Have people seen these?

http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650

http://d.puremagic.com/issues/show_bug.cgi?id=4890

This might need an emergency release... seems like a really bad one.

-Steve

Weird... The GC and GC-related thread code hasn't been touched in ages. I'll give it a look today.

Sent from my iPhone

On Sep 20, 2010, at 7:57 AM, Steve Schveighoffer <schveiguy at yahoo.com> wrote:

> Have people seen these?
> 
> http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
> 
> http://d.puremagic.com/issues/show_bug.cgi?id=4890
> 
> This might need an emergency release... seems like a really bad one.
> 
> -Steve
> 
> 
> 
> 
> _______________________________________________
> D-runtime mailing list
> D-runtime at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/d-runtime

Weird, the GC and GC-related threading code hasn't been touched in ages.  I'll give it a look today.

On Sep 20, 2010, at 7:57 AM, Steve Schveighoffer wrote:

> Have people seen these?
> 
> http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
> 
> http://d.puremagic.com/issues/show_bug.cgi?id=4890
> 
> This might need an emergency release... seems like a really bad one.
> 
> -Steve
> 
> 
> 
> 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

Okay, I've made a bunch of changes to how thread initialization and termination is done.  Tested to work on OSX with the deadlock case, but I think only Linux was deadlocking anyway.  I don't have my Linux or Windows VMs handy, so if someone could verify that threads still work on Windows (via the sample in the ticket below) that would be fantastic.  I'm building a new Linux VM on this machine to test there.

On Sep 20, 2010, at 7:57 AM, Steve Schveighoffer wrote:

> Have people seen these?
> 
> http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
> 
> http://d.puremagic.com/issues/show_bug.cgi?id=4890
> 
> This might need an emergency release... seems like a really bad one.
> 
> -Steve
> 
> 
> 
> 
> _______________________________________________
> D-runtime mailing list
> D-runtime at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/d-runtime

Tested with latest dmd/phobos/druntime on Linux

Still broken.  Any help I can offer?

-Steve



----- Original Message ----
> From: Sean Kelly <sean at invisibleduck.org>
> 
> Okay, I've made a bunch of changes to how thread initialization and termination
>is done.  Tested to work on OSX with the deadlock case, but I think only  Linux was deadlocking anyway.  I don't have my Linux or Windows VMs handy,  so if someone could verify that threads still work on Windows (via the sample in  the ticket below) that would be fantastic.  I'm building a new Linux VM on  this machine to test there.
> 
> On Sep 20, 2010, at 7:57 AM, Steve  Schveighoffer wrote:
> 
> > Have people seen these?
> > 
> > 
>http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
>
> > 
> > http://d.puremagic.com/issues/show_bug.cgi?id=4890
> > 
> >  This might need an emergency release... seems like a really bad one.
> > 
> > -Steve
> > 
> > 
> > 
> > 
> >  _______________________________________________
> > D-runtime mailing  list
> > D-runtime at puremagic.com
> >  http://lists.puremagic.com/mailman/listinfo/d-runtime
> 
> _______________________________________________
> phobos  mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
> 

Stack trace?

Sent from my iPhone

On Sep 20, 2010, at 11:48 AM, Steve Schveighoffer <schveiguy at yahoo.com> wrote:

> Tested with latest dmd/phobos/druntime on Linux
> 
> Still broken.  Any help I can offer?
> 
> -Steve
> 
> 
> 
> ----- Original Message ----
>> From: Sean Kelly <sean at invisibleduck.org>
>> 
>> Okay, I've made a bunch of changes to how thread initialization and termination is done.  Tested to work on OSX with the deadlock case, but I think only  Linux was deadlocking anyway.  I don't have my Linux or Windows VMs handy,  so if someone could verify that threads still work on Windows (via the sample in  the ticket below) that would be fantastic.  I'm building a new Linux VM on  this machine to test there.
>> 
>> On Sep 20, 2010, at 7:57 AM, Steve  Schveighoffer wrote:
>> 
>>> Have people seen these?
>>> 
>>> 
>> http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
>> 
>>> 
>>> http://d.puremagic.com/issues/show_bug.cgi?id=4890
>>> 
>>> This might need an emergency release... seems like a really bad one.
>>> 
>>> -Steve
>>> 
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> D-runtime mailing  list
>>> D-runtime at puremagic.com
>>> http://lists.puremagic.com/mailman/listinfo/d-runtime
>> 
>> _______________________________________________
>> phobos  mailing list
>> phobos at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/phobos
>> 
> 
> 
> 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

On Mon, 20 Sep 2010, Sean Kelly wrote:

You can check in a unit test that triggers the bug and watch the auto tester.

Not ideal, but functional.

If it still needs testing when I get home tonight, I'll give it a shot (it'll be 6-8 hours from now though).

> Okay, I've made a bunch of changes to how thread initialization and termination is done.  Tested to work on OSX with the deadlock case, but I think only Linux was deadlocking anyway.  I don't have my Linux or Windows VMs handy, so if someone could verify that threads still work on Windows (via the sample in the ticket below) that would be fantastic. I'm building a new Linux VM on this machine to test there.
> 
> On Sep 20, 2010, at 7:57 AM, Steve Schveighoffer wrote:
> 
> > Have people seen these?
> > 
> > http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
> > 
> > http://d.puremagic.com/issues/show_bug.cgi?id=4890
> > 
> > This might need an emergency release... seems like a really bad one.
> > 
> > -Steve
> > 
> > 
> > 
> > 
> > _______________________________________________
> > D-runtime mailing list
> > D-runtime at puremagic.com
> > http://lists.puremagic.com/mailman/listinfo/d-runtime
> 
> _______________________________________________
> D-runtime mailing list
> D-runtime at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/d-runtime
> 

Will the autotester detect a deadlock?

I'm not sure this is a good idea...

-Steve



----- Original Message ----
> From: Brad Roberts <braddr at puremagic.com>
> To: D's runtime library developers list <d-runtime at puremagic.com>
> Cc: Phobos <phobos at puremagic.com>
> Sent: Mon, September 20, 2010 4:06:55 PM
> Subject: Re: [phobos] [D-runtime] Possible critical deadlock issue for 2.049
> 
> On Mon, 20 Sep 2010, Sean Kelly wrote:
> 
> You can check in a unit test that  triggers the bug and watch the auto tester.
> 
> Not ideal, but  functional.
> 
> If it still needs testing when I get home tonight, I'll give  it a shot (it'll be 6-8 hours from now though).
> 
> > Okay, I've made  a bunch of changes to how thread initialization and
> > termination is  done.  Tested to work on OSX with the deadlock case, but
> > I think  only Linux was deadlocking anyway.  I don't have my Linux or
> >  Windows VMs handy, so if someone could verify that threads still work on
> > Windows (via the sample in the ticket below) that would be  fantastic.
> > I'm building a new Linux VM on this machine to test  there.
> > 
> > On Sep 20, 2010, at 7:57 AM, Steve Schveighoffer  wrote:
> > 
> > > Have people seen these?
> > > 
> > > 
>http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
>
> >  >
> > > http://d.puremagic.com/issues/show_bug.cgi?id=4890
> >  >
> > > This might need an emergency release... seems like a really  bad one.
> > > 
> > > -Steve
> > > 
> > > 
> >  >
> > > 
> > >  _______________________________________________
> > > D-runtime mailing  list
> > > D-runtime at puremagic.com
> > >  http://lists.puremagic.com/mailman/listinfo/d-runtime
> > 
> >  _______________________________________________
> > D-runtime mailing  list
> > D-runtime at puremagic.com
> > http://lists.puremagic.com/mailman/listinfo/d-runtime
> > 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
> 

Not so much.  Good point. :)  Though a test that has an alarm in it and self-kills if it doesn't finish in some controlled amount of time isn't that hard to do.  I'll add to my todo list something to detect this sort of problem at a high level in the tester as well.

On Mon, 20 Sep 2010, Steve Schveighoffer wrote:

> Date: Mon, 20 Sep 2010 13:10:53 -0700 (PDT)
> From: Steve Schveighoffer <schveiguy at yahoo.com>
> Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
> To: Discuss the phobos library for D <phobos at puremagic.com>
> Subject: Re: [phobos] [D-runtime] Possible critical deadlock issue for 2.049
> 
> Will the autotester detect a deadlock?
> 
> I'm not sure this is a good idea...
> 
> -Steve
> 
> 
> 
> ----- Original Message ----
> > From: Brad Roberts <braddr at puremagic.com>
> > To: D's runtime library developers list <d-runtime at puremagic.com>
> > Cc: Phobos <phobos at puremagic.com>
> > Sent: Mon, September 20, 2010 4:06:55 PM
> > Subject: Re: [phobos] [D-runtime] Possible critical deadlock issue for 2.049
> > 
> > On Mon, 20 Sep 2010, Sean Kelly wrote:
> > 
> > You can check in a unit test that  triggers the bug and watch the auto tester.
> > 
> > Not ideal, but  functional.
> > 
> > If it still needs testing when I get home tonight, I'll give  it a shot (it'll be 6-8 hours from now though).
> > 
> > > Okay, I've made  a bunch of changes to how thread initialization and
> > > termination is  done.  Tested to work on OSX with the deadlock case, but
> > > I think  only Linux was deadlocking anyway.  I don't have my Linux or
> > >  Windows VMs handy, so if someone could verify that threads still work on
> > > Windows (via the sample in the ticket below) that would be  fantastic.
> > > I'm building a new Linux VM on this machine to test  there.
> > > 
> > > On Sep 20, 2010, at 7:57 AM, Steve Schveighoffer  wrote:
> > > 
> > > > Have people seen these?
> > > > 
> > > > 
> >http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
> >
> > >  >
> > > > http://d.puremagic.com/issues/show_bug.cgi?id=4890
> > >  >
> > > > This might need an emergency release... seems like a really  bad one.
> > > > 
> > > > -Steve
> > > > 
> > > > 
> > >  >
> > > > 
> > > >  _______________________________________________
> > > > D-runtime mailing  list
> > > > D-runtime at puremagic.com
> > > >  http://lists.puremagic.com/mailman/listinfo/d-runtime
> > > 
> > >  _______________________________________________
> > > D-runtime mailing  list
> > > D-runtime at puremagic.com
> > > http://lists.puremagic.com/mailman/listinfo/d-runtime
> > > 
> > _______________________________________________
> > phobos mailing list
> > phobos at puremagic.com
> > http://lists.puremagic.com/mailman/listinfo/phobos
> > 
> 
> 
> 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
> 

Until now I've had unit testing of thread-related code be pretty light because most of the failure cases are hangs.  As standalone tests though, this seems a lot more reasonable.

On Sep 20, 2010, at 2:08 PM, Brad Roberts wrote:

> Not so much.  Good point. :)  Though a test that has an alarm in it and self-kills if it doesn't finish in some controlled amount of time isn't that hard to do.  I'll add to my todo list something to detect this sort of problem at a high level in the tester as well.
> 
> On Mon, 20 Sep 2010, Steve Schveighoffer wrote:
> 
>> Date: Mon, 20 Sep 2010 13:10:53 -0700 (PDT)
>> From: Steve Schveighoffer <schveiguy at yahoo.com>
>> Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
>> To: Discuss the phobos library for D <phobos at puremagic.com>
>> Subject: Re: [phobos] [D-runtime] Possible critical deadlock issue for 2.049
>> 
>> Will the autotester detect a deadlock?
>> 
>> I'm not sure this is a good idea...
>> 
>> -Steve
>> 
>> 
>> 
>> ----- Original Message ----
>>> From: Brad Roberts <braddr at puremagic.com>
>>> To: D's runtime library developers list <d-runtime at puremagic.com>
>>> Cc: Phobos <phobos at puremagic.com>
>>> Sent: Mon, September 20, 2010 4:06:55 PM
>>> Subject: Re: [phobos] [D-runtime] Possible critical deadlock issue for 2.049
>>> 
>>> On Mon, 20 Sep 2010, Sean Kelly wrote:
>>> 
>>> You can check in a unit test that  triggers the bug and watch the auto tester.
>>> 
>>> Not ideal, but  functional.
>>> 
>>> If it still needs testing when I get home tonight, I'll give  it a shot (it'll be 6-8 hours from now though).
>>> 
>>>> Okay, I've made  a bunch of changes to how thread initialization and termination is  done.  Tested to work on OSX with the deadlock case, but I think  only Linux was deadlocking anyway.  I don't have my Linux or Windows VMs handy, so if someone could verify that threads still work on Windows (via the sample in the ticket below) that would be  fantastic. I'm building a new Linux VM on this machine to test  there.
>>>> 
>>>> On Sep 20, 2010, at 7:57 AM, Steve Schveighoffer  wrote:
>>>> 
>>>>> Have people seen these?
>>>>> 
>>>>> 
>>> http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=117650
>>> 
>>>>> 
>>>>> http://d.puremagic.com/issues/show_bug.cgi?id=4890
>>>>> 
>>>>> This might need an emergency release... seems like a really  bad one.
>>>>> 
>>>>> -Steve
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> _______________________________________________
>>>>> D-runtime mailing  list
>>>>> D-runtime at puremagic.com
>>>>> http://lists.puremagic.com/mailman/listinfo/d-runtime
>>>> 
>>>> _______________________________________________
>>>> D-runtime mailing  list
>>>> D-runtime at puremagic.com
>>>> http://lists.puremagic.com/mailman/listinfo/d-runtime
>>>> 
>>> _______________________________________________
>>> phobos mailing list
>>> phobos at puremagic.com
>>> http://lists.puremagic.com/mailman/listinfo/phobos
>>> 
>> 
>> 
>> 
>> _______________________________________________
>> phobos mailing list
>> phobos at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/phobos
>> 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos