Due to second call to SMB_VFS_UNLINKAT() inside rmdir_internals() using the wrong parameters.
Have fix, need bugnumber and regression test.
Created attachment 16899 [details]
git-am fix for 4.15.next.
Cherry-picked from master. Note this must be applied *before* the fix for https://bugzilla.samba.org/show_bug.cgi?id=14879
This bug was referenced in samba master:
Reassigning to Jule for inclusion in 4.15.
Note to Jule:
Please note the patches for these bugs *MUST* be applied in the following
in order to apply correctly. Thanks !
Pushed to autobuild-v4-15-test.
This bug was referenced in samba v4-15-test:
Closing out bug report.
Created attachment 17023 [details]
patch for v4-14
73de1194c3c429ab93d722a852aa4f54213b112a s3: smbd: Fix recursive directory delete of a directory containing veto file and msdfs links.
- commit above was modified,changes in source3/smbd/close.c are not applicable
I'm probably not going to be able to get to this until Monday due to the USA Thanksgiving Holiday, sorry.
Comment on attachment 17023 [details]
patch for v4-14
Pavel, I don't see any code changes here. Just adding the tests and then removing the knownfail. Is this patch what you intended ?
Yes, that is intended.
If you check the patch for 4.15 there is only one code change:
For 4.14 this is not applicable, so I have just added the tests, so we can test the functionality.
Do you think that such patch should be skipped completely?
Oh I'm sorry, I misunderstood. Then the patch is on the whole OK, but just squash into one patch. There's no point in adding and then removing the knownfail. Just add the test without a knownfail, as 4.14.x doesn't fail here.
Created attachment 17025 [details]
squashed patch for v4-14
Adding a new patch squashed into a single commit
I see that the commit message for the squashed patch still says:
Let me know if this should be removed, or can be accepted