Proposal for adding O_DIRECT support into JDK 9

Lu, Yingqi yingqi.lu at intel.com
Wed Sep 13 16:52:39 UTC 2017


Hi Brian,

Thank you for your feedback! Please find the most recent version of the patch is available at http://cr.openjdk.java.net/~kkharbas/8164900/webrev.11/

In this version, I fixed the typos in the FileStore:getBlockSize. I also removed alignment checks with checkBufferPositionAligned and checkRemainingBufferSizeAligned from FileChannelImpl (only have them in IOUtil), but still keep checkChannelPositionAligned there. This way, I think the duplicated checks are removed. Please let me know if I miss anything.

Thanks,
Lucy

From: Brian Burkhalter [mailto:brian.burkhalter at oracle.com]
Sent: Tuesday, September 12, 2017 6:04 PM
To: Lu, Yingqi <yingqi.lu at intel.com>
Cc: nio-dev at openjdk.java.net
Subject: Re: Proposal for adding O_DIRECT support into JDK 9

Hi Lucy,

On Sep 12, 2017, at 3:58 PM, Lu, Yingqi <yingqi.lu at intel.com<mailto:yingqi.lu at intel.com>> wrote:


http://cr.openjdk.java.net/~kkharbas/8164900/webrev.10/

In this version, two issues have been addressed:

1.     Add description on the notion "block" in FileStore:getBlockSize

FileStore.java
116: "isthe smallest stoage" -> "is the smallest storage"
128: "if operation" -> "if the operation"


2.     Move alignment check from IOUtil to Util

There is one remaining issue: we check alignment on buffer size and channel position inside both FileChannelImpl and IOUtil. I looked at the code and think we might be able to remove the checks inside readFromNativeBuffer

readIntoNativeBuffer()


and writeFromNativeBuffer.

It does look as if either the FileChannelImpl or the IOUtil checks could be removed.

Thanks,

Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/nio-dev/attachments/20170913/39f58052/attachment.html>


More information about the nio-dev mailing list