Re: Segfault in Ruby/ODBC

Nathaniel Talbott [mailto:[e-mail]] wrote:

> My program is causing a segfault under load, and while I'm 
> going to dig in to figuring out why, I thought I'd go ahead 
> and see if anyone here could give me a hand. It appears that 
> the Ruby/ODBC extension is somehow behaving badly, and not 
> being a C guy, I'd appreciate any help you can give me.
> 
> The backtrace below was generated off of an optimized ruby 
> without debugging info... I may be able to do it again and 
> get better information, but not immediately. Also, the 
> program itself is large and proprietary, and as with most 
> SEGFAULTs, I can't reproduce the problem except with the full 
> program under load.

OK, I have better data now. First of all, I have a small program that will
reproduce the problem, provided you have access to an odbc database:

  require 'dbi'

  threads = []
  Thread.abort_on_exception = true
  10.times do |i|
    threads << Thread.new do
      db = DBI::connect("dbi:ODBC:dbname", "user", "password")
      loop do
        p "Looping #{i}..."
        db.tables
        db.execute("SELECT COUNT(*) FROM table").finish
        db.tables
        db.execute("SELECT * FROM table").finish
      end
    end
  end
  threads.each{|t| t.join}

Obviously you'd need to replace dbname, user, password and table with things
that make sense for your setup. I'm using unixODBC to connect via FreeTDS to
a MSSQL server on a Windows NT box. 

Also, I've recompiled Ruby and the odbc driver with debugging information,
and have a better backtrace:

  (gdb) run
  Starting program: /usr/bin/ruby t.rb
  [New Thread 16384 (LWP 16080)]
  "Looping 0..."
  "Looping 2..."
  "Looping 1..."

  Program received signal SIGSEGV, Segmentation fault.
  [Switching to Thread 16384 (LWP 16080)]
  0x400de1f5 in mallopt () from /lib/libc.so.6
  (gdb) bt
  #0  0x400de1f5 in mallopt () from /lib/libc.so.6
  #1  0x400dd383 in malloc () from /lib/libc.so.6
  #2  0x08070cf6 in ruby_xmalloc (size=16) at gc.c:108
  #3  0x080bf370 in st_init_table_with_size (type=0x81000dc, size=11) at
st.c:167
  #4  0x080bf3c8 in st_init_table (type=0x81000dc) at st.c:180
  #5  0x08073c87 in hash_alloc (klass=1075592068) at hash.c:188
  #6  0x08073cb8 in rb_hash_new () at hash.c:196
  #7  0x4022ec69 in do_fetch (q=0x819d6d0, mode=1) at odbc.c:3802
  #8  0x4022fce4 in stmt_fetch_hash1 (argc=1, argv=0xbfffabb0,
self=1075690112, bang=0)
      at odbc.c:4206
  #9  0x40230106 in stmt_each_hash (argc=0, argv=0x0, self=1075690112) at
odbc.c:4345
  #10 0x0806adb3 in call_cfunc (func=0x40230030 <stmt_each_hash>,
recv=1075690112, len=-1, argc=0,
      argv=0x0) at eval.c:4841
  #11 0x0805dc43 in rb_call0 (klass=1075735712, recv=1075690112, id=14073,
oid=14073, argc=0,
      argv=0x0, body=0x401e2164, nosuper=0) at eval.c:4975
  #12 0x0805e788 in rb_call (klass=1075735712, recv=1075690112, mid=14073,
argc=0, argv=0x0, scope=0)
      at eval.c:5196
  #13 0x0805866a in rb_eval (self=1075693732, n=0x401f75a0) at eval.c:3026
  #14 0x08057438 in rb_eval (self=1075693732, n=0x401f7a78) at eval.c:2753
  #15 0x08057979 in rb_eval (self=1075693732, n=0x401f6204) at eval.c:2839
  #16 0x0805e23f in rb_call0 (klass=1075705512, recv=1075693732, id=10217,
oid=10217, argc=0,
      argv=0x0, body=0x401f6204, nosuper=0) at eval.c:5103
  #17 0x0805e788 in rb_call (klass=1075705512, recv=1075693732, mid=10217,
argc=0, argv=0x0, scope=0)
      at eval.c:5196
  #18 0x0805866a in rb_eval (self=1075695672, n=0x401ab2dc) at eval.c:3026
  #19 0x0805e23f in rb_call0 (klass=1075842852, recv=1075695672, id=10217,
oid=10217, argc=0,
      argv=0x0, body=0x401ab2dc, nosuper=0) at eval.c:5103
  #20 0x0805e788 in rb_call (klass=1075842852, recv=1075695672, mid=10217,
argc=0, argv=0x0, scope=0)
      at eval.c:5196
  #21 0x0805866a in rb_eval (self=1075620448, n=0x401baf98) at eval.c:3026
  #22 0x0805c050 in rb_yield_0 (val=6, self=1075620448, klass=0, flags=0,
avalue=0) at eval.c:4229
  #23 0x0805c432 in rb_f_loop () at eval.c:4330
  #24 0x0806adcd in call_cfunc (func=0x805c418 <rb_f_loop>, recv=1075620448,
len=0, argc=0, argv=0x0)
      at eval.c:4844
  #25 0x0805dc43 in rb_call0 (klass=1075625128, recv=1075620448, id=3921,
oid=3921, argc=0,
      argv=0x0, body=0x401ca1a0, nosuper=0) at eval.c:4975
  #26 0x0805e788 in rb_call (klass=1075625128, recv=1075620448, mid=3921,
argc=0, argv=0x0, scope=1)
      at eval.c:5196
  #27 0x08058924 in rb_eval (self=1075620448, n=0x401bad7c) at eval.c:3041
  #28 0x08057438 in rb_eval (self=1075620448, n=0x401bad04) at eval.c:2753
  #29 0x0805c050 in rb_yield_0 (val=1075838972, self=1075620448, klass=0,
flags=2, avalue=2)
      at eval.c:4229
  #30 0x08068e60 in rb_thread_yield (arg=1075838972, th=0x816efd8) at
eval.c:9581
  #31 0x08068b04 in rb_thread_start_0 (fn=0x8068d94 <rb_thread_yield>,
arg=0x401ffffc, th=0x816efd8)
      at eval.c:9495
  #32 0x08068f53 in rb_thread_initialize (thread=1075839612,
args=1075838972) at eval.c:9610
  #33 0x0806ad94 in call_cfunc (func=0x8068f04 <rb_thread_initialize>,
recv=1075839612, len=-2,
      argc=0, argv=0x0) at eval.c:4838
  #34 0x0805dc43 in rb_call0 (klass=1075609928, recv=1075839612, id=2953,
oid=2953, argc=0,
      argv=0x0, body=0x401c80e4, nosuper=0) at eval.c:4975
  #35 0x0805e788 in rb_call (klass=1075609928, recv=1075839612, mid=2953,
argc=0, argv=0x0, scope=1)
      at eval.c:5196
  #36 0x0805e9ba in rb_funcall2 (recv=1075839612, mid=2953, argc=0,
argv=0x0) at eval.c:5271
  #37 0x0806124c in rb_obj_call_init (obj=1075839612, argc=0, argv=0x0) at
eval.c:6341
  #38 0x08068eb4 in rb_thread_s_new (argc=0, argv=0x0, klass=1075609928) at
eval.c:9594
  #39 0x0806adb3 in call_cfunc (func=0x8068e74 <rb_thread_s_new>,
recv=1075609928, len=-1, argc=0,
      argv=0x0) at eval.c:4841
  #40 0x0805dc43 in rb_call0 (klass=1075609908, recv=1075609928, id=3353,
oid=3353, argc=0,
      argv=0x0, body=0x401c810c, nosuper=0) at eval.c:4975
  #41 0x0805e788 in rb_call (klass=1075609908, recv=1075609928, mid=3353,
argc=0, argv=0x0, scope=0)
      at eval.c:5196
  #42 0x0805866a in rb_eval (self=1075620448, n=0x401bb254) at eval.c:3026
  #43 0x08057438 in rb_eval (self=1075620448, n=0x401bacf0) at eval.c:2753
  #44 0x08058506 in rb_eval (self=1075620448, n=0x401bacb4) at eval.c:3021
  #45 0x0805c050 in rb_yield_0 (val=1, self=1075620448, klass=0, flags=0,
avalue=0) at eval.c:4229
  #46 0x0805c306 in rb_yield (val=1) at eval.c:4282
  #47 0x08087af3 in int_dotimes (num=21) at numeric.c:1775
  #48 0x0806adcd in call_cfunc (func=0x8087aac <int_dotimes>, recv=21,
len=0, argc=0, argv=0x0)
      at eval.c:4844
  #49 0x0805dc43 in rb_call0 (klass=1075606888, recv=21, id=5385, oid=5385,
argc=0, argv=0x0,
      body=0x401c74a0, nosuper=0) at eval.c:4975
  #50 0x0805e788 in rb_call (klass=1075606888, recv=21, mid=5385, argc=0,
argv=0x0, scope=0)
      at eval.c:5196
  #51 0x0805866a in rb_eval (self=1075620448, n=0x401bb2e0) at eval.c:3026
  #52 0x08057438 in rb_eval (self=1075620448, n=0x401bb3bc) at eval.c:2753
  #53 0x08053a80 in eval_node (self=1075620448, node=0x401bb3bc) at
eval.c:1217
  #54 0x0805408a in ruby_exec () at eval.c:1387
  #55 0x0805410d in ruby_run () at eval.c:1408
  #56 0x08051f93 in main (argc=2, argv=0xbffff704, envp=0xbffff710) at
main.c:50
  (gdb) c
  Continuing.
  /usr/lib/ruby/site_ruby/1.8/DBD/ODBC/ODBC.rb:124: [BUG] Segmentation fault
  ruby 1.8.1 (2003-10-31) [i386-linux]


  Program received signal SIGABRT, Aborted.
  0x400955f1 in kill () from /lib/libc.so.6

I'm going to dig in to this and see if I can figure out what the problem is,
but any help would be appreciated. Let me know if there's any more
information I can provide that would be useful.

Thanks,


Nathaniel

<:((><