Re: BUG #14180: Segmentation fault on replication slave

Hi,

thanks for reporting this issue.

On 2016-06-07 09:16:18 +0000, boa(at)neogrid(dot)dk wrote:
> Program received signal SIGSEGV, Segmentation fault.
> _bt_restore_page (page=0x7f816fce2b40 "", from=0x55a0945abb70 "\036",
> len=<optimized out>) at
> /build/postgresql-9.5-xp9utH/postgresql-9.5-9.5.3/build/../src/backend/access/nbtree/nbtxlog.c:57
> 57
> /build/postgresql-9.5-xp9utH/postgresql-9.5-9.5.3/build/../src/backend/access/nbtree/nbtxlog.c:
> No such file or directory.
> (gdb) bt
> #0 _bt_restore_page (page=0x7f816fce2b40 "", from=0x55a0945abb70 "\036",
> len=<optimized out>) at
> /build/postgresql-9.5-xp9utH/postgresql-9.5-9.5.3/build/../src/backend/access/nbtree/nbtxlog.c:57
> #1 0x0000000000000000 in ?? ()
> (gdb) p from
> $1 = 0x55a0945abb70 "\036"
> (gdb) p end
> $2 = 0x55a0945ac928 "\305cO"
> (gdb) p i
> $3 = 3324
> (gdb) p len
> $4 = <optimized out>
> (gdb) p &itupdata
> $5 = (IndexTupleData *) 0x7ffe83ea84e0
> (gdb) p items
> $6 = {0x0 <repeats 408 times>}
> (gdb) p &items
> $7 = (Item (*)[408]) 0x7ffe83ea8820
> (gdb) p itemsz
> $8 = <optimized out>

Uhm, this is a bit odd. There's no backtrace, but types and such are
known? I guess you do have the debug symbols installed?

Greetings,

Andres Freund