On Wed, Sep 9, 2009 at 20:19, Gregory P. Smith
On Wed, Sep 9, 2009 at 9:07 AM, Thomas Wouters
On Sat, Jul 25, 2009 at 19:25, Gregory P. Smith
On Thu, Jul 23, 2009 at 4:28 PM, Thomas Wouters
So attached (and at http://codereview.appspot.com/96125/show ) is a preliminary fix, correcting the problem with os.fork(), os.forkpty()
os.fork1(). This doesn't expose a general API for C code to use, for two reasons: it's not easy, and I need this fix more than I need the API change :-) (I actually need this fix myself for Python 2.4, but it applies fairly cleanly.)
This looks good to me.
Anyone else want to take a look at this before I check it in? I updated
patch (in Rietveld) to contain some documentation about the hazards of mixing fork and threads, which is the best we can do at the moment, at least without seriously overhauling the threading APIs (which, granted, is not that bad an idea, considering the mess they're in.) I've now thoroughly tested the patch, and for most platforms it's strictly better. On AIX it *may* behave differently (possibly 'incorrectly' for specific cases) if something other than os.fork() calls the C fork() and calls PyOS_AfterFork(), since on AIX it used to nuke the thread lock. *I* think the new behaviour (not nuking the lock) is the correct thing to do, but since most places that release the import lock don't bother to check if
lock was even held, the old behaviour may have been succesfully masking
wrote: the the the
bug on AIX systems. Perhaps for the backport to 2.6 (which I think is in order, and also in accordance with the guidelines) I should leave the AIX workaround in? Anyone think it should not be removed from 3.x/2.7?
Your idea of making this an API called a 'fork lock' or something sounds good (though I think it needs a better name. PyBeforeFork & PyAfterFork?). The subprocess module, for example, disables garbage collection before forking and restores it afterwards to avoid http://bugs.python.org/issue1336. That type of thing could also be done in such a function.
Unfortunately it's rather hard to make those functions work correctly with the current API -- we can't provide functions you can just use as arguments to pthread_atfork because the global interpreter lock is not re-entrant and we have no way of testing whether the current thread holds the GIL. I also get the creepy-crawlies when I look at the various thread_* implementations and see the horribly unsafe things they do (and also, for instance, the PendingCall stuff in ceval.c :S) Unfortunately there's no good way to fix these things without breaking API compatibility, let alone ABI compatibility.
Take a look at http://code.google.com/p/python-atfork/ which I created to address the general fork+threading with locks held causing deadlocks issue with many standard library modules. Currently it only patches the logging module but I intend to support others. I want to get an atfork mechanism into 2.7/3.2 along with every lock in the standard library making proper use of it.
See also http://bugs.python.org/issue6721
I make no attempt to deal with C-level locks, only those acquired from python. It doesn't use pthread_atfork but instead models its behavior after that and wraps os.fork and os.forkpty so that they call the registered atfork methods as appropriate.
Well, sure, the *Python code* side of the problem is fixable, thanks to
Python being so awesome ;-P I'm strictly thinking of applications embedding
Python (or even extending it and calling into code that doesn't consider
Python.) Your python-atfork project looks like a terribly good idea, but it
won't fix the embedding/extending issues (nor is it intended to, right?)