David Bustos wrote:
> Quoth "C. Bergstr?m" on Mon, Dec 01, 2008 at 08:17:47PM +0100:
>
>> ../milestone/global.xml
>> svccfg_libscf.c:1884: scf_transaction_commit() failed with unexpected
>> error -1. Aborting.
>>
>
> This is strange, because svccfg should catch -1 and not abort. Indeed,
> there is a bug in the code which causes it to catch -1 and abort anyway.
>
> Fortunately we should be able to tell why scf_transaction_commit()
> failed by looking at the lcbdata field. Though I didn't see
> emsg_pg_deleted or emsg_pg_mod_perm in your output, so it's probably
> _BACKEND_READONLY, _BACKEND_ACCESS, _CONNECTION_BROKEN, or
> _NO_RESOURCES. Anyway, lcbdata is a pointer to the second arument of
> entity_pgroup_import(), so if you can get that pointer, we should be
> able to determine the exact error.
>
>
This what you mean?
Current function is entity_pgroup_import
1884 bad_error("scf_transaction_commit", r);
(dbx) dump
pg_changed = 0x45bf40 "%s changed unexpectedly (new property group "%s"
changed).\n"
cbdata = RECORD
p = 0x68cc10
ent = 0x4ab410
v = 0x68cc10
lcbdata = 0xfffffd7fffdfe4b0
pvt = 0xfffffd7fffdfe4b0
r = -1
issvc = 1
I'm not sure a proper fix, but I did a quick removal of the abort()
{ ( void ) fprintf ( ( & __iob [ 2 ] ) , "%s:%d: %s() failed with
unexpected " "error %d. Aborting.\n" , "svccfg_libscf.c" , 1884 , (
"scf_transaction_commit" ) , ( r ) ) ; };
svccfg_libscf.c:1884: scf_transaction_commit() failed with unexpected
error -1. Aborting.
svccfg: Could not delete svc:/TEMP/system/svc/global (repository
connection broken).
svccfg: Repository connection broken.
svccfg: Import of ../milestone/global.xml failed. Progress:
svccfg: Service "system/svc/global": not reached.
svccfg: Instance "default": not reached.
sh: line 2: 7024: Memory fault(coredump)
*** Error code 11
I also changed the makefile to include
LD_PRELOAD=libumem.so UMEM_DEBUG=default \
and I don't see any messages to stdout about any leaks so maybe that's
not it..
(btw. I really appreciate the help on this and I *know* it's my fault)
./C
