001/* 002 * Licensed to the Apache Software Foundation (ASF) under one 003 * or more contributor license agreements. See the NOTICE file 004 * distributed with this work for additional information 005 * regarding copyright ownership. The ASF licenses this file 006 * to you under the Apache License, Version 2.0 (the 007 * "License"); you may not use this file except in compliance 008 * with the License. You may obtain a copy of the License at 009 * 010 * http://www.apache.org/licenses/LICENSE-2.0 011 * 012 * Unless required by applicable law or agreed to in writing, software 013 * distributed under the License is distributed on an "AS IS" BASIS, 014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 015 * See the License for the specific language governing permissions and 016 * limitations under the License. 017 */ 018package org.apache.hadoop.hbase.io.hfile; 019 020import static org.apache.hadoop.hbase.io.ByteBuffAllocator.HEAP; 021import static org.apache.hadoop.hbase.io.hfile.BlockCompressedSizePredicator.BLOCK_COMPRESSED_SIZE_PREDICATOR; 022import static org.apache.hadoop.hbase.io.hfile.trace.HFileContextAttributesBuilderConsumer.CONTEXT_KEY; 023 024import io.opentelemetry.api.common.Attributes; 025import io.opentelemetry.api.common.AttributesBuilder; 026import io.opentelemetry.api.trace.Span; 027import io.opentelemetry.context.Context; 028import io.opentelemetry.context.Scope; 029import java.io.DataInputStream; 030import java.io.DataOutput; 031import java.io.DataOutputStream; 032import java.io.IOException; 033import java.nio.ByteBuffer; 034import java.util.ArrayList; 035import java.util.List; 036import java.util.Optional; 037import java.util.concurrent.atomic.AtomicReference; 038import java.util.concurrent.locks.Lock; 039import java.util.concurrent.locks.ReentrantLock; 040import org.apache.hadoop.conf.Configuration; 041import org.apache.hadoop.fs.FSDataInputStream; 042import org.apache.hadoop.fs.FSDataOutputStream; 043import org.apache.hadoop.hbase.Cell; 044import org.apache.hadoop.hbase.HConstants; 045import org.apache.hadoop.hbase.fs.HFileSystem; 046import org.apache.hadoop.hbase.io.ByteArrayOutputStream; 047import org.apache.hadoop.hbase.io.ByteBuffAllocator; 048import org.apache.hadoop.hbase.io.ByteBuffInputStream; 049import org.apache.hadoop.hbase.io.ByteBufferWriterDataOutputStream; 050import org.apache.hadoop.hbase.io.FSDataInputStreamWrapper; 051import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding; 052import org.apache.hadoop.hbase.io.encoding.EncodingState; 053import org.apache.hadoop.hbase.io.encoding.HFileBlockDecodingContext; 054import org.apache.hadoop.hbase.io.encoding.HFileBlockDefaultDecodingContext; 055import org.apache.hadoop.hbase.io.encoding.HFileBlockDefaultEncodingContext; 056import org.apache.hadoop.hbase.io.encoding.HFileBlockEncodingContext; 057import org.apache.hadoop.hbase.io.hfile.trace.HFileContextAttributesBuilderConsumer; 058import org.apache.hadoop.hbase.io.util.BlockIOUtils; 059import org.apache.hadoop.hbase.nio.ByteBuff; 060import org.apache.hadoop.hbase.nio.MultiByteBuff; 061import org.apache.hadoop.hbase.nio.SingleByteBuff; 062import org.apache.hadoop.hbase.regionserver.ShipperListener; 063import org.apache.hadoop.hbase.trace.HBaseSemanticAttributes.ReadType; 064import org.apache.hadoop.hbase.util.Bytes; 065import org.apache.hadoop.hbase.util.ChecksumType; 066import org.apache.hadoop.hbase.util.ClassSize; 067import org.apache.hadoop.hbase.util.EnvironmentEdgeManager; 068import org.apache.hadoop.util.ReflectionUtils; 069import org.apache.yetus.audience.InterfaceAudience; 070import org.slf4j.Logger; 071import org.slf4j.LoggerFactory; 072 073import org.apache.hbase.thirdparty.com.google.common.base.Preconditions; 074 075/** 076 * Cacheable Blocks of an {@link HFile} version 2 file. Version 2 was introduced in hbase-0.92.0. 077 * <p> 078 * Version 1 was the original file block. Version 2 was introduced when we changed the hbase file 079 * format to support multi-level block indexes and compound bloom filters (HBASE-3857). Support for 080 * Version 1 was removed in hbase-1.3.0. 081 * <h3>HFileBlock: Version 2</h3> In version 2, a block is structured as follows: 082 * <ul> 083 * <li><b>Header:</b> See Writer#putHeader() for where header is written; header total size is 084 * HFILEBLOCK_HEADER_SIZE 085 * <ul> 086 * <li>0. blockType: Magic record identifying the {@link BlockType} (8 bytes): e.g. 087 * <code>DATABLK*</code> 088 * <li>1. onDiskSizeWithoutHeader: Compressed -- a.k.a 'on disk' -- block size, excluding header, 089 * but including tailing checksum bytes (4 bytes) 090 * <li>2. uncompressedSizeWithoutHeader: Uncompressed block size, excluding header, and excluding 091 * checksum bytes (4 bytes) 092 * <li>3. prevBlockOffset: The offset of the previous block of the same type (8 bytes). This is used 093 * to navigate to the previous block without having to go to the block index 094 * <li>4: For minorVersions >=1, the ordinal describing checksum type (1 byte) 095 * <li>5: For minorVersions >=1, the number of data bytes/checksum chunk (4 bytes) 096 * <li>6: onDiskDataSizeWithHeader: For minorVersions >=1, the size of data 'on disk', including 097 * header, excluding checksums (4 bytes) 098 * </ul> 099 * </li> 100 * <li><b>Raw/Compressed/Encrypted/Encoded data:</b> The compression algorithm is the same for all 101 * the blocks in an {@link HFile}. If compression is NONE, this is just raw, serialized Cells. 102 * <li><b>Tail:</b> For minorVersions >=1, a series of 4 byte checksums, one each for the number 103 * of bytes specified by bytesPerChecksum. 104 * </ul> 105 * <h3>Caching</h3> Caches cache whole blocks with trailing checksums if any. We then tag on some 106 * metadata, the content of BLOCK_METADATA_SPACE which will be flag on if we are doing 'hbase' 107 * checksums and then the offset into the file which is needed when we re-make a cache key when we 108 * return the block to the cache as 'done'. See {@link Cacheable#serialize(ByteBuffer, boolean)} and 109 * {@link Cacheable#getDeserializer()}. 110 * <p> 111 * TODO: Should we cache the checksums? Down in Writer#getBlockForCaching(CacheConfig) where we make 112 * a block to cache-on-write, there is an attempt at turning off checksums. This is not the only 113 * place we get blocks to cache. We also will cache the raw return from an hdfs read. In this case, 114 * the checksums may be present. If the cache is backed by something that doesn't do ECC, say an 115 * SSD, we might want to preserve checksums. For now this is open question. 116 * <p> 117 * TODO: Over in BucketCache, we save a block allocation by doing a custom serialization. Be sure to 118 * change it if serialization changes in here. Could we add a method here that takes an IOEngine and 119 * that then serializes to it rather than expose our internals over in BucketCache? IOEngine is in 120 * the bucket subpackage. Pull it up? Then this class knows about bucketcache. Ugh. 121 */ 122@InterfaceAudience.Private 123public class HFileBlock implements Cacheable { 124 private static final Logger LOG = LoggerFactory.getLogger(HFileBlock.class); 125 public static final long FIXED_OVERHEAD = ClassSize.estimateBase(HFileBlock.class, false); 126 127 // Block Header fields. 128 129 // TODO: encapsulate Header related logic in this inner class. 130 static class Header { 131 // Format of header is: 132 // 8 bytes - block magic 133 // 4 bytes int - onDiskSizeWithoutHeader 134 // 4 bytes int - uncompressedSizeWithoutHeader 135 // 8 bytes long - prevBlockOffset 136 // The following 3 are only present if header contains checksum information 137 // 1 byte - checksum type 138 // 4 byte int - bytes per checksum 139 // 4 byte int - onDiskDataSizeWithHeader 140 static int BLOCK_MAGIC_INDEX = 0; 141 static int ON_DISK_SIZE_WITHOUT_HEADER_INDEX = 8; 142 static int UNCOMPRESSED_SIZE_WITHOUT_HEADER_INDEX = 12; 143 static int PREV_BLOCK_OFFSET_INDEX = 16; 144 static int CHECKSUM_TYPE_INDEX = 24; 145 static int BYTES_PER_CHECKSUM_INDEX = 25; 146 static int ON_DISK_DATA_SIZE_WITH_HEADER_INDEX = 29; 147 } 148 149 /** Type of block. Header field 0. */ 150 private BlockType blockType; 151 152 /** 153 * Size on disk excluding header, including checksum. Header field 1. 154 * @see Writer#putHeader(byte[], int, int, int, int) 155 */ 156 private int onDiskSizeWithoutHeader; 157 158 /** 159 * Size of pure data. Does not include header or checksums. Header field 2. 160 * @see Writer#putHeader(byte[], int, int, int, int) 161 */ 162 private int uncompressedSizeWithoutHeader; 163 164 /** 165 * The offset of the previous block on disk. Header field 3. 166 * @see Writer#putHeader(byte[], int, int, int, int) 167 */ 168 private long prevBlockOffset; 169 170 /** 171 * Size on disk of header + data. Excludes checksum. Header field 6, OR calculated from 172 * {@link #onDiskSizeWithoutHeader} when using HDFS checksum. 173 * @see Writer#putHeader(byte[], int, int, int, int) 174 */ 175 private final int onDiskDataSizeWithHeader; 176 // End of Block Header fields. 177 178 /** 179 * The in-memory representation of the hfile block. Can be on or offheap. Can be backed by a 180 * single ByteBuffer or by many. Make no assumptions. 181 * <p> 182 * Be careful reading from this <code>buf</code>. Duplicate and work on the duplicate or if not, 183 * be sure to reset position and limit else trouble down the road. 184 * <p> 185 * TODO: Make this read-only once made. 186 * <p> 187 * We are using the ByteBuff type. ByteBuffer is not extensible yet we need to be able to have a 188 * ByteBuffer-like API across multiple ByteBuffers reading from a cache such as BucketCache. So, 189 * we have this ByteBuff type. Unfortunately, it is spread all about HFileBlock. Would be good if 190 * could be confined to cache-use only but hard-to-do. 191 * <p> 192 * NOTE: this byteBuff including HFileBlock header and data, but excluding checksum. 193 */ 194 private ByteBuff bufWithoutChecksum; 195 196 /** 197 * Meta data that holds meta information on the hfileblock. 198 */ 199 private final HFileContext fileContext; 200 201 /** 202 * The offset of this block in the file. Populated by the reader for convenience of access. This 203 * offset is not part of the block header. 204 */ 205 private long offset = UNSET; 206 207 /** 208 * The on-disk size of the next block, including the header and checksums if present. UNSET if 209 * unknown. Blocks try to carry the size of the next block to read in this data member. Usually we 210 * get block sizes from the hfile index but sometimes the index is not available: e.g. when we 211 * read the indexes themselves (indexes are stored in blocks, we do not have an index for the 212 * indexes). Saves seeks especially around file open when there is a flurry of reading in hfile 213 * metadata. 214 */ 215 private int nextBlockOnDiskSize = UNSET; 216 217 private ByteBuffAllocator allocator; 218 219 /** 220 * On a checksum failure, do these many succeeding read requests using hdfs checksums before 221 * auto-reenabling hbase checksum verification. 222 */ 223 static final int CHECKSUM_VERIFICATION_NUM_IO_THRESHOLD = 3; 224 225 private static int UNSET = -1; 226 public static final boolean FILL_HEADER = true; 227 public static final boolean DONT_FILL_HEADER = false; 228 229 // How to get the estimate correctly? if it is a singleBB? 230 public static final int MULTI_BYTE_BUFFER_HEAP_SIZE = 231 (int) ClassSize.estimateBase(MultiByteBuff.class, false); 232 233 /** 234 * Space for metadata on a block that gets stored along with the block when we cache it. There are 235 * a few bytes stuck on the end of the HFileBlock that we pull in from HDFS. 8 bytes are for the 236 * offset of this block (long) in the file. Offset is important because is is used when we remake 237 * the CacheKey when we return block to the cache when done. There is also a flag on whether 238 * checksumming is being done by hbase or not. See class comment for note on uncertain state of 239 * checksumming of blocks that come out of cache (should we or should we not?). Finally there are 240 * 4 bytes to hold the length of the next block which can save a seek on occasion if available. 241 * (This EXTRA info came in with original commit of the bucketcache, HBASE-7404. It was formerly 242 * known as EXTRA_SERIALIZATION_SPACE). 243 */ 244 public static final int BLOCK_METADATA_SPACE = 245 Bytes.SIZEOF_BYTE + Bytes.SIZEOF_LONG + Bytes.SIZEOF_INT; 246 247 /** 248 * Each checksum value is an integer that can be stored in 4 bytes. 249 */ 250 static final int CHECKSUM_SIZE = Bytes.SIZEOF_INT; 251 252 static final byte[] DUMMY_HEADER_NO_CHECKSUM = 253 new byte[HConstants.HFILEBLOCK_HEADER_SIZE_NO_CHECKSUM]; 254 255 /** 256 * Used deserializing blocks from Cache. <code> 257 * ++++++++++++++ 258 * + HFileBlock + 259 * ++++++++++++++ 260 * + Checksums + <= Optional 261 * ++++++++++++++ 262 * + Metadata! + <= See note on BLOCK_METADATA_SPACE above. 263 * ++++++++++++++ 264 * </code> 265 * @see #serialize(ByteBuffer, boolean) 266 */ 267 public static final CacheableDeserializer<Cacheable> BLOCK_DESERIALIZER = new BlockDeserializer(); 268 269 public static final class BlockDeserializer implements CacheableDeserializer<Cacheable> { 270 private BlockDeserializer() { 271 } 272 273 @Override 274 public HFileBlock deserialize(ByteBuff buf, ByteBuffAllocator alloc) throws IOException { 275 // The buf has the file block followed by block metadata. 276 // Set limit to just before the BLOCK_METADATA_SPACE then rewind. 277 buf.limit(buf.limit() - BLOCK_METADATA_SPACE).rewind(); 278 // Get a new buffer to pass the HFileBlock for it to 'own'. 279 ByteBuff newByteBuff = buf.slice(); 280 // Read out the BLOCK_METADATA_SPACE content and shove into our HFileBlock. 281 buf.position(buf.limit()); 282 buf.limit(buf.limit() + HFileBlock.BLOCK_METADATA_SPACE); 283 boolean usesChecksum = buf.get() == (byte) 1; 284 long offset = buf.getLong(); 285 int nextBlockOnDiskSize = buf.getInt(); 286 return createFromBuff(newByteBuff, usesChecksum, offset, nextBlockOnDiskSize, null, alloc); 287 } 288 289 @Override 290 public int getDeserializerIdentifier() { 291 return DESERIALIZER_IDENTIFIER; 292 } 293 } 294 295 private static final int DESERIALIZER_IDENTIFIER; 296 static { 297 DESERIALIZER_IDENTIFIER = 298 CacheableDeserializerIdManager.registerDeserializer(BLOCK_DESERIALIZER); 299 } 300 301 private final int totalChecksumBytes; 302 303 /** 304 * Creates a new {@link HFile} block from the given fields. This constructor is used only while 305 * writing blocks and caching, and is sitting in a byte buffer and we want to stuff the block into 306 * cache. See {@link Writer#getBlockForCaching(CacheConfig)}. 307 * <p> 308 * TODO: The caller presumes no checksumming 309 * <p> 310 * TODO: HFile block writer can also off-heap ? 311 * </p> 312 * required of this block instance since going into cache; checksum already verified on underlying 313 * block data pulled in from filesystem. Is that correct? What if cache is SSD? 314 * @param blockType the type of this block, see {@link BlockType} 315 * @param onDiskSizeWithoutHeader see {@link #onDiskSizeWithoutHeader} 316 * @param uncompressedSizeWithoutHeader see {@link #uncompressedSizeWithoutHeader} 317 * @param prevBlockOffset see {@link #prevBlockOffset} 318 * @param buf block buffer with header 319 * ({@link HConstants#HFILEBLOCK_HEADER_SIZE} bytes) 320 * @param fillHeader when true, write the first 4 header fields into passed 321 * buffer. 322 * @param offset the file offset the block was read from 323 * @param onDiskDataSizeWithHeader see {@link #onDiskDataSizeWithHeader} 324 * @param fileContext HFile meta data 325 */ 326 public HFileBlock(BlockType blockType, int onDiskSizeWithoutHeader, 327 int uncompressedSizeWithoutHeader, long prevBlockOffset, ByteBuff buf, boolean fillHeader, 328 long offset, int nextBlockOnDiskSize, int onDiskDataSizeWithHeader, HFileContext fileContext, 329 ByteBuffAllocator allocator) { 330 this.blockType = blockType; 331 this.onDiskSizeWithoutHeader = onDiskSizeWithoutHeader; 332 this.uncompressedSizeWithoutHeader = uncompressedSizeWithoutHeader; 333 this.prevBlockOffset = prevBlockOffset; 334 this.offset = offset; 335 this.onDiskDataSizeWithHeader = onDiskDataSizeWithHeader; 336 this.nextBlockOnDiskSize = nextBlockOnDiskSize; 337 this.fileContext = fileContext; 338 this.allocator = allocator; 339 this.bufWithoutChecksum = buf; 340 if (fillHeader) { 341 overwriteHeader(); 342 } 343 this.bufWithoutChecksum.rewind(); 344 this.totalChecksumBytes = computeTotalChecksumBytes(); 345 } 346 347 /** 348 * Creates a block from an existing buffer starting with a header. Rewinds and takes ownership of 349 * the buffer. By definition of rewind, ignores the buffer position, but if you slice the buffer 350 * beforehand, it will rewind to that point. 351 * @param buf Has header, content, and trailing checksums if present. 352 */ 353 static HFileBlock createFromBuff(ByteBuff buf, boolean usesHBaseChecksum, final long offset, 354 final int nextBlockOnDiskSize, HFileContext fileContext, ByteBuffAllocator allocator) 355 throws IOException { 356 buf.rewind(); 357 final BlockType blockType = BlockType.read(buf); 358 final int onDiskSizeWithoutHeader = buf.getInt(Header.ON_DISK_SIZE_WITHOUT_HEADER_INDEX); 359 final int uncompressedSizeWithoutHeader = 360 buf.getInt(Header.UNCOMPRESSED_SIZE_WITHOUT_HEADER_INDEX); 361 final long prevBlockOffset = buf.getLong(Header.PREV_BLOCK_OFFSET_INDEX); 362 // This constructor is called when we deserialize a block from cache and when we read a block in 363 // from the fs. fileCache is null when deserialized from cache so need to make up one. 364 HFileContextBuilder fileContextBuilder = 365 fileContext != null ? new HFileContextBuilder(fileContext) : new HFileContextBuilder(); 366 fileContextBuilder.withHBaseCheckSum(usesHBaseChecksum); 367 int onDiskDataSizeWithHeader; 368 if (usesHBaseChecksum) { 369 byte checksumType = buf.get(Header.CHECKSUM_TYPE_INDEX); 370 int bytesPerChecksum = buf.getInt(Header.BYTES_PER_CHECKSUM_INDEX); 371 onDiskDataSizeWithHeader = buf.getInt(Header.ON_DISK_DATA_SIZE_WITH_HEADER_INDEX); 372 // Use the checksum type and bytes per checksum from header, not from fileContext. 373 fileContextBuilder.withChecksumType(ChecksumType.codeToType(checksumType)); 374 fileContextBuilder.withBytesPerCheckSum(bytesPerChecksum); 375 } else { 376 fileContextBuilder.withChecksumType(ChecksumType.NULL); 377 fileContextBuilder.withBytesPerCheckSum(0); 378 // Need to fix onDiskDataSizeWithHeader; there are not checksums after-block-data 379 onDiskDataSizeWithHeader = onDiskSizeWithoutHeader + headerSize(usesHBaseChecksum); 380 } 381 fileContext = fileContextBuilder.build(); 382 assert usesHBaseChecksum == fileContext.isUseHBaseChecksum(); 383 return new HFileBlockBuilder().withBlockType(blockType) 384 .withOnDiskSizeWithoutHeader(onDiskSizeWithoutHeader) 385 .withUncompressedSizeWithoutHeader(uncompressedSizeWithoutHeader) 386 .withPrevBlockOffset(prevBlockOffset).withOffset(offset) 387 .withOnDiskDataSizeWithHeader(onDiskDataSizeWithHeader) 388 .withNextBlockOnDiskSize(nextBlockOnDiskSize).withHFileContext(fileContext) 389 .withByteBuffAllocator(allocator).withByteBuff(buf.rewind()).withShared(!buf.hasArray()) 390 .build(); 391 } 392 393 /** 394 * Parse total on disk size including header and checksum. 395 * @param headerBuf Header ByteBuffer. Presumed exact size of header. 396 * @param checksumSupport true if checksum verification is in use. 397 * @return Size of the block with header included. 398 */ 399 private static int getOnDiskSizeWithHeader(final ByteBuff headerBuf, boolean checksumSupport) { 400 return headerBuf.getInt(Header.ON_DISK_SIZE_WITHOUT_HEADER_INDEX) + headerSize(checksumSupport); 401 } 402 403 /** 404 * @return the on-disk size of the next block (including the header size and any checksums if 405 * present) read by peeking into the next block's header; use as a hint when doing a read 406 * of the next block when scanning or running over a file. 407 */ 408 int getNextBlockOnDiskSize() { 409 return nextBlockOnDiskSize; 410 } 411 412 @Override 413 public BlockType getBlockType() { 414 return blockType; 415 } 416 417 @Override 418 public int refCnt() { 419 return bufWithoutChecksum.refCnt(); 420 } 421 422 @Override 423 public HFileBlock retain() { 424 bufWithoutChecksum.retain(); 425 return this; 426 } 427 428 /** 429 * Call {@link ByteBuff#release()} to decrease the reference count, if no other reference, it will 430 * return back the {@link ByteBuffer} to {@link org.apache.hadoop.hbase.io.ByteBuffAllocator} 431 */ 432 @Override 433 public boolean release() { 434 return bufWithoutChecksum.release(); 435 } 436 437 /** 438 * Calling this method in strategic locations where HFileBlocks are referenced may help diagnose 439 * potential buffer leaks. We pass the block itself as a default hint, but one can use 440 * {@link #touch(Object)} to pass their own hint as well. 441 */ 442 @Override 443 public HFileBlock touch() { 444 return touch(this); 445 } 446 447 @Override 448 public HFileBlock touch(Object hint) { 449 bufWithoutChecksum.touch(hint); 450 return this; 451 } 452 453 /** Returns get data block encoding id that was used to encode this block */ 454 short getDataBlockEncodingId() { 455 if (blockType != BlockType.ENCODED_DATA) { 456 throw new IllegalArgumentException("Querying encoder ID of a block " + "of type other than " 457 + BlockType.ENCODED_DATA + ": " + blockType); 458 } 459 return bufWithoutChecksum.getShort(headerSize()); 460 } 461 462 /** Returns the on-disk size of header + data part + checksum. */ 463 public int getOnDiskSizeWithHeader() { 464 return onDiskSizeWithoutHeader + headerSize(); 465 } 466 467 /** Returns the on-disk size of the data part + checksum (header excluded). */ 468 int getOnDiskSizeWithoutHeader() { 469 return onDiskSizeWithoutHeader; 470 } 471 472 /** Returns the uncompressed size of data part (header and checksum excluded). */ 473 public int getUncompressedSizeWithoutHeader() { 474 return uncompressedSizeWithoutHeader; 475 } 476 477 /** Returns the offset of the previous block of the same type in the file, or -1 if unknown */ 478 long getPrevBlockOffset() { 479 return prevBlockOffset; 480 } 481 482 /** 483 * Rewinds {@code buf} and writes first 4 header fields. {@code buf} position is modified as 484 * side-effect. 485 */ 486 private void overwriteHeader() { 487 bufWithoutChecksum.rewind(); 488 blockType.write(bufWithoutChecksum); 489 bufWithoutChecksum.putInt(onDiskSizeWithoutHeader); 490 bufWithoutChecksum.putInt(uncompressedSizeWithoutHeader); 491 bufWithoutChecksum.putLong(prevBlockOffset); 492 if (this.fileContext.isUseHBaseChecksum()) { 493 bufWithoutChecksum.put(fileContext.getChecksumType().getCode()); 494 bufWithoutChecksum.putInt(fileContext.getBytesPerChecksum()); 495 bufWithoutChecksum.putInt(onDiskDataSizeWithHeader); 496 } 497 } 498 499 /** 500 * Returns a buffer that does not include the header and checksum. 501 * @return the buffer with header skipped and checksum omitted. 502 */ 503 public ByteBuff getBufferWithoutHeader() { 504 ByteBuff dup = getBufferReadOnly(); 505 return dup.position(headerSize()).slice(); 506 } 507 508 /** 509 * Returns a read-only duplicate of the buffer this block stores internally ready to be read. 510 * Clients must not modify the buffer object though they may set position and limit on the 511 * returned buffer since we pass back a duplicate. This method has to be public because it is used 512 * in {@link CompoundBloomFilter} to avoid object creation on every Bloom filter lookup, but has 513 * to be used with caution. Buffer holds header, block content, and any follow-on checksums if 514 * present. 515 * @return the buffer of this block for read-only operations,the buffer includes header,but not 516 * checksum. 517 */ 518 public ByteBuff getBufferReadOnly() { 519 // TODO: ByteBuf does not support asReadOnlyBuffer(). Fix. 520 ByteBuff dup = this.bufWithoutChecksum.duplicate(); 521 assert dup.position() == 0; 522 return dup; 523 } 524 525 public ByteBuffAllocator getByteBuffAllocator() { 526 return this.allocator; 527 } 528 529 private void sanityCheckAssertion(long valueFromBuf, long valueFromField, String fieldName) 530 throws IOException { 531 if (valueFromBuf != valueFromField) { 532 throw new AssertionError(fieldName + " in the buffer (" + valueFromBuf 533 + ") is different from that in the field (" + valueFromField + ")"); 534 } 535 } 536 537 private void sanityCheckAssertion(BlockType valueFromBuf, BlockType valueFromField) 538 throws IOException { 539 if (valueFromBuf != valueFromField) { 540 throw new IOException("Block type stored in the buffer: " + valueFromBuf 541 + ", block type field: " + valueFromField); 542 } 543 } 544 545 /** 546 * Checks if the block is internally consistent, i.e. the first 547 * {@link HConstants#HFILEBLOCK_HEADER_SIZE} bytes of the buffer contain a valid header consistent 548 * with the fields. Assumes a packed block structure. This function is primary for testing and 549 * debugging, and is not thread-safe, because it alters the internal buffer pointer. Used by tests 550 * only. 551 */ 552 void sanityCheck() throws IOException { 553 // Duplicate so no side-effects 554 ByteBuff dup = this.bufWithoutChecksum.duplicate().rewind(); 555 sanityCheckAssertion(BlockType.read(dup), blockType); 556 557 sanityCheckAssertion(dup.getInt(), onDiskSizeWithoutHeader, "onDiskSizeWithoutHeader"); 558 559 sanityCheckAssertion(dup.getInt(), uncompressedSizeWithoutHeader, 560 "uncompressedSizeWithoutHeader"); 561 562 sanityCheckAssertion(dup.getLong(), prevBlockOffset, "prevBlockOffset"); 563 if (this.fileContext.isUseHBaseChecksum()) { 564 sanityCheckAssertion(dup.get(), this.fileContext.getChecksumType().getCode(), "checksumType"); 565 sanityCheckAssertion(dup.getInt(), this.fileContext.getBytesPerChecksum(), 566 "bytesPerChecksum"); 567 sanityCheckAssertion(dup.getInt(), onDiskDataSizeWithHeader, "onDiskDataSizeWithHeader"); 568 } 569 570 if (dup.limit() != onDiskDataSizeWithHeader) { 571 throw new AssertionError( 572 "Expected limit " + onDiskDataSizeWithHeader + ", got " + dup.limit()); 573 } 574 575 // We might optionally allocate HFILEBLOCK_HEADER_SIZE more bytes to read the next 576 // block's header, so there are two sensible values for buffer capacity. 577 int hdrSize = headerSize(); 578 dup.rewind(); 579 if ( 580 dup.remaining() != onDiskDataSizeWithHeader 581 && dup.remaining() != onDiskDataSizeWithHeader + hdrSize 582 ) { 583 throw new AssertionError("Invalid buffer capacity: " + dup.remaining() + ", expected " 584 + onDiskDataSizeWithHeader + " or " + (onDiskDataSizeWithHeader + hdrSize)); 585 } 586 } 587 588 @Override 589 public String toString() { 590 StringBuilder sb = new StringBuilder().append("[").append("blockType=").append(blockType) 591 .append(", fileOffset=").append(offset).append(", headerSize=").append(headerSize()) 592 .append(", onDiskSizeWithoutHeader=").append(onDiskSizeWithoutHeader) 593 .append(", uncompressedSizeWithoutHeader=").append(uncompressedSizeWithoutHeader) 594 .append(", prevBlockOffset=").append(prevBlockOffset).append(", isUseHBaseChecksum=") 595 .append(fileContext.isUseHBaseChecksum()); 596 if (fileContext.isUseHBaseChecksum()) { 597 sb.append(", checksumType=").append(ChecksumType.codeToType(this.bufWithoutChecksum.get(24))) 598 .append(", bytesPerChecksum=").append(this.bufWithoutChecksum.getInt(24 + 1)) 599 .append(", onDiskDataSizeWithHeader=").append(onDiskDataSizeWithHeader); 600 } else { 601 sb.append(", onDiskDataSizeWithHeader=").append(onDiskDataSizeWithHeader).append("(") 602 .append(onDiskSizeWithoutHeader).append("+") 603 .append(HConstants.HFILEBLOCK_HEADER_SIZE_NO_CHECKSUM).append(")"); 604 } 605 String dataBegin; 606 if (bufWithoutChecksum.hasArray()) { 607 dataBegin = Bytes.toStringBinary(bufWithoutChecksum.array(), 608 bufWithoutChecksum.arrayOffset() + headerSize(), 609 Math.min(32, bufWithoutChecksum.limit() - bufWithoutChecksum.arrayOffset() - headerSize())); 610 } else { 611 ByteBuff bufWithoutHeader = getBufferWithoutHeader(); 612 byte[] dataBeginBytes = 613 new byte[Math.min(32, bufWithoutHeader.limit() - bufWithoutHeader.position())]; 614 bufWithoutHeader.get(dataBeginBytes); 615 dataBegin = Bytes.toStringBinary(dataBeginBytes); 616 } 617 sb.append(", getOnDiskSizeWithHeader=").append(getOnDiskSizeWithHeader()) 618 .append(", totalChecksumBytes=").append(totalChecksumBytes()).append(", isUnpacked=") 619 .append(isUnpacked()).append(", buf=[").append(bufWithoutChecksum).append("]") 620 .append(", dataBeginsWith=").append(dataBegin).append(", fileContext=").append(fileContext) 621 .append(", nextBlockOnDiskSize=").append(nextBlockOnDiskSize).append("]"); 622 return sb.toString(); 623 } 624 625 /** 626 * Retrieves the decompressed/decrypted view of this block. An encoded block remains in its 627 * encoded structure. Internal structures are shared between instances where applicable. 628 */ 629 HFileBlock unpack(HFileContext fileContext, FSReader reader) throws IOException { 630 if (!fileContext.isCompressedOrEncrypted()) { 631 // TODO: cannot use our own fileContext here because HFileBlock(ByteBuffer, boolean), 632 // which is used for block serialization to L2 cache, does not preserve encoding and 633 // encryption details. 634 return this; 635 } 636 637 ByteBuff newBuf = allocateBufferForUnpacking(); // allocates space for the decompressed block 638 HFileBlock unpacked = shallowClone(this, newBuf); 639 640 boolean succ = false; 641 final Context context = 642 Context.current().with(CONTEXT_KEY, new HFileContextAttributesBuilderConsumer(fileContext)); 643 try (Scope ignored = context.makeCurrent()) { 644 HFileBlockDecodingContext ctx = blockType == BlockType.ENCODED_DATA 645 ? reader.getBlockDecodingContext() 646 : reader.getDefaultBlockDecodingContext(); 647 // Create a duplicated buffer without the header part. 648 int headerSize = this.headerSize(); 649 ByteBuff dup = this.bufWithoutChecksum.duplicate(); 650 dup.position(headerSize); 651 dup = dup.slice(); 652 // Decode the dup into unpacked#buf 653 ctx.prepareDecoding(unpacked.getOnDiskDataSizeWithHeader() - headerSize, 654 unpacked.getUncompressedSizeWithoutHeader(), unpacked.getBufferWithoutHeader(), dup); 655 succ = true; 656 return unpacked; 657 } finally { 658 if (!succ) { 659 unpacked.release(); 660 } 661 } 662 } 663 664 /** 665 * Always allocates a new buffer of the correct size. Copies header bytes from the existing 666 * buffer. Does not change header fields. Reserve room to keep checksum bytes too. 667 */ 668 private ByteBuff allocateBufferForUnpacking() { 669 int headerSize = headerSize(); 670 int capacityNeeded = headerSize + uncompressedSizeWithoutHeader; 671 672 ByteBuff source = bufWithoutChecksum.duplicate(); 673 ByteBuff newBuf = allocator.allocate(capacityNeeded); 674 675 // Copy header bytes into newBuf. 676 source.position(0); 677 newBuf.put(0, source, 0, headerSize); 678 679 // set limit to exclude next block's header 680 newBuf.limit(capacityNeeded); 681 return newBuf; 682 } 683 684 /** 685 * Return true when this block's buffer has been unpacked, false otherwise. Note this is a 686 * calculated heuristic, not tracked attribute of the block. 687 */ 688 public boolean isUnpacked() { 689 final int headerSize = headerSize(); 690 final int expectedCapacity = headerSize + uncompressedSizeWithoutHeader; 691 final int bufCapacity = bufWithoutChecksum.remaining(); 692 return bufCapacity == expectedCapacity || bufCapacity == expectedCapacity + headerSize; 693 } 694 695 /** 696 * Cannot be {@link #UNSET}. Must be a legitimate value. Used re-making the {@link BlockCacheKey} 697 * when block is returned to the cache. 698 * @return the offset of this block in the file it was read from 699 */ 700 long getOffset() { 701 if (offset < 0) { 702 throw new IllegalStateException("HFile block offset not initialized properly"); 703 } 704 return offset; 705 } 706 707 /** Returns a byte stream reading the data(excluding header and checksum) of this block */ 708 DataInputStream getByteStream() { 709 ByteBuff dup = this.bufWithoutChecksum.duplicate(); 710 dup.position(this.headerSize()); 711 return new DataInputStream(new ByteBuffInputStream(dup)); 712 } 713 714 @Override 715 public long heapSize() { 716 long size = FIXED_OVERHEAD; 717 size += fileContext.heapSize(); 718 if (bufWithoutChecksum != null) { 719 // Deep overhead of the byte buffer. Needs to be aligned separately. 720 size += ClassSize.align(bufWithoutChecksum.capacity() + MULTI_BYTE_BUFFER_HEAP_SIZE); 721 } 722 return ClassSize.align(size); 723 } 724 725 /** 726 * Will be override by {@link SharedMemHFileBlock} or {@link ExclusiveMemHFileBlock}. Return true 727 * by default. 728 */ 729 public boolean isSharedMem() { 730 return true; 731 } 732 733 /** 734 * Unified version 2 {@link HFile} block writer. The intended usage pattern is as follows: 735 * <ol> 736 * <li>Construct an {@link HFileBlock.Writer}, providing a compression algorithm. 737 * <li>Call {@link Writer#startWriting} and get a data stream to write to. 738 * <li>Write your data into the stream. 739 * <li>Call Writer#writeHeaderAndData(FSDataOutputStream) as many times as you need to. store the 740 * serialized block into an external stream. 741 * <li>Repeat to write more blocks. 742 * </ol> 743 * <p> 744 */ 745 static class Writer implements ShipperListener { 746 private enum State { 747 INIT, 748 WRITING, 749 BLOCK_READY 750 }; 751 752 private int maxSizeUnCompressed; 753 754 private BlockCompressedSizePredicator compressedSizePredicator; 755 756 /** Writer state. Used to ensure the correct usage protocol. */ 757 private State state = State.INIT; 758 759 /** Data block encoder used for data blocks */ 760 private final HFileDataBlockEncoder dataBlockEncoder; 761 762 private HFileBlockEncodingContext dataBlockEncodingCtx; 763 764 /** block encoding context for non-data blocks */ 765 private HFileBlockDefaultEncodingContext defaultBlockEncodingCtx; 766 767 /** 768 * The stream we use to accumulate data into a block in an uncompressed format. We reset this 769 * stream at the end of each block and reuse it. The header is written as the first 770 * {@link HConstants#HFILEBLOCK_HEADER_SIZE} bytes into this stream. 771 */ 772 private ByteArrayOutputStream baosInMemory; 773 774 /** 775 * Current block type. Set in {@link #startWriting(BlockType)}. Could be changed in 776 * {@link #finishBlock()} from {@link BlockType#DATA} to {@link BlockType#ENCODED_DATA}. 777 */ 778 private BlockType blockType; 779 780 /** 781 * A stream that we write uncompressed bytes to, which compresses them and writes them to 782 * {@link #baosInMemory}. 783 */ 784 private DataOutputStream userDataStream; 785 786 /** 787 * Bytes to be written to the file system, including the header. Compressed if compression is 788 * turned on. It also includes the checksum data that immediately follows the block data. 789 * (header + data + checksums) 790 */ 791 private ByteArrayOutputStream onDiskBlockBytesWithHeader; 792 793 /** 794 * The size of the checksum data on disk. It is used only if data is not compressed. If data is 795 * compressed, then the checksums are already part of onDiskBytesWithHeader. If data is 796 * uncompressed, then this variable stores the checksum data for this block. 797 */ 798 private byte[] onDiskChecksum = HConstants.EMPTY_BYTE_ARRAY; 799 800 /** 801 * Current block's start offset in the {@link HFile}. Set in 802 * {@link #writeHeaderAndData(FSDataOutputStream)}. 803 */ 804 private long startOffset; 805 806 /** 807 * Offset of previous block by block type. Updated when the next block is started. 808 */ 809 private long[] prevOffsetByType; 810 811 /** The offset of the previous block of the same type */ 812 private long prevOffset; 813 /** Meta data that holds information about the hfileblock **/ 814 private HFileContext fileContext; 815 816 private final ByteBuffAllocator allocator; 817 818 @Override 819 public void beforeShipped() { 820 if (getEncodingState() != null) { 821 getEncodingState().beforeShipped(); 822 } 823 } 824 825 EncodingState getEncodingState() { 826 return dataBlockEncodingCtx.getEncodingState(); 827 } 828 829 /** 830 * @param dataBlockEncoder data block encoding algorithm to use 831 */ 832 public Writer(Configuration conf, HFileDataBlockEncoder dataBlockEncoder, 833 HFileContext fileContext) { 834 this(conf, dataBlockEncoder, fileContext, ByteBuffAllocator.HEAP, fileContext.getBlocksize()); 835 } 836 837 public Writer(Configuration conf, HFileDataBlockEncoder dataBlockEncoder, 838 HFileContext fileContext, ByteBuffAllocator allocator, int maxSizeUnCompressed) { 839 if (fileContext.getBytesPerChecksum() < HConstants.HFILEBLOCK_HEADER_SIZE) { 840 throw new RuntimeException("Unsupported value of bytesPerChecksum. " + " Minimum is " 841 + HConstants.HFILEBLOCK_HEADER_SIZE + " but the configured value is " 842 + fileContext.getBytesPerChecksum()); 843 } 844 this.allocator = allocator; 845 this.dataBlockEncoder = 846 dataBlockEncoder != null ? dataBlockEncoder : NoOpDataBlockEncoder.INSTANCE; 847 this.dataBlockEncodingCtx = this.dataBlockEncoder.newDataBlockEncodingContext(conf, 848 HConstants.HFILEBLOCK_DUMMY_HEADER, fileContext); 849 // TODO: This should be lazily instantiated 850 this.defaultBlockEncodingCtx = new HFileBlockDefaultEncodingContext(conf, null, 851 HConstants.HFILEBLOCK_DUMMY_HEADER, fileContext); 852 // TODO: Set BAOS initial size. Use fileContext.getBlocksize() and add for header/checksum 853 baosInMemory = new ByteArrayOutputStream(); 854 prevOffsetByType = new long[BlockType.values().length]; 855 for (int i = 0; i < prevOffsetByType.length; ++i) { 856 prevOffsetByType[i] = UNSET; 857 } 858 // TODO: Why fileContext saved away when we have dataBlockEncoder and/or 859 // defaultDataBlockEncoder? 860 this.fileContext = fileContext; 861 this.compressedSizePredicator = (BlockCompressedSizePredicator) ReflectionUtils.newInstance( 862 conf.getClass(BLOCK_COMPRESSED_SIZE_PREDICATOR, UncompressedBlockSizePredicator.class), 863 new Configuration(conf)); 864 this.maxSizeUnCompressed = maxSizeUnCompressed; 865 } 866 867 /** 868 * Starts writing into the block. The previous block's data is discarded. 869 * @return the stream the user can write their data into 870 */ 871 DataOutputStream startWriting(BlockType newBlockType) throws IOException { 872 if (state == State.BLOCK_READY && startOffset != -1) { 873 // We had a previous block that was written to a stream at a specific 874 // offset. Save that offset as the last offset of a block of that type. 875 prevOffsetByType[blockType.getId()] = startOffset; 876 } 877 878 startOffset = -1; 879 blockType = newBlockType; 880 881 baosInMemory.reset(); 882 baosInMemory.write(HConstants.HFILEBLOCK_DUMMY_HEADER); 883 884 state = State.WRITING; 885 886 // We will compress it later in finishBlock() 887 userDataStream = new ByteBufferWriterDataOutputStream(baosInMemory); 888 if (newBlockType == BlockType.DATA) { 889 this.dataBlockEncoder.startBlockEncoding(dataBlockEncodingCtx, userDataStream); 890 } 891 return userDataStream; 892 } 893 894 /** 895 * Writes the Cell to this block 896 */ 897 void write(Cell cell) throws IOException { 898 expectState(State.WRITING); 899 this.dataBlockEncoder.encode(cell, dataBlockEncodingCtx, this.userDataStream); 900 } 901 902 /** 903 * Transitions the block writer from the "writing" state to the "block ready" state. Does 904 * nothing if a block is already finished. 905 */ 906 void ensureBlockReady() throws IOException { 907 Preconditions.checkState(state != State.INIT, "Unexpected state: " + state); 908 909 if (state == State.BLOCK_READY) { 910 return; 911 } 912 913 // This will set state to BLOCK_READY. 914 finishBlock(); 915 } 916 917 public boolean checkBoundariesWithPredicate() { 918 int rawBlockSize = encodedBlockSizeWritten(); 919 if (rawBlockSize >= maxSizeUnCompressed) { 920 return true; 921 } else { 922 return compressedSizePredicator.shouldFinishBlock(rawBlockSize); 923 } 924 } 925 926 /** 927 * Finish up writing of the block. Flushes the compressing stream (if using compression), fills 928 * out the header, does any compression/encryption of bytes to flush out to disk, and manages 929 * the cache on write content, if applicable. Sets block write state to "block ready". 930 */ 931 private void finishBlock() throws IOException { 932 if (blockType == BlockType.DATA) { 933 this.dataBlockEncoder.endBlockEncoding(dataBlockEncodingCtx, userDataStream, 934 baosInMemory.getBuffer(), blockType); 935 blockType = dataBlockEncodingCtx.getBlockType(); 936 } 937 userDataStream.flush(); 938 prevOffset = prevOffsetByType[blockType.getId()]; 939 940 // We need to cache the unencoded/uncompressed size before changing the block state 941 int rawBlockSize = 0; 942 if (this.getEncodingState() != null) { 943 rawBlockSize = encodedBlockSizeWritten(); 944 } 945 // We need to set state before we can package the block up for cache-on-write. In a way, the 946 // block is ready, but not yet encoded or compressed. 947 state = State.BLOCK_READY; 948 Bytes compressAndEncryptDat; 949 if (blockType == BlockType.DATA || blockType == BlockType.ENCODED_DATA) { 950 compressAndEncryptDat = 951 dataBlockEncodingCtx.compressAndEncrypt(baosInMemory.getBuffer(), 0, baosInMemory.size()); 952 } else { 953 compressAndEncryptDat = defaultBlockEncodingCtx.compressAndEncrypt(baosInMemory.getBuffer(), 954 0, baosInMemory.size()); 955 } 956 if (compressAndEncryptDat == null) { 957 compressAndEncryptDat = new Bytes(baosInMemory.getBuffer(), 0, baosInMemory.size()); 958 } 959 if (onDiskBlockBytesWithHeader == null) { 960 onDiskBlockBytesWithHeader = new ByteArrayOutputStream(compressAndEncryptDat.getLength()); 961 } 962 onDiskBlockBytesWithHeader.reset(); 963 onDiskBlockBytesWithHeader.write(compressAndEncryptDat.get(), 964 compressAndEncryptDat.getOffset(), compressAndEncryptDat.getLength()); 965 // Update raw and compressed sizes in the predicate 966 compressedSizePredicator.updateLatestBlockSizes(fileContext, rawBlockSize, 967 onDiskBlockBytesWithHeader.size()); 968 969 // Calculate how many bytes we need for checksum on the tail of the block. 970 int numBytes = (int) ChecksumUtil.numBytes(onDiskBlockBytesWithHeader.size(), 971 fileContext.getBytesPerChecksum()); 972 973 // Put the header for the on disk bytes; header currently is unfilled-out 974 putHeader(onDiskBlockBytesWithHeader, onDiskBlockBytesWithHeader.size() + numBytes, 975 baosInMemory.size(), onDiskBlockBytesWithHeader.size()); 976 977 if (onDiskChecksum.length != numBytes) { 978 onDiskChecksum = new byte[numBytes]; 979 } 980 ChecksumUtil.generateChecksums(onDiskBlockBytesWithHeader.getBuffer(), 0, 981 onDiskBlockBytesWithHeader.size(), onDiskChecksum, 0, fileContext.getChecksumType(), 982 fileContext.getBytesPerChecksum()); 983 } 984 985 /** 986 * Put the header into the given byte array at the given offset. 987 * @param onDiskSize size of the block on disk header + data + checksum 988 * @param uncompressedSize size of the block after decompression (but before optional data block 989 * decoding) including header 990 * @param onDiskDataSize size of the block on disk with header and data but not including the 991 * checksums 992 */ 993 private void putHeader(byte[] dest, int offset, int onDiskSize, int uncompressedSize, 994 int onDiskDataSize) { 995 offset = blockType.put(dest, offset); 996 offset = Bytes.putInt(dest, offset, onDiskSize - HConstants.HFILEBLOCK_HEADER_SIZE); 997 offset = Bytes.putInt(dest, offset, uncompressedSize - HConstants.HFILEBLOCK_HEADER_SIZE); 998 offset = Bytes.putLong(dest, offset, prevOffset); 999 offset = Bytes.putByte(dest, offset, fileContext.getChecksumType().getCode()); 1000 offset = Bytes.putInt(dest, offset, fileContext.getBytesPerChecksum()); 1001 Bytes.putInt(dest, offset, onDiskDataSize); 1002 } 1003 1004 private void putHeader(ByteBuff buff, int onDiskSize, int uncompressedSize, 1005 int onDiskDataSize) { 1006 buff.rewind(); 1007 blockType.write(buff); 1008 buff.putInt(onDiskSize - HConstants.HFILEBLOCK_HEADER_SIZE); 1009 buff.putInt(uncompressedSize - HConstants.HFILEBLOCK_HEADER_SIZE); 1010 buff.putLong(prevOffset); 1011 buff.put(fileContext.getChecksumType().getCode()); 1012 buff.putInt(fileContext.getBytesPerChecksum()); 1013 buff.putInt(onDiskDataSize); 1014 } 1015 1016 private void putHeader(ByteArrayOutputStream dest, int onDiskSize, int uncompressedSize, 1017 int onDiskDataSize) { 1018 putHeader(dest.getBuffer(), 0, onDiskSize, uncompressedSize, onDiskDataSize); 1019 } 1020 1021 /** 1022 * Similar to {@link #writeHeaderAndData(FSDataOutputStream)}, but records the offset of this 1023 * block so that it can be referenced in the next block of the same type. 1024 */ 1025 void writeHeaderAndData(FSDataOutputStream out) throws IOException { 1026 long offset = out.getPos(); 1027 if (startOffset != UNSET && offset != startOffset) { 1028 throw new IOException("A " + blockType + " block written to a " 1029 + "stream twice, first at offset " + startOffset + ", then at " + offset); 1030 } 1031 startOffset = offset; 1032 finishBlockAndWriteHeaderAndData(out); 1033 } 1034 1035 /** 1036 * Writes the header and the compressed data of this block (or uncompressed data when not using 1037 * compression) into the given stream. Can be called in the "writing" state or in the "block 1038 * ready" state. If called in the "writing" state, transitions the writer to the "block ready" 1039 * state. 1040 * @param out the output stream to write the 1041 */ 1042 protected void finishBlockAndWriteHeaderAndData(DataOutputStream out) throws IOException { 1043 ensureBlockReady(); 1044 long startTime = EnvironmentEdgeManager.currentTime(); 1045 out.write(onDiskBlockBytesWithHeader.getBuffer(), 0, onDiskBlockBytesWithHeader.size()); 1046 out.write(onDiskChecksum); 1047 HFile.updateWriteLatency(EnvironmentEdgeManager.currentTime() - startTime); 1048 } 1049 1050 /** 1051 * Returns the header or the compressed data (or uncompressed data when not using compression) 1052 * as a byte array. Can be called in the "writing" state or in the "block ready" state. If 1053 * called in the "writing" state, transitions the writer to the "block ready" state. This 1054 * returns the header + data + checksums stored on disk. 1055 * @return header and data as they would be stored on disk in a byte array 1056 */ 1057 byte[] getHeaderAndDataForTest() throws IOException { 1058 ensureBlockReady(); 1059 // This is not very optimal, because we are doing an extra copy. 1060 // But this method is used only by unit tests. 1061 byte[] output = new byte[onDiskBlockBytesWithHeader.size() + onDiskChecksum.length]; 1062 System.arraycopy(onDiskBlockBytesWithHeader.getBuffer(), 0, output, 0, 1063 onDiskBlockBytesWithHeader.size()); 1064 System.arraycopy(onDiskChecksum, 0, output, onDiskBlockBytesWithHeader.size(), 1065 onDiskChecksum.length); 1066 return output; 1067 } 1068 1069 /** 1070 * Releases resources used by this writer. 1071 */ 1072 void release() { 1073 if (dataBlockEncodingCtx != null) { 1074 dataBlockEncodingCtx.close(); 1075 dataBlockEncodingCtx = null; 1076 } 1077 if (defaultBlockEncodingCtx != null) { 1078 defaultBlockEncodingCtx.close(); 1079 defaultBlockEncodingCtx = null; 1080 } 1081 } 1082 1083 /** 1084 * Returns the on-disk size of the data portion of the block. This is the compressed size if 1085 * compression is enabled. Can only be called in the "block ready" state. Header is not 1086 * compressed, and its size is not included in the return value. 1087 * @return the on-disk size of the block, not including the header. 1088 */ 1089 int getOnDiskSizeWithoutHeader() { 1090 expectState(State.BLOCK_READY); 1091 return onDiskBlockBytesWithHeader.size() + onDiskChecksum.length 1092 - HConstants.HFILEBLOCK_HEADER_SIZE; 1093 } 1094 1095 /** 1096 * Returns the on-disk size of the block. Can only be called in the "block ready" state. 1097 * @return the on-disk size of the block ready to be written, including the header size, the 1098 * data and the checksum data. 1099 */ 1100 int getOnDiskSizeWithHeader() { 1101 expectState(State.BLOCK_READY); 1102 return onDiskBlockBytesWithHeader.size() + onDiskChecksum.length; 1103 } 1104 1105 /** 1106 * The uncompressed size of the block data. Does not include header size. 1107 */ 1108 int getUncompressedSizeWithoutHeader() { 1109 expectState(State.BLOCK_READY); 1110 return baosInMemory.size() - HConstants.HFILEBLOCK_HEADER_SIZE; 1111 } 1112 1113 /** 1114 * The uncompressed size of the block data, including header size. 1115 */ 1116 public int getUncompressedSizeWithHeader() { 1117 expectState(State.BLOCK_READY); 1118 return baosInMemory.size(); 1119 } 1120 1121 /** Returns true if a block is being written */ 1122 boolean isWriting() { 1123 return state == State.WRITING; 1124 } 1125 1126 /** 1127 * Returns the number of bytes written into the current block so far, or zero if not writing the 1128 * block at the moment. Note that this will return zero in the "block ready" state as well. 1129 * @return the number of bytes written 1130 */ 1131 public int encodedBlockSizeWritten() { 1132 return state != State.WRITING ? 0 : this.getEncodingState().getEncodedDataSizeWritten(); 1133 } 1134 1135 /** 1136 * Returns the number of bytes written into the current block so far, or zero if not writing the 1137 * block at the moment. Note that this will return zero in the "block ready" state as well. 1138 * @return the number of bytes written 1139 */ 1140 public int blockSizeWritten() { 1141 return state != State.WRITING ? 0 : this.getEncodingState().getUnencodedDataSizeWritten(); 1142 } 1143 1144 /** 1145 * Clones the header followed by the uncompressed data, even if using compression. This is 1146 * needed for storing uncompressed blocks in the block cache. Can be called in the "writing" 1147 * state or the "block ready" state. Returns only the header and data, does not include checksum 1148 * data. 1149 * @return Returns an uncompressed block ByteBuff for caching on write 1150 */ 1151 ByteBuff cloneUncompressedBufferWithHeader() { 1152 expectState(State.BLOCK_READY); 1153 ByteBuff bytebuff = allocator.allocate(baosInMemory.size()); 1154 baosInMemory.toByteBuff(bytebuff); 1155 int numBytes = (int) ChecksumUtil.numBytes(onDiskBlockBytesWithHeader.size(), 1156 fileContext.getBytesPerChecksum()); 1157 putHeader(bytebuff, onDiskBlockBytesWithHeader.size() + numBytes, baosInMemory.size(), 1158 onDiskBlockBytesWithHeader.size()); 1159 bytebuff.rewind(); 1160 return bytebuff; 1161 } 1162 1163 /** 1164 * Clones the header followed by the on-disk (compressed/encoded/encrypted) data. This is needed 1165 * for storing packed blocks in the block cache. Returns only the header and data, Does not 1166 * include checksum data. 1167 * @return Returns a copy of block bytes for caching on write 1168 */ 1169 private ByteBuff cloneOnDiskBufferWithHeader() { 1170 expectState(State.BLOCK_READY); 1171 ByteBuff bytebuff = allocator.allocate(onDiskBlockBytesWithHeader.size()); 1172 onDiskBlockBytesWithHeader.toByteBuff(bytebuff); 1173 bytebuff.rewind(); 1174 return bytebuff; 1175 } 1176 1177 private void expectState(State expectedState) { 1178 if (state != expectedState) { 1179 throw new IllegalStateException( 1180 "Expected state: " + expectedState + ", actual state: " + state); 1181 } 1182 } 1183 1184 /** 1185 * Takes the given {@link BlockWritable} instance, creates a new block of its appropriate type, 1186 * writes the writable into this block, and flushes the block into the output stream. The writer 1187 * is instructed not to buffer uncompressed bytes for cache-on-write. 1188 * @param bw the block-writable object to write as a block 1189 * @param out the file system output stream 1190 */ 1191 void writeBlock(BlockWritable bw, FSDataOutputStream out) throws IOException { 1192 bw.writeToBlock(startWriting(bw.getBlockType())); 1193 writeHeaderAndData(out); 1194 } 1195 1196 /** 1197 * Creates a new HFileBlock. Checksums have already been validated, so the byte buffer passed 1198 * into the constructor of this newly created block does not have checksum data even though the 1199 * header minor version is MINOR_VERSION_WITH_CHECKSUM. This is indicated by setting a 0 value 1200 * in bytesPerChecksum. This method copies the on-disk or uncompressed data to build the 1201 * HFileBlock which is used only while writing blocks and caching. 1202 * <p> 1203 * TODO: Should there be an option where a cache can ask that hbase preserve block checksums for 1204 * checking after a block comes out of the cache? Otehrwise, cache is responsible for blocks 1205 * being wholesome (ECC memory or if file-backed, it does checksumming). 1206 */ 1207 HFileBlock getBlockForCaching(CacheConfig cacheConf) { 1208 HFileContext newContext = new HFileContextBuilder().withBlockSize(fileContext.getBlocksize()) 1209 .withBytesPerCheckSum(0).withChecksumType(ChecksumType.NULL) // no checksums in cached data 1210 .withCompression(fileContext.getCompression()) 1211 .withDataBlockEncoding(fileContext.getDataBlockEncoding()) 1212 .withHBaseCheckSum(fileContext.isUseHBaseChecksum()) 1213 .withCompressTags(fileContext.isCompressTags()) 1214 .withIncludesMvcc(fileContext.isIncludesMvcc()) 1215 .withIncludesTags(fileContext.isIncludesTags()) 1216 .withColumnFamily(fileContext.getColumnFamily()).withTableName(fileContext.getTableName()) 1217 .build(); 1218 // Build the HFileBlock. 1219 HFileBlockBuilder builder = new HFileBlockBuilder(); 1220 ByteBuff buff; 1221 if (cacheConf.shouldCacheCompressed(blockType.getCategory())) { 1222 buff = cloneOnDiskBufferWithHeader(); 1223 } else { 1224 buff = cloneUncompressedBufferWithHeader(); 1225 } 1226 return builder.withBlockType(blockType) 1227 .withOnDiskSizeWithoutHeader(getOnDiskSizeWithoutHeader()) 1228 .withUncompressedSizeWithoutHeader(getUncompressedSizeWithoutHeader()) 1229 .withPrevBlockOffset(prevOffset).withByteBuff(buff).withFillHeader(FILL_HEADER) 1230 .withOffset(startOffset).withNextBlockOnDiskSize(UNSET) 1231 .withOnDiskDataSizeWithHeader(onDiskBlockBytesWithHeader.size() + onDiskChecksum.length) 1232 .withHFileContext(newContext).withByteBuffAllocator(cacheConf.getByteBuffAllocator()) 1233 .withShared(!buff.hasArray()).build(); 1234 } 1235 } 1236 1237 /** Something that can be written into a block. */ 1238 interface BlockWritable { 1239 /** The type of block this data should use. */ 1240 BlockType getBlockType(); 1241 1242 /** 1243 * Writes the block to the provided stream. Must not write any magic records. 1244 * @param out a stream to write uncompressed data into 1245 */ 1246 void writeToBlock(DataOutput out) throws IOException; 1247 } 1248 1249 /** 1250 * Iterator for reading {@link HFileBlock}s in load-on-open-section, such as root data index 1251 * block, meta index block, file info block etc. 1252 */ 1253 interface BlockIterator { 1254 /** 1255 * Get the next block, or null if there are no more blocks to iterate. 1256 */ 1257 HFileBlock nextBlock() throws IOException; 1258 1259 /** 1260 * Similar to {@link #nextBlock()} but checks block type, throws an exception if incorrect, and 1261 * returns the HFile block 1262 */ 1263 HFileBlock nextBlockWithBlockType(BlockType blockType) throws IOException; 1264 1265 /** 1266 * Now we use the {@link ByteBuffAllocator} to manage the nio ByteBuffers for HFileBlocks, so we 1267 * must deallocate all of the ByteBuffers in the end life. the BlockIterator's life cycle is 1268 * starting from opening an HFileReader and stopped when the HFileReader#close, so we will keep 1269 * track all the read blocks until we call {@link BlockIterator#freeBlocks()} when closing the 1270 * HFileReader. Sum bytes of those blocks in load-on-open section should be quite small, so 1271 * tracking them should be OK. 1272 */ 1273 void freeBlocks(); 1274 } 1275 1276 /** An HFile block reader with iteration ability. */ 1277 interface FSReader { 1278 /** 1279 * Reads the block at the given offset in the file with the given on-disk size and uncompressed 1280 * size. 1281 * @param offset of the file to read 1282 * @param onDiskSize the on-disk size of the entire block, including all applicable headers, 1283 * or -1 if unknown 1284 * @param pread true to use pread, otherwise use the stream read. 1285 * @param updateMetrics update the metrics or not. 1286 * @param intoHeap allocate the block's ByteBuff by {@link ByteBuffAllocator} or JVM heap. 1287 * For LRUBlockCache, we must ensure that the block to cache is an heap 1288 * one, because the memory occupation is based on heap now, also for 1289 * {@link CombinedBlockCache}, we use the heap LRUBlockCache as L1 cache to 1290 * cache small blocks such as IndexBlock or MetaBlock for faster access. So 1291 * introduce an flag here to decide whether allocate from JVM heap or not 1292 * so that we can avoid an extra off-heap to heap memory copy when using 1293 * LRUBlockCache. For most cases, we known what's the expected block type 1294 * we'll read, while for some special case (Example: 1295 * HFileReaderImpl#readNextDataBlock()), we cannot pre-decide what's the 1296 * expected block type, then we can only allocate block's ByteBuff from 1297 * {@link ByteBuffAllocator} firstly, and then when caching it in 1298 * {@link LruBlockCache} we'll check whether the ByteBuff is from heap or 1299 * not, if not then we'll clone it to an heap one and cache it. 1300 * @return the newly read block 1301 */ 1302 HFileBlock readBlockData(long offset, long onDiskSize, boolean pread, boolean updateMetrics, 1303 boolean intoHeap) throws IOException; 1304 1305 /** 1306 * Creates a block iterator over the given portion of the {@link HFile}. The iterator returns 1307 * blocks starting with offset such that offset <= startOffset < endOffset. Returned 1308 * blocks are always unpacked. Used when no hfile index available; e.g. reading in the hfile 1309 * index blocks themselves on file open. 1310 * @param startOffset the offset of the block to start iteration with 1311 * @param endOffset the offset to end iteration at (exclusive) 1312 * @return an iterator of blocks between the two given offsets 1313 */ 1314 BlockIterator blockRange(long startOffset, long endOffset); 1315 1316 /** Closes the backing streams */ 1317 void closeStreams() throws IOException; 1318 1319 /** Get a decoder for {@link BlockType#ENCODED_DATA} blocks from this file. */ 1320 HFileBlockDecodingContext getBlockDecodingContext(); 1321 1322 /** Get the default decoder for blocks from this file. */ 1323 HFileBlockDecodingContext getDefaultBlockDecodingContext(); 1324 1325 void setIncludesMemStoreTS(boolean includesMemstoreTS); 1326 1327 void setDataBlockEncoder(HFileDataBlockEncoder encoder, Configuration conf); 1328 1329 /** 1330 * To close the stream's socket. Note: This can be concurrently called from multiple threads and 1331 * implementation should take care of thread safety. 1332 */ 1333 void unbufferStream(); 1334 } 1335 1336 /** 1337 * Data-structure to use caching the header of the NEXT block. Only works if next read that comes 1338 * in here is next in sequence in this block. When we read, we read current block and the next 1339 * blocks' header. We do this so we have the length of the next block to read if the hfile index 1340 * is not available (rare, at hfile open only). 1341 */ 1342 private static class PrefetchedHeader { 1343 long offset = -1; 1344 byte[] header = new byte[HConstants.HFILEBLOCK_HEADER_SIZE]; 1345 final ByteBuff buf = new SingleByteBuff(ByteBuffer.wrap(header, 0, header.length)); 1346 1347 @Override 1348 public String toString() { 1349 return "offset=" + this.offset + ", header=" + Bytes.toStringBinary(header); 1350 } 1351 } 1352 1353 /** 1354 * Reads version 2 HFile blocks from the filesystem. 1355 */ 1356 static class FSReaderImpl implements FSReader { 1357 /** 1358 * The file system stream of the underlying {@link HFile} that does or doesn't do checksum 1359 * validations in the filesystem 1360 */ 1361 private FSDataInputStreamWrapper streamWrapper; 1362 1363 private HFileBlockDecodingContext encodedBlockDecodingCtx; 1364 1365 /** Default context used when BlockType != {@link BlockType#ENCODED_DATA}. */ 1366 private final HFileBlockDefaultDecodingContext defaultDecodingCtx; 1367 1368 /** 1369 * Cache of the NEXT header after this. Check it is indeed next blocks header before using it. 1370 * TODO: Review. This overread into next block to fetch next blocks header seems unnecessary 1371 * given we usually get the block size from the hfile index. Review! 1372 */ 1373 private AtomicReference<PrefetchedHeader> prefetchedHeader = 1374 new AtomicReference<>(new PrefetchedHeader()); 1375 1376 /** The size of the file we are reading from, or -1 if unknown. */ 1377 private long fileSize; 1378 1379 /** The size of the header */ 1380 protected final int hdrSize; 1381 1382 /** The filesystem used to access data */ 1383 private HFileSystem hfs; 1384 1385 private HFileContext fileContext; 1386 // Cache the fileName 1387 private String pathName; 1388 1389 private final ByteBuffAllocator allocator; 1390 1391 private final Lock streamLock = new ReentrantLock(); 1392 1393 private final boolean isPreadAllBytes; 1394 1395 private final long readWarnTime; 1396 1397 /** 1398 * If reading block cost time in milliseconds more than the threshold, a warning will be logged. 1399 */ 1400 public static final String FS_READER_WARN_TIME_MS = "hbase.fs.reader.warn.time.ms"; 1401 1402 FSReaderImpl(ReaderContext readerContext, HFileContext fileContext, ByteBuffAllocator allocator, 1403 Configuration conf) throws IOException { 1404 this.fileSize = readerContext.getFileSize(); 1405 this.hfs = readerContext.getFileSystem(); 1406 if (readerContext.getFilePath() != null) { 1407 this.pathName = readerContext.getFilePath().toString(); 1408 } 1409 this.fileContext = fileContext; 1410 this.hdrSize = headerSize(fileContext.isUseHBaseChecksum()); 1411 this.allocator = allocator; 1412 1413 this.streamWrapper = readerContext.getInputStreamWrapper(); 1414 // Older versions of HBase didn't support checksum. 1415 this.streamWrapper.prepareForBlockReader(!fileContext.isUseHBaseChecksum()); 1416 defaultDecodingCtx = new HFileBlockDefaultDecodingContext(conf, fileContext); 1417 encodedBlockDecodingCtx = defaultDecodingCtx; 1418 isPreadAllBytes = readerContext.isPreadAllBytes(); 1419 // Default warn threshold set to -1, it means skipping record the read block slow warning log. 1420 readWarnTime = conf.getLong(FS_READER_WARN_TIME_MS, -1L); 1421 } 1422 1423 @Override 1424 public BlockIterator blockRange(final long startOffset, final long endOffset) { 1425 final FSReader owner = this; // handle for inner class 1426 return new BlockIterator() { 1427 private volatile boolean freed = false; 1428 // Tracking all read blocks until we call freeBlocks. 1429 private List<HFileBlock> blockTracker = new ArrayList<>(); 1430 private long offset = startOffset; 1431 // Cache length of next block. Current block has the length of next block in it. 1432 private long length = -1; 1433 1434 @Override 1435 public HFileBlock nextBlock() throws IOException { 1436 if (offset >= endOffset) { 1437 return null; 1438 } 1439 HFileBlock b = readBlockData(offset, length, false, false, true); 1440 offset += b.getOnDiskSizeWithHeader(); 1441 length = b.getNextBlockOnDiskSize(); 1442 HFileBlock uncompressed = b.unpack(fileContext, owner); 1443 if (uncompressed != b) { 1444 b.release(); // Need to release the compressed Block now. 1445 } 1446 blockTracker.add(uncompressed); 1447 return uncompressed; 1448 } 1449 1450 @Override 1451 public HFileBlock nextBlockWithBlockType(BlockType blockType) throws IOException { 1452 HFileBlock blk = nextBlock(); 1453 if (blk.getBlockType() != blockType) { 1454 throw new IOException( 1455 "Expected block of type " + blockType + " but found " + blk.getBlockType()); 1456 } 1457 return blk; 1458 } 1459 1460 @Override 1461 public void freeBlocks() { 1462 if (freed) { 1463 return; 1464 } 1465 blockTracker.forEach(HFileBlock::release); 1466 blockTracker = null; 1467 freed = true; 1468 } 1469 }; 1470 } 1471 1472 /** 1473 * Does a positional read or a seek and read into the given byte buffer. We need take care that 1474 * we will call the {@link ByteBuff#release()} for every exit to deallocate the ByteBuffers, 1475 * otherwise the memory leak may happen. 1476 * @param dest destination buffer 1477 * @param size size of read 1478 * @param peekIntoNextBlock whether to read the next block's on-disk size 1479 * @param fileOffset position in the stream to read at 1480 * @param pread whether we should do a positional read 1481 * @param istream The input source of data 1482 * @return true to indicate the destination buffer include the next block header, otherwise only 1483 * include the current block data without the next block header. 1484 * @throws IOException if any IO error happen. 1485 */ 1486 protected boolean readAtOffset(FSDataInputStream istream, ByteBuff dest, int size, 1487 boolean peekIntoNextBlock, long fileOffset, boolean pread) throws IOException { 1488 if (!pread) { 1489 // Seek + read. Better for scanning. 1490 istream.seek(fileOffset); 1491 long realOffset = istream.getPos(); 1492 if (realOffset != fileOffset) { 1493 throw new IOException("Tried to seek to " + fileOffset + " to read " + size 1494 + " bytes, but pos=" + realOffset + " after seek"); 1495 } 1496 if (!peekIntoNextBlock) { 1497 BlockIOUtils.readFully(dest, istream, size); 1498 return false; 1499 } 1500 1501 // Try to read the next block header 1502 if (!BlockIOUtils.readWithExtra(dest, istream, size, hdrSize)) { 1503 // did not read the next block header. 1504 return false; 1505 } 1506 } else { 1507 // Positional read. Better for random reads; or when the streamLock is already locked. 1508 int extraSize = peekIntoNextBlock ? hdrSize : 0; 1509 if ( 1510 !BlockIOUtils.preadWithExtra(dest, istream, fileOffset, size, extraSize, isPreadAllBytes) 1511 ) { 1512 // did not read the next block header. 1513 return false; 1514 } 1515 } 1516 assert peekIntoNextBlock; 1517 return true; 1518 } 1519 1520 /** 1521 * Reads a version 2 block (version 1 blocks not supported and not expected). Tries to do as 1522 * little memory allocation as possible, using the provided on-disk size. 1523 * @param offset the offset in the stream to read at 1524 * @param onDiskSizeWithHeaderL the on-disk size of the block, including the header, or -1 if 1525 * unknown; i.e. when iterating over blocks reading in the file 1526 * metadata info. 1527 * @param pread whether to use a positional read 1528 * @param updateMetrics whether to update the metrics 1529 * @param intoHeap allocate ByteBuff of block from heap or off-heap. 1530 * @see FSReader#readBlockData(long, long, boolean, boolean, boolean) for more details about the 1531 * useHeap. 1532 */ 1533 @Override 1534 public HFileBlock readBlockData(long offset, long onDiskSizeWithHeaderL, boolean pread, 1535 boolean updateMetrics, boolean intoHeap) throws IOException { 1536 // Get a copy of the current state of whether to validate 1537 // hbase checksums or not for this read call. This is not 1538 // thread-safe but the one constraint is that if we decide 1539 // to skip hbase checksum verification then we are 1540 // guaranteed to use hdfs checksum verification. 1541 boolean doVerificationThruHBaseChecksum = streamWrapper.shouldUseHBaseChecksum(); 1542 FSDataInputStream is = streamWrapper.getStream(doVerificationThruHBaseChecksum); 1543 final Context context = Context.current().with(CONTEXT_KEY, 1544 new HFileContextAttributesBuilderConsumer(fileContext) 1545 .setSkipChecksum(doVerificationThruHBaseChecksum) 1546 .setReadType(pread ? ReadType.POSITIONAL_READ : ReadType.SEEK_PLUS_READ)); 1547 try (Scope ignored = context.makeCurrent()) { 1548 HFileBlock blk = readBlockDataInternal(is, offset, onDiskSizeWithHeaderL, pread, 1549 doVerificationThruHBaseChecksum, updateMetrics, intoHeap); 1550 if (blk == null) { 1551 HFile.LOG.warn("HBase checksum verification failed for file {} at offset {} filesize {}." 1552 + " Retrying read with HDFS checksums turned on...", pathName, offset, fileSize); 1553 1554 if (!doVerificationThruHBaseChecksum) { 1555 String msg = "HBase checksum verification failed for file " + pathName + " at offset " 1556 + offset + " filesize " + fileSize + " but this cannot happen because doVerify is " 1557 + doVerificationThruHBaseChecksum; 1558 HFile.LOG.warn(msg); 1559 throw new IOException(msg); // cannot happen case here 1560 } 1561 HFile.CHECKSUM_FAILURES.increment(); // update metrics 1562 1563 // If we have a checksum failure, we fall back into a mode where 1564 // the next few reads use HDFS level checksums. We aim to make the 1565 // next CHECKSUM_VERIFICATION_NUM_IO_THRESHOLD reads avoid 1566 // hbase checksum verification, but since this value is set without 1567 // holding any locks, it can so happen that we might actually do 1568 // a few more than precisely this number. 1569 is = this.streamWrapper.fallbackToFsChecksum(CHECKSUM_VERIFICATION_NUM_IO_THRESHOLD); 1570 doVerificationThruHBaseChecksum = false; 1571 blk = readBlockDataInternal(is, offset, onDiskSizeWithHeaderL, pread, 1572 doVerificationThruHBaseChecksum, updateMetrics, intoHeap); 1573 if (blk != null) { 1574 HFile.LOG.warn( 1575 "HDFS checksum verification succeeded for file {} at offset {} filesize" + " {}", 1576 pathName, offset, fileSize); 1577 } 1578 } 1579 if (blk == null && !doVerificationThruHBaseChecksum) { 1580 String msg = 1581 "readBlockData failed, possibly due to " + "checksum verification failed for file " 1582 + pathName + " at offset " + offset + " filesize " + fileSize; 1583 HFile.LOG.warn(msg); 1584 throw new IOException(msg); 1585 } 1586 1587 // If there is a checksum mismatch earlier, then retry with 1588 // HBase checksums switched off and use HDFS checksum verification. 1589 // This triggers HDFS to detect and fix corrupt replicas. The 1590 // next checksumOffCount read requests will use HDFS checksums. 1591 // The decrementing of this.checksumOffCount is not thread-safe, 1592 // but it is harmless because eventually checksumOffCount will be 1593 // a negative number. 1594 streamWrapper.checksumOk(); 1595 return blk; 1596 } 1597 } 1598 1599 /** 1600 * Check that {@code value} read from a block header seems reasonable, within a large margin of 1601 * error. 1602 * @return {@code true} if the value is safe to proceed, {@code false} otherwise. 1603 */ 1604 private boolean checkOnDiskSizeWithHeader(int value) { 1605 if (value < 0) { 1606 if (LOG.isTraceEnabled()) { 1607 LOG.trace( 1608 "onDiskSizeWithHeader={}; value represents a size, so it should never be negative.", 1609 value); 1610 } 1611 return false; 1612 } 1613 if (value - hdrSize < 0) { 1614 if (LOG.isTraceEnabled()) { 1615 LOG.trace("onDiskSizeWithHeader={}, hdrSize={}; don't accept a value that is negative" 1616 + " after the header size is excluded.", value, hdrSize); 1617 } 1618 return false; 1619 } 1620 return true; 1621 } 1622 1623 /** 1624 * Check that {@code value} provided by the calling context seems reasonable, within a large 1625 * margin of error. 1626 * @return {@code true} if the value is safe to proceed, {@code false} otherwise. 1627 */ 1628 private boolean checkCallerProvidedOnDiskSizeWithHeader(long value) { 1629 // same validation logic as is used by Math.toIntExact(long) 1630 int intValue = (int) value; 1631 if (intValue != value) { 1632 if (LOG.isTraceEnabled()) { 1633 LOG.trace("onDiskSizeWithHeaderL={}; value exceeds int size limits.", value); 1634 } 1635 return false; 1636 } 1637 if (intValue == -1) { 1638 // a magic value we expect to see. 1639 return true; 1640 } 1641 return checkOnDiskSizeWithHeader(intValue); 1642 } 1643 1644 /** 1645 * Check atomic reference cache for this block's header. Cache only good if next read coming 1646 * through is next in sequence in the block. We read next block's header on the tail of reading 1647 * the previous block to save a seek. Otherwise, we have to do a seek to read the header before 1648 * we can pull in the block OR we have to backup the stream because we over-read (the next 1649 * block's header). 1650 * @see PrefetchedHeader 1651 * @return The cached block header or null if not found. 1652 * @see #cacheNextBlockHeader(long, ByteBuff, int, int) 1653 */ 1654 private ByteBuff getCachedHeader(final long offset) { 1655 PrefetchedHeader ph = this.prefetchedHeader.get(); 1656 return ph != null && ph.offset == offset ? ph.buf : null; 1657 } 1658 1659 /** 1660 * Save away the next blocks header in atomic reference. 1661 * @see #getCachedHeader(long) 1662 * @see PrefetchedHeader 1663 */ 1664 private void cacheNextBlockHeader(final long offset, ByteBuff onDiskBlock, 1665 int onDiskSizeWithHeader, int headerLength) { 1666 PrefetchedHeader ph = new PrefetchedHeader(); 1667 ph.offset = offset; 1668 onDiskBlock.get(onDiskSizeWithHeader, ph.header, 0, headerLength); 1669 this.prefetchedHeader.set(ph); 1670 } 1671 1672 /** 1673 * Clear the cached value when its integrity is suspect. 1674 */ 1675 private void invalidateNextBlockHeader() { 1676 prefetchedHeader.set(null); 1677 } 1678 1679 private int getNextBlockOnDiskSize(ByteBuff onDiskBlock, int onDiskSizeWithHeader) { 1680 return onDiskBlock.getIntAfterPosition(onDiskSizeWithHeader + BlockType.MAGIC_LENGTH) 1681 + hdrSize; 1682 } 1683 1684 private ByteBuff allocate(int size, boolean intoHeap) { 1685 return intoHeap ? HEAP.allocate(size) : allocator.allocate(size); 1686 } 1687 1688 /** 1689 * Reads a version 2 block. 1690 * @param offset the offset in the stream to read at. 1691 * @param onDiskSizeWithHeaderL the on-disk size of the block, including the header and 1692 * checksums if present or -1 if unknown (as a long). Can be -1 if 1693 * we are doing raw iteration of blocks as when loading up file 1694 * metadata; i.e. the first read of a new file. Usually non-null 1695 * gotten from the file index. 1696 * @param pread whether to use a positional read 1697 * @param verifyChecksum Whether to use HBase checksums. If HBase checksum is switched 1698 * off, then use HDFS checksum. Can also flip on/off reading same 1699 * file if we hit a troublesome patch in an hfile. 1700 * @param updateMetrics whether need to update the metrics. 1701 * @param intoHeap allocate the ByteBuff of block from heap or off-heap. 1702 * @return the HFileBlock or null if there is a HBase checksum mismatch 1703 */ 1704 protected HFileBlock readBlockDataInternal(FSDataInputStream is, long offset, 1705 long onDiskSizeWithHeaderL, boolean pread, boolean verifyChecksum, boolean updateMetrics, 1706 boolean intoHeap) throws IOException { 1707 final Span span = Span.current(); 1708 final AttributesBuilder attributesBuilder = Attributes.builder(); 1709 Optional.of(Context.current()).map(val -> val.get(CONTEXT_KEY)) 1710 .ifPresent(c -> c.accept(attributesBuilder)); 1711 if (offset < 0) { 1712 throw new IOException("Invalid offset=" + offset + " trying to read " + "block (onDiskSize=" 1713 + onDiskSizeWithHeaderL + ")"); 1714 } 1715 if (!checkCallerProvidedOnDiskSizeWithHeader(onDiskSizeWithHeaderL)) { 1716 LOG.trace("Caller provided invalid onDiskSizeWithHeaderL={}", onDiskSizeWithHeaderL); 1717 onDiskSizeWithHeaderL = -1; 1718 } 1719 int onDiskSizeWithHeader = (int) onDiskSizeWithHeaderL; 1720 1721 // Try to use the cached header. Will serve us in rare case where onDiskSizeWithHeaderL==-1 1722 // and will save us having to seek the stream backwards to reread the header we 1723 // read the last time through here. 1724 ByteBuff headerBuf = getCachedHeader(offset); 1725 LOG.trace( 1726 "Reading {} at offset={}, pread={}, verifyChecksum={}, cachedHeader={}, " 1727 + "onDiskSizeWithHeader={}", 1728 this.fileContext.getHFileName(), offset, pread, verifyChecksum, headerBuf, 1729 onDiskSizeWithHeader); 1730 // This is NOT same as verifyChecksum. This latter is whether to do hbase 1731 // checksums. Can change with circumstances. The below flag is whether the 1732 // file has support for checksums (version 2+). 1733 boolean checksumSupport = this.fileContext.isUseHBaseChecksum(); 1734 long startTime = EnvironmentEdgeManager.currentTime(); 1735 if (onDiskSizeWithHeader == -1) { 1736 // The caller does not know the block size. Need to get it from the header. If header was 1737 // not cached (see getCachedHeader above), need to seek to pull it in. This is costly 1738 // and should happen very rarely. Currently happens on open of a hfile reader where we 1739 // read the trailer blocks to pull in the indices. Otherwise, we are reading block sizes 1740 // out of the hfile index. To check, enable TRACE in this file and you'll get an exception 1741 // in a LOG every time we seek. See HBASE-17072 for more detail. 1742 if (headerBuf == null) { 1743 if (LOG.isTraceEnabled()) { 1744 LOG.trace("Extra seek to get block size!", new RuntimeException()); 1745 } 1746 span.addEvent("Extra seek to get block size!", attributesBuilder.build()); 1747 headerBuf = HEAP.allocate(hdrSize); 1748 readAtOffset(is, headerBuf, hdrSize, false, offset, pread); 1749 headerBuf.rewind(); 1750 } 1751 onDiskSizeWithHeader = getOnDiskSizeWithHeader(headerBuf, checksumSupport); 1752 } 1753 1754 // The common case is that onDiskSizeWithHeader was produced by a read without checksum 1755 // validation, so give it a sanity check before trying to use it. 1756 if (!checkOnDiskSizeWithHeader(onDiskSizeWithHeader)) { 1757 if (verifyChecksum) { 1758 invalidateNextBlockHeader(); 1759 span.addEvent("Falling back to HDFS checksumming.", attributesBuilder.build()); 1760 return null; 1761 } else { 1762 throw new IOException("Invalid onDiskSizeWithHeader=" + onDiskSizeWithHeader); 1763 } 1764 } 1765 1766 int preReadHeaderSize = headerBuf == null ? 0 : hdrSize; 1767 // Allocate enough space to fit the next block's header too; saves a seek next time through. 1768 // onDiskBlock is whole block + header + checksums then extra hdrSize to read next header; 1769 // onDiskSizeWithHeader is header, body, and any checksums if present. preReadHeaderSize 1770 // says where to start reading. If we have the header cached, then we don't need to read 1771 // it again and we can likely read from last place we left off w/o need to backup and reread 1772 // the header we read last time through here. 1773 ByteBuff onDiskBlock = this.allocate(onDiskSizeWithHeader + hdrSize, intoHeap); 1774 boolean initHFileBlockSuccess = false; 1775 try { 1776 if (headerBuf != null) { 1777 onDiskBlock.put(0, headerBuf, 0, hdrSize).position(hdrSize); 1778 } 1779 boolean readNextHeader = readAtOffset(is, onDiskBlock, 1780 onDiskSizeWithHeader - preReadHeaderSize, true, offset + preReadHeaderSize, pread); 1781 onDiskBlock.rewind(); // in case of moving position when copying a cached header 1782 1783 // the call to validateChecksum for this block excludes the next block header over-read, so 1784 // no reason to delay extracting this value. 1785 int nextBlockOnDiskSize = -1; 1786 if (readNextHeader) { 1787 int parsedVal = getNextBlockOnDiskSize(onDiskBlock, onDiskSizeWithHeader); 1788 if (checkOnDiskSizeWithHeader(parsedVal)) { 1789 nextBlockOnDiskSize = parsedVal; 1790 } 1791 } 1792 if (headerBuf == null) { 1793 headerBuf = onDiskBlock.duplicate().position(0).limit(hdrSize); 1794 } 1795 1796 ByteBuff curBlock = onDiskBlock.duplicate().position(0).limit(onDiskSizeWithHeader); 1797 // Verify checksum of the data before using it for building HFileBlock. 1798 if (verifyChecksum && !validateChecksum(offset, curBlock, hdrSize)) { 1799 invalidateNextBlockHeader(); 1800 span.addEvent("Falling back to HDFS checksumming.", attributesBuilder.build()); 1801 return null; 1802 } 1803 1804 // TODO: is this check necessary or can we proceed with a provided value regardless of 1805 // what is in the header? 1806 int fromHeader = getOnDiskSizeWithHeader(headerBuf, checksumSupport); 1807 if (onDiskSizeWithHeader != fromHeader) { 1808 if (LOG.isTraceEnabled()) { 1809 LOG.trace("Passed in onDiskSizeWithHeader={} != {}, offset={}, fileContext={}", 1810 onDiskSizeWithHeader, fromHeader, offset, this.fileContext); 1811 } 1812 if (checksumSupport && verifyChecksum) { 1813 // This file supports HBase checksums and verification of those checksums was 1814 // requested. The block size provided by the caller (presumably from the block index) 1815 // does not match the block size written to the block header. treat this as 1816 // HBase-checksum failure. 1817 span.addEvent("Falling back to HDFS checksumming.", attributesBuilder.build()); 1818 invalidateNextBlockHeader(); 1819 return null; 1820 } 1821 throw new IOException("Passed in onDiskSizeWithHeader=" + onDiskSizeWithHeader + " != " 1822 + fromHeader + ", offset=" + offset + ", fileContext=" + this.fileContext); 1823 } 1824 1825 // remove checksum from buffer now that it's verified 1826 int sizeWithoutChecksum = curBlock.getInt(Header.ON_DISK_DATA_SIZE_WITH_HEADER_INDEX); 1827 curBlock.limit(sizeWithoutChecksum); 1828 long duration = EnvironmentEdgeManager.currentTime() - startTime; 1829 boolean tooSlow = this.readWarnTime >= 0 && duration > this.readWarnTime; 1830 if (updateMetrics) { 1831 HFile.updateReadLatency(duration, pread, tooSlow); 1832 } 1833 // The onDiskBlock will become the headerAndDataBuffer for this block. 1834 // If nextBlockOnDiskSizeWithHeader is not zero, the onDiskBlock already 1835 // contains the header of next block, so no need to set next block's header in it. 1836 HFileBlock hFileBlock = createFromBuff(curBlock, checksumSupport, offset, 1837 nextBlockOnDiskSize, fileContext, intoHeap ? HEAP : allocator); 1838 // Run check on uncompressed sizings. 1839 if (!fileContext.isCompressedOrEncrypted()) { 1840 hFileBlock.sanityCheckUncompressed(); 1841 } 1842 LOG.trace("Read {} in {} ms", hFileBlock, duration); 1843 if (!LOG.isTraceEnabled() && tooSlow) { 1844 LOG.warn("Read Block Slow: read {} cost {} ms, threshold = {} ms", hFileBlock, duration, 1845 this.readWarnTime); 1846 } 1847 span.addEvent("Read block", attributesBuilder.build()); 1848 // Cache next block header if we read it for the next time through here. 1849 if (nextBlockOnDiskSize != -1) { 1850 cacheNextBlockHeader(offset + hFileBlock.getOnDiskSizeWithHeader(), onDiskBlock, 1851 onDiskSizeWithHeader, hdrSize); 1852 } 1853 initHFileBlockSuccess = true; 1854 return hFileBlock; 1855 } finally { 1856 if (!initHFileBlockSuccess) { 1857 onDiskBlock.release(); 1858 } 1859 } 1860 } 1861 1862 @Override 1863 public void setIncludesMemStoreTS(boolean includesMemstoreTS) { 1864 this.fileContext = 1865 new HFileContextBuilder(this.fileContext).withIncludesMvcc(includesMemstoreTS).build(); 1866 } 1867 1868 @Override 1869 public void setDataBlockEncoder(HFileDataBlockEncoder encoder, Configuration conf) { 1870 encodedBlockDecodingCtx = encoder.newDataBlockDecodingContext(conf, fileContext); 1871 } 1872 1873 @Override 1874 public HFileBlockDecodingContext getBlockDecodingContext() { 1875 return this.encodedBlockDecodingCtx; 1876 } 1877 1878 @Override 1879 public HFileBlockDecodingContext getDefaultBlockDecodingContext() { 1880 return this.defaultDecodingCtx; 1881 } 1882 1883 /** 1884 * Generates the checksum for the header as well as the data and then validates it. If the block 1885 * doesn't uses checksum, returns false. 1886 * @return True if checksum matches, else false. 1887 */ 1888 private boolean validateChecksum(long offset, ByteBuff data, int hdrSize) { 1889 // If this is an older version of the block that does not have checksums, then return false 1890 // indicating that checksum verification did not succeed. Actually, this method should never 1891 // be called when the minorVersion is 0, thus this is a defensive check for a cannot-happen 1892 // case. Since this is a cannot-happen case, it is better to return false to indicate a 1893 // checksum validation failure. 1894 if (!fileContext.isUseHBaseChecksum()) { 1895 return false; 1896 } 1897 return ChecksumUtil.validateChecksum(data, pathName, offset, hdrSize); 1898 } 1899 1900 @Override 1901 public void closeStreams() throws IOException { 1902 streamWrapper.close(); 1903 } 1904 1905 @Override 1906 public void unbufferStream() { 1907 // To handle concurrent reads, ensure that no other client is accessing the streams while we 1908 // unbuffer it. 1909 if (streamLock.tryLock()) { 1910 try { 1911 this.streamWrapper.unbuffer(); 1912 } finally { 1913 streamLock.unlock(); 1914 } 1915 } 1916 } 1917 1918 @Override 1919 public String toString() { 1920 return "hfs=" + hfs + ", path=" + pathName + ", fileContext=" + fileContext; 1921 } 1922 } 1923 1924 /** An additional sanity-check in case no compression or encryption is being used. */ 1925 void sanityCheckUncompressed() throws IOException { 1926 if (onDiskSizeWithoutHeader != uncompressedSizeWithoutHeader + totalChecksumBytes()) { 1927 throw new IOException("Using no compression but " + "onDiskSizeWithoutHeader=" 1928 + onDiskSizeWithoutHeader + ", " + "uncompressedSizeWithoutHeader=" 1929 + uncompressedSizeWithoutHeader + ", numChecksumbytes=" + totalChecksumBytes()); 1930 } 1931 } 1932 1933 // Cacheable implementation 1934 @Override 1935 public int getSerializedLength() { 1936 if (bufWithoutChecksum != null) { 1937 // Include extra bytes for block metadata. 1938 return this.bufWithoutChecksum.limit() + BLOCK_METADATA_SPACE; 1939 } 1940 return 0; 1941 } 1942 1943 // Cacheable implementation 1944 @Override 1945 public void serialize(ByteBuffer destination, boolean includeNextBlockMetadata) { 1946 this.bufWithoutChecksum.get(destination, 0, getSerializedLength() - BLOCK_METADATA_SPACE); 1947 destination = addMetaData(destination, includeNextBlockMetadata); 1948 1949 // Make it ready for reading. flip sets position to zero and limit to current position which 1950 // is what we want if we do not want to serialize the block plus checksums if present plus 1951 // metadata. 1952 destination.flip(); 1953 } 1954 1955 /** 1956 * For use by bucketcache. This exposes internals. 1957 */ 1958 public ByteBuffer getMetaData(ByteBuffer bb) { 1959 bb = addMetaData(bb, true); 1960 bb.flip(); 1961 return bb; 1962 } 1963 1964 /** 1965 * Adds metadata at current position (position is moved forward). Does not flip or reset. 1966 * @return The passed <code>destination</code> with metadata added. 1967 */ 1968 private ByteBuffer addMetaData(final ByteBuffer destination, boolean includeNextBlockMetadata) { 1969 destination.put(this.fileContext.isUseHBaseChecksum() ? (byte) 1 : (byte) 0); 1970 destination.putLong(this.offset); 1971 if (includeNextBlockMetadata) { 1972 destination.putInt(this.nextBlockOnDiskSize); 1973 } 1974 return destination; 1975 } 1976 1977 // Cacheable implementation 1978 @Override 1979 public CacheableDeserializer<Cacheable> getDeserializer() { 1980 return HFileBlock.BLOCK_DESERIALIZER; 1981 } 1982 1983 @Override 1984 public int hashCode() { 1985 int result = 1; 1986 result = result * 31 + blockType.hashCode(); 1987 result = result * 31 + nextBlockOnDiskSize; 1988 result = result * 31 + (int) (offset ^ (offset >>> 32)); 1989 result = result * 31 + onDiskSizeWithoutHeader; 1990 result = result * 31 + (int) (prevBlockOffset ^ (prevBlockOffset >>> 32)); 1991 result = result * 31 + uncompressedSizeWithoutHeader; 1992 result = result * 31 + bufWithoutChecksum.hashCode(); 1993 return result; 1994 } 1995 1996 @Override 1997 public boolean equals(Object comparison) { 1998 if (this == comparison) { 1999 return true; 2000 } 2001 if (comparison == null) { 2002 return false; 2003 } 2004 if (!(comparison instanceof HFileBlock)) { 2005 return false; 2006 } 2007 2008 HFileBlock castedComparison = (HFileBlock) comparison; 2009 2010 if (castedComparison.blockType != this.blockType) { 2011 return false; 2012 } 2013 if (castedComparison.nextBlockOnDiskSize != this.nextBlockOnDiskSize) { 2014 return false; 2015 } 2016 // Offset is important. Needed when we have to remake cachekey when block is returned to cache. 2017 if (castedComparison.offset != this.offset) { 2018 return false; 2019 } 2020 if (castedComparison.onDiskSizeWithoutHeader != this.onDiskSizeWithoutHeader) { 2021 return false; 2022 } 2023 if (castedComparison.prevBlockOffset != this.prevBlockOffset) { 2024 return false; 2025 } 2026 if (castedComparison.uncompressedSizeWithoutHeader != this.uncompressedSizeWithoutHeader) { 2027 return false; 2028 } 2029 if ( 2030 ByteBuff.compareTo(this.bufWithoutChecksum, 0, this.bufWithoutChecksum.limit(), 2031 castedComparison.bufWithoutChecksum, 0, castedComparison.bufWithoutChecksum.limit()) != 0 2032 ) { 2033 return false; 2034 } 2035 return true; 2036 } 2037 2038 DataBlockEncoding getDataBlockEncoding() { 2039 if (blockType == BlockType.ENCODED_DATA) { 2040 return DataBlockEncoding.getEncodingById(getDataBlockEncodingId()); 2041 } 2042 return DataBlockEncoding.NONE; 2043 } 2044 2045 byte getChecksumType() { 2046 return this.fileContext.getChecksumType().getCode(); 2047 } 2048 2049 int getBytesPerChecksum() { 2050 return this.fileContext.getBytesPerChecksum(); 2051 } 2052 2053 /** Returns the size of data on disk + header. Excludes checksum. */ 2054 int getOnDiskDataSizeWithHeader() { 2055 return this.onDiskDataSizeWithHeader; 2056 } 2057 2058 /** 2059 * Return the number of bytes required to store all the checksums for this block. Each checksum 2060 * value is a 4 byte integer. <br/> 2061 * NOTE: ByteBuff returned by {@link HFileBlock#getBufferWithoutHeader()} and 2062 * {@link HFileBlock#getBufferReadOnly} or DataInputStream returned by 2063 * {@link HFileBlock#getByteStream()} does not include checksum. 2064 */ 2065 int totalChecksumBytes() { 2066 return totalChecksumBytes; 2067 } 2068 2069 private int computeTotalChecksumBytes() { 2070 // If the hfile block has minorVersion 0, then there are no checksum 2071 // data to validate. Similarly, a zero value in this.bytesPerChecksum 2072 // indicates that cached blocks do not have checksum data because 2073 // checksums were already validated when the block was read from disk. 2074 if (!fileContext.isUseHBaseChecksum() || this.fileContext.getBytesPerChecksum() == 0) { 2075 return 0; 2076 } 2077 return (int) ChecksumUtil.numBytes(onDiskDataSizeWithHeader, 2078 this.fileContext.getBytesPerChecksum()); 2079 } 2080 2081 /** 2082 * Returns the size of this block header. 2083 */ 2084 public int headerSize() { 2085 return headerSize(this.fileContext.isUseHBaseChecksum()); 2086 } 2087 2088 /** 2089 * Maps a minor version to the size of the header. 2090 */ 2091 public static int headerSize(boolean usesHBaseChecksum) { 2092 return usesHBaseChecksum 2093 ? HConstants.HFILEBLOCK_HEADER_SIZE 2094 : HConstants.HFILEBLOCK_HEADER_SIZE_NO_CHECKSUM; 2095 } 2096 2097 /** 2098 * Return the appropriate DUMMY_HEADER for the minor version 2099 */ 2100 // TODO: Why is this in here? 2101 byte[] getDummyHeaderForVersion() { 2102 return getDummyHeaderForVersion(this.fileContext.isUseHBaseChecksum()); 2103 } 2104 2105 /** 2106 * Return the appropriate DUMMY_HEADER for the minor version 2107 */ 2108 static private byte[] getDummyHeaderForVersion(boolean usesHBaseChecksum) { 2109 return usesHBaseChecksum ? HConstants.HFILEBLOCK_DUMMY_HEADER : DUMMY_HEADER_NO_CHECKSUM; 2110 } 2111 2112 /** 2113 * @return This HFileBlocks fileContext which will a derivative of the fileContext for the file 2114 * from which this block's data was originally read. 2115 */ 2116 public HFileContext getHFileContext() { 2117 return this.fileContext; 2118 } 2119 2120 /** 2121 * Convert the contents of the block header into a human readable string. This is mostly helpful 2122 * for debugging. This assumes that the block has minor version > 0. 2123 */ 2124 static String toStringHeader(ByteBuff buf) throws IOException { 2125 byte[] magicBuf = new byte[Math.min(buf.limit() - buf.position(), BlockType.MAGIC_LENGTH)]; 2126 buf.get(magicBuf); 2127 BlockType bt = BlockType.parse(magicBuf, 0, BlockType.MAGIC_LENGTH); 2128 int compressedBlockSizeNoHeader = buf.getInt(); 2129 int uncompressedBlockSizeNoHeader = buf.getInt(); 2130 long prevBlockOffset = buf.getLong(); 2131 byte cksumtype = buf.get(); 2132 long bytesPerChecksum = buf.getInt(); 2133 long onDiskDataSizeWithHeader = buf.getInt(); 2134 return " Header dump: magic: " + Bytes.toString(magicBuf) + " blockType " + bt 2135 + " compressedBlockSizeNoHeader " + compressedBlockSizeNoHeader 2136 + " uncompressedBlockSizeNoHeader " + uncompressedBlockSizeNoHeader + " prevBlockOffset " 2137 + prevBlockOffset + " checksumType " + ChecksumType.codeToType(cksumtype) 2138 + " bytesPerChecksum " + bytesPerChecksum + " onDiskDataSizeWithHeader " 2139 + onDiskDataSizeWithHeader; 2140 } 2141 2142 /** 2143 * Creates a new HFileBlockBuilder from the existing block and a new ByteBuff. The builder will be 2144 * loaded with all of the original fields from blk, except now using the newBuff and setting 2145 * isSharedMem based on the source of the passed in newBuff. An existing HFileBlock may have been 2146 * an {@link ExclusiveMemHFileBlock}, but the new buffer might call for a 2147 * {@link SharedMemHFileBlock}. Or vice versa. 2148 * @param blk the block to clone from 2149 * @param newBuff the new buffer to use 2150 */ 2151 private static HFileBlockBuilder createBuilder(HFileBlock blk, ByteBuff newBuff) { 2152 return new HFileBlockBuilder().withBlockType(blk.blockType) 2153 .withOnDiskSizeWithoutHeader(blk.onDiskSizeWithoutHeader) 2154 .withUncompressedSizeWithoutHeader(blk.uncompressedSizeWithoutHeader) 2155 .withPrevBlockOffset(blk.prevBlockOffset).withByteBuff(newBuff).withOffset(blk.offset) 2156 .withOnDiskDataSizeWithHeader(blk.onDiskDataSizeWithHeader) 2157 .withNextBlockOnDiskSize(blk.nextBlockOnDiskSize).withHFileContext(blk.fileContext) 2158 .withByteBuffAllocator(blk.allocator).withShared(!newBuff.hasArray()); 2159 } 2160 2161 private static HFileBlock shallowClone(HFileBlock blk, ByteBuff newBuf) { 2162 return createBuilder(blk, newBuf).build(); 2163 } 2164 2165 static HFileBlock deepCloneOnHeap(HFileBlock blk) { 2166 ByteBuff deepCloned = ByteBuff 2167 .wrap(ByteBuffer.wrap(blk.bufWithoutChecksum.toBytes(0, blk.bufWithoutChecksum.limit()))); 2168 return createBuilder(blk, deepCloned).build(); 2169 } 2170}