Re: Segfault in pg_stat_activity

Hi Peter, Magnus,


On 2015-07-27 21:20:33 +0200, Peter J. Holzer wrote:
> Since I've installed 9.5alpha1 a week ago, I've seen several segfaults
> in the log file:

Thanks for reporting the issue. I guess you have several ssl enabled
connections?

> Core was generated by `postgres: postgres template1 [local] SELECT                                   '.
> Program terminated with signal SIGSEGV, Segmentation fault.
> #0  0x00007fb94d314289 in memcpy (__len=200, __src=0x7fb948719ad8, __dest=0x7fb94f4c1bb0) at
/usr/include/x86_64-linux-gnu/bits/string3.h:51
> 51        return __builtin___memcpy_chk (__dest, __src, __len, __bos0 (__dest));
> (gdb) bt
> #0  0x00007fb94d314289 in memcpy (__len=200, __src=0x7fb948719ad8, __dest=0x7fb94f4c1bb0) at
/usr/include/x86_64-linux-gnu/bits/string3.h:51
> #1  pgstat_read_current_status () at
/tmp/buildd/postgresql-9.5-9.5~alpha1/build/../src/backend/postmaster/pgstat.c:3000
> #2  0x00007fb94d318db1 in pgstat_read_current_status () at
/tmp/buildd/postgresql-9.5-9.5~alpha1/build/../src/backend/postmaster/pgstat.c:2434

Uh. I think I see what's going on here. Magnus, could it be that you
didn't intend the += for localsslstatus?:
        /* Only valid entries get included into the local array */
        if (localentry->backendStatus.st_procpid > 0)
        {
            BackendIdGetTransactionIds(i,
                                       &localentry->backend_xid,
                                       &localentry->backend_xmin);

            localentry++;
            localappname += NAMEDATALEN;
            localactivity += pgstat_track_activity_query_size;
#ifdef USE_SSL
            localsslstatus += sizeof(PgBackendSSLStatus);
#endif
            localNumBackends++;
        }

because that'll forward localsslstatus not one, but uh, something like
200.

I think just replacing the above += with 1 might fix the issue. Can't
test just now.

Greetings,

Andres Freund