as I know, we've decided to make me "gatekeeper" of QE team for managing
lib389 and DS git repos.
Can you please update the status of this situation?
Also, there is two tickets with new reviewed testcases, that need to be
pushed. Can you please help me with that? Or if I'll get the rights
soon, I do it by myself. :)
git patch file (master) -- revised; fixed a bug pointed out by Thierry
(Thank you!!) and a minor memory leak.
On 09/30/2015 01:32 AM, 389 Project wrote:
> Comment (by tbordaz):
> Hi noriko,
> The fix looks good, just two remarks
> * pagedresults_set_search_result_pb is called each time search result is
> freed. Should not it be called in opshared.c:722 (I find this line
> suspicious should not it be SLAPI_PAGED_RESULTS_INDEX ?)
> * in pagedresults_set_search_result_pb, you call slapi_pblock_set(pb,
> SLAPI_PAGED_RESULTS_INDEX, &index); . My understanding it is to reset the
> pr_idx (index==-1). But later you test 'index'. Did you mean
> slapi_pblock_'''get'''(pb, SLAPI_PAGED_RESULTS_INDEX, &index).
On 09/29/2015 08:15 AM, 389 Project wrote:
> Unfortunately, it did not pass the IPA tests.
Bug Description: The cause of the problem is rather not a race condition but
accessing an already freed agreement in a plug-in:
> The crashed thread is deleting an agreement object, which calls
> It eventually calls op_shared_search with the deleted agreement
> base scope and filter "(|(objectclass=*)(objectclass=ldapsubentry))"
> Since it is a DSE entry it goes to dse_search, in which it calls
> replarea and crashes in slapi_sdn_copy by NULL dereference in from SDN...
Fix Description: This patch adds the check to agmt_get_replarea, in which if
the agreement is not in the agreement list, it returnes NULL repl area.
the NULL repl area is returned the callers back off with an error.