This is the mail archive of the pthreads-win32@sources.redhat.com mailing list for the pthreas-win32 project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

RE: New pthread_once implementation



> -----Original Message-----
> From: pthreads-win32-owner@sources.redhat.com [mailto:pthreads-win32-
> owner@sources.redhat.com] On Behalf Of Ross Johnson
> Sent: Friday, May 27, 2005 11:48 PM
> To: Vladimir Kliatchko
> Cc: 'Gottlob Frege'; Pthreads-Win32 list
> Subject: RE: New pthread_once implementation
> 
> On Fri, 2005-05-27 at 21:30 -0400, Vladimir Kliatchko wrote:
> > Nice catch. Let me see if I can fix it.
> >
> > Note that the same problem exists in the currently released event-based
> > implementation (cvs version 1.16):
> >
> > thread1 comes in, start initing
> > thread2 creates event, starts waiting
> > thread3 comes in starts waiting
> > thread1 is cancelled, signals event
> > thread2 wakes up, proceeds to the point right before the resetEvent
> > thread3 wakes up, closes event handle
> > thread2 resets closed handle
> 
> Relies on HANDLE uniqueness and assumes that an error will result. This
> is why the 2.6.0 version (and earlier) checks the return code and
> restores Win32 LastError if necessary - for GetLastError transparency.

Does Windows guarantee that the handles are not reused? What happens if a
thread closes a handle while another thread is blocked on it? Is any of this
in Microsoft documentation? Consider the following scenario for the
event-based implementation:

thread1 comes in, starts initing
thread2 creates event, starts waiting
thread3 comes in starts waiting
thread1 is cancelled, signals event
thread2 wakes up, proceeds to the point right before the resetEvent
thread3 wakes up, proceeds to the point right before it closes event handle
thread2 resets handle, clears state, proceeds to the point right before
setting state to 'done'
thread4 comes in proceeds to the point right before waiting on the event
thread3 closes event handle
threads 3, 4, and 2 race to close, wait, and set event

Basically it appears, if I am not confused, the handle may be closed while
(or before) another thread calls wait/reset/setevent. The handle may even be
closed while BOTH a thread is waiting and another thread is calling
SetEvent. Are all of these ok? Does it mean similar things are ok for the
semaphore based version and we just need to restore lasterror?

> 
> Remember, these are very rare situations following a cancellation. It
> might delay things a little but shouldn't break anything.
> 
> > Re: you previous message:
> > >"If only one thread ever comes in, and is canceled in the init_routine,
> > >then the semaphore is never cleaned up."
> >
> > If only one thread ever comes in, and is canceled in the init_routine,
> then
> > the semaphore is never created to begin with, right?
> >
> > Also, regarding my previous comment to Ross about very high cost of
> using
> > InterlockedExchangeAdd for MBR:
> > I did some simple benchmarking. Running pthread_once 50,000,000 on my
> pretty
> > slow single CPU machine takes about 2.1 seconds. Replacing
> > InterlockedExchangeAdd with simple read brings it down to 0.6 seconds.
> This
> > looks significant.
> 
> Using the PTW32_INTERLOCKED_COMPARE_EXCHANGE macro as in your latest (in
> CVS) version and building the library for inlined functions (nmake VC-
> inlined) and x86 architecture causes customised versions of
> InterlockedCompareExchange to be used, and this results in inlined asm.
> Same for PTW32_INTERLOCKED_EXCHANGE.
> 
> Also, on single-CPU x86, the library dynamically switches to using
> 'cmpxchg' rather than 'lock cmpxchg' to avoid locking the bus. This
> appears to match what the kernel32.dll versions do. On non-x86
> architectures the kernel32.dll versions are called, with call overhead.
> 
> PTW32_INTERLOCKED_EXCHANGE_ADD could be added, as could other
> architectures. See ptw32_InterlockedCompareExchange.c

I have rerun my benchmark with VC-inline. The difference is now less
significant 0.9 vs 0.6 but still noticeable. I guess cmpxchg even without
locking is quite expensive. On multi-CPU systems the difference should be
much higher due to the time it takes to lock the bus and to the contention
it may cause. It sounded as if you did not care much to try to optimize it.
I did not mean to suggest that we have to do it right now either. I just
wanted to get your opinion on whether we want to deal with this in the
future. 

> 
> > -----Original Message-----
> > From: Gottlob Frege [mailto:gottlobfrege@gmail.com]
> > Sent: Friday, May 27, 2005 12:42 PM
> > To: Ross Johnson
> > Cc: Vladimir Kliatchko
> > Subject: Re: New pthread_once implementation
> >
> > thread1 comes in, start initing
> > thread2 creates sema and waits
> > thread1 starts to cancel - resets control->state
> > thread3 comes in, goes into init
> > thread4 comes in, goes into else block
> > thread1 finishes cancel - releases semaphore
> > thread2 wakes up
> > thread2 decrements numSemaUsers to 0
> > thread4 increments numSemaUsers
> > thread4 does NOT set new semaphore
> > thread2 closes semaphore
> > thread4 tries to wait on closed semaphore...
> >
> >
> > On 5/27/05, Ross Johnson <ross.johnson@homemail.com.au> wrote:
> > > Guys,
> > >
> > > Is there anything you want to change before I cast a new release?
> > >
> > > http://sources.redhat.com/cgi-bin/cvsweb.cgi/pthreads/pthread_once.c?
> > > rev=1.18&content-type=text/x-cvsweb-markup&cvsroot=pthreads-win32
> > >
> > > Thanks.
> > > Ross
> > >
> > > On Fri, 2005-05-27 at 08:39 -0500, Tim Theisen wrote:
> > > > I picked up the latest and compiled with VC 7.1.  All tests passed.
> > > > Then, I ran 100 iterations of the once [1-4] tests.  These tests
> passed
> > > > as well.  So, it has my stamp of approval.
> > > >
> > > > ...Tim
> > > > --
> > > >        Tim Theisen                     Lead Research Software
> Engineer
> > > > Phone: +1 608 824 2848                 TomoTherapy Incorporated
> > > >   Fax: +1 608 824 2996                 1240 Deming Way
> > > >   Web: http://www.tomotherapy.com      Madison, WI 53717-1954
> > > >
> > > >
> > > >
> > > > -----Original Message-----
> > > > From: Ross Johnson [mailto:ross.johnson@homemail.com.au]
> > > > Sent: Friday, May 27, 2005 02:44
> > > > To: Tim Theisen
> > > > Cc: Vladimir Kliatchko; Gottlob Frege
> > > > Subject: New pthread_once implementation
> > > >
> > > >
> > > > Hi Tim,
> > > >
> > > > The current CVS head contains the latest and much simpler
> implementation
> > > > of pthread_once just presented on the mailing list. It passes on a
> UP
> > > > machine as usual but no-one has run it through an MP system yet.
> Could
> > > > you when you get time?
> > > >
> > > > Thanks.
> > > > Ross
> > > >
> > > >
> > > >
> > >
> > >
> >
> >



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]