[jira] [Comment Edited] (HIVE-19393) NonSyncDataInputBuffer.skipBytes hangs when the file is corrupted

["John Doe (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-19393?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16462983#comment-16462983 ] 

John Doe edited comment on HIVE-19393 at 5/3/18 8:30 PM:
---------------------------------------------------------

[~sershe] Sorry about the confusion. Actually, when the InputStream is corrupted, skip function returns -1 instead of 0, causing the infinite loop.


was (Author: dustinday):
[~sershe] Yes, you are right. Sorry about it.

> NonSyncDataInputBuffer.skipBytes hangs when the file is corrupted 
> ------------------------------------------------------------------
>
>                 Key: HIVE-19393
>                 URL: https://issues.apache.org/jira/browse/HIVE-19393
>             Project: Hive
>          Issue Type: Bug
>          Components: SQL
>    Affects Versions: 1.0.0
>            Reporter: John Doe
>            Priority: Minor
>
> When an InputStream is corrupted, the InputStream.skip can return -1, causing the while loop in NonSyncDataInputBuffer.skipBytes become infinite.
> {code:java}
>   public final int skipBytes(int count) throws IOException {
>     int skipped = 0;
>     long skip;
>     while (skipped < count && (skip = in.skip(count - skipped)) != 0) {
>       skipped += skip;
>     }
>     if (skipped < 0) {
>       throw new EOFException();
>     }
>     return skipped;
>   }
> {code}
> Similar bugs are Hadoop-8614, Yarn-2905, Yarn-163, Mapreduce-6990



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)