Size of a Block
+	 * \brief       Encoded size of a Block excluding Block Padding
+	 *
+	 * This value is stored in the Index. When doing random-access
+	 * reading, you should give this value to the Block decoder along
+	 * with uncompressed_size.
+	 */
+	lzma_vli unpadded_size;
+
+	/**
+	 * \brief       Uncompressed Size of a Block
 	 */
 	lzma_vli uncompressed_size;
 
@@ -80,7 +92,7 @@ extern void lzma_index_end(lzma_index *i, lzma_allocator *allocator);
  * \brief       Add a new Record to an Index
  *
  * \param       index             Pointer to a lzma_index structure
- * \param       total_size        Total Size of a Block
+ * \param       unpadded_size     Unpadded Size of a Block
  * \param       uncompressed_size Uncompressed Size of a Block, or
  *                                LZMA_VLI_UNKNOWN to indicate padding.
  *
@@ -92,7 +104,7 @@ extern void lzma_index_end(lzma_index *i, lzma_allocator *allocator);
  *              - LZMA_PROG_ERROR
  */
 extern lzma_ret lzma_index_append(lzma_index *i, lzma_allocator *allocator,
-		lzma_vli total_size, lzma_vli uncompressed_size)
+		lzma_vli unpadded_size, lzma_vli uncompressed_size)
 		lzma_attr_warn_unused_result;
 
 
diff --git a/src/liblzma/api/lzma/index_hash.h b/src/liblzma/api/lzma/index_hash.h
index 58fc8061..001e6b5c 100644
--- a/src/liblzma/api/lzma/index_hash.h
+++ b/src/liblzma/api/lzma/index_hash.h
@@ -57,7 +57,7 @@ extern void lzma_index_hash_end(
  * \brief       Add a new Record to an Index hash
  *
  * \param       index             Pointer to a lzma_index_hash structure
- * \param       total_size        Total Size of a Block
+ * \param       unpadded_size     Unpadded Size of a Block
  * \param       uncompressed_size Uncompressed Size of a Block
  *
  * \return      - LZMA_OK
@@ -67,7 +67,7 @@ extern void lzma_index_hash_end(
  *                used when lzma_index_hash_decode() has already been used.
  */
 extern lzma_ret lzma_index_hash_append(lzma_index_hash *index_hash,
-		lzma_vli total_size, lzma_vli uncompressed_size)
+		lzma_vli unpadded_size, lzma_vli uncompressed_size)
 		lzma_attr_warn_unused_result;
 
 
diff --git a/src/liblzma/common/block_decoder.c b/src/liblzma/common/block_decoder.c
index f9101c7d..2bfe0b92 100644
--- a/src/liblzma/common/block_decoder.c
+++ b/src/liblzma/common/block_decoder.c
@@ -33,13 +33,13 @@ struct lzma_coder_s {
 	lzma_next_coder next;
 
 	/// Decoding options; we also write Compressed Size and Uncompressed
-	/// Size back to this structure when the encoding has been finished.
+	/// Size back to this structure when the decoding has been finished.
 	lzma_block *options;
 
-	/// Compressed Size calculated while encoding
+	/// Compressed Size calculated while decoding
 	lzma_vli compressed_size;
 
-	/// Uncompressed Size calculated while encoding
+	/// Uncompressed Size calculated while decoding
 	lzma_vli uncompressed_size;
 
 	/// Maximum allowed Compressed Size; this takes into account the
@@ -110,6 +110,19 @@ block_decode(lzma_coder *coder, lzma_allocator *allocator,
 		if (ret != LZMA_STREAM_END)
 			return ret;
 
+		// Compressed and Uncompressed Sizes are now at their final
+		// values. Verify that they match the values given to us.
+		if (!is_size_valid(coder->compressed_size,
+					coder->options->compressed_size)
+				|| !is_size_valid(coder->uncompressed_size,
+					coder->options->uncompressed_size))
+			return LZMA_DATA_ERROR;
+
+		// Copy the values into coder->options. The caller
+		// may use this information to construct Index.
+		coder->options->compressed_size = coder->compressed_size;
+		coder->options->uncompressed_size = coder->uncompressed_size;
+
 		coder->sequence = SEQ_PADDING;
 	}
 
@@ -118,30 +131,19 @@ block_decode(lzma_coder *coder, lzma_allocator *allocator,
 	case SEQ_PADDING:
 		// Compressed Data is padded to a multiple of four bytes.
 		while (coder->compressed_size & 3) {
+			// We use compressed_size here just get the Padding
+			// right. The actual Compressed Size was stored to
+			// coder->options already, and won't be modified by
+			// us anymore.
+			++coder->compressed_size;
+
 			if (*in_pos >= in_size)
 				return LZMA_OK;
 
 			if (in[(*in_pos)++] != 0x00)
 				return LZMA_DATA_ERROR;
-
-			if (update_size(&coder->compressed_size, 1,
-					coder->compressed_limit))
-				return LZMA_DATA_ERROR;
 		}
 
-		// Compressed and Uncompressed Sizes are now at their final
-		// values. Verify that they match the values given to us.
-		if (!is_size_valid(coder->compressed_size,
-					coder->options->compressed_size)
-				|| !is_size_valid(coder->uncompressed_size,
-					coder->options->uncompressed_size))
-			return LZMA_DATA_ERROR;
-
-		// Copy the values into coder->options. The caller
-		// may use this information to construct Index.
-		coder->options->compressed_size = coder->compressed_size;
-		coder->options->uncompressed_size = coder->uncompressed_size;
-
 		if (coder->options->check == LZMA_CHECK_NONE)
 			return LZMA_STREAM_END;
 
@@ -193,14 +195,11 @@ lzma_block_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 {
 	lzma_next_coder_init(lzma_block_decoder_init, next, allocator);
 
-	// While lzma_block_total_size_get() is meant to calculate the Total
-	// Size, it also validates the options excluding the filters.
-	if (lzma_block_total_size_get(options) == 0)
-		return LZMA_PROG_ERROR;
-
-	// options->check is used for array indexing so we need to know that
-	// it is in the valid range.
-	if ((unsigned)(options->check) > LZMA_CHECK_ID_MAX)
+	// Validate the options. lzma_block_unpadded_size() does that for us
+	// except for Uncompressed Size and filters. Filters are validated
+	// by the raw decoder.
+	if (lzma_block_unpadded_size(options) == 0
+			|| !lzma_vli_is_valid(options->uncompressed_size))
 		return LZMA_PROG_ERROR;
 
 	// Allocate and initialize *next->coder if needed.
@@ -221,8 +220,8 @@ lzma_block_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 	next->coder->uncompressed_size = 0;
 
 	// If Compressed Size is not known, we calculate the maximum allowed
-	// value so that Total Size of the Block still is a valid VLI and
-	// a multiple of four.
+	// value so that encoded size of the Block (including Block Padding)
+	// is still a valid VLI and a multiple of four.
 	next->coder->compressed_limit
 			= options->compressed_size == LZMA_VLI_UNKNOWN
 				? (LZMA_VLI_MAX & ~LZMA_VLI_C(3))
diff --git a/src/liblzma/common/block_encoder.c b/src/liblzma/common/block_encoder.c
index 3c678f7d..6468cb44 100644
--- a/src/liblzma/common/block_encoder.c
+++ b/src/liblzma/common/block_encoder.c
@@ -27,8 +27,8 @@
 /// take into account the headers etc. to determine the exact maximum size
 /// of the Compressed Data field, but the complexity would give us nothing
 /// useful. Instead, limit the size of Compressed Data so that even with
-/// biggest possible Block Header and Check fields the total size of the
-/// Block stays as valid VLI. This way we don't produce incorrect output
+/// biggest possible Block Header and Check fields the total encoded size of
+/// the Block stays as valid VLI. This way we don't produce incorrect output
 /// if someone will really try creating a Block of 8 EiB.
 ///
 /// ~LZMA_VLI_C(3) is to guarantee that if we need padding at the end of
@@ -41,9 +41,9 @@ struct lzma_coder_s {
 	/// The filters in the chain; initialized with lzma_raw_decoder_init().
 	lzma_next_coder next;
 
-	/// Encoding options; we also write Total Size, Compressed Size, and
-	/// Uncompressed Size back to this structure when the encoding has
-	/// been finished.
+	/// Encoding options; we also write Unpadded Size, Compressed Size,
+	/// and Uncompressed Size back to this structure when the encoding
+	/// has been finished.
 	lzma_block *options;
 
 	enum {
@@ -58,8 +58,8 @@ struct lzma_coder_s {
 	/// Uncompressed Size calculated while encoding
 	lzma_vli uncompressed_size;
 
-	/// Position when writing out the Check field
-	size_t check_pos;
+	/// Position in Block Padding and the Check fields
+	size_t pos;
 
 	/// Check of the uncompressed data
 	lzma_check_state check;
@@ -106,6 +106,11 @@ block_encode(lzma_coder *coder, lzma_allocator *allocator,
 		assert(*in_pos == in_size);
 		assert(action == LZMA_FINISH);
 
+		// Copy the values into coder->options. The caller
+		// may use this information to construct Index.
+		coder->options->compressed_size = coder->compressed_size;
+		coder->options->uncompressed_size = coder->uncompressed_size;
+
 		coder->sequence = SEQ_PADDING;
 	}
 
@@ -113,28 +118,21 @@ block_encode(lzma_coder *coder, lzma_allocator *allocator,
 
 	case SEQ_PADDING:
 		// Pad Compressed Data to a multiple of four bytes.
-		while (coder->compressed_size & 3) {
+		while ((coder->compressed_size + coder->pos) & 3) {
 			if (*out_pos >= out_size)
 				return LZMA_OK;
 
 			out[*out_pos] = 0x00;
 			++*out_pos;
-
-			// No need to use check for overflow here since we
-			// have already checked in SEQ_CODE that Compressed
-			// Size will stay in proper limits.
-			++coder->compressed_size;
+			++coder->pos;
 		}
 
-		// Copy the values into coder->options. The caller
-		// may use this information to construct Index.
-		coder->options->compressed_size = coder->compressed_size;
-		coder->options->uncompressed_size = coder->uncompressed_size;
-
 		if (coder->options->check == LZMA_CHECK_NONE)
 			return LZMA_STREAM_END;
 
 		lzma_check_finish(&coder->check, coder->options->check);
+
+		coder->pos = 0;
 		coder->sequence = SEQ_CHECK;
 
 	// Fall through
@@ -144,11 +142,10 @@ block_encode(lzma_coder *coder, lzma_allocator *allocator,
 				= lzma_check_size(coder->options->check);
 
 		while (*out_pos < out_size) {
-			out[*out_pos] = coder->check.buffer.u8[
-					coder->check_pos];
+			out[*out_pos] = coder->check.buffer.u8[coder->pos];
 			++*out_pos;
 
-			if (++coder->check_pos == check_size)
+			if (++coder->pos == check_size)
 				return LZMA_STREAM_END;
 		}
 
@@ -199,9 +196,9 @@ lzma_block_encoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 	next->coder->options = options;
 	next->coder->compressed_size = 0;
 	next->coder->uncompressed_size = 0;
+	next->coder->pos = 0;
 
 	// Initialize the check
-	next->coder->check_pos = 0;
 	lzma_check_init(&next->coder->check, options->check);
 
 	// Initialize the requested filters.
diff --git a/src/liblzma/common/block_header_decoder.c b/src/liblzma/common/block_header_decoder.c
index 3b8e9f36..8421ac37 100644
--- a/src/liblzma/common/block_header_decoder.c
+++ b/src/liblzma/common/block_header_decoder.c
@@ -27,7 +27,7 @@ free_properties(lzma_block *options, lzma_allocator *allocator)
 	// Free allocated filter options. The last array member is not
 	// touched after the initialization in the beginning of
 	// lzma_block_header_decode(), so we don't need to touch that here.
-	for (size_t i = 0; i < LZMA_BLOCK_FILTERS_MAX; ++i) {
+	for (size_t i = 0; i < LZMA_FILTERS_MAX; ++i) {
 		lzma_free(options->filters[i].options, allocator);
 		options->filters[i].id = LZMA_VLI_UNKNOWN;
 		options->filters[i].options = NULL;
@@ -48,24 +48,19 @@ lzma_block_header_decode(lzma_block *options,
 
 	// Initialize the filter options array. This way the caller can
 	// safely free() the options even if an error occurs in this function.
-	for (size_t i = 0; i <= LZMA_BLOCK_FILTERS_MAX; ++i) {
+	for (size_t i = 0; i <= LZMA_FILTERS_MAX; ++i) {
 		options->filters[i].id = LZMA_VLI_UNKNOWN;
 		options->filters[i].options = NULL;
 	}
 
-	size_t in_size = options->header_size;
-
-	// Validate. The caller must have set options->header_size with
-	// lzma_block_header_size_decode() macro, so it is a programming error
-	// if these tests fail.
-	if (in_size < LZMA_BLOCK_HEADER_SIZE_MIN
-			|| in_size > LZMA_BLOCK_HEADER_SIZE_MAX
-			|| (in_size & 3)
-			|| lzma_block_header_size_decode(in[0]) != in_size)
+	// Validate Block Header Size and Check type. The caller must have
+	// already set these, so it is a programming error if this test fails.
+	if (lzma_block_header_size_decode(in[0]) != options->header_size
+			|| (unsigned int)(options->check) > LZMA_CHECK_ID_MAX)
 		return LZMA_PROG_ERROR;
 
 	// Exclude the CRC32 field.
-	in_size -= 4;
+	const size_t in_size = options->header_size - 4;
 
 	// Verify CRC32
 	if (lzma_crc32(in, in_size, 0) != integer_read_32(in + in_size))
@@ -83,15 +78,9 @@ lzma_block_header_decode(lzma_block *options,
 		return_if_error(lzma_vli_decode(&options->compressed_size,
 				NULL, in, &in_pos, in_size));
 
-		if (options->compressed_size > LZMA_VLI_MAX / 4 - 1)
-			return LZMA_DATA_ERROR;
-
-		options->compressed_size = (options->compressed_size + 1) * 4;
-
-		// Check that Total Size (that is, size of
-		// Block Header + Compressed Data + Check) is
-		// representable as a VLI.
-		if (lzma_block_total_size_get(options) == 0)
+		// Validate Compressed Size. This checks that it isn't zero
+		// and that the total size of the Block is a valid VLI.
+		if (lzma_block_unpadded_size(options) == 0)
 			return LZMA_DATA_ERROR;
 	} else {
 		options->compressed_size = LZMA_VLI_UNKNOWN;
diff --git a/src/liblzma/common/block_header_encoder.c b/src/liblzma/common/block_header_encoder.c
index 9326350b..b9980363 100644
--- a/src/liblzma/common/block_header_encoder.c
+++ b/src/liblzma/common/block_header_encoder.c
@@ -25,21 +25,20 @@ extern LZMA_API lzma_ret
 lzma_block_header_size(lzma_block *options)
 {
 	// Block Header Size + Block Flags + CRC32.
-	size_t size = 1 + 1 + 4;
+	uint32_t size = 1 + 1 + 4;
 
 	// Compressed Size
 	if (options->compressed_size != LZMA_VLI_UNKNOWN) {
-		if (options->compressed_size > LZMA_VLI_MAX / 4 - 1
-				|| options->compressed_size == 0
-				|| (options->compressed_size & 3))
+		const uint32_t add = lzma_vli_size(options->compressed_size);
+		if (add == 0 || options->compressed_size == 0)
 			return LZMA_PROG_ERROR;
 
-		size += lzma_vli_size(options->compressed_size / 4 - 1);
+		size += add;
 	}
 
 	// Uncompressed Size
 	if (options->uncompressed_size != LZMA_VLI_UNKNOWN) {
-		const size_t add = lzma_vli_size(options->uncompressed_size);
+		const uint32_t add = lzma_vli_size(options->uncompressed_size);
 		if (add == 0)
 			return LZMA_PROG_ERROR;
 
@@ -51,10 +50,9 @@ lzma_block_header_size(lzma_block *options)
 			|| options->filters[0].id == LZMA_VLI_UNKNOWN)
 		return LZMA_PROG_ERROR;
 
-	for (size_t i = 0; options->filters[i].id != LZMA_VLI_UNKNOWN;
-			++i) {
+	for (size_t i = 0; options->filters[i].id != LZMA_VLI_UNKNOWN; ++i) {
 		// Don't allow too many filters.
-		if (i == 4)
+		if (i == LZMA_FILTERS_MAX)
 			return LZMA_PROG_ERROR;
 
 		uint32_t add;
@@ -65,12 +63,13 @@ lzma_block_header_size(lzma_block *options)
 	}
 
 	// Pad to a multiple of four bytes.
-	options->header_size = (size + 3) & ~(size_t)(3);
+	options->header_size = (size + 3) & ~UINT32_C(3);
 
-	// NOTE: We don't verify that Total Size of the Block stays within
-	// limits. This is because it is possible that we are called with
-	// exaggerated values to reserve space for Block Header, and later
-	// called again with lower, real values.
+	// NOTE: We don't verify that the encoded size of the Block stays
+	// within limits. This is because it is possible that we are called
+	// with exaggerated Compressed Size (e.g. LZMA_VLI_MAX) to reserve
+	// space for Block Header, and later called again with lower,
+	// real values.
 
 	return LZMA_OK;
 }
@@ -79,9 +78,9 @@ lzma_block_header_size(lzma_block *options)
 extern LZMA_API lzma_ret
 lzma_block_header_encode(const lzma_block *options, uint8_t *out)
 {
-	if ((options->header_size & 3)
-			|| options->header_size < LZMA_BLOCK_HEADER_SIZE_MIN
-			|| options->header_size > LZMA_BLOCK_HEADER_SIZE_MAX)
+	// Valdidate everything but filters.
+	if (lzma_block_unpadded_size(options) == 0
+			|| !lzma_vli_is_valid(options->uncompressed_size))
 		return LZMA_PROG_ERROR;
 
 	// Indicate the size of the buffer _excluding_ the CRC32 field.
@@ -90,32 +89,28 @@ lzma_block_header_encode(const lzma_block *options, uint8_t *out)
 	// Store the Block Header Size.
 	out[0] = out_size / 4;
 
-	// We write Block Flags a little later.
+	// We write Block Flags in pieces.
+	out[1] = 0x00;
 	size_t out_pos = 2;
 
 	// Compressed Size
 	if (options->compressed_size != LZMA_VLI_UNKNOWN) {
-		// Compressed Size must be non-zero, fit into a 63-bit
-		// integer and be a multiple of four. Also the Total Size
-		// of the Block must fit into 63-bit integer.
-		if (options->compressed_size == 0
-				|| (options->compressed_size & 3)
-				|| options->compressed_size
-					> LZMA_VLI_MAX
-				|| lzma_block_total_size_get(options) == 0)
-			return LZMA_PROG_ERROR;
-
 		return_if_error(lzma_vli_encode(
-				options->compressed_size / 4 - 1, NULL,
+				options->compressed_size, NULL,
 				out, &out_pos, out_size));
+
+		out[1] |= 0x40;
 	}
 
 	// Uncompressed Size
-	if (options->uncompressed_size != LZMA_VLI_UNKNOWN)
+	if (options->uncompressed_size != LZMA_VLI_UNKNOWN) {
 		return_if_error(lzma_vli_encode(
 				options->uncompressed_size, NULL,
 				out, &out_pos, out_size));
 
+		out[1] |= 0x80;
+	}
+
 	// Filter Flags
 	if (options->filters == NULL
 			|| options->filters[0].id == LZMA_VLI_UNKNOWN)
@@ -124,24 +119,16 @@ lzma_block_header_encode(const lzma_block *options, uint8_t *out)
 	size_t filter_count = 0;
 	do {
 		// There can be at maximum of four filters.
-		if (filter_count == 4)
+		if (filter_count == LZMA_FILTERS_MAX)
 			return LZMA_PROG_ERROR;
 
 		return_if_error(lzma_filter_flags_encode(
 				options->filters + filter_count,
 				out, &out_pos, out_size));
 
-	} while (options->filters[++filter_count].id
-			!= LZMA_VLI_UNKNOWN);
-
-	// Block Flags
-	out[1] = filter_count - 1;
+	} while (options->filters[++filter_count].id != LZMA_VLI_UNKNOWN);
 
-	if (options->compressed_size != LZMA_VLI_UNKNOWN)
-		out[1] |= 0x40;
-
-	if (options->uncompressed_size != LZMA_VLI_UNKNOWN)
-		out[1] |= 0x80;
+	out[1] |= filter_count - 1;
 
 	// Padding
 	memzero(out + out_pos, out_size - out_pos);
diff --git a/src/liblzma/common/block_util.c b/src/liblzma/common/block_util.c
index 7b46ba32..66e1cad9 100644
--- a/src/liblzma/common/block_util.c
+++ b/src/liblzma/common/block_util.c
@@ -18,10 +18,11 @@
 ///////////////////////////////////////////////////////////////////////////////
 
 #include "common.h"
+#include "index.h"
 
 
 extern LZMA_API lzma_ret
-lzma_block_total_size_set(lzma_block *options, lzma_vli total_size)
+lzma_block_compressed_size(lzma_block *options, lzma_vli total_size)
 {
 	// Validate.
 	if (options->header_size < LZMA_BLOCK_HEADER_SIZE_MIN
@@ -45,29 +46,47 @@ lzma_block_total_size_set(lzma_block *options, lzma_vli total_size)
 
 
 extern LZMA_API lzma_vli
-lzma_block_total_size_get(const lzma_block *options)
+lzma_block_unpadded_size(const lzma_block *options)
 {
-	// Validate the values that we are interested in.
+	// Validate the values that we are interested in i.e. all but
+	// Uncompressed Size and the filters.
+	//
+	// NOTE: This function is used for validation too, so it is
+	// essential that these checks are always done even if
+	// Compressed Size is unknown.
 	if (options->header_size < LZMA_BLOCK_HEADER_SIZE_MIN
 			|| options->header_size > LZMA_BLOCK_HEADER_SIZE_MAX
 			|| (options->header_size & 3)
-			|| (unsigned)(options->check) > LZMA_CHECK_ID_MAX)
+			|| !lzma_vli_is_valid(options->compressed_size)
+			|| options->compressed_size == 0
+			|| (unsigned int)(options->check) > LZMA_CHECK_ID_MAX)
 		return 0;
 
 	// If Compressed Size is unknown, return that we cannot know
-	// Total Size either.
+	// size of the Block either.
 	if (options->compressed_size == LZMA_VLI_UNKNOWN)
 		return LZMA_VLI_UNKNOWN;
 
-	const lzma_vli total_size = options->compressed_size
-			+ options->header_size
-			+ lzma_check_size(options->check);
+	// Calculate Unpadded Size and validate it.
+	const lzma_vli unpadded_size = options->compressed_size
+				+ options->header_size
+				+ lzma_check_size(options->check);
 
-	// Validate the calculated Total Size.
-	if (options->compressed_size > LZMA_VLI_MAX
-			|| (options->compressed_size & 3)
-			|| total_size > LZMA_VLI_MAX)
+	assert(unpadded_size >= UNPADDED_SIZE_MIN);
+	if (unpadded_size > UNPADDED_SIZE_MAX)
 		return 0;
 
-	return total_size;
+	return unpadded_size;
+}
+
+
+extern LZMA_API lzma_vli
+lzma_block_total_size(const lzma_block *options)
+{
+	lzma_vli unpadded_size = lzma_block_unpadded_size(options);
+
+	if (unpadded_size != 0 && unpadded_size != LZMA_VLI_UNKNOWN)
+		unpadded_size = vli_ceil4(unpadded_size);
+
+	return unpadded_size;
 }
diff --git a/src/liblzma/common/common.h b/src/liblzma/common/common.h
index 275cf05f..0ee8574c 100644
--- a/src/liblzma/common/common.h
+++ b/src/liblzma/common/common.h
@@ -66,10 +66,6 @@
 	| LZMA_CONCATENATED )
 
 
-///////////
-// Types //
-///////////
-
 /// Type of encoder/decoder specific data; the actual structure is defined
 /// differently in different coders.
 typedef struct lzma_coder_s lzma_coder;
@@ -187,10 +183,6 @@ struct lzma_internal_s {
 };
 
 
-///////////////
-// Functions //
-///////////////
-
 /// Allocates memory
 extern void *lzma_alloc(size_t size, lzma_allocator *allocator)
 		lzma_attribute((malloc));
diff --git a/src/liblzma/common/filter_common.c b/src/liblzma/common/filter_common.c
index 71ceeca0..03b6859a 100644
--- a/src/liblzma/common/filter_common.c
+++ b/src/liblzma/common/filter_common.c
@@ -164,7 +164,7 @@ validate_chain(const lzma_filter *filters, size_t *count)
 	// There must be 1-4 filters. The last filter must be usable as
 	// the last filter in the chain. At maximum of three filters are
 	// allowed to change the size of the data.
-	if (i > LZMA_BLOCK_FILTERS_MAX || !last_ok || changes_size_count > 3)
+	if (i > LZMA_FILTERS_MAX || !last_ok || changes_size_count > 3)
 		return LZMA_OPTIONS_ERROR;
 
 	*count = i;
@@ -182,7 +182,7 @@ lzma_raw_coder_init(lzma_next_coder *next, lzma_allocator *allocator,
 	return_if_error(validate_chain(options, &count));
 
 	// Set the filter functions and copy the options pointer.
-	lzma_filter_info filters[LZMA_BLOCK_FILTERS_MAX + 1];
+	lzma_filter_info filters[LZMA_FILTERS_MAX + 1];
 	if (is_encoder) {
 		for (size_t i = 0; i < count; ++i) {
 			// The order of the filters is reversed in the
diff --git a/src/liblzma/common/index.c b/src/liblzma/common/index.c
index f965749f..1fe65650 100644
--- a/src/liblzma/common/index.c
+++ b/src/liblzma/common/index.c
@@ -20,24 +20,34 @@
 #include "index.h"
 
 
-/// Number of Records to allocate at once.
+/// Number of Records to allocate at once in the unrolled list.
 #define INDEX_GROUP_SIZE 256
 
 
 typedef struct lzma_index_group_s lzma_index_group;
 struct lzma_index_group_s {
-	/// Next group
+	/// Previous group
 	lzma_index_group *prev;
 
-	/// Previous group
+	/// Next group
 	lzma_index_group *next;
 
 	/// Index of the last Record in this group
 	size_t last;
 
-	/// Total Size fields as cumulative sum relative to the beginning
-	/// of the group. The total size of the group is total_sums[last].
-	lzma_vli total_sums[INDEX_GROUP_SIZE];
+	/// Unpadded Size fields as special cumulative sum relative to the
+	/// beginning of the group. It's special in sense that the previous
+	/// value is rounded up the next multiple of four with before
+	/// calculating the new value. The total encoded size of the Blocks
+	/// in the group is unpadded_sums[last] rounded up to the next
+	/// multiple of four.
+	///
+	/// For example, if the Unpadded Sizes are 39, 57, and 81, the stored
+	/// values are 39, 97 (40 + 57), and 181 (100 + 181). The total
+	/// encoded size of these Blocks is 184.
+	///
+	/// This encoding is nice from point of view of lzma_index_locate().
+	lzma_vli unpadded_sums[INDEX_GROUP_SIZE];
 
 	/// Uncompressed Size fields as cumulative sum relative to the
 	/// beginning of the group. The uncompressed size of the group is
@@ -56,19 +66,13 @@ struct lzma_index_s {
 	/// Uncompressed size of the Stream
 	lzma_vli uncompressed_size;
 
-	/// Number of non-padding records. This is needed by Index encoder.
+	/// Number of non-padding records. This is needed for Index encoder.
 	lzma_vli count;
 
 	/// Size of the List of Records field; this is updated every time
 	/// a new non-padding Record is added.
 	lzma_vli index_list_size;
 
-	/// This is zero if no Indexes have been combined with
-	/// lzma_index_cat(). With combined Indexes, this contains the sizes
-	/// of all but latest the Streams, including possible Stream Padding
-	/// fields.
-	lzma_vli padding_size;
-
 	/// First group of Records
 	lzma_index_group *head;
 
@@ -80,8 +84,8 @@ struct lzma_index_s {
 		/// Group where the current read position is.
 		lzma_index_group *group;
 
-		/// The most recently read record in *group
-		lzma_vli record;
+		/// The most recently read Record in *group
+		size_t record;
 
 		/// Uncompressed offset of the beginning of *group relative
 		/// to the beginning of the Stream
@@ -102,6 +106,10 @@ struct lzma_index_s {
 		/// Stream. This is needed when a new Index is concatenated
 		/// to this lzma_index structure.
 		lzma_vli index_list_size;
+
+		/// Total size of all but the last Stream and all Stream
+		/// Padding fields.
+		lzma_vli streams_size;
 	} old;
 };
 
@@ -136,12 +144,12 @@ lzma_index_init(lzma_index *i, lzma_allocator *allocator)
 	i->uncompressed_size = 0;
 	i->count = 0;
 	i->index_list_size = 0;
-	i->padding_size = 0;
 	i->head = NULL;
 	i->tail = NULL;
 	i->current.group = NULL;
 	i->old.count = 0;
 	i->old.index_list_size = 0;
+	i->old.streams_size = 0;
 
 	return i;
 }
@@ -195,12 +203,12 @@ lzma_index_file_size(const lzma_index *i)
 {
 	// If multiple Streams are concatenated, the Stream Header, Index,
 	// and Stream Footer fields of all but the last Stream are already
-	// included in padding_size. Thus, we need to calculate only the
+	// included in old.streams_size. Thus, we need to calculate only the
 	// size of the last Index, not all Indexes.
-	return i->total_size + i->padding_size
+	return i->old.streams_size + LZMA_STREAM_HEADER_SIZE + i->total_size
 			+ index_size(i->count - i->old.count,
 				i->index_list_size - i->old.index_list_size)
-			+ LZMA_STREAM_HEADER_SIZE * 2;
+			+ LZMA_STREAM_HEADER_SIZE;
 }
 
 
@@ -219,10 +227,11 @@ lzma_index_padding_size(const lzma_index *i)
 }
 
 
-/// Helper function for index_append()
+/// Appends a new Record to the Index. If needed, this allocates a new
+/// Record group.
 static lzma_ret
 index_append_real(lzma_index *i, lzma_allocator *allocator,
-		lzma_vli total_size, lzma_vli uncompressed_size,
+		lzma_vli unpadded_size, lzma_vli uncompressed_size,
 		bool is_padding)
 {
 	// Add the new record.
@@ -237,7 +246,7 @@ index_append_real(lzma_index *i, lzma_allocator *allocator,
 		g->prev = i->tail;
 		g->next = NULL;
 		g->last = 0;
-		g->total_sums[0] = total_size;
+		g->unpadded_sums[0] = unpadded_size;
 		g->uncompressed_sums[0] = uncompressed_size;
 		g->paddings[0] = is_padding;
 
@@ -252,9 +261,9 @@ index_append_real(lzma_index *i, lzma_allocator *allocator,
 
 	} else {
 		// i->tail has space left for at least one record.
-		i->tail->total_sums[i->tail->last + 1]
-				= i->tail->total_sums[i->tail->last]
-					+ total_size;
+		i->tail->unpadded_sums[i->tail->last + 1]
+				= unpadded_size + vli_ceil4(
+					i->tail->unpadded_sums[i->tail->last]);
 		i->tail->uncompressed_sums[i->tail->last + 1]
 				= i->tail->uncompressed_sums[i->tail->last]
 					+ uncompressed_size;
@@ -266,13 +275,14 @@ index_append_real(lzma_index *i, lzma_allocator *allocator,
 }
 
 
-static lzma_ret
-index_append(lzma_index *i, lzma_allocator *allocator, lzma_vli total_size,
-		lzma_vli uncompressed_size, bool is_padding)
+extern LZMA_API lzma_ret
+lzma_index_append(lzma_index *i, lzma_allocator *allocator,
+		lzma_vli unpadded_size, lzma_vli uncompressed_size)
 {
-	if (total_size > LZMA_VLI_MAX
+	if (unpadded_size < UNPADDED_SIZE_MIN
+			|| unpadded_size > UNPADDED_SIZE_MAX
 			|| uncompressed_size > LZMA_VLI_MAX)
-		return LZMA_DATA_ERROR;
+		return LZMA_PROG_ERROR;
 
 	// This looks a bit ugly. We want to first validate that the Index
 	// and Stream stay in valid limits after adding this Record. After
@@ -280,65 +290,38 @@ index_append(lzma_index *i, lzma_allocator *allocator, lzma_vli total_size,
 	// slightly more correct to validate before allocating, YMMV).
 	lzma_ret ret;
 
-	if (is_padding) {
-		assert(uncompressed_size == 0);
+	// First update the overall info so we can validate it.
+	const lzma_vli index_list_size_add = lzma_vli_size(unpadded_size)
+			+ lzma_vli_size(uncompressed_size);
 
-		// First update the info so we can validate it.
-		i->padding_size += total_size;
-
-		if (i->padding_size > LZMA_VLI_MAX
-				|| lzma_index_file_size(i) > LZMA_VLI_MAX)
-			ret = LZMA_DATA_ERROR; // Would grow past the limits.
-		else
-			ret = index_append_real(i, allocator,
-					total_size, uncompressed_size, true);
-
-		// If something went wrong, undo the updated value.
-		if (ret != LZMA_OK)
-			i->padding_size -= total_size;
+	const lzma_vli total_size = vli_ceil4(unpadded_size);
 
-	} else {
-		// First update the overall info so we can validate it.
-		const lzma_vli index_list_size_add
-				= lzma_vli_size(total_size / 4 - 1)
-				+ lzma_vli_size(uncompressed_size);
-
-		i->total_size += total_size;
-		i->uncompressed_size += uncompressed_size;
-		++i->count;
-		i->index_list_size += index_list_size_add;
-
-		if (i->total_size > LZMA_VLI_MAX
-				|| i->uncompressed_size > LZMA_VLI_MAX
-				|| lzma_index_size(i) > LZMA_BACKWARD_SIZE_MAX
-				|| lzma_index_file_size(i) > LZMA_VLI_MAX)
-			ret = LZMA_DATA_ERROR; // Would grow past the limits.
-		else
-			ret = index_append_real(i, allocator,
-					total_size, uncompressed_size, false);
+	i->total_size += total_size;
+	i->uncompressed_size += uncompressed_size;
+	++i->count;
+	i->index_list_size += index_list_size_add;
 
-		if (ret != LZMA_OK) {
-			// Something went wrong. Undo the updates.
-			i->total_size -= total_size;
-			i->uncompressed_size -= uncompressed_size;
-			--i->count;
-			i->index_list_size -= index_list_size_add;
-		}
+	if (i->total_size > LZMA_VLI_MAX
+			|| i->uncompressed_size > LZMA_VLI_MAX
+			|| lzma_index_size(i) > LZMA_BACKWARD_SIZE_MAX
+			|| lzma_index_file_size(i) > LZMA_VLI_MAX)
+		ret = LZMA_DATA_ERROR; // Would grow past the limits.
+	else
+		ret = index_append_real(i, allocator, unpadded_size,
+				uncompressed_size, false);
+
+	if (ret != LZMA_OK) {
+		// Something went wrong. Undo the updates.
+		i->total_size -= total_size;
+		i->uncompressed_size -= uncompressed_size;
+		--i->count;
+		i->index_list_size -= index_list_size_add;
 	}
 
 	return ret;
 }
 
 
-extern LZMA_API lzma_ret
-lzma_index_append(lzma_index *i, lzma_allocator *allocator,
-		lzma_vli total_size, lzma_vli uncompressed_size)
-{
-	return index_append(i, allocator,
-			total_size, uncompressed_size, false);
-}
-
-
 /// Initialize i->current to point to the first Record.
 static bool
 init_current(lzma_index *i)
@@ -370,10 +353,10 @@ previous_group(lzma_index *i)
 	i->current.record = i->current.group->last;
 
 	// Then update the offsets.
-	i->current.stream_offset -= i->current.group
-			->total_sums[i->current.group->last];
-	i->current.uncompressed_offset -= i->current.group
-			->uncompressed_sums[i->current.group->last];
+	i->current.stream_offset -= vli_ceil4(i->current.group->unpadded_sums[
+			i->current.group->last]);
+	i->current.uncompressed_offset -= i->current.group->uncompressed_sums[
+			i->current.group->last];
 
 	return;
 }
@@ -386,8 +369,8 @@ next_group(lzma_index *i)
 	assert(i->current.group->next != NULL);
 
 	// Update the offsets first.
-	i->current.stream_offset += i->current.group
-			->total_sums[i->current.group->last];
+	i->current.stream_offset += vli_ceil4(i->current.group->unpadded_sums[
+			i->current.group->last]);
 	i->current.uncompressed_offset += i->current.group
 			->uncompressed_sums[i->current.group->last];
 
@@ -403,30 +386,39 @@ next_group(lzma_index *i)
 static void
 set_info(const lzma_index *i, lzma_index_record *info)
 {
-	info->total_size = i->current.group->total_sums[i->current.record];
+	// First copy the cumulative sizes from the current Record of the
+	// current group.
+	info->unpadded_size
+			= i->current.group->unpadded_sums[i->current.record];
+	info->total_size = vli_ceil4(info->unpadded_size);
 	info->uncompressed_size = i->current.group->uncompressed_sums[
 			i->current.record];
 
+	// Copy the start offsets of this group.
 	info->stream_offset = i->current.stream_offset;
 	info->uncompressed_offset = i->current.uncompressed_offset;
 
 	// If it's not the first Record in this group, we need to do some
 	// adjustements.
 	if (i->current.record > 0) {
-		// _sums[] are cumulative, thus we need to substract the
-		// _previous _sums[] to get the sizes of this Record.
-		info->total_size -= i->current.group
-				->total_sums[i->current.record - 1];
-		info->uncompressed_size -= i->current.group
+		// Since the _sums[] are cumulative, we substract the sums of
+		// the previous Record to get the sizes of the current Record,
+		// and add the sums of the previous Record to the offsets.
+		// With unpadded_sums[] we need to take into account that it
+		// uses a bit weird way to do the cumulative summing
+		const lzma_vli total_sum
+				= vli_ceil4(i->current.group->unpadded_sums[
+					i->current.record - 1]);
+
+		const lzma_vli uncompressed_sum = i->current.group
 				->uncompressed_sums[i->current.record - 1];
 
-		// i->current.{total,uncompressed}_offsets have the offset
-		// of the beginning of the group, thus we need to add the
-		// appropriate amount to get the offsetes of this Record.
-		info->stream_offset += i->current.group
-				->total_sums[i->current.record - 1];
-		info->uncompressed_offset += i->current.group
-				->uncompressed_sums[i->current.record - 1];
+		info->total_size -= total_sum;
+		info->unpadded_size -= total_sum;
+		info->uncompressed_size -= uncompressed_sum;
+
+		info->stream_offset += total_sum;
+		info->uncompressed_offset += uncompressed_sum;
 	}
 
 	return;
@@ -548,11 +540,22 @@ lzma_index_cat(lzma_index *restrict dest, lzma_index *restrict src,
 
 	// Check that the combined size of the Indexes stays within limits.
 	{
+		const lzma_vli dest_size = index_size_unpadded(
+				dest->count, dest->index_list_size);
+		const lzma_vli src_size = index_size_unpadded(
+				src->count, src->index_list_size);
+		if (vli_ceil4(dest_size + src_size) > LZMA_BACKWARD_SIZE_MAX)
+			return LZMA_DATA_ERROR;
+	}
+
+	// Check that the combined size of the "files" (combined total
+	// encoded sizes) stays within limits.
+	{
 		const lzma_vli dest_size = lzma_index_file_size(dest);
 		const lzma_vli src_size = lzma_index_file_size(src);
-		if (dest_size + src_size > LZMA_VLI_UNKNOWN
+		if (dest_size + src_size > LZMA_VLI_MAX
 				|| dest_size + src_size + padding
-					> LZMA_VLI_UNKNOWN)
+					> LZMA_VLI_MAX)
 			return LZMA_DATA_ERROR;
 	}
 
@@ -561,17 +564,37 @@ lzma_index_cat(lzma_index *restrict dest, lzma_index *restrict src,
 	//
 	// NOTE: This cannot overflow, because Index Size is always
 	// far smaller than LZMA_VLI_MAX, and adding two VLIs
-	// (Index Size and padding) doesn't overflow. It may become
-	// an invalid VLI if padding is huge, but that is caught by
-	// index_append().
+	// (Index Size and padding) doesn't overflow.
 	padding += index_size(dest->count - dest->old.count,
 				dest->index_list_size
 					- dest->old.index_list_size)
 			+ LZMA_STREAM_HEADER_SIZE * 2;
 
+	// While the above cannot overflow, but it may become an invalid VLI.
+	if (padding > LZMA_VLI_MAX)
+		return LZMA_DATA_ERROR;
+
 	// Add the padding Record.
-	return_if_error(index_append(
-			dest, allocator, padding, 0, true));
+	{
+		lzma_ret ret;
+
+		// First update the info so we can validate it.
+		dest->old.streams_size += padding;
+
+		if (dest->old.streams_size > LZMA_VLI_MAX
+				|| lzma_index_file_size(dest) > LZMA_VLI_MAX)
+			ret = LZMA_DATA_ERROR; // Would grow past the limits.
+		else
+			ret = index_append_real(dest, allocator,
+					padding, 0, true);
+
+		// If something went wrong, undo the updated value and return
+		// the error.
+		if (ret != LZMA_OK) {
+			dest->old.streams_size -= padding;
+			return ret;
+		}
+	}
 
 	// Avoid wasting lots of memory if src->head has only a few records
 	// that fit into dest->tail. That is, combine two groups if possible.
@@ -581,9 +604,10 @@ lzma_index_cat(lzma_index *restrict dest, lzma_index *restrict src,
 	if (src->head != NULL && src->head->last + 1
 			<= INDEX_GROUP_SIZE - dest->tail->last - 1) {
 		// Copy the first Record.
-		dest->tail->total_sums[dest->tail->last + 1]
-			= dest->tail->total_sums[dest->tail->last]
-				+ src->head->total_sums[0];
+		dest->tail->unpadded_sums[dest->tail->last + 1]
+				= vli_ceil4(dest->tail->unpadded_sums[
+					dest->tail->last])
+				+ src->head->unpadded_sums[0];
 
 		dest->tail->uncompressed_sums[dest->tail->last + 1]
 			= dest->tail->uncompressed_sums[dest->tail->last]
@@ -596,10 +620,11 @@ lzma_index_cat(lzma_index *restrict dest, lzma_index *restrict src,
 
 		// Copy the rest.
 		for (size_t i = 1; i < src->head->last; ++i) {
-			dest->tail->total_sums[dest->tail->last + 1]
-				= dest->tail->total_sums[dest->tail->last]
-					+ src->head->total_sums[i + 1]
-					- src->head->total_sums[i];
+			dest->tail->unpadded_sums[dest->tail->last + 1]
+				= vli_ceil4(dest->tail->unpadded_sums[
+						dest->tail->last])
+					+ src->head->unpadded_sums[i + 1]
+					- src->head->unpadded_sums[i];
 
 			dest->tail->uncompressed_sums[dest->tail->last + 1]
 				= dest->tail->uncompressed_sums[
@@ -636,13 +661,13 @@ lzma_index_cat(lzma_index *restrict dest, lzma_index *restrict src,
 	dest->old.count = dest->count + src->old.count;
 	dest->old.index_list_size
 			= dest->index_list_size + src->old.index_list_size;
+	dest->old.streams_size += src->old.streams_size;
 
 	// Update overall information.
 	dest->total_size += src->total_size;
 	dest->uncompressed_size += src->uncompressed_size;
 	dest->count += src->count;
 	dest->index_list_size += src->index_list_size;
-	dest->padding_size += src->padding_size;
 
 	// *src has nothing left but the base structure.
 	lzma_free(src, allocator);
@@ -690,7 +715,7 @@ lzma_index_dup(const lzma_index *src, lzma_allocator *allocator)
 
 		// Copy the arrays so that we don't read uninitialized memory.
 		const size_t count = src_group->last + 1;
-		memcpy(dest_group->total_sums, src_group->total_sums,
+		memcpy(dest_group->unpadded_sums, src_group->unpadded_sums,
 				sizeof(lzma_vli) * count);
 		memcpy(dest_group->uncompressed_sums,
 				src_group->uncompressed_sums,
@@ -729,8 +754,8 @@ lzma_index_equal(const lzma_index *a, const lzma_index *b)
 	while (ag != NULL && bg != NULL) {
 		const size_t count = ag->last + 1;
 		if (ag->last != bg->last
-				|| memcmp(ag->total_sums,
-					bg->total_sums,
+				|| memcmp(ag->unpadded_sums,
+					bg->unpadded_sums,
 					sizeof(lzma_vli) * count) != 0
 				|| memcmp(ag->uncompressed_sums,
 					bg->uncompressed_sums,
diff --git a/src/liblzma/common/index.h b/src/liblzma/common/index.h
index df897367..79719dd7 100644
--- a/src/liblzma/common/index.h
+++ b/src/liblzma/common/index.h
@@ -23,14 +23,11 @@
 #include "common.h"
 
 
-/// Maximum encoded value of Total Size.
-#define TOTAL_SIZE_ENCODED_MAX (LZMA_VLI_MAX / 4 - 1)
+/// Minimum Unpadded Size
+#define UNPADDED_SIZE_MIN LZMA_VLI_C(5)
 
-/// Convert the real Total Size value to a value that is stored to the Index.
-#define total_size_encode(size) ((size) / 4 - 1)
-
-/// Convert the encoded Total Size value from Index to the real Total Size.
-#define total_size_decode(size) (((size) + 1) * 4)
+/// Maximum Unpadded Size
+#define UNPADDED_SIZE_MAX (LZMA_VLI_MAX & ~LZMA_VLI_C(3))
 
 
 /// Get the size of the Index Padding field. This is needed by Index encoder
@@ -38,6 +35,16 @@
 extern uint32_t lzma_index_padding_size(const lzma_index *i);
 
 
+/// Round the variable-length integer to the next multiple of four.
+static inline lzma_vli
+vli_ceil4(lzma_vli vli)
+{
+	assert(vli <= LZMA_VLI_MAX);
+	return (vli + 3) & ~LZMA_VLI_C(3);
+}
+
+
+/// Calculate the size of the Index field excluding Index Padding
 static inline lzma_vli
 index_size_unpadded(lzma_vli count, lzma_vli index_list_size)
 {
@@ -46,20 +53,20 @@ index_size_unpadded(lzma_vli count, lzma_vli index_list_size)
 }
 
 
+/// Calculate the size of the Index field including Index Padding
 static inline lzma_vli
 index_size(lzma_vli count, lzma_vli index_list_size)
 {
-	// Round up to a mulitiple of four.
-	return (index_size_unpadded(count, index_list_size) + 3)
-			& ~LZMA_VLI_C(3);
+	return vli_ceil4(index_size_unpadded(count, index_list_size));
 }
 
 
+/// Calculate the total size of the Stream
 static inline lzma_vli
-index_stream_size(
-		lzma_vli total_size, lzma_vli count, lzma_vli index_list_size)
+index_stream_size(lzma_vli blocks_size,
+		lzma_vli count, lzma_vli index_list_size)
 {
-	return LZMA_STREAM_HEADER_SIZE + total_size
+	return LZMA_STREAM_HEADER_SIZE + blocks_size
 			+ index_size(count, index_list_size)
 			+ LZMA_STREAM_HEADER_SIZE;
 }
diff --git a/src/liblzma/common/index_decoder.c b/src/liblzma/common/index_decoder.c
index ae66595a..5faac161 100644
--- a/src/liblzma/common/index_decoder.c
+++ b/src/liblzma/common/index_decoder.c
@@ -25,7 +25,7 @@ struct lzma_coder_s {
 	enum {
 		SEQ_INDICATOR,
 		SEQ_COUNT,
-		SEQ_TOTAL,
+		SEQ_UNPADDED,
 		SEQ_UNCOMPRESSED,
 		SEQ_PADDING_INIT,
 		SEQ_PADDING,
@@ -38,8 +38,8 @@ struct lzma_coder_s {
 	/// Number of Records left to decode.
 	lzma_vli count;
 
-	/// The most recent Total Size field
-	lzma_vli total_size;
+	/// The most recent Unpadded Size field
+	lzma_vli unpadded_size;
 
 	/// The most recent Uncompressed Size field
 	lzma_vli uncompressed_size;
@@ -91,14 +91,14 @@ index_decode(lzma_coder *coder, lzma_allocator *allocator,
 		ret = LZMA_OK;
 		coder->pos = 0;
 		coder->sequence = coder->count == 0
-				? SEQ_PADDING_INIT : SEQ_TOTAL;
+				? SEQ_PADDING_INIT : SEQ_UNPADDED;
 		break;
 	}
 
-	case SEQ_TOTAL:
+	case SEQ_UNPADDED:
 	case SEQ_UNCOMPRESSED: {
-		lzma_vli *size = coder->sequence == SEQ_TOTAL
-				? &coder->total_size
+		lzma_vli *size = coder->sequence == SEQ_UNPADDED
+				? &coder->unpadded_size
 				: &coder->uncompressed_size;
 
 		ret = lzma_vli_decode(size, &coder->pos,
@@ -109,27 +109,26 @@ index_decode(lzma_coder *coder, lzma_allocator *allocator,
 		ret = LZMA_OK;
 		coder->pos = 0;
 
-		if (coder->sequence == SEQ_TOTAL) {
-			// Validate that encoded Total Size isn't too big.
-			if (coder->total_size > TOTAL_SIZE_ENCODED_MAX)
+		if (coder->sequence == SEQ_UNPADDED) {
+			// Validate that encoded Unpadded Size isn't too small
+			// or too big.
+			if (coder->unpadded_size < UNPADDED_SIZE_MIN
+					|| coder->unpadded_size
+						> UNPADDED_SIZE_MAX)
 				return LZMA_DATA_ERROR;
 
-			// Convert the encoded Total Size to the real
-			// Total Size.
-			coder->total_size = total_size_decode(
-					coder->total_size);
 			coder->sequence = SEQ_UNCOMPRESSED;
 		} else {
 			// Add the decoded Record to the Index.
 			return_if_error(lzma_index_append(
 					coder->index, allocator,
-					coder->total_size,
+					coder->unpadded_size,
 					coder->uncompressed_size));
 
 			// Check if this was the last Record.
 			coder->sequence = --coder->count == 0
 					? SEQ_PADDING_INIT
-					: SEQ_TOTAL;
+					: SEQ_UNPADDED;
 		}
 
 		break;
diff --git a/src/liblzma/common/index_encoder.c b/src/liblzma/common/index_encoder.c
index 3005f835..522dbb53 100644
--- a/src/liblzma/common/index_encoder.c
+++ b/src/liblzma/common/index_encoder.c
@@ -26,7 +26,7 @@ struct lzma_coder_s {
 	enum {
 		SEQ_INDICATOR,
 		SEQ_COUNT,
-		SEQ_TOTAL,
+		SEQ_UNPADDED,
 		SEQ_UNCOMPRESSED,
 		SEQ_NEXT,
 		SEQ_PADDING,
@@ -97,18 +97,20 @@ index_encode(lzma_coder *coder,
 			break;
 		}
 
-		// Total Size must be a multiple of four.
-		if (coder->record.total_size & 3)
+		// Unpadded Size must be within valid limits.
+		if (coder->record.unpadded_size < UNPADDED_SIZE_MIN
+				|| coder->record.unpadded_size
+					> UNPADDED_SIZE_MAX)
 			return LZMA_PROG_ERROR;
 
-		coder->sequence = SEQ_TOTAL;
+		coder->sequence = SEQ_UNPADDED;
 
 	// Fall through
 
-	case SEQ_TOTAL:
+	case SEQ_UNPADDED:
 	case SEQ_UNCOMPRESSED: {
-		const lzma_vli size = coder->sequence == SEQ_TOTAL
-				? total_size_encode(coder->record.total_size)
+		const lzma_vli size = coder->sequence == SEQ_UNPADDED
+				? coder->record.unpadded_size
 				: coder->record.uncompressed_size;
 
 		ret = lzma_vli_encode(size, &coder->pos,
diff --git a/src/liblzma/common/index_hash.c b/src/liblzma/common/index_hash.c
index 5e581838..162094d1 100644
--- a/src/liblzma/common/index_hash.c
+++ b/src/liblzma/common/index_hash.c
@@ -23,8 +23,8 @@
 
 
 typedef struct {
-	/// Sum of the Total Size fields
-	lzma_vli total_size;
+	/// Sum of the Block sizes (including Block Padding)
+	lzma_vli blocks_size;
 
 	/// Sum of the Uncompressed Size fields
 	lzma_vli uncompressed_size;
@@ -35,7 +35,7 @@ typedef struct {
 	/// Size of the List of Index Records as bytes
 	lzma_vli index_list_size;
 
-	/// Check calculated from Total Sizes and Uncompressed Sizes.
+	/// Check calculated from Unpadded Sizes and Uncompressed Sizes.
 	lzma_check_state check;
 
 } lzma_index_hash_info;
@@ -45,7 +45,7 @@ struct lzma_index_hash_s {
 	enum {
 		SEQ_BLOCK,
 		SEQ_COUNT,
-		SEQ_TOTAL,
+		SEQ_UNPADDED,
 		SEQ_UNCOMPRESSED,
 		SEQ_PADDING_INIT,
 		SEQ_PADDING,
@@ -61,8 +61,8 @@ struct lzma_index_hash_s {
 	/// Number of Records not fully decoded
 	lzma_vli remaining;
 
-	/// Total Size currently being read from an Index Record.
-	lzma_vli total_size;
+	/// Unpadded Size currently being read from an Index Record.
+	lzma_vli unpadded_size;
 
 	/// Uncompressed Size currently being read from an Index Record.
 	lzma_vli uncompressed_size;
@@ -86,15 +86,15 @@ lzma_index_hash_init(lzma_index_hash *index_hash, lzma_allocator *allocator)
 	}
 
 	index_hash->sequence = SEQ_BLOCK;
-	index_hash->blocks.total_size = 0;
+	index_hash->blocks.blocks_size = 0;
 	index_hash->blocks.uncompressed_size = 0;
 	index_hash->blocks.count = 0;
 	index_hash->blocks.index_list_size = 0;
-	index_hash->records.total_size = 0;
+	index_hash->records.blocks_size = 0;
 	index_hash->records.uncompressed_size = 0;
 	index_hash->records.count = 0;
 	index_hash->records.index_list_size = 0;
-	index_hash->total_size = 0;
+	index_hash->unpadded_size = 0;
 	index_hash->uncompressed_size = 0;
 	index_hash->pos = 0;
 	index_hash->crc32 = 0;
@@ -128,16 +128,16 @@ lzma_index_hash_size(const lzma_index_hash *index_hash)
 
 /// Updates the sizes and the hash without any validation.
 static lzma_ret
-hash_append(lzma_index_hash_info *info, lzma_vli total_size,
+hash_append(lzma_index_hash_info *info, lzma_vli unpadded_size,
 		lzma_vli uncompressed_size)
 {
-	info->total_size += total_size;
+	info->blocks_size += vli_ceil4(unpadded_size);
 	info->uncompressed_size += uncompressed_size;
-	info->index_list_size += lzma_vli_size(total_size_encode(total_size))
+	info->index_list_size += lzma_vli_size(unpadded_size)
 			+ lzma_vli_size(uncompressed_size);
 	++info->count;
 
-	const lzma_vli sizes[2] = { total_size, uncompressed_size };
+	const lzma_vli sizes[2] = { unpadded_size, uncompressed_size };
 	lzma_check_update(&info->check, LZMA_CHECK_BEST,
 			(const uint8_t *)(sizes), sizeof(sizes));
 
@@ -146,26 +146,27 @@ hash_append(lzma_index_hash_info *info, lzma_vli total_size,
 
 
 extern LZMA_API lzma_ret
-lzma_index_hash_append(lzma_index_hash *index_hash, lzma_vli total_size,
+lzma_index_hash_append(lzma_index_hash *index_hash, lzma_vli unpadded_size,
 		lzma_vli uncompressed_size)
 {
 	// Validate the arguments.
-	if (index_hash->sequence != SEQ_BLOCK || total_size == 0
-			|| total_size > LZMA_VLI_MAX || (total_size & 3)
+	if (index_hash->sequence != SEQ_BLOCK
+			|| unpadded_size < UNPADDED_SIZE_MIN
+			|| unpadded_size > UNPADDED_SIZE_MAX
 			|| uncompressed_size > LZMA_VLI_MAX)
 		return LZMA_PROG_ERROR;
 
 	// Update the hash.
 	return_if_error(hash_append(&index_hash->blocks,
-			total_size, uncompressed_size));
+			unpadded_size, uncompressed_size));
 
 	// Validate the properties of *info are still in allowed limits.
-	if (index_hash->blocks.total_size > LZMA_VLI_MAX
+	if (index_hash->blocks.blocks_size > LZMA_VLI_MAX
 			|| index_hash->blocks.uncompressed_size > LZMA_VLI_MAX
 			|| index_size(index_hash->blocks.count,
 					index_hash->blocks.index_list_size)
 				> LZMA_BACKWARD_SIZE_MAX
-			|| index_stream_size(index_hash->blocks.total_size,
+			|| index_stream_size(index_hash->blocks.blocks_size,
 					index_hash->blocks.count,
 					index_hash->blocks.index_list_size)
 				> LZMA_VLI_MAX)
@@ -216,14 +217,14 @@ lzma_index_hash_decode(lzma_index_hash *index_hash, const uint8_t *in,
 
 		// Handle the special case when there are no Blocks.
 		index_hash->sequence = index_hash->remaining == 0
-				? SEQ_PADDING_INIT : SEQ_TOTAL;
+				? SEQ_PADDING_INIT : SEQ_UNPADDED;
 		break;
 	}
 
-	case SEQ_TOTAL:
+	case SEQ_UNPADDED:
 	case SEQ_UNCOMPRESSED: {
-		lzma_vli *size = index_hash->sequence == SEQ_TOTAL
-				? &index_hash->total_size
+		lzma_vli *size = index_hash->sequence == SEQ_UNPADDED
+				? &index_hash->unpadded_size
 				: &index_hash->uncompressed_size;
 
 		ret = lzma_vli_decode(size, &index_hash->pos,
@@ -234,18 +235,17 @@ lzma_index_hash_decode(lzma_index_hash *index_hash, const uint8_t *in,
 		ret = LZMA_OK;
 		index_hash->pos = 0;
 
-		if (index_hash->sequence == SEQ_TOTAL) {
-			if (index_hash->total_size > TOTAL_SIZE_ENCODED_MAX)
+		if (index_hash->sequence == SEQ_UNPADDED) {
+			if (index_hash->unpadded_size < UNPADDED_SIZE_MIN
+					|| index_hash->unpadded_size
+						> UNPADDED_SIZE_MAX)
 				return LZMA_DATA_ERROR;
 
-			index_hash->total_size = total_size_decode(
-					index_hash->total_size);
-
 			index_hash->sequence = SEQ_UNCOMPRESSED;
 		} else {
 			// Update the hash.
 			return_if_error(hash_append(&index_hash->records,
-					index_hash->total_size,
+					index_hash->unpadded_size,
 					index_hash->uncompressed_size));
 
 			// Verify that we don't go over the known sizes. Note
@@ -254,8 +254,8 @@ lzma_index_hash_decode(lzma_index_hash *index_hash, const uint8_t *in,
 			// that values in index_hash->blocks are already
 			// validated and we are fine as long as we don't
 			// exceed them in index_hash->records.
-			if (index_hash->blocks.total_size
-					< index_hash->records.total_size
+			if (index_hash->blocks.blocks_size
+					< index_hash->records.blocks_size
 					|| index_hash->blocks.uncompressed_size
 					< index_hash->records.uncompressed_size
 					|| index_hash->blocks.index_list_size
@@ -264,7 +264,7 @@ lzma_index_hash_decode(lzma_index_hash *index_hash, const uint8_t *in,
 
 			// Check if this was the last Record.
 			index_hash->sequence = --index_hash->remaining == 0
-					? SEQ_PADDING_INIT : SEQ_TOTAL;
+					? SEQ_PADDING_INIT : SEQ_UNPADDED;
 		}
 
 		break;
@@ -288,8 +288,8 @@ lzma_index_hash_decode(lzma_index_hash *index_hash, const uint8_t *in,
 		}
 
 		// Compare the sizes.
-		if (index_hash->blocks.total_size
-				!= index_hash->records.total_size
+		if (index_hash->blocks.blocks_size
+				!= index_hash->records.blocks_size
 				|| index_hash->blocks.uncompressed_size
 				!= index_hash->records.uncompressed_size
 				|| index_hash->blocks.index_list_size
diff --git a/src/liblzma/common/stream_decoder.c b/src/liblzma/common/stream_decoder.c
index e137685f..9be47893 100644
--- a/src/liblzma/common/stream_decoder.c
+++ b/src/liblzma/common/stream_decoder.c
@@ -190,7 +190,7 @@ stream_decode(lzma_coder *coder, lzma_allocator *allocator,
 		// Set up a buffer to hold the filter chain. Block Header
 		// decoder will initialize all members of this array so
 		// we don't need to do it here.
-		lzma_filter filters[LZMA_BLOCK_FILTERS_MAX + 1];
+		lzma_filter filters[LZMA_FILTERS_MAX + 1];
 		coder->block_options.filters = filters;
 
 		// Decode the Block Header.
@@ -216,7 +216,7 @@ stream_decode(lzma_coder *coder, lzma_allocator *allocator,
 
 		// Free the allocated filter options since they are needed
 		// only to initialize the Block decoder.
-		for (size_t i = 0; i < LZMA_BLOCK_FILTERS_MAX; ++i)
+		for (size_t i = 0; i < LZMA_FILTERS_MAX; ++i)
 			lzma_free(filters[i].options, allocator);
 
 		coder->block_options.filters = NULL;
@@ -243,7 +243,7 @@ stream_decode(lzma_coder *coder, lzma_allocator *allocator,
 		// Block decoded successfully. Add the new size pair to
 		// the Index hash.
 		return_if_error(lzma_index_hash_append(coder->index_hash,
-				lzma_block_total_size_get(
+				lzma_block_unpadded_size(
 					&coder->block_options),
 				coder->block_options.uncompressed_size));
 
@@ -270,7 +270,7 @@ stream_decode(lzma_coder *coder, lzma_allocator *allocator,
 
 	// Fall through
 
-	case SEQ_STREAM_FOOTER:
+	case SEQ_STREAM_FOOTER: {
 		// Copy the Stream Footer to the internal buffer.
 		lzma_bufcpy(in, in_pos, in_size, coder->buffer, &coder->pos,
 				LZMA_STREAM_HEADER_SIZE);
@@ -306,6 +306,7 @@ stream_decode(lzma_coder *coder, lzma_allocator *allocator,
 			return LZMA_STREAM_END;
 
 		coder->sequence = SEQ_STREAM_PADDING;
+	}
 
 	// Fall through
 
diff --git a/src/liblzma/common/stream_encoder.c b/src/liblzma/common/stream_encoder.c
index 0376fd3b..e52ad692 100644
--- a/src/liblzma/common/stream_encoder.c
+++ b/src/liblzma/common/stream_encoder.c
@@ -157,11 +157,11 @@ stream_encode(lzma_coder *coder, lzma_allocator *allocator,
 			return ret;
 
 		// Add a new Index Record.
-		const lzma_vli total_size = lzma_block_total_size_get(
+		const lzma_vli unpadded_size = lzma_block_unpadded_size(
 				&coder->block_options);
-		assert(total_size != 0);
+		assert(unpadded_size != 0);
 		return_if_error(lzma_index_append(coder->index, allocator,
-				total_size,
+				unpadded_size,
 				coder->block_options.uncompressed_size));
 
 		coder->sequence = SEQ_BLOCK_INIT;
diff --git a/src/liblzma/lz/lz_decoder.h b/src/liblzma/lz/lz_decoder.h
index d2a77ba4..53ee1c1e 100644
--- a/src/liblzma/lz/lz_decoder.h
+++ b/src/liblzma/lz/lz_decoder.h
@@ -157,14 +157,14 @@ dict_repeat(lzma_dict *dict, uint32_t distance, uint32_t *len)
 		uint32_t copy_size = dict->size - copy_pos;
 
 		if (copy_size < left) {
-			memcpy(dict->buf + dict->pos, dict->buf + copy_pos,
+			memmove(dict->buf + dict->pos, dict->buf + copy_pos,
 					copy_size);
 			dict->pos += copy_size;
 			copy_size = left - copy_size;
 			memcpy(dict->buf + dict->pos, dict->buf, copy_size);
 			dict->pos += copy_size;
 		} else {
-			memcpy(dict->buf + dict->pos, dict->buf + copy_pos,
+			memmove(dict->buf + dict->pos, dict->buf + copy_pos,
 					left);
 			dict->pos += left;
 		}
diff --git a/src/liblzma/subblock/subblock_decoder.c b/src/liblzma/subblock/subblock_decoder.c
index 7cf06988..3096b442 100644
--- a/src/liblzma/subblock/subblock_decoder.c
+++ b/src/liblzma/subblock/subblock_decoder.c
@@ -211,7 +211,7 @@ decode_buffer(lzma_coder *coder, lzma_allocator *allocator,
 			break;
 		}
 
-		case FLAG_END_SUBFILTER:
+		case FLAG_END_SUBFILTER: {
 			if (coder->padding != 0 || (in[*in_pos] & 0x0F)
 					|| coder->subfilter.code == NULL
 					|| !coder->got_output_with_subfilter)
@@ -250,6 +250,7 @@ decode_buffer(lzma_coder *coder, lzma_allocator *allocator,
 			++*in_pos;
 
 			break;
+		}
 
 		default:
 			return LZMA_DATA_ERROR;