mbox series

[0/1] Add callback address and state to nfsd client info

Message ID 1620999041-9341-1-git-send-email-dwysocha@redhat.com (mailing list archive)
Headers show
Series Add callback address and state to nfsd client info | expand

Message

David Wysochanski May 14, 2021, 1:30 p.m. UTC
For troubleshooting, it is useful to show the callback address and state,
even though we do have this equivalent info inside Chuck's ftrace patches.
Note there is a show_cb_state() inside fs/nfsd/trace.h and this code
has a similar function.  It may be better to consolidate these two
if these additions are ok for nfsd client info, but not sure where
a good header is to place it - do we need a new file, maybe
fs/nfsd/nfs4callback.h?

Dave Wysochanski (1):
  nfsd4: Expose the callback address and state of each NFS4 client

 fs/nfsd/nfs4state.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

Comments

J. Bruce Fields May 14, 2021, 2:15 p.m. UTC | #1
On Fri, May 14, 2021 at 09:30:40AM -0400, Dave Wysochanski wrote:
> For troubleshooting, it is useful to show the callback address and state,
> even though we do have this equivalent info inside Chuck's ftrace patches.

Good idea.

> Note there is a show_cb_state() inside fs/nfsd/trace.h and this code
> has a similar function.  It may be better to consolidate these two
> if these additions are ok for nfsd client info, but not sure where
> a good header is to place it - do we need a new file, maybe
> fs/nfsd/nfs4callback.h?

nfs4state.c already includes trace.h, do we need anything more?

I'll admit I've just been adding things wherever seems expedient for a
while, so there may be some more logical way to organize nfsd headers.

--b.

> 
> Dave Wysochanski (1):
>   nfsd4: Expose the callback address and state of each NFS4 client
> 
>  fs/nfsd/nfs4state.c | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
> 
> -- 
> 1.8.3.1
David Wysochanski May 14, 2021, 4:41 p.m. UTC | #2
On Fri, May 14, 2021 at 10:15 AM Bruce Fields <bfields@fieldses.org> wrote:
>
> On Fri, May 14, 2021 at 09:30:40AM -0400, Dave Wysochanski wrote:
> > For troubleshooting, it is useful to show the callback address and state,
> > even though we do have this equivalent info inside Chuck's ftrace patches.
>
> Good idea.
>
> > Note there is a show_cb_state() inside fs/nfsd/trace.h and this code
> > has a similar function.  It may be better to consolidate these two
> > if these additions are ok for nfsd client info, but not sure where
> > a good header is to place it - do we need a new file, maybe
> > fs/nfsd/nfs4callback.h?
>
> nfs4state.c already includes trace.h, do we need anything more?
>

Probably not.  I am testing a renamed function (I find that
"<typename>2str" is more common in the kernel) "cb_state2str"
defined in fs/nfsd/trace.c and declaration in fs/nfsd/trace.h

If that makes sense I'll send a v2.




> I'll admit I've just been adding things wherever seems expedient for a
> while, so there may be some more logical way to organize nfsd headers.
>
> --b.
>
> >
> > Dave Wysochanski (1):
> >   nfsd4: Expose the callback address and state of each NFS4 client
> >
> >  fs/nfsd/nfs4state.c | 17 +++++++++++++++++
> >  1 file changed, 17 insertions(+)
> >
> > --
> > 1.8.3.1
>