In rpc_server/srv_samr_nt.c:_samr_delete_dom_group(), the grp variable is
uninitialised when it is passed to smb_delete_group().
Bad things will happen if either lp_delgroup_script or
lp_winbind_enable_local_accounts is set.
Created attachment 939 [details]
take group name from SID lookup
I think what is supposed to happen is for the group name to come from
how can grp it be uninitialized when we just called
grp=getgrpgid() a few lines before. WHat am I not
(In reply to comment #2)
> how can grp it be uninitialized when we just called
> grp=getgrpgid() a few lines before. WHat am I not
> seeing here?
I don't see this call anywhere in _samr_delete_dom_group(). I'm looking
at SVN revision 5184.
James, you're looking at trunk, right? srv_samr_nt.c seems wildly different
between 3_0 and trunk, and this needs merging. I'm closing this bug with
"later", when the merge is done.
ok. That pretty much settles it then. I had wondered
if trunk had more work on the group stuff. I'm going to
merge since code from 3.0 since trunk is obviously wrong.
(In reply to comment #4)
> James, you're looking at trunk, right? srv_samr_nt.c seems wildly different
> between 3_0 and trunk, and this needs merging. I'm closing this bug with
> "later", when the merge is done.
Yes, I'm looking at trunk. Doesn't trunk get merged into 3_0 for releases?
trunk/ is not necessarily merged into trunk, it is more like a testbed for
ideas. For example I've started the third attempt to get winbind non-blocking,
all three with completely different ideas. Or, lately stuff that mimir has done
has been revoked again. If you want to get something quality-checked by the very
valuable IRIX compiler warnings, please look at the branch SAMBA_3_0. This is
the one we really care about.
no. trunk uis basically 3.1. The SAMBA_3_0 is teh stable
development tree which gets merged to the SAMBA_3_0_RELEASE
tree for release.
(In reply to comment #8)
> no. trunk uis basically 3.1. The SAMBA_3_0 is teh stable
> development tree which gets merged to the SAMBA_3_0_RELEASE
> tree for release.
I see .. thanks for setting me straight ..
originally reported against 3.0.11pre1. Moving back to version to 3.0.10
to remove preX and rcX versions.