Add block processor data structure

The interface is designed for parallel, asynchronuous processing of data
blocks with an I/O callback that handles the serialized result.

The underlying implementation is currently still synchronuous.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

5 files changed, 259 insertions, 0 deletions

diff --git a/include/block_processor.h b/include/block_processor.h
new file mode 100644
index 0000000..4944969
--- /dev/null
+++ b/include/block_processor.h
@@ -0,0 +1,111 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * block_processor.h
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#ifndef BLOCK_PROCESSOR_H
+#define BLOCK_PROCESSOR_H
+
+#include "config.h"
+#include "compress.h"
+
+enum {
+	/* only calculate checksum, do NOT compress the data */
+	BLK_DONT_COMPRESS = 0x0001,
+
+	/* set by compressor worker if the block was actually compressed */
+	BLK_IS_COMPRESSED = 0x0002,
+
+	/* do not calculate block checksum */
+	BLK_DONT_CHECKSUM = 0x0004,
+
+	/* set by compressor worker if compression failed */
+	BLK_COMPRESS_ERROR = 0x0008,
+
+	/* first user setable block flag */
+	BLK_USER = 0x0080
+};
+
+typedef struct block_t {
+	/* used internally, ignored and overwritten when enqueueing blocks */
+	struct block_t *next;
+	uint32_t sequence_number;
+
+	/* Size of the data area */
+	uint32_t size;
+
+	/* checksum of the input data */
+	uint32_t checksum;
+
+	/* user settable file block index */
+	uint32_t index;
+
+	/* user pointer associated with the block */
+	void *user;
+
+	/* user settable flag field */
+	uint32_t flags;
+
+	/* raw data to be processed */
+	uint8_t data[];
+} block_t;
+
+typedef struct block_processor_t block_processor_t;
+
+/*
+  Gets called for each processed block. May be called from a different thread
+  than the one that calls enqueue, but only from one thread at a time.
+  Guaranteed to be called on blocks in the order that they are submitted
+  to enqueue.
+
+  A non-zero return value is interpreted as fatal error.
+ */
+typedef int (*block_cb)(void *user, block_t *blk);
+
+block_processor_t *block_processor_create(size_t max_block_size,
+					  compressor_t *cmp,
+					  unsigned int num_workers,
+					  void *user,
+					  block_cb callback);
+
+void block_processor_destroy(block_processor_t *proc);
+
+/*
+  Add a block to be processed. Returns non-zero on error and prints a message
+  to stderr.
+
+  The function takes over ownership of the submitted block. It is freed with
+  a after processing and calling the block callback.
+
+  Even on failure, the workers may still be running and
+  block_processor_finish must be called before cleaning up.
+*/
+int block_processor_enqueue(block_processor_t *proc, block_t *block);
+
+/*
+  Wait for the compressor workers to finish. Returns zero on success, non-zero
+  if an internal error occoured or one of the block callbacks returned a
+  non-zero value.
+ */
+int block_processor_finish(block_processor_t *proc);
+
+/*
+  Convenience function to create a block structure and optionally fill it with
+  content.
+
+  filename is used for printing error messages. If fd is a valid file
+  descriptor (>= 0), the function attempts to populate the payload data
+  from the input file.
+ */
+block_t *create_block(const char *filename, int fd, size_t size,
+		      void *user, uint32_t flags);
+
+/*
+  Convenience function to process a data block. Returns 0 on success,
+  prints to stderr on failure.
+ */
+int process_block(block_t *block, compressor_t *cmp,
+		  uint8_t *scratch, size_t scratch_size);
+
+#endif /* BLOCK_PROCESSOR_H */
diff --git a/lib/Makemodule.am b/lib/Makemodule.am
index 82c910c..7c2c99d 100644
--- a/lib/Makemodule.am
+++ b/lib/Makemodule.am
@@ -18,6 +18,8 @@ libtar_a_CFLAGS = $(AM_CFLAGS)
 libtar_a_CPPFLAGS = $(AM_CPPFLAGS)
 
 libcompress_a_SOURCES = lib/comp/compressor.c lib/comp/internal.h
+libcompress_a_SOURCES += lib/comp/block_processor.c include/block_processor.h
+libcompress_a_SOURCES += lib/comp/create_block.c lib/comp/process_block.c
 libcompress_a_SOURCES += include/compress.h
 libcompress_a_CFLAGS = $(AM_CFLAGS)
 libcompress_a_CPPFLAGS = $(AM_CPPFLAGS)
diff --git a/lib/comp/block_processor.c b/lib/comp/block_processor.c
new file mode 100644
index 0000000..5938b3a
--- /dev/null
+++ b/lib/comp/block_processor.c
@@ -0,0 +1,72 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * block_processor.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+
+#include "block_processor.h"
+#include "util.h"
+
+#include <string.h>
+#include <stdlib.h>
+#include <stdio.h>
+
+struct block_processor_t {
+	size_t max_block_size;
+	compressor_t *cmp;
+	block_cb cb;
+	void *user;
+	int status;
+
+	uint8_t scratch[];
+};
+
+block_processor_t *block_processor_create(size_t max_block_size,
+					  compressor_t *cmp,
+					  unsigned int num_workers,
+					  void *user,
+					  block_cb callback)
+{
+	block_processor_t *proc = calloc(1, sizeof(*proc) + max_block_size);
+	(void)num_workers;
+
+	if (proc == NULL) {
+		perror("Creating block processor");
+		return NULL;
+	}
+
+	proc->max_block_size = max_block_size;
+	proc->cmp = cmp;
+	proc->cb = callback;
+	proc->user = user;
+	return proc;
+}
+
+void block_processor_destroy(block_processor_t *proc)
+{
+	free(proc);
+}
+
+int block_processor_enqueue(block_processor_t *proc, block_t *block)
+{
+	if (process_block(block, proc->cmp,
+			  proc->scratch, proc->max_block_size))
+		goto fail;
+
+	if (proc->cb(proc->user, block))
+		goto fail;
+
+	free(block);
+	return 0;
+fail:
+	free(block);
+	proc->status = -1;
+	return -1;
+}
+
+int block_processor_finish(block_processor_t *proc)
+{
+	return proc->status;
+}
diff --git a/lib/comp/create_block.c b/lib/comp/create_block.c
new file mode 100644
index 0000000..e410091
--- /dev/null
+++ b/lib/comp/create_block.c
@@ -0,0 +1,37 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * create_block.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+
+#include "block_processor.h"
+#include "util.h"
+
+#include <string.h>
+#include <stdlib.h>
+#include <stdio.h>
+
+block_t *create_block(const char *filename, int fd, size_t size,
+		      void *user, uint32_t flags)
+{
+	block_t *blk = calloc(1, sizeof(*blk) + size);
+
+	if (blk == NULL) {
+		perror(filename);
+		return NULL;
+	}
+
+	if (fd >= 0) {
+		if (read_data(filename, fd, blk->data, size)) {
+			free(blk);
+			return NULL;
+		}
+	}
+
+	blk->size = size;
+	blk->user = user;
+	blk->flags = flags;
+	return blk;
+}
diff --git a/lib/comp/process_block.c b/lib/comp/process_block.c
new file mode 100644
index 0000000..76cd07d
--- /dev/null
+++ b/lib/comp/process_block.c
@@ -0,0 +1,37 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * process_block.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+
+#include "block_processor.h"
+#include "util.h"
+
+#include <string.h>
+
+int process_block(block_t *block, compressor_t *cmp,
+		  uint8_t *scratch, size_t scratch_size)
+{
+	ssize_t ret;
+
+	if (!(block->flags & BLK_DONT_CHECKSUM))
+		block->checksum = update_crc32(0, block->data, block->size);
+
+	if (!(block->flags & BLK_DONT_COMPRESS)) {
+		ret = cmp->do_block(cmp, block->data, block->size,
+				    scratch, scratch_size);
+
+		if (ret < 0)
+			return -1;
+
+		if (ret > 0) {
+			memcpy(block->data, scratch, ret);
+			block->size = ret;
+			block->flags |= BLK_IS_COMPRESSED;
+		}
+	}
+
+	return 0;
+}