Found by Codenomicon at the SNIA plugfest. Patches for 4.0.0 and 3.6.next to follow. Jeremy.
Created attachment 7956 [details] git-am fix for 4.0.0rc2 and also 3.6.next.
Comment on attachment 7956 [details] git-am fix for 4.0.0rc2 and also 3.6.next. Patch also applies cleanly to 3.6.next.
Re-assigning to Karolin for inclusion in 3.6.next and 4.0.0rc2. Jeremy.
Comment on attachment 7956 [details] git-am fix for 4.0.0rc2 and also 3.6.next. Looks good. We may want to backport e6c600aa2c751e694917322378417816c3e58eb6 (s4:torture/smb2: improve the smb2.create.blob test) too
Pushed to autobuild-v4-0-test and v3-6-test. Re-assigning to Jeremy because of Metze's backport comment. Thanks!
Created attachment 8036 [details] Test for v4-0-test
Comment on attachment 8036 [details] Test for v4-0-test LGTM.
Re-assigning to Karolin to push Metze's 4.0.0rc3 test.
(In reply to comment #8) > Re-assigning to Karolin to push Metze's 4.0.0rc3 test. Pushed to autobuild-v4-0-test. Closing out bug report. Thanks!
Created attachment 8110 [details] Patch for v4-0-test and v3-6-test This fixes a problem in https://bugzilla.samba.org/attachment.cgi?id=7956 where we send uninitialized padding. I think this is needed for 3.6.9
Karolin, this is needed for 3.6.9 otherwise we introduce a regression.
(In reply to comment #10) > Created attachment 8110 [details] > Patch for v4-0-test and v3-6-test > > This fixes a problem in https://bugzilla.samba.org/attachment.cgi?id=7956 > where we send uninitialized padding. > > I think this is needed for 3.6.9 A bit more information on the patch: - blob_offset = 0x14 + tag_length; + blob_offset = 0x10 + tag_length; blob_pad = smb2_create_blob_padding(blob_offset, 8); The resulting blob_offset will be the same, as we later have this: if (blob_pad > 0) { memset(buffer->data+ofs+blob_offset, 0, blob_pad); blob_offset += blob_pad; } The important part is that blob_pad changes from 0 to 4 and we initialize the 4 byte padding (for the typical case where tag_length is 4).
Do you plan to backport the torture test for 3.6 too?
(In reply to comment #13) > Do you plan to backport the torture test for 3.6 too? I don't think that is needed
(In reply to comment #11) > Karolin, this is needed for 3.6.9 otherwise we introduce a regression. Okay, thanks for the heads-up. Samba 3.6.9 is scheduled for tomorrow, please review this patch asap. Additionally, I would like to propose to delay the release to avoid including last minute patches and allow proper testing/review. Comments welcome. Thanks!
Comment on attachment 8110 [details] Patch for v4-0-test and v3-6-test ack
Assigning to Karolin for inclusion into v3-6-test and v4-0-test
It looks like the test case is simple and can be applied without problems to 3.6. I would like to have the test which verifies that it is fixed in 3.6 too so QA can verify it.
(In reply to comment #18) > It looks like the test case is simple and can be applied without problems to > 3.6. I would like to have the test which verifies that it is fixed in 3.6 too > so QA can verify it. If you propose a patch (or if the v4-0-test one applies) I'm happy to ack it...
(In reply to comment #15) > (In reply to comment #11) > > Karolin, this is needed for 3.6.9 otherwise we introduce a regression. > > Okay, thanks for the heads-up. > > Samba 3.6.9 is scheduled for tomorrow, please review this patch asap. > Additionally, I would like to propose to delay the release to avoid including > last minute patches and allow proper testing/review. > > Comments welcome. > > Thanks! Talked to Metze and it seems to be an obvious fix which is needed for 3.6.9. Going ahead with the release.
Pushed to v3-6-test and autobuild-v4-0-test. Will be included in 3.6.9. Re-assigning to Andreas as he seems to work on a test.
The "Test for v4-0-test" patch from comment #6 applies cleanly to v3.6-test.
(In reply to comment #22) > The "Test for v4-0-test" patch from comment #6 applies cleanly to v3.6-test. Ok, then I'm fine if it gets backported to v3-6-test.
Comment on attachment 8110 [details] Patch for v4-0-test and v3-6-test From the "better late than never" dept. +1 from me (it was over my weekend :-). Jeremy.
Karolin, please also pick the test/torture patch to v3-6-test
(In reply to comment #25) > Karolin, please also pick the test/torture patch to v3-6-test (In reply to comment #25) > Karolin, please also pick the test/torture patch to v3-6-test Pushed. Closing out bug report. Thanks!