You are viewing a plain text version of this content. The canonical link for it is here.
Posted to modperl@perl.apache.org by Gisle Aas <gi...@ActiveState.com> on 2004/12/08 12:46:02 UTC
Fix perl/ithreads random failures
I had to do tests with an older version of mod_perl_1.99 that still
had the perl/ithreads* tests. I see that these tests was removed from
the distribution in [1], but I think that if you apply the attached
patch you can start distribution these tests again.
[1] http://cvs.apache.org/viewcvs.cgi/modperl-2.0/lib/ModPerl/Manifest.pm?r1=1.9&r2=1.10
Regards,
Gisle
Index: t/response/TestPerl/ithreads.pm
--- t/response/TestPerl/ithreads.pm.~1~ Wed Dec 8 03:43:55 2004
+++ t/response/TestPerl/ithreads.pm Wed Dec 8 03:43:55 2004
@@ -57,7 +57,10 @@
$counter_shar += $counter_shar for 1..10;
});
$counter_priv += $counter_priv for 1..10;
- $counter_shar += $counter_shar for 1..10;
+ {
+ lock $counter_shar;
+ $counter_shar += $counter_shar for 1..10;
+ }
$thr->join;
ok t_cmp(2**20, $counter_shar, "shared counter");
ok t_cmp(2**10, $counter_priv, "private counter");
End of Patch.
--
Report problems: http://perl.apache.org/bugs/
Mail list info: http://perl.apache.org/maillist/modperl.html
List etiquette: http://perl.apache.org/maillist/email-etiquette.html
Re: Fix perl/ithreads random failures
Posted by Stas Bekman <st...@stason.org>.
Gisle Aas wrote:
> I had to do tests with an older version of mod_perl_1.99 that still
> had the perl/ithreads* tests. I see that these tests was removed from
> the distribution in [1], but I think that if you apply the attached
> patch you can start distribution these tests again.
>
> [1] http://cvs.apache.org/viewcvs.cgi/modperl-2.0/lib/ModPerl/Manifest.pm?r1=1.9&r2=1.10
>
> Regards,
> Gisle
>
>
> Index: t/response/TestPerl/ithreads.pm
> --- t/response/TestPerl/ithreads.pm.~1~ Wed Dec 8 03:43:55 2004
> +++ t/response/TestPerl/ithreads.pm Wed Dec 8 03:43:55 2004
> @@ -57,7 +57,10 @@
> $counter_shar += $counter_shar for 1..10;
> });
> $counter_priv += $counter_priv for 1..10;
> - $counter_shar += $counter_shar for 1..10;
> + {
> + lock $counter_shar;
> + $counter_shar += $counter_shar for 1..10;
> + }
> $thr->join;
> ok t_cmp(2**20, $counter_shar, "shared counter");
> ok t_cmp(2**10, $counter_priv, "private counter");
> End of Patch.
Right on, Gisle. I was locking it inside the child-threads, but forgot
about the parent. Thanks.
--
__________________________________________________________________
Stas Bekman JAm_pH ------> Just Another mod_perl Hacker
http://stason.org/ mod_perl Guide ---> http://perl.apache.org
mailto:stas@stason.org http://use.perl.org http://apacheweek.com
http://modperlbook.org http://apache.org http://ticketmaster.com
--
Report problems: http://perl.apache.org/bugs/
Mail list info: http://perl.apache.org/maillist/modperl.html
List etiquette: http://perl.apache.org/maillist/email-etiquette.html
Re: Fix perl/ithreads random failures
Posted by Stas Bekman <st...@stason.org>.
Steve Hay wrote:
> Gisle Aas wrote:
>
>
>>Steve Hay <st...@uk.radan.com> writes:
>>
>>
>>
>>
>>>Gisle Aas wrote:
>>>
>>>
>>>
>>>
>>>>I had to do tests with an older version of mod_perl_1.99 that still
>>>>had the perl/ithreads* tests. I see that these tests was removed from
>>>>the distribution in [1], but I think that if you apply the attached
>>>>patch you can start distribution these tests again.
>>>>
>>>>[1] http://cvs.apache.org/viewcvs.cgi/modperl-2.0/lib/ModPerl/Manifest.pm?r1=1.9&r2=1.10
>>>>
>>>>
>>>>
>>>
>>>mp2 is now in svn, not cvs.
>>>
>>>
>>
>>Isn't CVS in sync as well?
>>
>>I read <http://perl.apache.org/contribute/> and it only mentioned CVS
>>and had a pointer to the CVS web interface.
>>
>>
>>
>>
>>>Check-out the latest svn revision as described at:
>>>
>>>http://perl.apache.org/download/source.html#Development_mod_perl_2_0_Source_Distribution
>>>
>>>
>>
>>This one also has pointers to the CVS snapshots and the CVS web
>>interface so I assume that it is kept in sync.
>>
>>
>
> Other people can say for sure whether cvs is kept in sync, but my
> impression was that it was not. I could be wrong. Anyway, svn is
> definitely the way forward.
It's not in sync. The only reason it's still working is because I have
tons of unfinished cvs checkouts, which I need to find the time to process
and then we can close cvs.
> I know that not all the docs have been updated yet, hence there are
> various references to cvs around the place still.
Right, sorry, I'll be updating those pages today. for now please refer to:
http://perl.apache.org/download/source.html#Development_mod_perl_2_0_Source_Distribution
which is up-to-date.
--
__________________________________________________________________
Stas Bekman JAm_pH ------> Just Another mod_perl Hacker
http://stason.org/ mod_perl Guide ---> http://perl.apache.org
mailto:stas@stason.org http://use.perl.org http://apacheweek.com
http://modperlbook.org http://apache.org http://ticketmaster.com
--
Report problems: http://perl.apache.org/bugs/
Mail list info: http://perl.apache.org/maillist/modperl.html
List etiquette: http://perl.apache.org/maillist/email-etiquette.html
Re: Fix perl/ithreads random failures
Posted by Gisle Aas <gi...@ActiveState.com>.
Steve Hay <st...@uk.radan.com> writes:
> What was the problem that you had on HP/UX? Was it crashing or just
> failing (as in "not ok")?
It was not crashing, just ending up the the wrong value. The
TEST_VERBOSE=1 output was this.
perl/ithreads.....1..4
# Running under perl version 5.008006 for hpux
# Current time local: Wed Dec 8 02:47:14 2004
# Current time GMT: Wed Dec 8 10:47:14 2004
# Using Test.pm version 1.25
ok 1
# testing : thread callback returned value
# expected: 2
# received: 2
ok 2
# testing : shared counter
# expected: 1048576
# received: 524288
not ok 3
# testing : private counter
# expected: 1024
# received: 1024
ok 4
FAILED test 3
Failed 1/4 tests, 75.00% okay
perl/ithreads2....# connecting to localhost:8540
1..4
# Running under perl version 5.008006 for hpux
# Current time local: Wed Dec 8 02:47:15 2004
# Current time GMT: Wed Dec 8 10:47:15 2004
# Using Test.pm version 1.25
ok 1
# testing : thread callback returned value
# expected: 2
# received: 2
ok 2
# testing : shared counter
# expected: 1048576
# received: 262144
not ok 3
# testing : private counter
# expected: 1024
# received: 1024
ok 4
FAILED test 3
Failed 1/4 tests, 75.00% okay
Failed Test Stat Wstat Total Fail Failed List of Failed
------------------------------------------------------------------------------------------------------------------------------
perl/ithreads.t 4 1 25.00% 3
perl/ithreads2.t 4 1 25.00% 3
--
Report problems: http://perl.apache.org/bugs/
Mail list info: http://perl.apache.org/maillist/modperl.html
List etiquette: http://perl.apache.org/maillist/email-etiquette.html
Re: Fix perl/ithreads random failures
Posted by Steve Hay <st...@uk.radan.com>.
Gisle Aas wrote:
>Steve Hay <st...@uk.radan.com> writes:
>
>
>
>>Gisle Aas wrote:
>>
>>
>>
>>>I had to do tests with an older version of mod_perl_1.99 that still
>>>had the perl/ithreads* tests. I see that these tests was removed from
>>>the distribution in [1], but I think that if you apply the attached
>>>patch you can start distribution these tests again.
>>>
>>>[1] http://cvs.apache.org/viewcvs.cgi/modperl-2.0/lib/ModPerl/Manifest.pm?r1=1.9&r2=1.10
>>>
>>>
>>>
>>mp2 is now in svn, not cvs.
>>
>>
>
>Isn't CVS in sync as well?
>
>I read <http://perl.apache.org/contribute/> and it only mentioned CVS
>and had a pointer to the CVS web interface.
>
>
>
>> Check-out the latest svn revision as described at:
>>
>>http://perl.apache.org/download/source.html#Development_mod_perl_2_0_Source_Distribution
>>
>>
>
>This one also has pointers to the CVS snapshots and the CVS web
>interface so I assume that it is kept in sync.
>
>
Other people can say for sure whether cvs is kept in sync, but my
impression was that it was not. I could be wrong. Anyway, svn is
definitely the way forward.
I know that not all the docs have been updated yet, hence there are
various references to cvs around the place still.
>
>
>>and you can try that rather than having to go back to an older version.
>>
>>
>
>The patch was reletive to an older version by from glancing at the CVS
>head it looked that it would apply there as well.
>
>
Yes, the patch applied to the current svn version.
>
>
>>The test file, of course, is in the dev source, but is currently marked
>>as under construction:
>>
>>http://svn.apache.org/repos/asf/perl/modperl/trunk/t/perl/ithreads.t
>>
>>Anyway, I tried your patch (with the ithreads test re-instated) and
>>sadly it didn't fix it for me :( The complete diff of changes that I
>>made to my svn checkout (revision 111269) are at the end of this mail.
>>
>>The following test sequence still fails every time for me:
>>
>>perl t/TEST -verbose t/modules/reload.t t/perl/api.t t/perl/ithreads.t
>>
>>on WinXP with Apache 2.0.50 and Perl 5.8.5.
>>
>>
>
>So what is the output when this fails?
>
>I tested on HP-UX/IA-64 and the fix cured the problem there. The fix
>is also obviously correct as both the parent and the child thread
>needs to lock for it to take effect.
>
I can't argue about whether it is correct or not, it's just that it
didn't fix the failures that I've been experiencing for some time on WinXP.
On WinXP the test sequence that I posted in my last mail causes the
Apache.exe process to crash. Nothing is written to the error_log, and
only this appears in the console (even with the -verbose option):
t\perl\ithreads.....request has failed (the response code was: 500)
see t/logs/error_log for more details
dubious
Test returned status 9 (wstat 2304, 0x900)
There is no popup application error dialogue box, hence no opportunity
to step into the source and even see where it crashed.
This has been discussed several times on the dev list (which is probably
where we should be now...), most recently here:
http://marc.theaimsgroup.com/?t=110141084300003&r=1&w=2
Stas can't even reliably reproduce the problem on Linux, so hasn't been
able to debug it himself at all.
What was the problem that you had on HP/UX? Was it crashing or just
failing (as in "not ok")?
- Steve
------------------------------------------------
Radan Computational Ltd.
We would like to take this opportunity to wish all our customers, suppliers and colleagues seasons greetings. We will not be sending corporate greetings cards this year. Instead, we will be making a donation to charity.
The information contained in this message and any files transmitted with it are confidential and intended for the addressee(s) only. If you have received this message in error or there are any problems, please notify the sender immediately. The unauthorized use, disclosure, copying or alteration of this message is strictly forbidden. Note that any views or opinions presented in this email are solely those of the author and do not necessarily represent those of Radan Computational Ltd. The recipient(s) of this message should check it and any attached files for viruses: Radan Computational will accept no liability for any damage caused by any virus transmitted by this email.
Re: Fix perl/ithreads random failures
Posted by Gisle Aas <gi...@ActiveState.com>.
Steve Hay <st...@uk.radan.com> writes:
> Gisle Aas wrote:
>
> >I had to do tests with an older version of mod_perl_1.99 that still
> >had the perl/ithreads* tests. I see that these tests was removed from
> >the distribution in [1], but I think that if you apply the attached
> >patch you can start distribution these tests again.
> >
> >[1] http://cvs.apache.org/viewcvs.cgi/modperl-2.0/lib/ModPerl/Manifest.pm?r1=1.9&r2=1.10
> >
> mp2 is now in svn, not cvs.
Isn't CVS in sync as well?
I read <http://perl.apache.org/contribute/> and it only mentioned CVS
and had a pointer to the CVS web interface.
> Check-out the latest svn revision as described at:
>
> http://perl.apache.org/download/source.html#Development_mod_perl_2_0_Source_Distribution
This one also has pointers to the CVS snapshots and the CVS web
interface so I assume that it is kept in sync.
> and you can try that rather than having to go back to an older version.
The patch was reletive to an older version by from glancing at the CVS
head it looked that it would apply there as well.
> The test file, of course, is in the dev source, but is currently marked
> as under construction:
>
> http://svn.apache.org/repos/asf/perl/modperl/trunk/t/perl/ithreads.t
>
> Anyway, I tried your patch (with the ithreads test re-instated) and
> sadly it didn't fix it for me :( The complete diff of changes that I
> made to my svn checkout (revision 111269) are at the end of this mail.
>
> The following test sequence still fails every time for me:
>
> perl t/TEST -verbose t/modules/reload.t t/perl/api.t t/perl/ithreads.t
>
> on WinXP with Apache 2.0.50 and Perl 5.8.5.
So what is the output when this fails?
I tested on HP-UX/IA-64 and the fix cured the problem there. The fix
is also obviously correct as both the parent and the child thread
needs to lock for it to take effect.
Regards,
Gisle
--
Report problems: http://perl.apache.org/bugs/
Mail list info: http://perl.apache.org/maillist/modperl.html
List etiquette: http://perl.apache.org/maillist/email-etiquette.html
Re: Fix perl/ithreads random failures
Posted by Steve Hay <st...@uk.radan.com>.
Gisle Aas wrote:
>I had to do tests with an older version of mod_perl_1.99 that still
>had the perl/ithreads* tests. I see that these tests was removed from
>the distribution in [1], but I think that if you apply the attached
>patch you can start distribution these tests again.
>
>[1] http://cvs.apache.org/viewcvs.cgi/modperl-2.0/lib/ModPerl/Manifest.pm?r1=1.9&r2=1.10
>
mp2 is now in svn, not cvs. Check-out the latest svn revision as
described at:
http://perl.apache.org/download/source.html#Development_mod_perl_2_0_Source_Distribution
and you can try that rather than having to go back to an older version.
The test file, of course, is in the dev source, but is currently marked
as under construction:
http://svn.apache.org/repos/asf/perl/modperl/trunk/t/perl/ithreads.t
Anyway, I tried your patch (with the ithreads test re-instated) and
sadly it didn't fix it for me :( The complete diff of changes that I
made to my svn checkout (revision 111269) are at the end of this mail.
The following test sequence still fails every time for me:
perl t/TEST -verbose t/modules/reload.t t/perl/api.t t/perl/ithreads.t
on WinXP with Apache 2.0.50 and Perl 5.8.5.
- Steve
Index: t/response/TestPerl/ithreads.pm
===================================================================
--- t/response/TestPerl/ithreads.pm (revision 111269)
+++ t/response/TestPerl/ithreads.pm (working copy)
@@ -57,7 +57,10 @@
$counter_shar += $counter_shar for
1..10;
});
$counter_priv += $counter_priv for 1..10;
- $counter_shar += $counter_shar for 1..10;
+ {
+ lock $counter_shar;
+ $counter_shar += $counter_shar for 1..10;
+ }
$thr->join;
ok t_cmp($counter_shar, 2**20, "shared counter");
ok t_cmp($counter_priv, 2**10, "private counter");
Index: t/perl/ithreads.t
===================================================================
--- t/perl/ithreads.t (revision 111269)
+++ t/perl/ithreads.t (working copy)
@@ -11,10 +11,9 @@
# perl < 5.6.0 fails to compile code with 'shared' attributes, so we
must skip
# it here.
-#unless ($] >= 5.008001 && $Config{useithreads}) {
-# plan tests => 1, need
-# {"perl 5.8.1 or higher w/ithreads enabled is required" => 0};
-#}
-plan tests => 1, under_construction;
+unless ($] >= 5.008001 && $Config{useithreads}) {
+ plan tests => 1, need
+ {"perl 5.8.1 or higher w/ithreads enabled is required" => 0};
+}
print GET_BODY_ASSERT "/TestPerl__ithreads";
------------------------------------------------
Radan Computational Ltd.
We would like to take this opportunity to wish all our customers, suppliers and colleagues seasons greetings. We will not be sending corporate greetings cards this year. Instead, we will be making a donation to charity.
The information contained in this message and any files transmitted with it are confidential and intended for the addressee(s) only. If you have received this message in error or there are any problems, please notify the sender immediately. The unauthorized use, disclosure, copying or alteration of this message is strictly forbidden. Note that any views or opinions presented in this email are solely those of the author and do not necessarily represent those of Radan Computational Ltd. The recipient(s) of this message should check it and any attached files for viruses: Radan Computational will accept no liability for any damage caused by any virus transmitted by this email.
Re: Fix perl/ithreads random failures
Posted by Stas Bekman <st...@stason.org>.
Gisle Aas wrote:
> I had to do tests with an older version of mod_perl_1.99 that still
> had the perl/ithreads* tests. I see that these tests was removed from
> the distribution in [1], but I think that if you apply the attached
> patch you can start distribution these tests again.
>
> [1] http://cvs.apache.org/viewcvs.cgi/modperl-2.0/lib/ModPerl/Manifest.pm?r1=1.9&r2=1.10
>
> Regards,
> Gisle
>
>
> Index: t/response/TestPerl/ithreads.pm
> --- t/response/TestPerl/ithreads.pm.~1~ Wed Dec 8 03:43:55 2004
> +++ t/response/TestPerl/ithreads.pm Wed Dec 8 03:43:55 2004
> @@ -57,7 +57,10 @@
> $counter_shar += $counter_shar for 1..10;
> });
> $counter_priv += $counter_priv for 1..10;
> - $counter_shar += $counter_shar for 1..10;
> + {
> + lock $counter_shar;
> + $counter_shar += $counter_shar for 1..10;
> + }
> $thr->join;
> ok t_cmp(2**20, $counter_shar, "shared counter");
> ok t_cmp(2**10, $counter_priv, "private counter");
> End of Patch.
Thanks Gisle, committed.
--
__________________________________________________________________
Stas Bekman JAm_pH ------> Just Another mod_perl Hacker
http://stason.org/ mod_perl Guide ---> http://perl.apache.org
mailto:stas@stason.org http://use.perl.org http://apacheweek.com
http://modperlbook.org http://apache.org http://ticketmaster.com
--
Report problems: http://perl.apache.org/bugs/
Mail list info: http://perl.apache.org/maillist/modperl.html
List etiquette: http://perl.apache.org/maillist/email-etiquette.html