Refactored calc_file_checksum()

Author: CherkashinSergey

src/data.c

@@ -1418,75 +1418,13 @@ get_wal_file(const char *from_path, const char *to_path)
  * but created in process of backup, such as stream XLOG files,
  * PG_TABLESPACE_MAP_FILE and PG_BACKUP_LABEL_FILE.
  */
-bool
+void
 calc_file_checksum(pgFile *file)
 {
-	FILE	   *in;
-	size_t		read_len = 0;
-	int			errno_tmp;
-	char		buf[BLCKSZ];
-	struct stat	st;
-	pg_crc32	crc;
-
 	Assert(S_ISREG(file->mode));
-	INIT_TRADITIONAL_CRC32(crc);
-
-	/* reset size summary */
-	file->read_size = 0;
-	file->write_size = 0;
-
-	/* open backup mode file for read */
-	in = fopen(file->path, PG_BINARY_R);
-	if (in == NULL)
-	{
-		FIN_TRADITIONAL_CRC32(crc);
-		file->crc = crc;
-
-		/* maybe deleted, it's not error */
-		if (errno == ENOENT)
-			return false;
-
-		elog(ERROR, "cannot open source file \"%s\": %s", file->path,
-			 strerror(errno));
-	}
-
-	/* stat source file to change mode of destination file */
-	if (fstat(fileno(in), &st) == -1)
-	{
-		fclose(in);
-		elog(ERROR, "cannot stat \"%s\": %s", file->path,
-			 strerror(errno));
-	}
-
-	for (;;)
-	{
-		read_len = fread(buf, 1, sizeof(buf), in);
-
-		if(read_len == 0)
-			break;
-
-		/* update CRC */
-		COMP_TRADITIONAL_CRC32(crc, buf, read_len);
-
-		file->write_size += read_len;
-		file->read_size += read_len;
-	}
-
-	errno_tmp = errno;
-	if (!feof(in))
-	{
-		fclose(in);
-		elog(ERROR, "cannot read backup mode file \"%s\": %s",
-			 file->path, strerror(errno_tmp));
-	}
-
-	/* finish CRC calculation and store into pgFile */
-	FIN_TRADITIONAL_CRC32(crc);
-	file->crc = crc;
-
-	fclose(in);
 
-	return true;
+	file->crc = pgFileGetCRC(file->path, false, false, &file->read_size);
+	file->write_size = file->read_size;
 }
 
 /*
@@ -1779,11 +1717,11 @@ fileEqualCRC(const char *path1, const char *path2, bool path2_is_compressed)
 	else
 #endif
 	{
-		crc2 = pgFileGetCRC(path2);
+		crc2 = pgFileGetCRC(path2, false, true, NULL);
 	}
 
 	/* Get checksum of original file */
-	crc1 = pgFileGetCRC(path1);
+	crc1 = pgFileGetCRC(path1, false, true, NULL);
 
 	return EQ_CRC32C(crc1, crc2);
 }

src/dir.c

@@ -259,36 +259,55 @@ pgFileDelete(pgFile *file)
 }
 
 pg_crc32
-pgFileGetCRC(const char *file_path, bool use_crc32c)
+pgFileGetCRC(const char *file_path, bool use_crc32c, bool raise_on_deleted,
+			 size_t *bytes_read)
 {
 	FILE	   *fp;
 	pg_crc32	crc = 0;
 	char		buf[1024];
 	size_t		len;
+	size_t		total = 0;
 	int			errno_tmp;
 
+	INIT_FILE_CRC32(use_crc32c, crc);
+
 	/* open file in binary read mode */
 	fp = fopen(file_path, PG_BINARY_R);
 	if (fp == NULL)
-		elog(ERROR, "cannot open file \"%s\": %s",
-			file_path, strerror(errno));
+	{
+		if (!raise_on_deleted && errno == ENOENT)
+		{
+			FIN_FILE_CRC32(use_crc32c, crc);
+			return crc;
+		}
+		else
+			elog(ERROR, "cannot open file \"%s\": %s",
+				file_path, strerror(errno));
+	}
 
-	/* calc CRC of backup file */
-	INIT_FILE_CRC32(use_crc32c, crc);
-	while ((len = fread(buf, 1, sizeof(buf), fp)) == sizeof(buf))
+	/* calc CRC of file */
+	for (;;)
 	{
 		if (interrupted)
 			elog(ERROR, "interrupted during CRC calculation");
+
+		len = fread(buf, 1, sizeof(buf), fp);
+		if(len == 0)
+			break;
+		/* update CRC */
 		COMP_FILE_CRC32(use_crc32c, crc, buf, len);
+		total += len;
 	}
+
+	if (bytes_read)
+		*bytes_read = total;
+
 	errno_tmp = errno;
 	if (!feof(fp))
 		elog(WARNING, "cannot read \"%s\": %s", file_path,
 			strerror(errno_tmp));
-	if (len > 0)
-		COMP_FILE_CRC32(use_crc32c, crc, buf, len);
-	FIN_FILE_CRC32(use_crc32c, crc);
 
+	FIN_FILE_CRC32(use_crc32c, crc);
 	fclose(fp);
 
 	return crc;

src/merge.c

@@ -524,7 +524,7 @@ merge_files(void *arg)
 				 * do that.
 				 */
 				file->write_size = pgFileSize(to_path_tmp);
-				file->crc = pgFileGetCRC(to_path_tmp, false);
+				file->crc = pgFileGetCRC(to_path_tmp, false, true, NULL);
 			}
 		}
 		else

src/pg_probackup.h

@@ -531,7 +531,8 @@ extern pgFile *pgFileNew(const char *path, bool omit_symlink);
 extern pgFile *pgFileInit(const char *path);
 extern void pgFileDelete(pgFile *file);
 extern void pgFileFree(void *file);
-extern pg_crc32 pgFileGetCRC(const char *file_path, bool use_crc32c);
+extern pg_crc32 pgFileGetCRC(const char *file_path, bool use_crc32c,
+							 bool raise_on_deleted, size_t *bytes_read);
 extern int pgFileComparePath(const void *f1, const void *f2);
 extern int pgFileComparePathDesc(const void *f1, const void *f2);
 extern int pgFileCompareLinked(const void *f1, const void *f2);
@@ -552,7 +553,7 @@ extern void push_wal_file(const char *from_path, const char *to_path,
 						  bool is_compress, bool overwrite);
 extern void get_wal_file(const char *from_path, const char *to_path);
 
-extern bool calc_file_checksum(pgFile *file);
+extern void calc_file_checksum(pgFile *file);
 
 extern bool check_file_pages(pgFile* file,
 							 XLogRecPtr stop_lsn,

src/util.c

@@ -334,7 +334,7 @@ set_min_recovery_point(pgFile *file, const char *backup_path, XLogRecPtr stop_ba
 	writeControlFile(&ControlFile, fullpath);
 
 	/* Update pg_control checksum in backup_list */
-	file->crc = pgFileGetCRC(fullpath, false);
+	file->crc = pgFileGetCRC(fullpath, false, true, NULL);
 
 	pg_free(buffer);
 }

src/validate.c

@@ -224,7 +224,8 @@ pgBackupValidateFiles(void *arg)
 			 * To avoid this problem we need to use different algorithm, CRC-32 in
 			 * this case.
 			 */
-			crc = pgFileGetCRC(file->path, arguments->backup_version <= 20021);
+			crc = pgFileGetCRC(file->path, arguments->backup_version <= 20021,
+							   true, NULL);
 			if (crc != file->crc)
 			{
 				elog(WARNING, "Invalid CRC of backup file \"%s\" : %X. Expected %X",