HBASE-4278 Race condition in Slab.java that occurs due to spinlock unlocking

early (Li Pi)


git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1162886 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Zhihong Yu 2011-08-29 16:00:47 +00:00
parent 5626936a70
commit 53e7f73ab9
2 changed files with 7 additions and 2 deletions

View File

@ -220,6 +220,8 @@ Release 0.91.0 - Unreleased
HBASE-3229 HBASE-3229 Table creation, though using "async" call to master,
can actually run for a while and cause RPC timeout (Ming Ma)
HBASE-4252 TestLogRolling's low-probability failure (Jieshan Bean)
HBASE-4278 Race condition in Slab.java that occurs due to spinlock unlocking
early (Li Pi)
IMPROVEMENTS
HBASE-3290 Max Compaction Size (Nicolas Spiegelberg via Stack)

View File

@ -113,8 +113,11 @@ class Slab implements org.apache.hadoop.hbase.io.HeapSize {
*/
ByteBuffer alloc(int bufferSize) {
int newCapacity = Preconditions.checkPositionIndex(bufferSize, blockSize);
while (buffers.isEmpty()); // Spinlock
ByteBuffer returnedBuffer = buffers.remove();
ByteBuffer returnedBuffer = buffers.poll();
while(returnedBuffer == null){
returnedBuffer = buffers.poll();
}
returnedBuffer.clear().limit(newCapacity);
return returnedBuffer;
}