From: "Xavier Noëlle" <xavier.noelle@...>
Date: 2010-01-21T02:10:38+09:00
Subject: Multi-threaded DBI causing segfault

Hello,
I'm new to this mailing list, so do not hesitate to tell me if
everything's ok with my post.

I'm using DBI to connect to a MySQL database. I tried to reduce my
code as much as possible, knowing that:
 - constants are valid (not '****') and lead to valid databases, valid
tables and the select returns a valid result
 - removing the finish() does not solve the problem
 - limiting the result size (eg: 'LIMIT 0, 30') does not solve the problem
 - NB_THREADS is set to 50 to make it crash quickly (but, starting
from 2, any value makes it crash)
 - as you can see below, I'm using Ruby 1.8.7 on Ubuntu 9.10 64 bits
 - the file and line number in which the segmentation fault is
reported is variable (sometimes, it happens my own files)

The error output is as follow:
/usr/lib/ruby/1.8/dbd/mysql/statement.rb:36: [BUG] Segmentation fault
ruby 1.8.7 (2009-06-12 patchlevel 174) [x86_64-linux]

Question: is it a real bug or am I doing something wrong ?
If I'm wrong, where ?
If I'm not, is there a workaround to thread the execution of code
using DBI without a segfault ?

Thanks for any hint about my problem :-)

=================
#! /usr/bin/env ruby

require 'rubygems'
require 'dbi'

PRODUCTS_DB = '*****'

PRODUCTS_DB_HOST = '*****'
PRODUCTS_DB_USER = '*****'
PRODUCTS_DB_PASS = '****'

def getDbh()
  return DBI.connect("DBI:Mysql:#{PRODUCTS_DB}:"+
                     "#{PRODUCTS_DB_HOST}", PRODUCTS_DB_USER,
                     PRODUCTS_DB_PASS)
end

NB_THREADS = 50
threads = Array.new(NB_THREADS)

NB_THREADS.times() do |i|
  puts i
  threads[i] = Thread.new() do
    dbh = getDbh()

    while true
      hdl = dbh.execute('SELECT * FROM ****** LIMIT 0, 30') do |dummy|
        dummy[0]
      end
      hdl.finish()
      puts 'There'
    end
  end
end

threads.each() do |t|
  t.join()
end
=================

-- 
Xavier NOELLE