Thread: ECPG regression test failures on Solaris 10/x86_64 with Sun studio 11

On solaris 10U2 using the Sun compiler we get segmentation faults in
nearly all of the ECPG regression tests on a 64bit built. the resulting
cores look similiar to:


Current function is ECPGget_variable  91           var->pointer = va_arg(APREF, char *);
(dbx) where
=>[1] ECPGget_variable(ap = 0xfffffd7fffdff6d0, type = ECPGt_char, var =
0x41d9d0, indicator = '\001'), line 91 in "execute.c" [2] create_statement(lineno = 28, compat = 0, force_indicator =
1,
connection = 0x413210, stmt = 0xfffffd7fffdff818, query = 0x4013e8
"select  current_database ()     ", ap = 0xfffffd7fffdff828), line 192
in "execute.c" [3] ECPGdo(lineno = 28, compat = 0, force_indicator = 1,
connection_name = (nil), query = 0x4013e8 "select  current_database ()  ", ... = 0x1, ...), line 1548 in "execute.c"
[4]main(), line 30 in "test2.pgc"
 


Stefan

What version of PG do you use  and what is your ./configure options?
Zdenek

Stefan Kaltenbrunner napsal(a):
> On solaris 10U2 using the Sun compiler we get segmentation faults in
> nearly all of the ECPG regression tests on a 64bit built. the resulting
> cores look similiar to:
> 
> 
> Current function is ECPGget_variable
>    91           var->pointer = va_arg(APREF, char *);
> (dbx) where
> =>[1] ECPGget_variable(ap = 0xfffffd7fffdff6d0, type = ECPGt_char, var =
> 0x41d9d0, indicator = '\001'), line 91 in "execute.c"
>   [2] create_statement(lineno = 28, compat = 0, force_indicator = 1,
> connection = 0x413210, stmt = 0xfffffd7fffdff818, query = 0x4013e8
> "select  current_database ()     ", ap = 0xfffffd7fffdff828), line 192
> in "execute.c"
>   [3] ECPGdo(lineno = 28, compat = 0, force_indicator = 1,
> connection_name = (nil), query = 0x4013e8 "select  current_database ()
>    ", ... = 0x1, ...), line 1548 in "execute.c"
>   [4] main(), line 30 in "test2.pgc"
> 
> 
> Stefan
> 
> ---------------------------(end of broadcast)---------------------------
> TIP 9: In versions below 8.0, the planner will ignore your desire to
>        choose an index scan if your joining column's datatypes do not
>        match

Zdenek Kotala wrote:
> What version of PG do you use  and what is your ./configure options?

This is on -HEAD and with the following box and setup:

http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=clownfish&dt=2007-01-10%2014:18:23

one does not need the --with-perl and --with-python flags to trigger
that - ./configure --enable-integer-datetimes --without-readline
-enable-nls --enable-debug --enable-cassert is enough for me.

now that tom commited the BYTE_ORDER fix I will force a run on clownfish
to get a proper report up.


Stefan