[PATCHES] smb3 create replay (for multi-channel)
Michael Adam
obnox at samba.org
Thu Mar 3 08:33:25 UTC 2016
On 2016-03-02 at 16:45 -0800, Jeremy Allison wrote:
> On Wed, Mar 02, 2016 at 10:45:11PM +0100, Michael Adam wrote:
> >
> > Attached find the updated patchset.
> > Differences with the original one:
> >
> > - some cosmetic clean-up of commit messages
> > and test names
> > - added three new tests:
> > - replay-dht2-oplock-lease
> > (create with oplock, replay with lease)
> > - replay-dht2-lease-oplock
> > (create with lease, replay with oplock)
> > - replay-dht2-lease3
> > (replay with a different lease key)
> > - fixed treatment of leases in create replay
> > in corner cases
>
> Pushing this, but autobuild is failing with:
>
> [252(789)/1892 at 20m43s] samba3.blackbox.smbclient_machine_auth.plain (nt4_member:local)
> ERROR: Testsuite[samba3.blackbox.smbclient_machine_auth.plain (nt4_member:local)]
> REASON: unable to set up environment nt4_member:local - exiting
>
> Don't think this is to do with your new code.
>
> I'll keep trying tomorrow.
Ok. I investigated, and it was in fact related.
May bad that I had not run a full make test with the
latest patchset but subsets of the tests... :-/
There was a NULL dereference of 'op' in smb2_create_send
right at the and, because I used op instead of result->op
in one (new) place. Usually these are the same at this
place, but for IPC or PRINT shares, we ended up here without
op == result->op but op == NULL.
Attached find the 2-line patch to be squashed into the
'implement create replay' test to fix the crash.
I am running a full make test with this now and will
(given the in-retrospect obviousness) push once this has
passed.
Cheers - Michael
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20160303/7b3333fb/signature.sig>
More information about the samba-technical
mailing list