ACK/Cmnt: [bionic:linux-azure-4.15][PATCH 0/2] LP:#1882268 - [linux-azure] Request for two CIFS commits in 16.04
stefan.bader at canonical.com
Wed Oct 28 09:34:51 UTC 2020
On 27.10.20 12:42, Marcelo Henrique Cerri wrote:
> On Tue, Oct 27, 2020 at 10:20:37AM +0100, Stefan Bader wrote:
>> On 26.10.20 15:34, Marcelo Henrique Cerri wrote:
>>> BugLink: https://bugs.launchpad.net/bugs/1882268
>>> The commit id 2084ed57167c ("CIFS: Only send SMB2_NEGOTIATE command on
>>> new TCP connections") is a clean cherry pick from upstream while
>>> 84a1f5b1cc6f ("cifs: Fix potential softlockups while refreshing DFS
>>> cache") had to be backported to 4.15.
>>> Both changes are already included in the 5.4 linux-azure
>>> kernel. 2084ed57167c was included upstream in 5.1 and 84a1f5b1cc6f was
>>> applied via an upstream stable update (LP:#1860125). 2084ed57167c
>>> fixes commit id e99c63e4d86d ("SMB3: Fix deadlock in validate
>>> negotiate hits reconnect") that was included to our 4.15 kernel also
>>> via an upstream stable update.
>>> The original author of 84a1f5b1cc6f reviewed and approved my backport
>>> for 4.15. Microsoft reported positive results with the test kernel and
>>> I also smoke tested it using dbench as described by LP:#1896642.
>>> Since we don't have a direct reproducer but we have positive feedback
>>> from Microsoft I recommend that we include these fixes only to the
>>> 4.15 linux-azure kernels for now. Later we can evaluate if we should
>>> include them to the generic 4.15 kernel too.
>>> Paulo Alcantara (SUSE) (1):
>>> cifs: Fix potential softlockups while refreshing DFS cache
>>> Pavel Shilovsky (1):
>>> CIFS: Only send SMB2_NEGOTIATE command on new TCP connections
>>> fs/cifs/smb2pdu.c | 38 ++++++++++++++++++++++++++++----------
>>> fs/cifs/smb2transport.c | 8 ++++++++
>>> 2 files changed, 36 insertions(+), 10 deletions(-)
>> The bug report gets a bit confusing with all the linux-azure tasks being more or
>> less invalid. Maybe we could drop those and only keep linux-azure-4.15...
> Hi, Stefan.
> I agree it's a bit confusing. The bug was initially nominated for
> xenial:linux-azure (which is the more relevant 4.15 azure kernel). But
> the work needs to be done against bionic:linux-azure-4.15. In the past
> I used to remove those nominations, but it causes so much trouble
> when, for some reason, we need to add those nominations back again
> that now I prefer to simply mark them as invalid.
> If you prefer I can remove them though, just let me know.
I think in this case it probably slims things down quite a bit. The problem you
refer to happens when you want to get series nominations back. Which does not
work after deleting it in one place. You then have to remove all nominations of
that series (remembering their state) and add them back together.
In this case we would remove the other source and all nominations. You could
then even remove the nomination that is not needed for linux-azure-4.15. As
there is only one source left, any series nomination removed is completely gone
and so it can be just added back normally. The problem is only with bug reports
that have multiple sources.
Ok, I have cleaned everything up. Note that you could nominate again for xenial
>> Acked-by: Stefan Bader <stefan.bader at canonical.com>
>> kernel-team mailing list
>> kernel-team at lists.ubuntu.com
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 833 bytes
Desc: OpenPGP digital signature
More information about the kernel-team