Re: [LTP] [PATCH] fctnl23: can't set a read lease on a r/w open file

[Robert W. <ro...@us...>]

Roman,

  Below is some info from the test author regarding fcntl23.  Apparently,
you may be encountering a bug depending on your kernel level.

-Robbie
<ro...@us...>

----- Forwarded by Robert Williamson/Austin/IBM on 05/26/2005 02:29 PM
-----

Jacky Malcles <Jac...@bu...> wrote on 05/20/2005 01:54:18 AM:

> Hi Prashant,
>
> wanna let you know this:
>
> 1)
> Note: forwarded message folling
>
> from Ricky Ng-Adam
> Hello!
>
> This was the answer from Linus Torvalds. Really, there seems to have
> been a bug but I never had time to follow up on it. Have you tried with
> the latest kernels? I thought there had been changes around those
> pieces the code in more recent kernels.
>
> Good luck,
>
> Note: forwarded message attached.
>
>
> __________________________________________________
> Subject:
> Re: Fw: [2.6.10-rc3:fs/locks.c] fcntl F_SETLEASE/[ F_WRLCK| F_RDLCK]
> inconsistencies
> From:
> Linus Torvalds <tor...@os...>
> Date:
> Tue, 7 Dec 2004 18:52:42 -0800 (PST)
> To:
> Andrew Morton <ak...@os...>
> CC:
> an...@th..., Ricky Ng-Adam <rn...@ya...>, Matthew
Wilcox
> <wi...@de...>
>
> On Tue, 7 Dec 2004, Andrew Morton wrote:
>
>  >> Thoughts?
>  >>
>
> Well, The "0" can definitely not be just changed to a "1" for the
> i_writecount check, but maybe something like
>
>          int maxwrites = (file->f_mode & FMODE_WRITE) ? 1 : 0;
>
>          if (arg == F_RDLCK && atomic_read(&inode->i_writecount) >
maxwrites)
>                  ...
>
> might work.
>
> Who uses F_SETLEASE? Just the samba/NFSv4 servers? It really should be up
> to those, unless there are other real users..
>
>                  Linus
> __________________________________________________
>
> ========================================================================
>
> 2)
>
> I did a try on a 2.6.12-rc4 kernel and noted:
> open()  |  lease requested
>    flag   | F_RDLCK  | F_WRLCK
> ---------+----------+----------
> O_RDONLY | okay     |  okay
> O_WRONLY | EAGAIN   |  okay
> O_RDWR   | EAGAIN   |  okay
>
> seems that the implementation does not include the related patch...
>
> 3)
>
> there are reasons, for details:
> see on kernel.org the discussioon
> as example from J.Bruce Fielsd <bf...@fi...>
> "
> On Tue, May 03, 2005 at 12:21:24PM -0400, William A.(Andy) Adamson wrote:
>
>  >> the other side of the coin would be break_lease.
>
>
> Yeah, I'm a little confused as to why anyone would have the expectation
> that read leases would not conflict with write opens by the same
> process, given that break_lease() has never functioned that way, so
> later write opens by the same process have always broken any read lease.
>
> Are there applications that actually depend on the old behaviour?  Is
> there any documentation that blesses it?  All I can find is the fcntl
> man page, and as far as I can tell an implementation that makes read
> leases conflict with all write opens (by the same process or not) is
> consistent with that man page.
>
> --b."
>
> ========================================================================
>
> so what of having 3 Testcases?
>
> fcntl23 - O_RDWR
> fcntl27 -> O_RDONLY
> fcntl28 -> O_WRONLY
>
> regards.
> Jacky
>
>
>
>
>
> pra...@wi... wrote:
> >
> > In addition to this I am testing this on 2.6.8 kernel (NFS mounted
one),
> > where I think the concept of F_FDLOCK has been changed a bit.
> >
> > My RHEL 3  and fedora are 2.4.X kernels
> >
> > Regards,
> > Prashant P Yendigeri,
> > WindRiver ODC,
> > Tower 7 'c' wing,Electronic city,
> > Bangalore
> > Ext : 6732
> >
> > -----Original Message-----
> > From: Prashant Yendigeri (WT01 - EMBEDDED & PRODUCT ENGINEERING
> > SOLUTIONS)
> > Sent: Friday, May 20, 2005 10:18 AM
> > To: 'Jacky Malcles'
> > Subject: RE: fcntl23 TC
> >
> > Hello Jacky,
> > I am testing this on RHEL 3 and Fedora.Well on these fcntl23 passes
> > without a hitch.But I have built my own kernel and rootfs and running
on
> > intel (P4) board, mounted via NFS. Here it is failing.Do you see any
NFS
> > issues here ?
> > or
> > we can leave test code as it is.
> >
> > Thanks.
> >
> > Regards,
> > Prashant P Yendigeri,
> > WindRiver ODC,
> > Tower 7 'c' wing,Electronic city,
> > Bangalore
> > Ext : 6732
> >
> > -----Original Message-----
> > From: Jacky Malcles [mailto:Jac...@bu...]
> > Sent: Thursday, May 19, 2005 8:49 PM
> > To: Prashant Yendigeri (WT01 - EMBEDDED & PRODUCT ENGINEERING
SOLUTIONS)
> > Cc: ro...@us...
> > Subject: fcntl23 TC
> >
> > Prashant,
> >
> > If you agree,
> > I suggest that fcntl23 remains opening the file with O_RDWR flag
> >
> > and instead of checking if it's getting a F_RDLCK..
> >                 checking of the errno that is returned..
> >
> >
> > regards,
> >
> > --
> >   Jacky Malcles            B1-403   Email : Jac...@bu...
> >   Bull SA, 1 rue de Provence, B.P 208, 38432 Echirolles CEDEX, FRANCE
> >   Tel : 04.76.29.73.14
> >
> >
> >
> > Confidentiality Notice
> >
> > The information contained in this electronic message and any
> attachments to this message are intended
> > for the exclusive use of the addressee(s) and may contain
> confidential or privileged information. If
> > you are not the intended recipient, please notify the sender at
> Wipro or Mai...@wi... immediately
> > and destroy all copies of this message and any attachments.
> >
>
>
> --
>   Jacky Malcles            B1-403   Email : Jac...@bu...
>   Bull SA, 1 rue de Provence, B.P 208, 38432 Echirolles CEDEX, FRANCE
>   Tel : 04.76.29.73.14