Thread: pg_control_init() bug

Hello,

I noticed that pg_control_init() is failing an assertion on 13devel:

    TRAP: FailedAssertion("(atti->attalign) == 's'", File: "heaptuple.c", Line: 160)
    postgres: bossartn postgres [local] SELECT(ExceptionalCondition+0x55)[0x8e02b5]
    postgres: bossartn postgres [local] SELECT(heap_compute_data_size+0x224)[0x486eb4]
    postgres: bossartn postgres [local] SELECT(heap_form_tuple+0x82)[0x488262]
    postgres: bossartn postgres [local] SELECT(pg_control_init+0x259)[0x902ca9]
    postgres: bossartn postgres [local] SELECT(ExecMakeTableFunctionResult+0x345)[0x656a65]
    postgres: bossartn postgres [local] SELECT[0x667e03]
    postgres: bossartn postgres [local] SELECT(ExecScan+0x3ab)[0x6572db]
    postgres: bossartn postgres [local] SELECT(standard_ExecutorRun+0x137)[0x64f097]
    postgres: bossartn postgres [local] SELECT[0x7c5d8b]
    postgres: bossartn postgres [local] SELECT(PortalRun+0x1d1)[0x7c71c1]
    postgres: bossartn postgres [local] SELECT[0x7c34d1]
    postgres: bossartn postgres [local] SELECT(PostgresMain+0x624)[0x7c4aa4]
    postgres: bossartn postgres [local] SELECT[0x74109a]
    postgres: bossartn postgres [local] SELECT(main+0x440)[0x69f220]
    /lib64/libc.so.6(__libc_start_main+0xfd)[0x7fbb78309d1d]
    postgres: bossartn postgres [local] SELECT[0x47b039]
    2019-12-10 20:02:03.100 UTC [20859] LOG:  server process (PID 20868) was terminated by signal 6: Aborted
    2019-12-10 20:02:03.100 UTC [20859] DETAIL:  Failed process was running: SELECT * FROM pg_control_init();
    2019-12-10 20:02:03.100 UTC [20859] LOG:  terminating any other active server processes

The attached patch seems to clear this up.  I think this was missed in
2e4db241.

Nathan

"Bossart, Nathan" <bossartn@amazon.com> writes:
> I noticed that pg_control_init() is failing an assertion on 13devel:

Hmm, yeah.  In a non-assert build I get

regression=# select * from pg_control_init();
ERROR:  function return row and query-specified return row do not match
DETAIL:  Returned row contains 12 attributes, but query expects 11.

> The attached patch seems to clear this up.  I think this was missed in
> 2e4db241.

Evidently.  Thanks for the report!

            regards, tom lane

On Tue, Dec 10, 2019 at 05:46:48PM -0500, Tom Lane wrote:
> "Bossart, Nathan" <bossartn@amazon.com> writes:
> > I noticed that pg_control_init() is failing an assertion on 13devel:
> 
> Hmm, yeah.  In a non-assert build I get
> 
> regression=# select * from pg_control_init();
> ERROR:  function return row and query-specified return row do not match
> DETAIL:  Returned row contains 12 attributes, but query expects 11.
> 
> > The attached patch seems to clear this up.  I think this was missed in
> > 2e4db241.
> 
> Evidently.  Thanks for the report!

FYI, this has been applied:

    Fix tuple column count in pg_control_init().
    
    Oversight in commit 2e4db241b.
    
    Nathan Bossart
    
    Discussion: https://postgr.es/m/1B616360-396A-4482-AA28-375566C86160@amazon.com
    
    Branch
    ------
    master
    
    Details
    -------
    https://git.postgresql.org/pg/commitdiff/8729fa72483f8a9acf299508bb2cbae1aa9a29b8
    
    Modified Files
    --------------
    src/backend/utils/misc/pg_controldata.c | 2 +-
    1 file changed, 1 insertion(+), 1 deletion(-)

-- 
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

+ As you are, so once was I.  As I am, so you will be. +
+                      Ancient Roman grave inscription +