Re: Segmentation fault while COPY in 7.3

Yes, we found a double pfree in 7.3.  There will be a fix in 7.3.1.

---------------------------------------------------------------------------

Medi Montaseri wrote:
> Something is funny about this pfree()....here is a similar crash on a 7.2.3
>
> (gdb)
> #0  0x00420874 in heap_freetuple ()
> #1  0x004a8390 in acquire_sample_rows ()
> #2  0x004a75c8 in analyze_rel ()
> #3  0x0049f690 in vacuum ()
> #4  0x005585d8 in ProcessUtility ()
> #5  0x00553c78 in pg_exec_query_string ()
> #6  0x00555850 in PostgresMain ()
> #7  0x00524660 in DoBackend ()
> #8  0x00523d08 in BackendStartup ()
> #9  0x00521c18 in ServerLoop ()
> #10 0x005211c8 in PostmasterMain ()
> #11 0x004df3d8 in main ()
> #12 0x2ae34928 in __libc_start_main () from /lib/libc.so.6
> (gdb)
>
>
>
> Nicolai Tufar wrote:
>
> >Help!
> >Backend crashes on any COPY command I issue:
> >PostgreSQL is 7.3 release, is compiled from sources.
> >The message I get is:
> >
> >apb=# COPY maras2.mrk_yyn (bsvr_no, yyn_tur, yyn_no, yyn_syf, yyn_trh) TO
> >stdout;
> >server closed the connection unexpectedly
> >        This probably means the server terminated abnormally
> >        before or while processing the request.
> >The connection to the server was lost. Attempting reset: Failed.
> >!#
> >
> >
> >The output of strace on backend is following:
> >
> ># strace -p 18883
> >recv(7, "QCOPY maras2.mrk_yyn (bsvr_no, y"..., 8192, 0) = 77
> >gettimeofday({1038713111, 629131}, NULL) = 0
> >_llseek(11, 0, [0], SEEK_SET)           = 0
> >read(11, "\0\0\0\0\20\0\0\0\1\0\0\0\24\0\360\37\360\37\1 b1\5\0\1"..., 8192)
> >= 8192
> >read(11, "\0\0\0\0x#\375\"\20\0\0\0008\0h\35\360\37\1 \370\235\220"...,
> >8192) = 8192
> >_llseek(12, 0, [0], SEEK_SET)           = 0
> >read(12, "\0\0\0\0\0#\375\"\20\0\0\0008\0@\34\0 \1 \200\237\0\1\0"..., 8192)
> >= 8192
> >_llseek(21, 40960, [40960], SEEK_SET)   = 0
> >read(21, "\0\0\0\0\4\251\3#\20\0\0\0004\4p\17\360\37\1 \340\237 "..., 8192)
> >= 8192
> >_llseek(22, 106496, [106496], SEEK_SET) = 0
> >read(22, "\0\0\0\0(\326\377\"\20\0\0\0\10\1\200\1\0 \1 \200\237\0"..., 8192)
> >= 8192
> >read(17, "\0\0\0\0\20\0\0\0\1\0\0\0\360\0\20\1\0 \1 p\237 \1\340"..., 8192)
> >= 8192
> >open("/data/pgsql/base/2015749/2015757", O_RDWR|O_LARGEFILE) = 39
> >_llseek(39, 0, [20037632], SEEK_END)    = 0
> >_llseek(39, 0, [0], SEEK_SET)           = 0
> >read(39, "\0\0\0\0\270\7\4#\20\0\0\0\300\1\350\1\0 \1 \270\237\216"...,
> >8192) = 8192
> >--- SIGSEGV (Segmentation fault) ---
> >+++ killed by SIGSEGV +++
> >#
> >
> >I have recompiled postgres with debug information enabled, run gdb, attached
> >to backend process
> >and caught SIGSEGV as following:
> >
> >(gdb) continue
> >Continuing.
> >
> >Program received signal SIGSEGV, Segmentation fault.
> >0x081697bf in pfree (pointer=0x828a7b0) at mcxt.c:480
> >480             (*header->context->methods->free_p) (header->context,
> >pointer);
> >(gdb)
> >
> >
> >The output of bt is:
> >
> >(gdb) bt
> >#0  0x081697bf in pfree (pointer=0x828a7b0) at mcxt.c:480
> >#1  0x080b37be in CopyTo (rel=0x402be088, attnumlist=0x828a3f8, binary=0
> >'\0', oids=0 '\0',
> >    fp=0x0, delim=0x81c2a57 "\t", null_print=0x81b2c22 "\\N") at copy.c:671
> >#2  0x080b32f6 in DoCopy (stmt=0x827b398) at copy.c:491
> >#3  0x08118a6a in pg_exec_query_string (query_string=0x827af48, dest=Remote,
> >    parse_context=0x8245378) at postgres.c:789
> >#4  0x08119b89 in PostgresMain (argc=4, argv=0xbfffd2a0, username=0x8240a81
> >"postgres")
> >    at postgres.c:2016
> >#5  0x08101d7c in DoBackend (port=0x8240950) at postmaster.c:2293
> >#6  0x081016c2 in BackendStartup (port=0x8240950) at postmaster.c:1915
> >#7  0x08100875 in ServerLoop () at postmaster.c:1000
> >#8  0x08100326 in PostmasterMain (argc=1, argv=0x82270c0) at
> >postmaster.c:779
> >#9  0x080debab in main (argc=1, argv=0xbfffdc34) at main.c:210
> >#10 0x42017589 in __libc_start_main () from /lib/i686/libc.so.6
> >(gdb)
> >
> >
> >
> >Thanks in advance,
> >Nic
> >
> >
> >---------------------------(end of broadcast)---------------------------
> >TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
> >
> >
>
>
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 5: Have you checked our extensive FAQ?
>
> http://www.postgresql.org/users-lounge/docs/faq.html
>

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073