[HDFS-10656] Optimize conversion of byte arrays back to path string

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.8.0, 2.7.4, 3.0.0-alpha1
• Component/s: hdfs
• Target Version/s: 2.9.0
• Hadoop Flags: Reviewed
• Assignee: Daryn Sharp
• Reporter: Daryn Sharp
• Created: 19/Jul/16 22:56
• Updated: 06/Jan/17 08:06
• Resolved: 03/Aug/16 16:55

Description

DFSUtil.byteArray2PathString generates excessive object allocation.

1. each byte array is encoded to a string (copy)
2. string appended to a builder which extracts the chars from the intermediate string (copy) and adds to its own char array
3. builder's char array is re-alloced if over 16 chars (copy)
4. builder's toString creates another string (copy)

Instead of allocating all these objects and performing multiple byte/char encoding/decoding conversions, the byte array can be built in-place with a single final conversion to a string.

Issue Links

• requires: HDFS-10655 Fix path related byte array conversion bugs