Bug #787712 threaded python deadlocks

[Warren Togami Jr.]

Hey folks,

https://bugzilla.redhat.com/show_bug.cgi?id=787712
I am hitting this bug very frequently on Fedora 16 where python deadlocks.
 I tried to apply the patch in this bug to python.src.rpm currently in
Fedora 16, but my local rpmbuild fails with a test error.  After disabling
%check rpmbuild fails with these missing files:

    File not found:
/home/warren/rpmbuild/BUILDROOT/python-2.7.2-5.2.fc16.fork.x86_64/usr/lib64/python2.7/lib-dynload/ossaudiodev.so
    File not found:
/home/warren/rpmbuild/BUILDROOT/python-2.7.2-5.2.fc16.fork.x86_64/usr/lib64/python2.7/plat-linux2

It seems python is missing some BuildRequires.

Anyhow, who maintains python these days?  Could we please go ahead with a
F16 test update?  This is a pretty serious issue. =(

Warren Togami
wtogami(a)gmail.com

[Bohuslav Kabrda]

----- Original Message -----

>  Hey folks, 
>  https://bugzilla.redhat.com/show_bug.cgi?id=787712
>  I am hitting this bug very frequently on Fedora 16 where python
>  deadlocks. I tried to apply the patch in this bug to python.src.rpm
>  currently in Fedora 16, but my local rpmbuild fails with a test
>  error. After disabling %check rpmbuild fails with these missing
>  files: 
>  File not found:
>
> /home/warren/rpmbuild/BUILDROOT/python-2.7.2-5.2.fc16.fork.x86_64/usr/lib64/python2.7/lib-dynload/ossaudiodev.so
>  File not found:
>
> /home/warren/rpmbuild/BUILDROOT/python-2.7.2-5.2.fc16.fork.x86_64/usr/lib64/python2.7/plat-linux2
>  It seems python is missing some BuildRequires. 
>  Anyhow, who maintains python these days? Could we please go ahead
>  with a F16 test update? This is a pretty serious issue. =( 
>  Warren Togami
>  wtogami(a)gmail.com
>  _______________________________________________
>  python-devel mailing list
>  python-devel(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/python-devel 
Hi Warren, 
I reproduced the bug on F16 and here is what I found, comparing to the original Koji
built: 

Koji: 
... 
checking MACHDEP... linux2 
... 
Python build finished, but the necessary bits to build these modules were not found: 
bsddb185 sunaudiodev 

My local mock: 
... 
checking MACHDEP... linux3 
... 
Python build finished, but the necessary bits to build these modules were not found: 
bsddb185 dl imageop 
linuxaudiodev ossaudiodev sunaudiodev 
To find the necessary bits, look in setup.py in detect_modules() for the module's
name. 

I'm not sure, but it seems, that plat-linux2 is replaced by plat-linux-3 because of
the MACHDEP change - I'm not sure, but was the previous build of Python in Koji on
2.x.x kernel? If yes, this would explain this, because that is how MACHDEP is constructed
in configure. 

So the solution is to run make with MACHDEP="linux2" so that everything gets
compiled, but the directory plat-linux2 won't be present anymore, which is I think
fine and should be replaced by plat-linux3 in the %files section. 

<sigh>And after digging so deep, I found out that this was already discussed here
[1] and fixed in this revision [2]. So after applying [2], the directory should stay
plat-linux2 and everything should work, also.</sigh> 

-- 

Regards, 
Bohuslav "Slavek" Kabrda. 

[1] http://bugs.python.org/issue12326 
[2] http://hg.python.org/cpython/rev/265da863d017

[Warren Togami Jr.]

https://bugzilla.redhat.com/show_bug.cgi?id=787712
Hi, thanks for looking into this.  I am however more concerned that
this patch goes into both F17 and F16-updates.  Who can do it?  Can I?

Warren


On Fri, Feb 24, 2012 at 2:33 AM, Bohuslav Kabrda <bkabrda(a)redhat.com&gt; wrote:
>  ________________________________
>
>  Hey folks,
>
>  https://bugzilla.redhat.com/show_bug.cgi?id=787712
>  I am hitting this bug very frequently on Fedora 16 where python deadlocks.
>   I tried to apply the patch in this bug to python.src.rpm currently in
>  Fedora 16, but my local rpmbuild fails with a test error.  After disabling
>  %check rpmbuild fails with these missing files:
>
>      File not found:
>
> /home/warren/rpmbuild/BUILDROOT/python-2.7.2-5.2.fc16.fork.x86_64/usr/lib64/python2.7/lib-dynload/ossaudiodev.so
>      File not found:
>
> /home/warren/rpmbuild/BUILDROOT/python-2.7.2-5.2.fc16.fork.x86_64/usr/lib64/python2.7/plat-linux2
>
>  It seems python is missing some BuildRequires.
>
>  Anyhow, who maintains python these days?  Could we please go ahead with a
>  F16 test update?  This is a pretty serious issue. =(
>
>  Warren Togami
>  wtogami(a)gmail.com
>
>  _______________________________________________
>  python-devel mailing list
>  python-devel(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/python-devel
>
>
>  Hi Warren,
>  I reproduced the bug on F16 and here is what I found, comparing to the
>  original Koji built:
>
>  Koji:
>  ...
>  checking MACHDEP... linux2
>  ...
>  Python build finished, but the necessary bits to build these modules were
>  not found:
>  bsddb185           sunaudiodev
>
>  My local mock:
>  ...
>  checking MACHDEP... linux3
>  ...
>  Python build finished, but the necessary bits to build these modules were
>  not found:
>  bsddb185           dl                 imageop
>  linuxaudiodev      ossaudiodev        sunaudiodev
>  To find the necessary bits, look in setup.py in detect_modules() for the
>  module's name.
>
>  I'm not sure, but it seems, that plat-linux2 is replaced by plat-linux-3
>  because of the MACHDEP change - I'm not sure, but was the previous build of
>  Python in Koji on 2.x.x kernel? If yes, this would explain this, because
>  that is how MACHDEP is constructed in configure.
>
>  So the solution is to run make with MACHDEP="linux2" so that everything
>  gets compiled, but the directory plat-linux2 won't be present anymore, which
>  is I think fine and should be replaced by plat-linux3 in the %files section.
>
>  <sigh>And after digging so deep, I found out that this was already
>  discussed here [1] and fixed in this revision [2]. So after applying [2],
>  the directory should stay plat-linux2 and everything should work,
>  also.</sigh>
>
>  --
>  Regards,
>  Bohuslav "Slavek" Kabrda.
>
>  [1] http://bugs.python.org/issue12326
>  [2] http://hg.python.org/cpython/rev/265da863d017
>
>  _______________________________________________
>  python-devel mailing list
>  python-devel(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/python-devel

[Thomas Spura]

On Sat, Feb 25, 2012 at 4:37 AM, Warren Togami Jr. <wtogami(a)gmail.com&gt; wrote:
>  https://bugzilla.redhat.com/show_bug.cgi?id=787712
>  Hi, thanks for looking into this.  I am however more concerned that
>  this patch goes into both F17 and F16-updates.  Who can do it?  Can I? 
Any provenpackager, yes.

I just have applied that patch here and get a failure in test_ssl:
FAIL: test_openssl_version (test.test_ssl.BasicSocketTests)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/builddir/build/BUILD/Python-2.7.2/Lib/test/test_ssl.py", line
151, in test_openssl_version
    (s, t))
AssertionError: ('OpenSSL 1.0.1-fips-beta2 19 Jan 2012', (1, 0, 0, 0, 3))

So openssl ssems to report to be 1.0.0 and not 1.0.1.
(not with 'openssl version' but OPENSSL_VERSION_NUMBER seems to be wrong)

Will have a look at this later today or tomorrow...

(This threading issue also applies to python3, where I'll have a look too...)

Greetings,
   Tom

[Thomas Spura]

On Sat, Feb 25, 2012 at 12:07 PM, Thomas Spura
<tomspur(a)fedoraproject.org&gt; wrote:
>  I just have applied that patch here and get a failure in test_ssl:
>  FAIL: test_openssl_version (test.test_ssl.BasicSocketTests)
>  ----------------------------------------------------------------------
>  Traceback (most recent call last):
>   File "/builddir/build/BUILD/Python-2.7.2/Lib/test/test_ssl.py", line
>  151, in test_openssl_version
>     (s, t))
>  AssertionError: ('OpenSSL 1.0.1-fips-beta2 19 Jan 2012', (1, 0, 0, 0, 3))
>
>  So openssl ssems to report to be 1.0.0 and not 1.0.1.
>  (not with 'openssl version' but OPENSSL_VERSION_NUMBER seems to be wrong) 
Already reported here:
https://bugzilla.redhat.com/show_bug.cgi?id=794950

pushed the patch, the build needs to wait till openssl works again as
expected...

Greetings,
   Tom