[linux-cifs-client] Re: pending cifs stable fixes ok but delay until 2.6.17 goes out?

[Andrew Morton]

Steven French <sfrench at us.ibm.com> wrote:
>
> Just wanted to double check our tentative decision to defer pushing the current cifs-stable
> tree bug fixes into mainline until 2.6.17 comes out.  The tree
>       rsync://rsync.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6-stable.git
> recently finished testing out ok, looks fine and contains only low risk fixes, but it looks
> like it is too late to recommend pushing it upstream right?   There is a much bigger
> change (Note: that the backlevel server support for cifs client which is needed
> in order to deprecate smbfs, and enable better security negotation, and which is
> much larger,  and thus in a distinct tree temporarily while I am testing,
> is not included in the list below).   The following fixes could be pushed easily enough
> and include:
>       - a small memory leak on every session reconnect (after e.g. server or network failure)
>       - working support for advisory posix locking to current samba (with /proc/fs/cifs/Experimental on)
>       - a fix for a rare oops
>       - a fix for long share names in certain asian languages
>       - a fix for passing O_TRUNC which works around a bug in a less common NAS appliance
> 
> and would update the following files:
> 
>  fs/cifs/CHANGES     |    7 +++
>  fs/cifs/cifsfs.h    |    2 -
>  fs/cifs/cifsproto.h |    2 -
>  fs/cifs/cifssmb.c   |   40 ++++++++++++++++++---
>  fs/cifs/connect.c   |   97 +++++++++++++++++++++++++++++++++++++++++++++++-----
>  fs/cifs/file.c      |   12 ++++--
>  6 files changed, 140 insertions(+), 20 deletions(-)

No, I'd suggest that you push these changes up to Linus asap.  They
probably fix more bugs than they add ;)

If regressions are added then we can fix them up in 2.6.17.x.  But please
do be aware that it's largely up to you, the originator, to forward the
necessary patches to stable at kernel.org.