// But in gcd(a - b, b), a - b is even and b is odd, so we can divide out powers of two.

      // We bend over backwards to avoid branching, adapting a technique from
      // http://graphics.stanford.edu/~seander/bithacks.html#IntegerMinOrMax

      int delta = a - b; // can't overflow, since a and b are nonnegative

      // But in gcd(a - b, b), a - b is even and b is odd, so we can divide out powers of two.

      // We bend over backwards to avoid branching, adapting a technique from
      // http://graphics.stanford.edu/~seander/bithacks.html#IntegerMinOrMax

      long delta = a - b; // can't overflow, since a and b are nonnegative

 * </ul>
 *
 * <p>This class uses (possibly repeated) invocations of {@link java.lang.System#gc()} to cause
 * finalization to happen. However, a call to {@code System.gc()} is specified to be no more than a
 * hint, so this technique may fail at the whim of the JDK implementation, for example if a user
 * specified the JVM flag {@code -XX:+DisableExplicitGC}. But in practice, it works very well for
 * ordinary tests.
 *

  /*
   * Leaving K and V as upper bounds rather than the actual key and value types allows type
   * parameters to be left implicit more often. CacheBuilder uses the same technique.
   */

  private MultimapBuilder() {}

  private static final int DEFAULT_EXPECTED_KEYS = 8;

  /** Uses a hash table to map keys to value collections. */

 * </ul>
 *
 * <p>This class uses (possibly repeated) invocations of {@link java.lang.System#gc()} to cause
 * finalization to happen. However, a call to {@code System.gc()} is specified to be no more than a
 * hint, so this technique may fail at the whim of the JDK implementation, for example if a user
 * specified the JVM flag {@code -XX:+DisableExplicitGC}. But in practice, it works very well for
 * ordinary tests.
 *

 * to that of {@code ConcurrentHashMap} in a reusable form, and extends it for semaphores and
 * read-write locks. Conceptually, lock striping is the technique of dividing a lock into many
 * <i>stripes</i>, increasing the granularity of a single lock and allowing independent operations
 * to lock different stripes and proceed concurrently, instead of creating contention for a single

         * the redirected target has different capabilities. But
         * the way we have been doing that is to call exists() which
         * calls this method so another technique will be necessary
         * to support DFS referral _to_ Win95/98/ME.
         */

        if( tree.session.transport.hasCapability( ServerMessageBlock.CAP_NT_SMBS )) {

            /*

         * the redirected target has different capabilities. But
         * the way we have been doing that is to call exists() which
         * calls this method so another technique will be necessary
         * to support DFS referral _to_ Win95/98/ME.
         */

        if ( th.isSMB2() ) {
            // just open and close. withOpen will store the attributes