(ITS#3426) HEAD syncprov segfaults on delete to replica when a group ACL is used

[quanah@stanford.edu]

Full_Name: Quanah Gibson-Mount
Version: HEAD
OS: Solaris 8
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (171.66.182.82)


When doing a delete of the suspect entry in ITS#3425, it caused the provider to
segfault:

Program received signal SIGSEGV, Segmentation fault.
0x0003d438 in backend_group (op=0x7c800fc0, target=0x7c8010a8,
gr_ndn=0x7c7ff888, op_ndn=0x7c801054, group_oc=0x1cee50,
    group_at=0x1f9b80) at backend.c:1201
1201                    if (g->ga_be != op->o_bd || g->ga_oc != group_oc ||
(gdb) bt
#0  0x0003d438 in backend_group (op=0x7c800fc0, target=0x7c8010a8,
gr_ndn=0x7c7ff888, op_ndn=0x7c801054, group_oc=0x1cee50,
    group_at=0x1f9b80) at backend.c:1201
#1  0x0005b38c in acl_mask (a=0x1c49b8, mask=0x7c80033c, op=0x7c800fc0,
e=0x7c8010a8, desc=0x1bbd08, val=0x0, nmatch=100,
    matches=0x7c7ff9d0, count=3, state=0x0) at acl.c:1313
#2  0x00058040 in access_allowed_mask (op=0x7c800fc0, e=0x7c8010a8,
desc=0x1bbd08, val=0x0, access=ACL_READ, state=0x0,
    maskp=0x0) at acl.c:370
#3  0x00040ea0 in slap_send_search_entry (op=0x7c800fc0, rs=0x7c8010e0) at
result.c:742
#4  0x00129968 in syncprov_sendresp (op=0xb3d488, opc=0xb89dcc, so=0x14a1e68,
e=0x0, mode=3, queue=1) at syncprov.c:808
#5  0x0012b748 in syncprov_op_response (op=0xb3d488, rs=0x7c801ac8) at
syncprov.c:1321
#6  0x0003f0d0 in send_ldap_response (op=0xb3d488, rs=0x7c801ac8) at
result.c:310
#7  0x000402e4 in slap_send_ldap_result (op=0xb3d488, rs=0x7c801ac8) at
result.c:607
#8  0x000e1bdc in bdb_delete (op=0xb3d488, rs=0x7c801ac8) at delete.c:578
#9  0x000a41f4 in over_op_func (op=0xb3d488, rs=0x7c801ac8, which=op_delete) at
backover.c:284
#10 0x000a43c0 in over_op_delete (op=0xb3d488, rs=0x7c801ac8) at backover.c:340
#11 0x0004d55c in fe_op_delete (op=0xb3d488, rs=0x7c801ac8) at delete.c:209
#12 0x0004d1ac in do_delete (op=0xb3d488, rs=0x7c801ac8) at delete.c:95
#13 0x000296f8 in connection_operation (ctx=0x7c801ba0, arg_v=0xb3d488) at
connection.c:978
#14 0xff32da88 in ldap_int_thread_pool_wrapper (xpool=0x1be498) at tpool.c:467
#15 0xfed5b024 in _thread_start () from /usr/lib/libthread.so.1
#16 0xfed5b024 in _thread_start () from /usr/lib/libthread.so.1

This appears to be a bug in the group ACL code.