Google Git
Sign in
apache / hbase-site / 1f0bfa6196fcf97b0302890df5bb4e21963b0077 / . / devapidocs / src-html / org / apache / hadoop / hbase / io / hfile / ChecksumUtil.html
blob: 1ac40ad1d68974735ac3328588bcd5f7136493f6 [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/*<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.io.hfile;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.nio.ByteBuffer;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import org.apache.hadoop.fs.ChecksumException;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import org.apache.hadoop.hbase.nio.ByteBuff;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.nio.SingleByteBuff;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.util.ChecksumType;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.util.DataChecksum;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.slf4j.Logger;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.slf4j.LoggerFactory;<a name="line.29"></a>
<span class="sourceLineNo">030</span><a name="line.30"></a>
<span class="sourceLineNo">031</span>/**<a name="line.31"></a>
<span class="sourceLineNo">032</span> * Utility methods to compute and validate checksums.<a name="line.32"></a>
<span class="sourceLineNo">033</span> */<a name="line.33"></a>
<span class="sourceLineNo">034</span>@InterfaceAudience.Private<a name="line.34"></a>
<span class="sourceLineNo">035</span>public class ChecksumUtil {<a name="line.35"></a>
<span class="sourceLineNo">036</span> public static final Logger LOG = LoggerFactory.getLogger(ChecksumUtil.class);<a name="line.36"></a>
<span class="sourceLineNo">037</span><a name="line.37"></a>
<span class="sourceLineNo">038</span> public static final int CHECKSUM_BUF_SIZE = 256;<a name="line.38"></a>
<span class="sourceLineNo">039</span><a name="line.39"></a>
<span class="sourceLineNo">040</span> /**<a name="line.40"></a>
<span class="sourceLineNo">041</span> * This is used by unit tests to make checksum failures throw an exception instead of returning<a name="line.41"></a>
<span class="sourceLineNo">042</span> * null. Returning a null value from checksum validation will cause the higher layer to retry that<a name="line.42"></a>
<span class="sourceLineNo">043</span> * read with hdfs-level checksums. Instead, we would like checksum failures to cause the entire<a name="line.43"></a>
<span class="sourceLineNo">044</span> * unit test to fail.<a name="line.44"></a>
<span class="sourceLineNo">045</span> */<a name="line.45"></a>
<span class="sourceLineNo">046</span> private static boolean generateExceptions = false;<a name="line.46"></a>
<span class="sourceLineNo">047</span><a name="line.47"></a>
<span class="sourceLineNo">048</span> /**<a name="line.48"></a>
<span class="sourceLineNo">049</span> * Generates a checksum for all the data in indata. The checksum is written to outdata.<a name="line.49"></a>
<span class="sourceLineNo">050</span> * @param indata input data stream<a name="line.50"></a>
<span class="sourceLineNo">051</span> * @param startOffset starting offset in the indata stream from where to compute checkums<a name="line.51"></a>
<span class="sourceLineNo">052</span> * from<a name="line.52"></a>
<span class="sourceLineNo">053</span> * @param endOffset ending offset in the indata stream upto which checksums needs to be<a name="line.53"></a>
<span class="sourceLineNo">054</span> * computed<a name="line.54"></a>
<span class="sourceLineNo">055</span> * @param outdata the output buffer where checksum values are written<a name="line.55"></a>
<span class="sourceLineNo">056</span> * @param outOffset the starting offset in the outdata where the checksum values are<a name="line.56"></a>
<span class="sourceLineNo">057</span> * written<a name="line.57"></a>
<span class="sourceLineNo">058</span> * @param checksumType type of checksum<a name="line.58"></a>
<span class="sourceLineNo">059</span> * @param bytesPerChecksum number of bytes per checksum value<a name="line.59"></a>
<span class="sourceLineNo">060</span> */<a name="line.60"></a>
<span class="sourceLineNo">061</span> static void generateChecksums(byte[] indata, int startOffset, int endOffset, byte[] outdata,<a name="line.61"></a>
<span class="sourceLineNo">062</span> int outOffset, ChecksumType checksumType, int bytesPerChecksum) throws IOException {<a name="line.62"></a>
<span class="sourceLineNo">063</span><a name="line.63"></a>
<span class="sourceLineNo">064</span> if (checksumType == ChecksumType.NULL) {<a name="line.64"></a>
<span class="sourceLineNo">065</span> return; // No checksum for this block.<a name="line.65"></a>
<span class="sourceLineNo">066</span> }<a name="line.66"></a>
<span class="sourceLineNo">067</span><a name="line.67"></a>
<span class="sourceLineNo">068</span> DataChecksum checksum =<a name="line.68"></a>
<span class="sourceLineNo">069</span> DataChecksum.newDataChecksum(checksumType.getDataChecksumType(), bytesPerChecksum);<a name="line.69"></a>
<span class="sourceLineNo">070</span><a name="line.70"></a>
<span class="sourceLineNo">071</span> checksum.calculateChunkedSums(ByteBuffer.wrap(indata, startOffset, endOffset - startOffset),<a name="line.71"></a>
<span class="sourceLineNo">072</span> ByteBuffer.wrap(outdata, outOffset, outdata.length - outOffset));<a name="line.72"></a>
<span class="sourceLineNo">073</span> }<a name="line.73"></a>
<span class="sourceLineNo">074</span><a name="line.74"></a>
<span class="sourceLineNo">075</span> /**<a name="line.75"></a>
<span class="sourceLineNo">076</span> * Like the hadoop's {@link DataChecksum#verifyChunkedSums(ByteBuffer, ByteBuffer, String, long)},<a name="line.76"></a>
<span class="sourceLineNo">077</span> * this method will also verify checksum of each chunk in data. the difference is: this method can<a name="line.77"></a>
<span class="sourceLineNo">078</span> * accept {@link ByteBuff} as arguments, we can not add it in hadoop-common so defined here.<a name="line.78"></a>
<span class="sourceLineNo">079</span> * @param dataChecksum to calculate the checksum.<a name="line.79"></a>
<span class="sourceLineNo">080</span> * @param data as the input<a name="line.80"></a>
<span class="sourceLineNo">081</span> * @param checksums to compare<a name="line.81"></a>
<span class="sourceLineNo">082</span> * @param pathName indicate that the data is read from which file.<a name="line.82"></a>
<span class="sourceLineNo">083</span> * @return a flag indicate the checksum match or mismatch.<a name="line.83"></a>
<span class="sourceLineNo">084</span> * @see org.apache.hadoop.util.DataChecksum#verifyChunkedSums(ByteBuffer, ByteBuffer, String,<a name="line.84"></a>
<span class="sourceLineNo">085</span> * long)<a name="line.85"></a>
<span class="sourceLineNo">086</span> */<a name="line.86"></a>
<span class="sourceLineNo">087</span> private static boolean verifyChunkedSums(DataChecksum dataChecksum, ByteBuff data,<a name="line.87"></a>
<span class="sourceLineNo">088</span> ByteBuff checksums, String pathName) {<a name="line.88"></a>
<span class="sourceLineNo">089</span> // Almost all of the HFile Block are about 64KB, and it would be a SingleByteBuff, use the<a name="line.89"></a>
<span class="sourceLineNo">090</span> // Hadoop's verify checksum directly, because it'll use the native checksum, which has no extra<a name="line.90"></a>
<span class="sourceLineNo">091</span> // byte[] allocation or copying. (HBASE-21917)<a name="line.91"></a>
<span class="sourceLineNo">092</span> if (data instanceof SingleByteBuff &amp;&amp; checksums instanceof SingleByteBuff) {<a name="line.92"></a>
<span class="sourceLineNo">093</span> // the checksums ByteBuff must also be an SingleByteBuff because it's duplicated from data.<a name="line.93"></a>
<span class="sourceLineNo">094</span> ByteBuffer dataBB = (ByteBuffer) (data.nioByteBuffers()[0]).duplicate()<a name="line.94"></a>
<span class="sourceLineNo">095</span> .position(data.position()).limit(data.limit());<a name="line.95"></a>
<span class="sourceLineNo">096</span> ByteBuffer checksumBB = (ByteBuffer) (checksums.nioByteBuffers()[0]).duplicate()<a name="line.96"></a>
<span class="sourceLineNo">097</span> .position(checksums.position()).limit(checksums.limit());<a name="line.97"></a>
<span class="sourceLineNo">098</span> try {<a name="line.98"></a>
<span class="sourceLineNo">099</span> dataChecksum.verifyChunkedSums(dataBB, checksumBB, pathName, 0);<a name="line.99"></a>
<span class="sourceLineNo">100</span> return true;<a name="line.100"></a>
<span class="sourceLineNo">101</span> } catch (ChecksumException e) {<a name="line.101"></a>
<span class="sourceLineNo">102</span> return false;<a name="line.102"></a>
<span class="sourceLineNo">103</span> }<a name="line.103"></a>
<span class="sourceLineNo">104</span> }<a name="line.104"></a>
<span class="sourceLineNo">105</span><a name="line.105"></a>
<span class="sourceLineNo">106</span> // If the block is a MultiByteBuff. we use a small byte[] to update the checksum many times for<a name="line.106"></a>
<span class="sourceLineNo">107</span> // reducing GC pressure. it's a rare case.<a name="line.107"></a>
<span class="sourceLineNo">108</span> int checksumTypeSize = dataChecksum.getChecksumType().size;<a name="line.108"></a>
<span class="sourceLineNo">109</span> if (checksumTypeSize == 0) {<a name="line.109"></a>
<span class="sourceLineNo">110</span> return true;<a name="line.110"></a>
<span class="sourceLineNo">111</span> }<a name="line.111"></a>
<span class="sourceLineNo">112</span> // we have 5 checksum type now: NULL,DEFAULT,MIXED,CRC32,CRC32C. the former three need 0 byte,<a name="line.112"></a>
<span class="sourceLineNo">113</span> // and the other two need 4 bytes.<a name="line.113"></a>
<span class="sourceLineNo">114</span> assert checksumTypeSize == 4;<a name="line.114"></a>
<span class="sourceLineNo">115</span><a name="line.115"></a>
<span class="sourceLineNo">116</span> int bytesPerChecksum = dataChecksum.getBytesPerChecksum();<a name="line.116"></a>
<span class="sourceLineNo">117</span> int startDataPos = data.position();<a name="line.117"></a>
<span class="sourceLineNo">118</span> data.mark();<a name="line.118"></a>
<span class="sourceLineNo">119</span> checksums.mark();<a name="line.119"></a>
<span class="sourceLineNo">120</span> try {<a name="line.120"></a>
<span class="sourceLineNo">121</span> // allocate an small buffer for reducing young GC (HBASE-21917), and copy 256 bytes from<a name="line.121"></a>
<span class="sourceLineNo">122</span> // ByteBuff to update the checksum each time. if we upgrade to an future JDK and hadoop<a name="line.122"></a>
<span class="sourceLineNo">123</span> // version which support DataCheckSum#update(ByteBuffer), we won't need to update the checksum<a name="line.123"></a>
<span class="sourceLineNo">124</span> // multiple times then.<a name="line.124"></a>
<span class="sourceLineNo">125</span> byte[] buf = new byte[CHECKSUM_BUF_SIZE];<a name="line.125"></a>
<span class="sourceLineNo">126</span> byte[] sum = new byte[checksumTypeSize];<a name="line.126"></a>
<span class="sourceLineNo">127</span> while (data.remaining() &gt; 0) {<a name="line.127"></a>
<span class="sourceLineNo">128</span> int n = Math.min(data.remaining(), bytesPerChecksum);<a name="line.128"></a>
<span class="sourceLineNo">129</span> checksums.get(sum);<a name="line.129"></a>
<span class="sourceLineNo">130</span> dataChecksum.reset();<a name="line.130"></a>
<span class="sourceLineNo">131</span> for (int remain = n, len; remain &gt; 0; remain -= len) {<a name="line.131"></a>
<span class="sourceLineNo">132</span> // Copy 256 bytes from ByteBuff to update the checksum each time, if the remaining<a name="line.132"></a>
<span class="sourceLineNo">133</span> // bytes is less than 256, then just update the remaining bytes.<a name="line.133"></a>
<span class="sourceLineNo">134</span> len = Math.min(CHECKSUM_BUF_SIZE, remain);<a name="line.134"></a>
<span class="sourceLineNo">135</span> data.get(buf, 0, len);<a name="line.135"></a>
<span class="sourceLineNo">136</span> dataChecksum.update(buf, 0, len);<a name="line.136"></a>
<span class="sourceLineNo">137</span> }<a name="line.137"></a>
<span class="sourceLineNo">138</span> int calculated = (int) dataChecksum.getValue();<a name="line.138"></a>
<span class="sourceLineNo">139</span> int stored = (sum[0] &lt;&lt; 24 &amp; 0xff000000) | (sum[1] &lt;&lt; 16 &amp; 0xff0000)<a name="line.139"></a>
<span class="sourceLineNo">140</span> | (sum[2] &lt;&lt; 8 &amp; 0xff00) | (sum[3] &amp; 0xff);<a name="line.140"></a>
<span class="sourceLineNo">141</span> if (calculated != stored) {<a name="line.141"></a>
<span class="sourceLineNo">142</span> if (LOG.isTraceEnabled()) {<a name="line.142"></a>
<span class="sourceLineNo">143</span> long errPos = data.position() - startDataPos - n;<a name="line.143"></a>
<span class="sourceLineNo">144</span> LOG.trace("Checksum error: {} at {} expected: {} got: {}", pathName, errPos, stored,<a name="line.144"></a>
<span class="sourceLineNo">145</span> calculated);<a name="line.145"></a>
<span class="sourceLineNo">146</span> }<a name="line.146"></a>
<span class="sourceLineNo">147</span> return false;<a name="line.147"></a>
<span class="sourceLineNo">148</span> }<a name="line.148"></a>
<span class="sourceLineNo">149</span> }<a name="line.149"></a>
<span class="sourceLineNo">150</span> } finally {<a name="line.150"></a>
<span class="sourceLineNo">151</span> data.reset();<a name="line.151"></a>
<span class="sourceLineNo">152</span> checksums.reset();<a name="line.152"></a>
<span class="sourceLineNo">153</span> }<a name="line.153"></a>
<span class="sourceLineNo">154</span> return true;<a name="line.154"></a>
<span class="sourceLineNo">155</span> }<a name="line.155"></a>
<span class="sourceLineNo">156</span><a name="line.156"></a>
<span class="sourceLineNo">157</span> /**<a name="line.157"></a>
<span class="sourceLineNo">158</span> * Validates that the data in the specified HFileBlock matches the checksum. Generates the<a name="line.158"></a>
<span class="sourceLineNo">159</span> * checksums for the data and then validate that it matches those stored in the end of the data.<a name="line.159"></a>
<span class="sourceLineNo">160</span> * @param buf Contains the data in following order: HFileBlock header, data, checksums.<a name="line.160"></a>
<span class="sourceLineNo">161</span> * @param pathName Path of the HFile to which the {@code data} belongs. Only used for logging.<a name="line.161"></a>
<span class="sourceLineNo">162</span> * @param offset offset of the data being validated. Only used for logging.<a name="line.162"></a>
<span class="sourceLineNo">163</span> * @param hdrSize Size of the block header in {@code data}. Only used for logging.<a name="line.163"></a>
<span class="sourceLineNo">164</span> * @return True if checksum matches, else false.<a name="line.164"></a>
<span class="sourceLineNo">165</span> */<a name="line.165"></a>
<span class="sourceLineNo">166</span> static boolean validateChecksum(ByteBuff buf, String pathName, long offset, int hdrSize) {<a name="line.166"></a>
<span class="sourceLineNo">167</span> ChecksumType ctype = ChecksumType.codeToType(buf.get(HFileBlock.Header.CHECKSUM_TYPE_INDEX));<a name="line.167"></a>
<span class="sourceLineNo">168</span> if (ctype == ChecksumType.NULL) {<a name="line.168"></a>
<span class="sourceLineNo">169</span> return true;// No checksum validations needed for this block.<a name="line.169"></a>
<span class="sourceLineNo">170</span> }<a name="line.170"></a>
<span class="sourceLineNo">171</span><a name="line.171"></a>
<span class="sourceLineNo">172</span> // read in the stored value of the checksum size from the header.<a name="line.172"></a>
<span class="sourceLineNo">173</span> int bytesPerChecksum = buf.getInt(HFileBlock.Header.BYTES_PER_CHECKSUM_INDEX);<a name="line.173"></a>
<span class="sourceLineNo">174</span> DataChecksum dataChecksum =<a name="line.174"></a>
<span class="sourceLineNo">175</span> DataChecksum.newDataChecksum(ctype.getDataChecksumType(), bytesPerChecksum);<a name="line.175"></a>
<span class="sourceLineNo">176</span> assert dataChecksum != null;<a name="line.176"></a>
<span class="sourceLineNo">177</span> int onDiskDataSizeWithHeader =<a name="line.177"></a>
<span class="sourceLineNo">178</span> buf.getInt(HFileBlock.Header.ON_DISK_DATA_SIZE_WITH_HEADER_INDEX);<a name="line.178"></a>
<span class="sourceLineNo">179</span> LOG.trace(<a name="line.179"></a>
<span class="sourceLineNo">180</span> "dataLength={}, sizeWithHeader={}, checksumType={}, file={}, "<a name="line.180"></a>
<span class="sourceLineNo">181</span> + "offset={}, headerSize={}, bytesPerChecksum={}",<a name="line.181"></a>
<span class="sourceLineNo">182</span> buf.capacity(), onDiskDataSizeWithHeader, ctype.getName(), pathName, offset, hdrSize,<a name="line.182"></a>
<span class="sourceLineNo">183</span> bytesPerChecksum);<a name="line.183"></a>
<span class="sourceLineNo">184</span> ByteBuff data = buf.duplicate().position(0).limit(onDiskDataSizeWithHeader);<a name="line.184"></a>
<span class="sourceLineNo">185</span> ByteBuff checksums = buf.duplicate().position(onDiskDataSizeWithHeader).limit(buf.limit());<a name="line.185"></a>
<span class="sourceLineNo">186</span> return verifyChunkedSums(dataChecksum, data, checksums, pathName);<a name="line.186"></a>
<span class="sourceLineNo">187</span> }<a name="line.187"></a>
<span class="sourceLineNo">188</span><a name="line.188"></a>
<span class="sourceLineNo">189</span> /**<a name="line.189"></a>
<span class="sourceLineNo">190</span> * Returns the number of bytes needed to store the checksums for a specified data size<a name="line.190"></a>
<span class="sourceLineNo">191</span> * @param datasize number of bytes of data<a name="line.191"></a>
<span class="sourceLineNo">192</span> * @param bytesPerChecksum number of bytes in a checksum chunk<a name="line.192"></a>
<span class="sourceLineNo">193</span> * @return The number of bytes needed to store the checksum values<a name="line.193"></a>
<span class="sourceLineNo">194</span> */<a name="line.194"></a>
<span class="sourceLineNo">195</span> static long numBytes(long datasize, int bytesPerChecksum) {<a name="line.195"></a>
<span class="sourceLineNo">196</span> return numChunks(datasize, bytesPerChecksum) * HFileBlock.CHECKSUM_SIZE;<a name="line.196"></a>
<span class="sourceLineNo">197</span> }<a name="line.197"></a>
<span class="sourceLineNo">198</span><a name="line.198"></a>
<span class="sourceLineNo">199</span> /**<a name="line.199"></a>
<span class="sourceLineNo">200</span> * Returns the number of checksum chunks needed to store the checksums for a specified data size<a name="line.200"></a>
<span class="sourceLineNo">201</span> * @param datasize number of bytes of data<a name="line.201"></a>
<span class="sourceLineNo">202</span> * @param bytesPerChecksum number of bytes in a checksum chunk<a name="line.202"></a>
<span class="sourceLineNo">203</span> * @return The number of checksum chunks<a name="line.203"></a>
<span class="sourceLineNo">204</span> */<a name="line.204"></a>
<span class="sourceLineNo">205</span> static long numChunks(long datasize, int bytesPerChecksum) {<a name="line.205"></a>
<span class="sourceLineNo">206</span> long numChunks = datasize / bytesPerChecksum;<a name="line.206"></a>
<span class="sourceLineNo">207</span> if (datasize % bytesPerChecksum != 0) {<a name="line.207"></a>
<span class="sourceLineNo">208</span> numChunks++;<a name="line.208"></a>
<span class="sourceLineNo">209</span> }<a name="line.209"></a>
<span class="sourceLineNo">210</span> return numChunks;<a name="line.210"></a>
<span class="sourceLineNo">211</span> }<a name="line.211"></a>
<span class="sourceLineNo">212</span><a name="line.212"></a>
<span class="sourceLineNo">213</span> /**<a name="line.213"></a>
<span class="sourceLineNo">214</span> * Mechanism to throw an exception in case of hbase checksum failure. This is used by unit tests<a name="line.214"></a>
<span class="sourceLineNo">215</span> * only.<a name="line.215"></a>
<span class="sourceLineNo">216</span> * @param value Setting this to true will cause hbase checksum verification failures to generate<a name="line.216"></a>
<span class="sourceLineNo">217</span> * exceptions.<a name="line.217"></a>
<span class="sourceLineNo">218</span> */<a name="line.218"></a>
<span class="sourceLineNo">219</span> public static void generateExceptionForChecksumFailureForTest(boolean value) {<a name="line.219"></a>
<span class="sourceLineNo">220</span> generateExceptions = value;<a name="line.220"></a>
<span class="sourceLineNo">221</span> }<a name="line.221"></a>
<span class="sourceLineNo">222</span>}<a name="line.222"></a>
</pre>
</div>
</body>
</html>