It looks like the IB BMI layer is ending up double-freeing the
method_addr structure on the BMI_ib_set_info function, but it only
happens when the Metadata server is also a data server.
If you look at the following GDB output, the last two entries have the
same method_addr, and I can't figure out a good way to tell in
BMI_set_info if the method_address has already been freed. It also looks
like the id_string has been mangled or freed somewhere earlier as well.
Breakpoint 2, dealloc_ref_st (deadref=0x100ec110) at
../src/io/bmi/reference-list.c:239
239 free(deadref);
2: *deadref = {bmi_addr = 269402384, id_string = 0x100ec138 "",
method_addr = 0x100ed1d8,
interface = 0x1008dcbc, list_link = {next = 0x100ec0c4, prev =
0x100af198}, ref_count = 0}
1: deadref = (ref_st_p) 0x100ec110
(gdb) c
Continuing.
Breakpoint 2, dealloc_ref_st (deadref=0x100ec0b0) at
../src/io/bmi/reference-list.c:239
239 free(deadref);
2: *deadref = {bmi_addr = 269402288, id_string = 0x100ec0d8
"\020\016Á\b/da1:3339,tcp://da1.ib:3338",
method_addr = 0x100b7f98, interface = 0x1008dcbc, list_link = {next =
0x100ec074, prev = 0x100ec124},
ref_count = 0}
1: deadref = (ref_st_p) 0x100ec0b0
(gdb) c
Continuing.
Breakpoint 2, dealloc_ref_st (deadref=0x100ec060) at
../src/io/bmi/reference-list.c:239
239 free(deadref);
2: *deadref = {bmi_addr = 269402208, id_string = 0x100ec088
"\020\016ˬ/da2:3339,tcp://da2.ib:3338",
method_addr = 0x100b7d18, interface = 0x1008dcbc, list_link = {next =
0x100b7d44, prev = 0x100ec0c4},
ref_count = 0}
1: deadref = (ref_st_p) 0x100ec060
(gdb) c
Continuing.
Breakpoint 2, dealloc_ref_st (deadref=0x100b7d30) at
../src/io/bmi/reference-list.c:239
239 free(deadref);
2: *deadref = {bmi_addr = 269188400, id_string = 0x100b7d58
"\020\v}\020/da2:3339",
method_addr = 0x100b7d18, interface = 0x1008dcbc, list_link = {next =
0x100af198, prev = 0x100ec074},
ref_count = 0}
1: deadref = (ref_st_p) 0x100b7d30
(gdb) c
Continuing.
_______________________________________________
Pvfs2-developers mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers