This is the mail archive of the
cygwin-patches@cygwin.com
mailing list for the Cygwin project.
Re: src/winsup/cygwin ChangeLog thread.cc thread.h ...
Ok this is a quick-and-it-couldbe-cleaner patch.
It's interim - this weekend I'll make time to roll the logic throughout
thread.cc. The patch doesn't introduce any new issues though, and it is
the correct IMO step to solving the issue(s) I was trying to address
with my last lets-break-cygwin patch.
I have _no_ idea why it worked at all after I built that .dll :}. The
fault for those wanting the grisly details was that I changed the
semantics of verifyableobject_isvalid without updating the tests against
the return code. Doh.
I'm having some trouble with cvs+ssh with this patch .. though I'm not
sure why. For a little while I though it might be chris's tuesday
sleep(1) change, because I was getting strange results from pspec> I'm
not sure though.
Anyway, I don't have time to complete a binary search now...
What I have established is that the faulty change (other than my
thread.cc snafu) is sometime between 1am tuesday 25th and now. In other
words, a dll built from cvs @tuesday 1am, with the most recent thread.cc
and thread.h and this patch seems to run ok. The cond_wait bug seems
particularly ticklish however, and that may be the cvs+ssh problem I was
seeing.
So, you can ignore this blurb :].
I'm not checking this patch in _yet_ as I'm still confirming that
everything is really ok. I'll have a little time in the office tomorrow
to follow up, it's bedtime now though.
Rob
----- Original Message -----
From: "egor duda" <deo@logos-m.ru>
To: "Robert Collins" <robert.collins@itdomain.com.au>
Cc: <cygwin-patches@cygwin.com>
Sent: Thursday, September 27, 2001 11:00 PM
Subject: Re: src/winsup/cygwin ChangeLog thread.cc thread.h ...
> Hi!
>
> Thursday, 27 September, 2001 Robert Collins
robert.collins@itdomain.com.au wrote:
>
> >> rscc> * thread.cc (pthread_cond::BroadCast): Use address
with
> RC> verifyable_object_isvalid().
> >> rscc> (pthread_cond::Signal): Ditto.
> >>
> >> [...]
> >>
> >> Robert, i have problems with your last patch. at program startup
> >> read_etc_passwd() is called recursively and second call blocks at
> >> pthread_mutex_lock()
fixmutex.patch