Обсуждение: Re: [COMMITTERS] pgsql: SQL/MED catalog manipulation facilities This doesn't do any
petere@postgresql.org (Peter Eisentraut) writes:
> Log Message:
> -----------
> SQL/MED catalog manipulation facilities
The buildfarm says this patch has got serious portability issues.
grebe, dugong, and ermine are all crashing here:
--- 109,115 ---- (3 rows) ALTER FOREIGN DATA WRAPPER foo OPTIONS (a '1', b '2');
! server closed the connection unexpectedly
! This probably means the server terminated abnormally
! before or while processing the request.
! connection to server was lost
(Unfortunately I can't reproduce this on any of my own machines.)
The Windows MSVC machines are all pretty unhappy too, presumably because
you didn't even try to teach the MSVC build system about the newly added
shared libraries.
regards, tom lane
Re: [COMMITTERS] pgsql: SQL/MED catalog manipulation facilities This doesn't do any
От
"Sergey E. Koposov"
Дата:
Hi,
On Fri, 19 Dec 2008, Tom Lane wrote:
>> SQL/MED catalog manipulation facilities
>
> The buildfarm says this patch has got serious portability issues.
>
> grebe, dugong, and ermine are all crashing here:
>
> --- 109,115 ----
> (3 rows)
>
> ALTER FOREIGN DATA WRAPPER foo OPTIONS (a '1', b '2');
> ! server closed the connection unexpectedly
> ! This probably means the server terminated abnormally
> ! before or while processing the request.
> ! connection to server was lost
>
> (Unfortunately I can't reproduce this on any of my own machines.)
Here is the gdb backtrace from dugong:
-------------------
Core was generated by ostgres: math regression [local] ALTER FOREIGN DATA WRAPPER '.
Program terminated with signal 11, Segmentation fault.
#0 0x40000000000560b0 in heap_compute_data_size ( tupleDesc=0x200000000287fbb8, values=0x60000000001ae060,
isnull=0x60000000001ae0b8"") at heaptuple.c:103
103 if (ATT_IS_PACKABLE(att[i]) &&
(gdb) bt
#0 0x40000000000560b0 in heap_compute_data_size ( tupleDesc=0x200000000287fbb8, values=0x60000000001ae060,
isnull=0x60000000001ae0b8"") at heaptuple.c:103
#1 0x400000000005cac0 in heap_form_tuple (tupleDescriptor=0x200000000287fbb8, values=0x60000000001ae060,
isnull=0x60000000001ae0b8"") at heaptuple.c:728
#2 0x400000000005d700 in heap_modify_tuple (tuple=0x600000000019e9f8, tupleDesc=0x200000000287fbb8,
replValues=0x60000fffffb2e318, replIsnull=0x60000fffffb2e340 "", doReplace=0x60000fffffb2e345 "") at
heaptuple.c:853
#3 0x4000000000399480 in AlterForeignDataWrapper (stmt=0x6000000000160e78) at foreigncmds.c:480
#4 0x40000000007b6bc0 in ProcessUtility (parsetree=0x6000000000160e78, queryString=0x60000000001603b8 "ALTER
FOREIGNDATA WRAPPER foo OPTIONS (a '1', b '2');", params=0x0, isTopLevel=1 '\001', dest=0x6000000000160f10,
completionTag=0x60000fffffb2e96e"") at utility.c:469
#5 0x40000000007b2e20 in PortalRunUtility (portal=0x60000000001a5fc8, utilityStmt=0x6000000000160e78, isTopLevel=1
'\001', dest=0x6000000000160f10, completionTag=0x60000fffffb2e96e "") at pquery.c:1183
#6 0x40000000007b3380 in PortalRunMulti (portal=0x60000000001a5fc8, isTopLevel=1 '\001', dest=0x6000000000160f10,
altdest=0x6000000000160f10, completionTag=0x60000fffffb2e96e "") at pquery.c:1286
#7 0x40000000007b1150 in PortalRun (portal=0x60000000001a5fc8, count=9223372036854775807, isTopLevel=1 '\001',
dest=0x6000000000160f10,
---Type <return> to continue, or q <return> to quit--- altdest=0x6000000000160f10, completionTag=0x60000fffffb2e96e
"") at pquery.c:815
#8 0x400000000079eed0 in exec_simple_query ( query_string=0x60000000001603b8 "ALTER FOREIGN DATA WRAPPER foo
OPTIONS(a '1', b '2');") at postgres.c:991
#9 0x40000000007ab570 in PostgresMain (argc=6, argv=0x60000000000c2ab0, username=0x60000000000c2800 "math") at
postgres.c:3600
#10 0x40000000006dadb0 in BackendRun (port=0x60000000000e3d40) at postmaster.c:3258
#11 0x40000000006d8de0 in BackendStartup (port=0x60000000000e3d40) at postmaster.c:2872
#12 0x40000000006d1700 in ServerLoop () at postmaster.c:1283
#13 0x40000000006d0480 in PostmasterMain (argc=6, argv=0x60000000000c0d20) at postmaster.c:1031
#14 0x4000000000533e50 in main (argc=6, argv=0x60000000000c0d20) at main.c:188
(gdb)
------------------------
And the full backtrace if needed:
---------------
(gdb) bt full
#0 0x40000000000560b0 in heap_compute_data_size ( tupleDesc=0x200000000287fbb8, values=0x60000000001ae060,
isnull=0x60000000001ae0b8"") at heaptuple.c:103 val = 1761288 data_length = 78 i = 4
numberOfAttributes= 5 att = (Form_pg_attribute *) 0x200000000287fbe0
#1 0x400000000005cac0 in heap_form_tuple (tupleDescriptor=0x200000000287fbb8, values=0x60000000001ae060,
isnull=0x60000000001ae0b8"") at heaptuple.c:728 tuple = (HeapTuple) 0x60000000001adfd0 td =
(HeapTupleHeader)0x60000000001ae0a0 len = 32 data_len = 8 hoff = 32 hasnull = 1 '\001'
att = (Form_pg_attribute *) 0x200000000287fbe0 numberOfAttributes = 5 i = 5 __func__ =
"heap_form_tuple"
#2 0x400000000005d700 in heap_modify_tuple (tuple=0x600000000019e9f8, tupleDesc=0x200000000287fbb8,
replValues=0x60000fffffb2e318, replIsnull=0x60000fffffb2e340 "", doReplace=0x60000fffffb2e345 "") at
heaptuple.c:853
---Type <return> to continue, or q <return> to quit--- numberOfAttributes = 5 attoff = 5 values =
(Datum*) 0x60000000001ae060 isnull = (bool *) 0x60000000001ae0b8 "" newTuple = (HeapTuple) 0x91800195972
#3 0x4000000000399480 in AlterForeignDataWrapper (stmt=0x6000000000160e78) at foreigncmds.c:480 rel =
(Relation)0x200000000287f908 tp = (HeapTuple) 0x600000000019e9f8 repl_val = {0, 0, 0, 0, 1761288}
repl_null= "\000\000\000\000" repl_repl = "\000\000\000\000\001" fdwId = 28313 isnull = 1 '\001'
datum = 6917529027642843144 fdwlib = (ForeignDataWrapperLibrary *) 0x600000000016d7f0 __func__ =
"AlterForeignDataWrapper"
#4 0x40000000007b6bc0 in ProcessUtility (parsetree=0x6000000000160e78, queryString=0x60000000001603b8 "ALTER
FOREIGNDATA WRAPPER foo OPTIONS (a '1', b '2');", params=0x0, isTopLevel=1 '\001', dest=0x6000000000160f10,
completionTag=0x60000fffffb2e96e"") at utility.c:469 __func__ = "ProcessUtility"
#5 0x40000000007b2e20 in PortalRunUtility (portal=0x60000000001a5fc8, utilityStmt=0x6000000000160e78, isTopLevel=1
'\001',
---Type <return> to continue, or q <return> to quit--- dest=0x6000000000160f10, completionTag=0x60000fffffb2e96e "")
at pquery.c:1183 active_snapshot_set = 1 '\001' __func__ = "PortalRunUtility"
#6 0x40000000007b3380 in PortalRunMulti (portal=0x60000000001a5fc8, isTopLevel=1 '\001', dest=0x6000000000160f10,
altdest=0x6000000000160f10, completionTag=0x60000fffffb2e96e "") at pquery.c:1286 stmt = (Node *)
0x6000000000160e78 stmtlist_item = (ListCell *) 0x6000000000161120
#7 0x40000000007b1150 in PortalRun (portal=0x60000000001a5fc8, count=9223372036854775807, isTopLevel=1 '\001',
dest=0x6000000000160f10, altdest=0x6000000000160f10, completionTag=0x60000fffffb2e96e "") at pquery.c:815
save_exception_stack= (sigjmp_buf *) 0x60000fffffb2eac0 save_context_stack = (ErrorContextCallback *) 0x0
local_sigjmp_buf= {{__jmpbuf = {6917546619822073200, 2305843009217053184, 0, 2674341019140927, 0, 0, 0, 0,
4611686018435452112,0, 0, 0, 0, 0, -4611686018427386609, 0, 5832769, 6917537823729074712, 0, 6917546619822073296,
0<repeats 44 times>, 63, 4611686018440740537, 6917529027641081860, 209, 0, 4611686018440740536},
__mask_was_saved= 0, __saved_mask = {__val = {6917529027642472976, 6917529027641081861, 4611686018440948192,
6917529027641081856, 6917529027642472976, 6917529027641081860, 6917529027642527480, 0,
6917529027642022104,6917529027642022104, 80, 6917529027642527504,
---Type <return> to continue, or q <return> to quit--- 4611686018440740536, 6917529027642527584, 80,
6917529027641081856}}}} result = 0 '\0' saveTopTransactionResourceOwner = (ResourceOwner)
0x60000000001855d8 saveTopTransactionContext = (MemoryContext) 0x60000000000e59f0 saveActivePortal =
(Portal)0x0 saveResourceOwner = (ResourceOwner) 0x60000000001855d8 savePortalContext = (MemoryContext)
0x0 saveMemoryContext = (MemoryContext) 0x60000000000e59f0 __func__ = "PortalRun"
#8 0x400000000079eed0 in exec_simple_query ( query_string=0x60000000001603b8 "ALTER FOREIGN DATA WRAPPER foo
OPTIONS(a '1', b '2');") at postgres.c:991 parsetree = (Node *) 0x6000000000160e78 portal = (Portal)
0x60000000001a5fc8 snapshot_set = 0 '\0' commandTag = 0x4000000000cbe750 "ALTER FOREIGN DATA WRAPPER"
completionTag =
"\000\000Q\000\000\000\000\000\000`\234\002\000\000\000\000\003\026\000\000\000\000000\000\000\000\000\000\000?\003\026\000\000\000\000\000\000\000\017\000\000\0006\000\000\000s\001\001\000?\003\026\000\000"
querytree_list = (List *) 0x60000000001610e8 plantree_list = (List *) 0x6000000000161148 receiver =
(DestReceiver*) 0x6000000000160f10 format = 0 dest = DestRemote
---Type <return> to continue, or q <return> to quit--- oldcontext = (MemoryContext) 0x60000000000e59f0
parsetree_list= (List *) 0x6000000000160ed8 parsetree_item = (ListCell *) 0x6000000000160eb0
save_log_statement_stats= 0 '\0' was_logged = 0 '\0' isTopLevel = 1 '\001' msec_str =
"\000\000\000 000\000\000\000\000\000\000\003\026\000\000\000\000000\000\000\000\017\000 000\000\000" __func__ =
"exec_simple_query"
#9 0x40000000007ab570 in PostgresMain (argc=6, argv=0x60000000000c2ab0, username=0x60000000000c2800 "math") at
postgres.c:3600 query_string = 0x60000000001603b8 "ALTER FOREIGN DATA WRAPPER foo OPTIONS (a '1', b '2');"
flag= -1 dbname = 0x60000000000e3450 "regression" userDoption = 0x0 secure = 0 '\0' errs =
0 debug_flag = -1 guc_names = (List *) 0x0 guc_values = (List *) 0x0 ctx = PGC_SUSET
gucsource= PGC_S_CLIENT am_superuser = 1 '\001'
---Type <return> to continue, or q <return> to quit--- firstchar = 81 stack_base = 0 '\0'
input_message= { data = 0x60000000001603b8 "ALTER FOREIGN DATA WRAPPER foo OPTIONS (a '1', b '2');", len = 55, maxlen
=1024, cursor = 55} local_sigjmp_buf = {{__jmpbuf = {6917546619822074288, 2305843009217053184, 0,
2674341019124543,0, 0, 0, 0, 4611686018435427728, 0, 0, 0, 0, 0, -4611686018427386867, 0, 5832769,
6917537823729074568,0, 6917546619822074560, 0 <repeats 40 times>, 6917546619822075412, 6917546619822075420,
6917546619822075443, 6917546619822075412, 6917546619822075443, 0, 0, 0, 0, 0}, __mask_was_saved = 1,
__saved_mask= {__val = {18446744066192964099, 6917529027641081856, 0, 6917529027641899184, 352, 512,
6917529027641898808,58, 6752, 8144, 6917529027641899344, 5, 6917529027641873152, 81604378636, 0,
6917529027641081861}}}} send_ready_for_query = 0 '\0' __func__ = "PostgresMain"
#10 0x40000000006dadb0 in BackendRun (port=0x60000000000e3d40) at postmaster.c:3258 av = (char **)
0x60000000000c2ab0 maxac = 27 ac = 6 secs = 283072531 usecs = 545137
---Type <return> to continue, or q <return> to quit--- protobuf = "-v196608\000\000\000@
(\f\000\000\000\000 000(\f\000\000\000\000?\000\000" i = 6 __func__ = "BackendRun"
#11 0x40000000006d8de0 in BackendStartup (port=0x60000000000e3d40) at postmaster.c:2872 bn = (Backend *)
0x60000000000e3410 pid = 0 __func__ = "BackendStartup"
#12 0x40000000006d1700 in ServerLoop () at postmaster.c:1283 port = (Port *) 0x60000000000e3d40 i = 0
rmask = {fds_bits = {16, 0 <repeats 15 times>}} selres = 1 readmask = {fds_bits = {16, 0 <repeats 15
times>}} nSockets = 5 now = 1229757331 last_touch_time = 1229757314 __func__ = "ServerLoop"
#13 0x40000000006d0480 in PostmasterMain (argc=6, argv=0x60000000000c0d20) at postmaster.c:1031 opt = -1
status = 0 userDoption = 0x60000000000c0e00 "/home/math/pgsql_cvs/pgsql/src/test/re---Type <return> to
continue,or q <return> to quit---
gress/./tmp_check/data" i = 64 __func__ = "PostmasterMain"
#14 0x4000000000533e50 in main (argc=6, argv=0x60000000000c0d20) at main.c:188
No locals.
-----------------
Regards, Sergey
*******************************************************************
Sergey E. Koposov
Max Planck Institute for Astronomy/Sternberg Astronomical Institute
Tel: +49-6221-528-349
Web: http://lnfm1.sai.msu.ru/~math
E-mail: math@sai.msu.ru
Re: [COMMITTERS] pgsql: SQL/MED catalog manipulation facilities This doesn't do any
От
Heikki Linnakangas
Дата:
Tom Lane wrote: > petere@postgresql.org (Peter Eisentraut) writes: >> Log Message: >> ----------- >> SQL/MED catalog manipulation facilities > > The buildfarm says this patch has got serious portability issues. > > grebe, dugong, and ermine are all crashing here: > > --- 109,115 ---- > (3 rows) > > ALTER FOREIGN DATA WRAPPER foo OPTIONS (a '1', b '2'); > ! server closed the connection unexpectedly > ! This probably means the server terminated abnormally > ! before or while processing the request. > ! connection to server was lost > > (Unfortunately I can't reproduce this on any of my own machines.) Neither can I. All the non-Windows crashes in the buildfarm seem to be on 64-bit machines, although not all 64-bit members crash. The callers of transformGenericOptions-function in foreigncmds.c use OidIsValid to check the return value of transformGenericOptions, but it returns an array, not an Oid. I committed a fix for that, we'll see if it heals the buildfarm. -- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com
Re: [COMMITTERS] pgsql: SQL/MED catalog manipulation facilities This doesn't do any
От
Alvaro Herrera
Дата:
Tom Lane wrote: > petere@postgresql.org (Peter Eisentraut) writes: > > Log Message: > > ----------- > > SQL/MED catalog manipulation facilities > > The buildfarm says this patch has got serious portability issues. FWIW I recently learned that you can set MALLOC_PERTURB_ to a nonzero value to cause glibc's malloc() to fill memory with the byte value therein, and clear it with the bitwise inverse on free. This could be helpful to find cases of uninitialized memory usage. -- Alvaro Herrera http://www.CommandPrompt.com/ PostgreSQL Replication, Consulting, Custom Development, 24x7 support
Re: [COMMITTERS] pgsql: SQL/MED catalog manipulation facilities This doesn't do any
От
Martin Pihlak
Дата:
Heikki Linnakangas wrote:
> The callers of transformGenericOptions-function in foreigncmds.c use
> OidIsValid to check the return value of transformGenericOptions, but it
> returns an array, not an Oid. I committed a fix for that, we'll see if
> it heals the buildfarm.
>
Thanks for that. There was an additional use of the Datum as Oid in
AlterForeignDataWrapper, namely the assignment to repl_vals.
Another bug was uncovered on orca -- a missing semicolon in gram.y:3004
Fixes attached.
regards,
Martin
*** a/src/backend/commands/foreigncmds.c
--- b/src/backend/commands/foreigncmds.c
***************
*** 466,472 **** AlterForeignDataWrapper(AlterFdwStmt *stmt)
NULL, fdwlib->validateOptionList);
if (PointerIsValid(DatumGetPointer(datum)))
! repl_val[Anum_pg_foreign_data_wrapper_fdwoptions - 1] = ObjectIdGetDatum(datum);
else
repl_null[Anum_pg_foreign_data_wrapper_fdwoptions - 1] = true;
--- 466,472 ----
NULL, fdwlib->validateOptionList);
if (PointerIsValid(DatumGetPointer(datum)))
! repl_val[Anum_pg_foreign_data_wrapper_fdwoptions - 1] = datum;
else
repl_null[Anum_pg_foreign_data_wrapper_fdwoptions - 1] = true;
*** a/src/backend/parser/gram.y
--- b/src/backend/parser/gram.y
***************
*** 3001,3007 **** CreateUserMappingStmt: CREATE USER MAPPING FOR auth_ident SERVER name create_gen
auth_ident:
CURRENT_USER { $$ = "current_user"; }
| USER { $$ = "current_user"; }
! | RoleId { $$ = (strcmp($1, "public") == 0) ? NULL : $1 }
;
/*****************************************************************************
--- 3001,3007 ----
auth_ident:
CURRENT_USER { $$ = "current_user"; }
| USER { $$ = "current_user"; }
! | RoleId { $$ = (strcmp($1, "public") == 0) ? NULL : $1; }
;
/*****************************************************************************
Martin Pihlak <martin.pihlak@gmail.com> writes:
> Thanks for that. There was an additional use of the Datum as Oid in
> AlterForeignDataWrapper, namely the assignment to repl_vals.
Ah, I just finished rediscovering that myself.
> Another bug was uncovered on orca -- a missing semicolon in gram.y:3004
Yeah, newer bisons are pickier about that. Will apply.
regards, tom lane