RFR : 8164900 was (Re: Proposal for adding O_DIRECT support into JDK 9)

[Seán Coffey]

One comment from me. I don't want to block any integration and it could 
be looked at later if necessary.

Is the new exception handling a bit sparse in places with this 
enhancement ? Take FileChannelImpl.java for example. There are 12 new 
sites where IOException can be thrown. While calculations are made to 
determine if the Exception is thrown or not, none of the real detail 
gets transmitted to the end user. If exceptions were to be seen in these 
areas then I think the details on channel position and bytes remained 
etc.. are useful for end user debugging.

regards,
Sean.


On 30/08/2017 19:44, Lu, Yingqi wrote:
>
> Oh, great! Will do!
>
> *From:*Brian Burkhalter [mailto:brian.burkhalter at oracle.com]
> *Sent:* Wednesday, August 30, 2017 11:42 AM
> *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 Aug 30, 2017, at 11:37 AM, Lu, Yingqi <yingqi.lu at intel.com 
> <mailto:yingqi.lu at intel.com>> wrote:
>
>
>
>     I will check.
>
>     I was also wondering whether there is an equivalent to mincore()
>     or something which could enable DirectIOTest on Windows.
>
> Thanks!
>
>
>
>     Very good point! I only tested on UFS. If we decide not to disable
>     the test, what is the best method to check? fstype? I am sorry
>     that I do not have much experiences with Solaris OS.
>
>     I wonder whether we need to disable this test on Solaris or add a
>     check of the file system type to determine whether it is either
>     NFS or UFS?
>
> You might be able to check using FileStore#type() [1].
>
> Thanks,
>
> Brian
>
> [1] 
> http://download.java.net/java/jdk9/docs/api/java/nio/file/FileStore.html#type--
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/nio-dev/attachments/20170830/e642ddc2/attachment.html>