Joseph Sutton notes that we need to: Make a deep copy of the message elements in msg_diff() so that if either of the input messages are deallocated early, the result does not refer to non-existing elements.
This bug was referenced in samba master: a99a76722d6046a5d63032e3d2bb3f791da948a6 c2bbe774ce03661666a1f48922a9ab681ef4f64b 19a2af02f57d99db8ed3c6b028c3abdf4b553700
are we going to backport?
I'll backport and upload. Yes, we will need this to allow backport of other things like bronze-bit tests.
Created attachment 16813 [details] patch from master backported to 4.15 (only) The 4.14 and earlier branches need a backport, not just a cherry-pick, but we should do this much at least.
Created attachment 16814 [details] patch from master backported to 4.14 I've run 'make test TESTS=samba.tests.segfault' to confirm the backport
Assigning to Jule for 4.14 and 4.15
Pushed to autobuild-v4-{15,14}-test.
Thanks. I'll be backporting a patch with a new LDB version number in it shortly, so an external ldb will also include the change.
This bug was referenced in samba v4-15-test: db294baff360f1c44c05798f6cda4584166adfd7 e52ddfbe5728487cb2c8b8ceeb2f63c5c15a7541 eb28bd54ac5c36a9392ef19c49bf97a18b81974e
This bug was referenced in samba v4-14-test: 9f79d4256f8f24127f06f0bf25092c5ca84a7d59 9d61f2f2f3eb3fd79bb2d78da8d64b0f8b66d3aa 5a90b3e832cda88339c5cebca7043e842b348e47
(In reply to Andrew Bartlett from comment #8) If the patch with the new LDB version should also be in the next 4.14 release, then I need the patch until Tuesday.
This bug was referenced in samba v4-14-stable (Release samba-4.14.8): 9f79d4256f8f24127f06f0bf25092c5ca84a7d59 9d61f2f2f3eb3fd79bb2d78da8d64b0f8b66d3aa 5a90b3e832cda88339c5cebca7043e842b348e47
This bug was referenced in samba master: 76899e236149ff3b86cd9032a3c6bdafe3a2f036
The remaining step is to bump the ldb version number in 4.15, see the patch on bug 14848
(In reply to Jule Anger from comment #11) Sorry, the patch to bump the version number on bug 14848 was not approved in time so Samba 4.14.8 shipped with a different internal ldb to the 'system' one we require. So far the changes only really matter for our tests, but this isn't ideal. I continue to advocate to remove this ldb release step, but without success so far.
This bug was referenced in samba v4-14-test: c1d2a0570dfc697bbdda6047f10da4ea9cf261f8
This bug was referenced in samba v4-15-test: a795e0c84597aa045d011e663dbad3cdabf0f1e6
This bug was referenced in samba v4-15-stable (Release samba-4.15.1): db294baff360f1c44c05798f6cda4584166adfd7 e52ddfbe5728487cb2c8b8ceeb2f63c5c15a7541 eb28bd54ac5c36a9392ef19c49bf97a18b81974e a795e0c84597aa045d011e663dbad3cdabf0f1e6
This bug was referenced in samba v4-14-stable (Release samba-4.14.9): c1d2a0570dfc697bbdda6047f10da4ea9cf261f8
This bug was referenced in samba v4-13-test: f47f0f9f4595ab9e83479b493a1110e602aee4fc 400d04533ab3525f29721dffa137faf096c274e8 0c36416e319e307022d11a915a4ac508c0ab2eea 74e65d7c06c5eda79105f43d87efcaec09dfbb77
This bug was referenced in samba v4-13-stable (Release samba-4.13.13): f47f0f9f4595ab9e83479b493a1110e602aee4fc 400d04533ab3525f29721dffa137faf096c274e8 0c36416e319e307022d11a915a4ac508c0ab2eea 74e65d7c06c5eda79105f43d87efcaec09dfbb77