[
https://issues.apache.org/jira/browse/HDFS-12245?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16808866#comment-16808866
]
Adam Antal commented on HDFS-12245:
-----------------------------------
[~jojochuang], on a second thought maybe this change was not good, since 2 <<
14 - 1 is interpreted as 2 << (14 - 1) = 2 << 13 = (2 ^ 1) ^ 13 = 2 ^ 14 =
16384, as this was in my patch v1. You can verify the bitshifting in this
online compiler: http://tpcg.io/H8a4bn. So the patch v2 is invalid.
Here is my patch v3 with your suggestion, but with the right numbers.
> Update INodeId javadoc
> ----------------------
>
> Key: HDFS-12245
> URL: https://issues.apache.org/jira/browse/HDFS-12245
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: namenode
> Reporter: Wei-Chiu Chuang
> Assignee: Adam Antal
> Priority: Major
> Attachments: HDFS-12245.001.patch, HDFS-12245.002.patch
>
>
> The INodeId javadoc states that id 1 to 1000 is reserved and root inode id
> start from 1001. That is no longer true after HDFS-4434.
> Also, it's a little weird in INodeId
> {code}
> public static final long LAST_RESERVED_ID = 2 << 14 - 1;
> public static final long ROOT_INODE_ID = LAST_RESERVED_ID + 1;
> {code}
> It seems the intent was for LAST_RESERVED_ID to be (2^14) - 1 = 32767. But
> due to Java operator precedence, LAST_RESERVED_ID = 2^(14-1) = 16384. Maybe
> it doesn't matter, not sure.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
