1 /*
2 * Licensed to the Apache Software Foundation (ASF) under one
3 * or more contributor license agreements. See the NOTICE file
4 * distributed with this work for additional information
5 * regarding copyright ownership. The ASF licenses this file
6 * to you under the Apache License, Version 2.0 (the
7 * "License"); you may not use this file except in compliance
8 * with the License. You may obtain a copy of the License at
9 *
10 * http://www.apache.org/licenses/LICENSE-2.0
11 *
12 * Unless required by applicable law or agreed to in writing, software
13 * distributed under the License is distributed on an "AS IS" BASIS,
14 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15 * See the License for the specific language governing permissions and
16 * limitations under the License.
17 */
18 package org.apache.hadoop.hbase.io.hfile;
19
20 import java.io.ByteArrayOutputStream;
21 import java.io.IOException;
22 import java.nio.ByteBuffer;
23
24 import org.apache.commons.logging.Log;
25 import org.apache.commons.logging.LogFactory;
26 import org.apache.hadoop.fs.ChecksumException;
27 import org.apache.hadoop.hbase.classification.InterfaceAudience;
28 import org.apache.hadoop.fs.Path;
29 import org.apache.hadoop.hbase.util.ChecksumType;
30 import org.apache.hadoop.util.DataChecksum;
31
32 /**
33 * Utility methods to compute and validate checksums.
34 */
35 @InterfaceAudience.Private
36 public class ChecksumUtil {
37 public static final Log LOG = LogFactory.getLog(ChecksumUtil.class);
38
39 /** This is used to reserve space in a byte buffer */
40 private static byte[] DUMMY_VALUE = new byte[128 * HFileBlock.CHECKSUM_SIZE];
41
42 /**
43 * This is used by unit tests to make checksum failures throw an
44 * exception instead of returning null. Returning a null value from
45 * checksum validation will cause the higher layer to retry that
46 * read with hdfs-level checksums. Instead, we would like checksum
47 * failures to cause the entire unit test to fail.
48 */
49 private static boolean generateExceptions = false;
50
51 /**
52 * Generates a checksum for all the data in indata. The checksum is
53 * written to outdata.
54 * @param indata input data stream
55 * @param startOffset starting offset in the indata stream from where to
56 * compute checkums from
57 * @param endOffset ending offset in the indata stream upto
58 * which checksums needs to be computed
59 * @param outdata the output buffer where checksum values are written
60 * @param outOffset the starting offset in the outdata where the
61 * checksum values are written
62 * @param checksumType type of checksum
63 * @param bytesPerChecksum number of bytes per checksum value
64 */
65 static void generateChecksums(byte[] indata, int startOffset, int endOffset,
66 byte[] outdata, int outOffset, ChecksumType checksumType,
67 int bytesPerChecksum) throws IOException {
68
69 if (checksumType == ChecksumType.NULL) {
70 return; // No checksum for this block.
71 }
72
73 DataChecksum checksum = DataChecksum.newDataChecksum(
74 checksumType.getDataChecksumType(), bytesPerChecksum);
75
76 checksum.calculateChunkedSums(
77 ByteBuffer.wrap(indata, startOffset, endOffset - startOffset),
78 ByteBuffer.wrap(outdata, outOffset, outdata.length - outOffset));
79 }
80
81 /**
82 * Validates that the data in the specified HFileBlock matches the checksum. Generates the
83 * checksums for the data and then validate that it matches those stored in the end of the data.
84 * @param buffer Contains the data in following order: HFileBlock header, data, checksums.
85 * @param path Path of the HFile to which the {@code data} belongs. Only used for logging.
86 * @param offset offset of the data being validated. Only used for logging.
87 * @param hdrSize Size of the block header in {@code data}. Only used for logging.
88 * @return True if checksum matches, else false.
89 */
90 static boolean validateChecksum(ByteBuffer buffer, Path path, long offset, int hdrSize)
91 throws IOException {
92 // A ChecksumType.NULL indicates that the caller is not interested in validating checksums,
93 // so we always return true.
94 ChecksumType cktype =
95 ChecksumType.codeToType(buffer.get(HFileBlock.Header.CHECKSUM_TYPE_INDEX));
96 if (cktype == ChecksumType.NULL) {
97 return true; // No checksum validations needed for this block.
98 }
99
100 // read in the stored value of the checksum size from the header.
101 int bytesPerChecksum = buffer.getInt(HFileBlock.Header.BYTES_PER_CHECKSUM_INDEX);
102
103 DataChecksum dataChecksum = DataChecksum.newDataChecksum(
104 cktype.getDataChecksumType(), bytesPerChecksum);
105 assert dataChecksum != null;
106 int onDiskDataSizeWithHeader =
107 buffer.getInt(HFileBlock.Header.ON_DISK_DATA_SIZE_WITH_HEADER_INDEX);
108 if (LOG.isTraceEnabled()) {
109 LOG.info("dataLength=" + buffer.capacity()
110 + ", sizeWithHeader=" + onDiskDataSizeWithHeader
111 + ", checksumType=" + cktype.getName()
112 + ", file=" + path.toString()
113 + ", offset=" + offset
114 + ", headerSize=" + hdrSize
115 + ", bytesPerChecksum=" + bytesPerChecksum);
116 }
117 try {
118 ByteBuffer data = (ByteBuffer) buffer.duplicate().position(0).limit(onDiskDataSizeWithHeader);
119 ByteBuffer checksums = (ByteBuffer) buffer.duplicate().position(onDiskDataSizeWithHeader)
120 .limit(buffer.capacity());
121 dataChecksum.verifyChunkedSums(data, checksums, path.toString(), 0);
122 } catch (ChecksumException e) {
123 return false;
124 }
125 return true; // checksum is valid
126 }
127
128 /**
129 * Returns the number of bytes needed to store the checksums for
130 * a specified data size
131 * @param datasize number of bytes of data
132 * @param bytesPerChecksum number of bytes in a checksum chunk
133 * @return The number of bytes needed to store the checksum values
134 */
135 static long numBytes(long datasize, int bytesPerChecksum) {
136 return numChunks(datasize, bytesPerChecksum) *
137 HFileBlock.CHECKSUM_SIZE;
138 }
139
140 /**
141 * Returns the number of checksum chunks needed to store the checksums for
142 * a specified data size
143 * @param datasize number of bytes of data
144 * @param bytesPerChecksum number of bytes in a checksum chunk
145 * @return The number of checksum chunks
146 */
147 static long numChunks(long datasize, int bytesPerChecksum) {
148 long numChunks = datasize/bytesPerChecksum;
149 if (datasize % bytesPerChecksum != 0) {
150 numChunks++;
151 }
152 return numChunks;
153 }
154
155 /**
156 * Write dummy checksums to the end of the specified bytes array
157 * to reserve space for writing checksums later
158 * @param baos OutputStream to write dummy checkum values
159 * @param numBytes Number of bytes of data for which dummy checksums
160 * need to be generated
161 * @param bytesPerChecksum Number of bytes per checksum value
162 */
163 static void reserveSpaceForChecksums(ByteArrayOutputStream baos,
164 int numBytes, int bytesPerChecksum) throws IOException {
165 long numChunks = numChunks(numBytes, bytesPerChecksum);
166 long bytesLeft = numChunks * HFileBlock.CHECKSUM_SIZE;
167 while (bytesLeft > 0) {
168 long count = Math.min(bytesLeft, DUMMY_VALUE.length);
169 baos.write(DUMMY_VALUE, 0, (int)count);
170 bytesLeft -= count;
171 }
172 }
173
174 /**
175 * Mechanism to throw an exception in case of hbase checksum
176 * failure. This is used by unit tests only.
177 * @param value Setting this to true will cause hbase checksum
178 * verification failures to generate exceptions.
179 */
180 public static void generateExceptionForChecksumFailureForTest(boolean value) {
181 generateExceptions = value;
182 }
183 }
184