This is the mail archive of the
cygwin-apps
mailing list for the Cygwin project.
Last pre-vacation patch (was Re: STC for libapr fcntl failure)
- From: Corinna Vinschen <corinna-cygwin at cygwin dot com>
- To: cygwin-apps at cygwin dot com
- Date: Fri, 26 Apr 2013 20:33:51 +0200
- Subject: Last pre-vacation patch (was Re: STC for libapr fcntl failure)
- References: <517AA478 dot 3010904 at acm dot org>
- Reply-to: cygwin-apps at cygwin dot com
On Apr 26 08:59, David Rothenberger wrote:
> The libapr1 tests are failing for Cygwin 64-bit when trying to use
> fcntl(3) for process mutexes. Calling fcntl() to acquire the lock
> currently fails with a "lock: Bad address" error.
>
> I've attached a simple test case that demonstrates the problem. Just
> run "make" to build and run it.
Hi David,
this is a bug in newlib which defines fcntl as a function taking three
int parameters, rather than two ints and the rest as variadic args
as it would have been correct.
I worked around that in Cygwin for now(*) and I'm uploading a new
1.7.19-3 package while I'm writing this mail. Should be available
in 10 minutes or so.
The STC you sent to the main list is very likely not actually a flock
problem, but rather a signal problem. Please test it on 1.7.17 again.
If should work there and flock hasn't changed for a while.
Other than, I'm on vaca until mid-May. I'm available on and off
via email, but I quite certainly won't move a finger ;)
Obviously I would still be glad if you guys work further on this stuff
and, if it's not asked too much, try to track down the problems you
encounter and send patches, if possible. I will care for that again as
soon as I'm back.
See you later,
Corinna
(*) Actually, that will be the final fix for Cygwin. Nevertheless, I
would be glad if you could nudge me after my vacation again, to
remind me that this problem also needs a newlib fix.
--
Corinna Vinschen Please, send mails regarding Cygwin to
Cygwin Maintainer cygwin AT cygwin DOT com
Red Hat