Re: [rfc patch 0/4] splice: cleanups and fixes

[Subrata Modak <subrata@linux.vnet.ibm.com>]

Hi Miklos ,

On Sun, 2008-06-22 at 08:16 +0200, Miklos Szeredi wrote:
> > In the backdrop of the changes in splice(), would you also like to help
> > us in reviewing the existing testcase for splice() in LTP:
> > 
> > http://ltp.cvs.sourceforge.net/ltp/ltp/testcases/kernel/syscalls/splice/,
> 
> Well, there's probably lot more different tests that could be done:
> 
>  - splicing from various sources (socket, pipe, regular file)
>  - splicing to various sources (...)
>  - splicing in full page(s)
>  - splicing in partial page(s)
>  - filling the destination pipe (blocking/nonblocking)
>  - splicing in a big buffer (bigger than the pipe buffer)
>  - splicing out full buffers
>  - splicing out partial buffers
>  - splicing out multiple buffers
>  - splicing out more data then is in the pipe (blocking/nonblocking)
>  - parallel splicing (in and/or out)
> 

Thanks for pointing out the areas that can be tested in addition to what
is existing. The list is very comprehensive. I will see if i can get
somebody to work on this. Will put you in CC for review. Hope you do not
mind that.

> And then there's vmsplice(), which I haven't even looked at.

Yes, please do look into this and give us more feedback like the splice
above. You can find it here:

http://ltp.cvs.sourceforge.net/ltp/ltp/testcases/kernel/syscalls/vmsplice/.

> 
> > It would also be great if you can send us some patch as well in the
> > regard.
> 
> Unfortunately this would be a much bigger project than the cleanup I
> did on the kernel splice code, and I don't really have the time to do
> it :(

We appreciate the already feedback you have given us. Hope you will also
help in future review of the testcases.

Regards--
Subrata

> 
> Miklos