[PATCH 00/17] RFC: proposed changes to the NFS mount command

[PATCH 00/17] RFC: proposed changes to the NFS mount command

[Chuck Lever]

Hi Neil-

I've got a series of patches to the NFS mount command that prepare it
for using a C string to pass mount options to the kernel, rather than
a binary nfs_mount_data structure.

The goal of these clean-ups is to allow the creation of a ./configure
option that will disable building nfsmount.c and nfs4mount.c, and
enable building a version of these files that will introduce support
for passing mount options as a string (that work is yet to be done).

They certainly need some testing yet, but can you take a look at these
and tell me if they are generally acceptable or if I need to take a
different direction?  Hopefully they are straightforward and won't take
much of your time.

Thanks in advance for your advice!

-- 
corporate:    <chuck dot lever at oracle dot com>

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: [PATCH 00/17] RFC: proposed changes to the NFS mount command

[Neil Brown]

On Sunday July 15, chuck.lever@oracle.com wrote:
> Hi Neil-
> 
> I've got a series of patches to the NFS mount command that prepare it
> for using a C string to pass mount options to the kernel, rather than
> a binary nfs_mount_data structure.

I've wondered why that was happening... Can't we just make a version
x+1 with a big string buffer so new options get put there is regular
string format and old options stay the way they are?

But I guess that - long term - it is better to be just like other
filesystems.


> 
> The goal of these clean-ups is to allow the creation of a ./configure
> option that will disable building nfsmount.c and nfs4mount.c, and
> enable building a version of these files that will introduce support
> for passing mount options as a string (that work is yet to be done).

Does this mean that you cannot compile nfs-utils to support both old
and new kernels?  Is that really a good plan?
What is the over-all migration strategy?


> 
> They certainly need some testing yet, but can you take a look at these
> and tell me if they are generally acceptable or if I need to take a
> different direction?  Hopefully they are straightforward and won't take
> much of your time.

Some of these are obviously good cleanups.  Would it inconvenience you
if I just picked off the ones I like for now and leave you to rebase?

I've replied to the ones I had specific concerns about.  Others are OK
providing it really does make sense to have an either-or
configuration.

NeilBrown

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: [PATCH 00/17] RFC: proposed changes to the NFS mount command

[Christoph Hellwig]

On Mon, Jul 16, 2007 at 02:05:34PM +1000, Neil Brown wrote:
> I've wondered why that was happening... Can't we just make a version
> x+1 with a big string buffer so new options get put there is regular
> string format and old options stay the way they are?
> 
> But I guess that - long term - it is better to be just like other
> filesystems.

Having anything but a plain string means the filesystem is a complete
oddball.  And with the planned removal of smbfs nfs would be the only
oddball left.

> > option that will disable building nfsmount.c and nfs4mount.c, and
> > enable building a version of these files that will introduce support
> > for passing mount options as a string (that work is yet to be done).
> 
> Does this mean that you cannot compile nfs-utils to support both old
> and new kernels?  Is that really a good plan?
> What is the over-all migration strategy?

Making the different argument passing variants exclusive sounds like
a bad idea indeed.


-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: [PATCH 00/17] RFC: proposed changes to the NFS mount command

[Chuck Lever]

[-- Attachment #1: Type: text/plain, Size: 1247 bytes --]



Christoph Hellwig wrote:
> On Mon, Jul 16, 2007 at 02:05:34PM +1000, Neil Brown wrote:
>> I've wondered why that was happening... Can't we just make a version
>> x+1 with a big string buffer so new options get put there is regular
>> string format and old options stay the way they are?
>>
>> But I guess that - long term - it is better to be just like other
>> filesystems.
> 
> Having anything but a plain string means the filesystem is a complete
> oddball.  And with the planned removal of smbfs nfs would be the only
> oddball left.
> 
>>> option that will disable building nfsmount.c and nfs4mount.c, and
>>> enable building a version of these files that will introduce support
>>> for passing mount options as a string (that work is yet to be done).
>> Does this mean that you cannot compile nfs-utils to support both old
>> and new kernels?  Is that really a good plan?
>> What is the over-all migration strategy?
> 
> Making the different argument passing variants exclusive sounds like
> a bad idea indeed.

Yeah, Neil has convinced me that mount.nfs must support binary and 
string option processing simultaneously.

The 17 refactoring patches are still probably good clean ups overall.  I 
will leave them stand on their own merits.

[-- Attachment #2: chuck.lever.vcf --]
[-- Type: text/x-vcard, Size: 290 bytes --]

begin:vcard
fn:Chuck Lever
n:Lever;Chuck
org:Oracle Corporation;Corporate Architecture: Linux Projects Group
adr:;;1015 Granger Avenue;Ann Arbor;MI;48104;USA
title:Principal Member of Staff
tel;work:+1 248 614 5091
x-mozilla-html:FALSE
url:http://oss.oracle.com/~cel
version:2.1
end:vcard


[-- Attachment #3: Type: text/plain, Size: 286 bytes --]

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/

[-- Attachment #4: Type: text/plain, Size: 140 bytes --]

_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: [PATCH 00/17] RFC: proposed changes to the NFS mount command

[Chuck Lever]

[-- Attachment #1: Type: text/plain, Size: 2019 bytes --]

Neil Brown wrote:
> On Sunday July 15, chuck.lever@oracle.com wrote:
>> Hi Neil-
>>
>> I've got a series of patches to the NFS mount command that prepare it
>> for using a C string to pass mount options to the kernel, rather than
>> a binary nfs_mount_data structure.
> 
> I've wondered why that was happening... Can't we just make a version
> x+1 with a big string buffer so new options get put there is regular
> string format and old options stay the way they are?
> 
> But I guess that - long term - it is better to be just like other
> filesystems.
> 
> 
>> The goal of these clean-ups is to allow the creation of a ./configure
>> option that will disable building nfsmount.c and nfs4mount.c, and
>> enable building a version of these files that will introduce support
>> for passing mount options as a string (that work is yet to be done).
> 
> Does this mean that you cannot compile nfs-utils to support both old
> and new kernels?  Is that really a good plan?
> What is the over-all migration strategy?

The NFS client string mount option support is destined for 2.6.23.

My plan was to build for either binary mount options or string mount 
options, switched via a ./configure option.  But it's a good point that 
currently, the mount command will support all versions of 
nfs_mount_data, so it should probably support binary and string options 
simultaneously too.

I will reconsider my design.  Thanks for the useful feedback.

>> They certainly need some testing yet, but can you take a look at these
>> and tell me if they are generally acceptable or if I need to take a
>> different direction?  Hopefully they are straightforward and won't take
>> much of your time.
> 
> Some of these are obviously good cleanups.  Would it inconvenience you
> if I just picked off the ones I like for now and leave you to rebase?

I would like to do some testing, and maybe some review from others on 
the list (hint hint cough).  Can you let me know (offline) which ones 
you want to take, and I will test those first?

[-- Attachment #2: chuck.lever.vcf --]
[-- Type: text/x-vcard, Size: 290 bytes --]

begin:vcard
fn:Chuck Lever
n:Lever;Chuck
org:Oracle Corporation;Corporate Architecture: Linux Projects Group
adr:;;1015 Granger Avenue;Ann Arbor;MI;48104;USA
title:Principal Member of Staff
tel;work:+1 248 614 5091
x-mozilla-html:FALSE
url:http://oss.oracle.com/~cel
version:2.1
end:vcard


[-- Attachment #3: Type: text/plain, Size: 286 bytes --]

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/

[-- Attachment #4: Type: text/plain, Size: 140 bytes --]

_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs