Delivery-Date: Sun Sep 21 08:57:14 2008
Message-Id: <200809211559.m8LFxO5R000642@bourbon.usc.edu>
To: cs551@merlot.usc.edu
Subject: Re: doubt regarding signal and wait
Date: Sun, 21 Sep 2008 08:59:23 -0700
From: Bill Cheng <william@bourbon.usc.edu>

Someone wrote:

  > on the page 82 of "pthreads programming" it says,
  > "If all waiting threads are of the same priority, they are released in
  > a first-in first-out order for each pthread_cond_signal call that's
  > issued."
  > 
  > So I guess it should be in FIFO fashion. (I agree that we should write
  > code to make sure that it works under any order though.)

Hmm...  That's not how I remembered it!  Clearly, I was
wrong.  Thanks for pointing out what's in the book!

But since the book is really not the pthread spec, I would
not be so sure if Solaris implements everything according
to the book.
--
Bill Cheng // bill.cheng@usc.edu <URL:http://merlot.usc.edu/william/usc/>


  On Sat, Sep 20, 2008 at 8:32 PM, Bill Cheng <william@bourbon.usc.edu> wrote:
  > Someone wrote:
  >
  >  > In what fashion does the signal wake up the waiting thread.Is it FIFO or is
  >  > it OS Dependent ?
  >
  > It's unspecified.  So, it may be OS dependent.  Or it may
  > depend on other stuff that you have no control of.
  >
  >  > In my case,
  >  > Server 1 is started first and than Server 2.
  >  > Since initially Queue is empty, both will sleep for some customer to arrive
  >  > in the Queue.
  >  >
  >  > But Server 2 (and not Server 1) gets chance to serve the first Customer.
  >
  > You should write code to make sure that it works under
  > any order since you cannot predict which thread will
  > wake up first.
  > --
  > Bill Cheng // bill.cheng@usc.edu <URL:http://merlot.usc.edu/william/usc/>