Remove uncompressed size tracking from the filter encoders.

It's not strictly needed there, and just complicates the
code. LZ encoder never even had this feature.

The primary reason to have uncompressed size tracking in
filter encoders was validating that the application
doesn't give different amount of input that it had
promised. A side effect was to validate internal workings
of liblzma.

Uncompressed size tracking is still present in the Block
encoder. Maybe it should be added to LZMA_Alone and raw
encoders too. It's simpler to have one coder just to
validate the uncompressed size instead of having it
in every filter.

3 files changed, 12 insertions, 80 deletions

diff --git a/src/liblzma/common/copy_coder.c b/src/liblzma/common/copy_coder.c
index 41f327d1..0bd674f6 100644
--- a/src/liblzma/common/copy_coder.c
+++ b/src/liblzma/common/copy_coder.c
@@ -42,33 +42,12 @@ copy_encode(lzma_coder *coder, lzma_allocator *allocator,
 				in, in_pos, in_size, out, out_pos, out_size,
 				action);
 
-	// If we get here, we are the last filter in the chain.
-	assert(coder->uncompressed_size <= LZMA_VLI_VALUE_MAX);
-
-	const size_t in_avail = in_size - *in_pos;
-
-	// Check that we don't have too much input.
-	if ((lzma_vli)(in_avail) > coder->uncompressed_size)
-		return LZMA_DATA_ERROR;
-
-	// Check that once LZMA_FINISH has been given, the amount of input
-	// matches uncompressed_size, which is always known.
-	if (action == LZMA_FINISH
-			&& coder->uncompressed_size != (lzma_vli)(in_avail))
-		return LZMA_DATA_ERROR;
-
 	// We are the last coder in the chain.
 	// Just copy as much data as possible.
-	const size_t in_used = bufcpy(
-			in, in_pos, in_size, out, out_pos, out_size);
-
-	// Update uncompressed_size if it is known.
-	if (coder->uncompressed_size != LZMA_VLI_VALUE_UNKNOWN)
-		coder->uncompressed_size -= in_used;
+	bufcpy(in, in_pos, in_size, out, out_pos, out_size);
 
 	// LZMA_SYNC_FLUSH and LZMA_FINISH are the same thing for us.
-	if ((action != LZMA_RUN && *in_pos == in_size)
-			|| coder->uncompressed_size == 0)
+	if (action != LZMA_RUN && *in_pos == in_size)
 		return LZMA_STREAM_END;
 
 	return LZMA_OK;
diff --git a/src/liblzma/simple/simple_coder.c b/src/liblzma/simple/simple_coder.c
index f60dff34..6ecd119e 100644
--- a/src/liblzma/simple/simple_coder.c
+++ b/src/liblzma/simple/simple_coder.c
@@ -39,44 +39,23 @@ copy_or_code(lzma_coder *coder, lzma_allocator *allocator,
 	if (coder->next.code == NULL) {
 		const size_t in_avail = in_size - *in_pos;
 
-		if (coder->is_encoder) {
-			if (action == LZMA_FINISH) {
-				// If uncompressed size is known and the
-				// amount of available input doesn't match
-				// the uncompressed size, return an error.
-				if (coder->uncompressed_size
-						!= LZMA_VLI_VALUE_UNKNOWN
-						&& coder->uncompressed_size
-							!= in_avail)
-					return LZMA_DATA_ERROR;
-
-			} else if (coder->uncompressed_size
-					< (lzma_vli)(in_avail)) {
-				// There is too much input available.
-				return LZMA_DATA_ERROR;
-			}
-		} else {
+		if (!coder->is_encoder) {
 			// Limit in_size so that we don't copy too much.
 			if ((lzma_vli)(in_avail) > coder->uncompressed_size)
 				in_size = *in_pos + (size_t)(
 						coder->uncompressed_size);
 		}
 
-		// Store the old position so we can update uncompressed_size.
 		const size_t out_start = *out_pos;
-
-		// Copy the data
 		bufcpy(in, in_pos, in_size, out, out_pos, out_size);
 
-		// Update uncompressed_size.
-		if (coder->uncompressed_size != LZMA_VLI_VALUE_UNKNOWN)
-			coder->uncompressed_size -= *out_pos - out_start;
-
 		// Check if end of stream was reached.
 		if (coder->is_encoder) {
 			if (action == LZMA_FINISH && *in_pos == in_size)
 				coder->end_was_reached = true;
-		} else {
+		} else if (coder->uncompressed_size
+				!= LZMA_VLI_VALUE_UNKNOWN) {
+			coder->uncompressed_size -= *out_pos - out_start;
 			if (coder->uncompressed_size == 0)
 				coder->end_was_reached = true;
 		}
diff --git a/src/liblzma/subblock/subblock_encoder.c b/src/liblzma/subblock/subblock_encoder.c
index e1af4a45..96129d87 100644
--- a/src/liblzma/subblock/subblock_encoder.c
+++ b/src/liblzma/subblock/subblock_encoder.c
@@ -41,6 +41,7 @@ do { \
 struct lzma_coder_s {
 	lzma_next_coder next;
 	bool next_finished;
+	bool use_eopm;
 
 	enum {
 		SEQ_FILL,
@@ -62,8 +63,6 @@ struct lzma_coder_s {
 
 	lzma_options_subblock *options;
 
-	lzma_vli uncompressed_size;
-
 	size_t pos;
 	uint32_t tmp;
 
@@ -235,18 +234,6 @@ subblock_buffer(lzma_coder *coder, lzma_allocator *allocator,
 		size_t in_size, uint8_t *restrict out,
 		size_t *restrict out_pos, size_t out_size, lzma_action action)
 {
-	// Verify that there is a sane amount of input.
-	if (coder->uncompressed_size != LZMA_VLI_VALUE_UNKNOWN) {
-		const lzma_vli in_avail = in_size - *in_pos;
-		if (action == LZMA_FINISH) {
-			if (in_avail != coder->uncompressed_size)
-				return LZMA_DATA_ERROR;
-		} else {
-			if (in_avail > coder->uncompressed_size)
-				return LZMA_DATA_ERROR;
-		}
-	}
-
 	// Check if we need to do something special with the Subfilter.
 	if (coder->options != NULL && coder->options->allow_subfilters) {
 		switch (coder->options->subfilter_mode) {
@@ -304,18 +291,12 @@ subblock_buffer(lzma_coder *coder, lzma_allocator *allocator,
 			assert(coder->subfilter.subcoder.code == NULL);
 
 			// No Subfilter is enabled, just copy the data as is.
-			// NOTE: uncompressed_size cannot overflow because we
-			// have checked/ it in the beginning of this function.
-			const size_t in_used = bufcpy(in, in_pos, in_size,
+			coder->alignment.in_pending += bufcpy(
+					in, in_pos, in_size,
 					coder->subblock.data,
 					&coder->subblock.size,
 					coder->subblock.limit);
 
-			if (coder->uncompressed_size != LZMA_VLI_VALUE_UNKNOWN)
-				coder->uncompressed_size -= in_used;
-
-			coder->alignment.in_pending += in_used;
-
 		} else {
 			const size_t in_start = *in_pos;
 			lzma_ret ret;
@@ -351,11 +332,6 @@ subblock_buffer(lzma_coder *coder, lzma_allocator *allocator,
 			if (in_used > 0)
 				coder->subfilter.got_input = true;
 
-			// NOTE: uncompressed_size cannot overflow because we
-			// have checked it in the beginning of this function.
-			if (coder->uncompressed_size != LZMA_VLI_VALUE_UNKNOWN)
-				coder->uncompressed_size -= *in_pos - in_start;
-
 			coder->alignment.in_pending += in_used;
 
 			if (ret == LZMA_STREAM_END) {
@@ -509,14 +485,11 @@ subblock_buffer(lzma_coder *coder, lzma_allocator *allocator,
 				break;
 			}
 
-			if (coder->uncompressed_size
-					== LZMA_VLI_VALUE_UNKNOWN) {
+			if (coder->use_eopm) {
 				// NOTE: No need to use write_byte() here
 				// since we are finishing.
 				out[*out_pos] = 0x10;
 				++*out_pos;
-			} else if (coder->uncompressed_size != 0) {
-				return LZMA_DATA_ERROR;
 			}
 
 			return LZMA_STREAM_END;
@@ -782,7 +755,8 @@ lzma_subblock_encoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 	next->coder->next_finished = false;
 	next->coder->sequence = SEQ_FILL;
 	next->coder->options = filters[0].options;
-	next->coder->uncompressed_size = filters[0].uncompressed_size;
+	next->coder->use_eopm = filters[0].uncompressed_size
+			== LZMA_VLI_VALUE_UNKNOWN;
 	next->coder->pos = 0;
 
 	next->coder->alignment.in_pending = 0;