[jira] [Commented] (ARROW-2489) [Plasma] test_plasma.py crashes

2018-04-23 Thread ASF GitHub Bot (JIRA)

[ 
https://issues.apache.org/jira/browse/ARROW-2489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16448061#comment-16448061
 ] 

ASF GitHub Bot commented on ARROW-2489:
---

pitrou commented on issue #1933: ARROW-2489: [Plasma] Fix PlasmaClient ABI 
variation
URL: https://github.com/apache/arrow/pull/1933#issuecomment-383559208
 
 
   I think the AppVeyor failure is unrelated. It also succeeded on my AppVeyor 
account: https://ci.appveyor.com/project/pitrou/arrow/build/1.0.331


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


> [Plasma] test_plasma.py crashes
> ---
>
> Key: ARROW-2489
> URL: https://issues.apache.org/jira/browse/ARROW-2489
> Project: Apache Arrow
>  Issue Type: Bug
>  Components: GPU, Plasma (C++), Python
>Reporter: Antoine Pitrou
>Priority: Major
>  Labels: pull-request-available
>
> This is new here:
> {code}$ py.test   --tb=native pyarrow/tests/test_plasma.py 
> = test 
> session starts 
> ==
> platform linux -- Python 3.6.5, pytest-3.3.2, py-1.5.2, pluggy-0.6.0
> rootdir: /home/antoine/arrow/python, inifile: setup.cfg
> plugins: xdist-1.22.0, timeout-1.2.1, repeat-0.4.1, forked-0.2, 
> faulthandler-1.3.1
> collected 23 items
>   
>
> pyarrow/tests/test_plasma.py *** Error in 
> `/home/antoine/miniconda3/envs/pyarrow/bin/python': double free or corruption 
> (!prev): 0x01699520 ***
> [...]
> Current thread 0x7fe7e8570700 (most recent call first):
>   File "/home/antoine/arrow/python/pyarrow/tests/test_plasma.py", line 211 in 
> test_connection_failure_raises_exception
> [...]
> {code}
> Here is the C backtrace under gdb:
> {code}
> #0  0x769d0428 in __GI_raise (sig=sig@entry=6) at 
> ../sysdeps/unix/sysv/linux/raise.c:54
> #1  0x769d202a in __GI_abort () at abort.c:89
> #2  0x76a127ea in __libc_message (do_abort=do_abort@entry=2, 
> fmt=fmt@entry=0x76b2bed8 "*** Error in `%s': %s: 0x%s ***\n")
> at ../sysdeps/posix/libc_fatal.c:175
> #3  0x76a1b37a in malloc_printerr (ar_ptr=, 
> ptr=, str=0x76b2c008 "double free or corruption (!prev)", 
> action=3)
> at malloc.c:5006
> #4  _int_free (av=, p=, have_lock=0) at 
> malloc.c:3867
> #5  0x76a1f53c in __GI___libc_free (mem=) at 
> malloc.c:2968
> #6  0x7fffbdfcc504 in std::_Sp_counted_ptr (__gnu_cxx::_Lock_policy)2>::_M_dispose (this=0x9defb0)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:373
> #7  0x7fffbdfc903c in 
> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release (this=0x9defb0) 
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:149
> #8  0x7fffbdfc82b9 in 
> std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count 
> (this=0x7fffc1214510, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:666
> #9  0x7fffbdfc8276 in std::__shared_ptr (__gnu_cxx::_Lock_policy)2>::~__shared_ptr (this=0x7fffc1214508, 
> __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:914
> #10 0x7fffbdfc8fc4 in std::shared_ptr::~shared_ptr 
> (this=0x7fffc1214508, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr.h:93
> #11 0x7fffbdfc8fde in 
> __Pyx_call_destructor (x=...)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:281
> #12 0x7fffbdfbc317 in __pyx_tp_dealloc_7pyarrow_6plasma_PlasmaClient 
> (o=0x7fffc12144f0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:10383
> #13 0x7fffbdfb8986 in __pyx_pf_7pyarrow_6plasma_2connect (__pyx_self=0x0, 
> __pyx_v_store_socket_name=0x7fffbc922c48, 
> __pyx_v_manager_socket_name=0x77fa0ab0, __pyx_v_release_delay=0, 
> __pyx_v_num_retries=1)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:9147
> #14 0x7fffbdfb7dec in __pyx_pw_7pyarrow_6plasma_3connect (__pyx_self=0x0, 
> __pyx_args=0x7fffbc4d9688, __pyx_kwds=0x0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:8978
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (ARROW-2489) [Plasma] test_plasma.py crashes

2018-04-23 Thread ASF GitHub Bot (JIRA)

[ 
https://issues.apache.org/jira/browse/ARROW-2489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16447821#comment-16447821
 ] 

ASF GitHub Bot commented on ARROW-2489:
---

pitrou opened a new pull request #1933: ARROW-2489: [Plasma] Fix PlasmaClient 
ABI variation
URL: https://github.com/apache/arrow/pull/1933
 
 
   When compiled with GPU support, the PlasmaClient ABI would differ, leading 
to a crash in the Python bindings to Plasma.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


> [Plasma] test_plasma.py crashes
> ---
>
> Key: ARROW-2489
> URL: https://issues.apache.org/jira/browse/ARROW-2489
> Project: Apache Arrow
>  Issue Type: Bug
>  Components: GPU, Plasma (C++), Python
>Reporter: Antoine Pitrou
>Priority: Major
>  Labels: pull-request-available
>
> This is new here:
> {code}$ py.test   --tb=native pyarrow/tests/test_plasma.py 
> = test 
> session starts 
> ==
> platform linux -- Python 3.6.5, pytest-3.3.2, py-1.5.2, pluggy-0.6.0
> rootdir: /home/antoine/arrow/python, inifile: setup.cfg
> plugins: xdist-1.22.0, timeout-1.2.1, repeat-0.4.1, forked-0.2, 
> faulthandler-1.3.1
> collected 23 items
>   
>
> pyarrow/tests/test_plasma.py *** Error in 
> `/home/antoine/miniconda3/envs/pyarrow/bin/python': double free or corruption 
> (!prev): 0x01699520 ***
> [...]
> Current thread 0x7fe7e8570700 (most recent call first):
>   File "/home/antoine/arrow/python/pyarrow/tests/test_plasma.py", line 211 in 
> test_connection_failure_raises_exception
> [...]
> {code}
> Here is the C backtrace under gdb:
> {code}
> #0  0x769d0428 in __GI_raise (sig=sig@entry=6) at 
> ../sysdeps/unix/sysv/linux/raise.c:54
> #1  0x769d202a in __GI_abort () at abort.c:89
> #2  0x76a127ea in __libc_message (do_abort=do_abort@entry=2, 
> fmt=fmt@entry=0x76b2bed8 "*** Error in `%s': %s: 0x%s ***\n")
> at ../sysdeps/posix/libc_fatal.c:175
> #3  0x76a1b37a in malloc_printerr (ar_ptr=, 
> ptr=, str=0x76b2c008 "double free or corruption (!prev)", 
> action=3)
> at malloc.c:5006
> #4  _int_free (av=, p=, have_lock=0) at 
> malloc.c:3867
> #5  0x76a1f53c in __GI___libc_free (mem=) at 
> malloc.c:2968
> #6  0x7fffbdfcc504 in std::_Sp_counted_ptr (__gnu_cxx::_Lock_policy)2>::_M_dispose (this=0x9defb0)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:373
> #7  0x7fffbdfc903c in 
> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release (this=0x9defb0) 
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:149
> #8  0x7fffbdfc82b9 in 
> std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count 
> (this=0x7fffc1214510, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:666
> #9  0x7fffbdfc8276 in std::__shared_ptr (__gnu_cxx::_Lock_policy)2>::~__shared_ptr (this=0x7fffc1214508, 
> __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:914
> #10 0x7fffbdfc8fc4 in std::shared_ptr::~shared_ptr 
> (this=0x7fffc1214508, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr.h:93
> #11 0x7fffbdfc8fde in 
> __Pyx_call_destructor (x=...)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:281
> #12 0x7fffbdfbc317 in __pyx_tp_dealloc_7pyarrow_6plasma_PlasmaClient 
> (o=0x7fffc12144f0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:10383
> #13 0x7fffbdfb8986 in __pyx_pf_7pyarrow_6plasma_2connect (__pyx_self=0x0, 
> __pyx_v_store_socket_name=0x7fffbc922c48, 
> __pyx_v_manager_socket_name=0x77fa0ab0, __pyx_v_release_delay=0, 
> __pyx_v_num_retries=1)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:9147
> #14 0x7fffbdfb7dec in __pyx_pw_7pyarrow_6plasma_3connect (__pyx_self=0x0, 
> __pyx_args=0x7fffbc4d9688, __pyx_kwds=0x0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:8978
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (ARROW-2489) [Plasma] test_plasma.py crashes

2018-04-21 Thread Robert Nishihara (JIRA)

[ 
https://issues.apache.org/jira/browse/ARROW-2489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16446988#comment-16446988
 ] 

Robert Nishihara commented on ARROW-2489:
-

Ok, that makes sense.

> [Plasma] test_plasma.py crashes
> ---
>
> Key: ARROW-2489
> URL: https://issues.apache.org/jira/browse/ARROW-2489
> Project: Apache Arrow
>  Issue Type: Bug
>  Components: GPU, Plasma (C++), Python
>Reporter: Antoine Pitrou
>Priority: Major
>
> This is new here:
> {code}$ py.test   --tb=native pyarrow/tests/test_plasma.py 
> = test 
> session starts 
> ==
> platform linux -- Python 3.6.5, pytest-3.3.2, py-1.5.2, pluggy-0.6.0
> rootdir: /home/antoine/arrow/python, inifile: setup.cfg
> plugins: xdist-1.22.0, timeout-1.2.1, repeat-0.4.1, forked-0.2, 
> faulthandler-1.3.1
> collected 23 items
>   
>
> pyarrow/tests/test_plasma.py *** Error in 
> `/home/antoine/miniconda3/envs/pyarrow/bin/python': double free or corruption 
> (!prev): 0x01699520 ***
> [...]
> Current thread 0x7fe7e8570700 (most recent call first):
>   File "/home/antoine/arrow/python/pyarrow/tests/test_plasma.py", line 211 in 
> test_connection_failure_raises_exception
> [...]
> {code}
> Here is the C backtrace under gdb:
> {code}
> #0  0x769d0428 in __GI_raise (sig=sig@entry=6) at 
> ../sysdeps/unix/sysv/linux/raise.c:54
> #1  0x769d202a in __GI_abort () at abort.c:89
> #2  0x76a127ea in __libc_message (do_abort=do_abort@entry=2, 
> fmt=fmt@entry=0x76b2bed8 "*** Error in `%s': %s: 0x%s ***\n")
> at ../sysdeps/posix/libc_fatal.c:175
> #3  0x76a1b37a in malloc_printerr (ar_ptr=, 
> ptr=, str=0x76b2c008 "double free or corruption (!prev)", 
> action=3)
> at malloc.c:5006
> #4  _int_free (av=, p=, have_lock=0) at 
> malloc.c:3867
> #5  0x76a1f53c in __GI___libc_free (mem=) at 
> malloc.c:2968
> #6  0x7fffbdfcc504 in std::_Sp_counted_ptr (__gnu_cxx::_Lock_policy)2>::_M_dispose (this=0x9defb0)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:373
> #7  0x7fffbdfc903c in 
> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release (this=0x9defb0) 
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:149
> #8  0x7fffbdfc82b9 in 
> std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count 
> (this=0x7fffc1214510, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:666
> #9  0x7fffbdfc8276 in std::__shared_ptr (__gnu_cxx::_Lock_policy)2>::~__shared_ptr (this=0x7fffc1214508, 
> __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:914
> #10 0x7fffbdfc8fc4 in std::shared_ptr::~shared_ptr 
> (this=0x7fffc1214508, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr.h:93
> #11 0x7fffbdfc8fde in 
> __Pyx_call_destructor (x=...)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:281
> #12 0x7fffbdfbc317 in __pyx_tp_dealloc_7pyarrow_6plasma_PlasmaClient 
> (o=0x7fffc12144f0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:10383
> #13 0x7fffbdfb8986 in __pyx_pf_7pyarrow_6plasma_2connect (__pyx_self=0x0, 
> __pyx_v_store_socket_name=0x7fffbc922c48, 
> __pyx_v_manager_socket_name=0x77fa0ab0, __pyx_v_release_delay=0, 
> __pyx_v_num_retries=1)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:9147
> #14 0x7fffbdfb7dec in __pyx_pw_7pyarrow_6plasma_3connect (__pyx_self=0x0, 
> __pyx_args=0x7fffbc4d9688, __pyx_kwds=0x0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:8978
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (ARROW-2489) [Plasma] test_plasma.py crashes

2018-04-21 Thread Uwe L. Korn (JIRA)

[ 
https://issues.apache.org/jira/browse/ARROW-2489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16446982#comment-16446982
 ] 

Uwe L. Korn commented on ARROW-2489:


The size of the exposed class would always be fixed, only the class that 
contains the actual implementation would have a differing size. As this 
_internal_ class would only be used and visible inside of {{libplasma}}, all 
users of it would use the correct size.

> [Plasma] test_plasma.py crashes
> ---
>
> Key: ARROW-2489
> URL: https://issues.apache.org/jira/browse/ARROW-2489
> Project: Apache Arrow
>  Issue Type: Bug
>  Components: GPU, Plasma (C++), Python
>Reporter: Antoine Pitrou
>Priority: Major
>
> This is new here:
> {code}$ py.test   --tb=native pyarrow/tests/test_plasma.py 
> = test 
> session starts 
> ==
> platform linux -- Python 3.6.5, pytest-3.3.2, py-1.5.2, pluggy-0.6.0
> rootdir: /home/antoine/arrow/python, inifile: setup.cfg
> plugins: xdist-1.22.0, timeout-1.2.1, repeat-0.4.1, forked-0.2, 
> faulthandler-1.3.1
> collected 23 items
>   
>
> pyarrow/tests/test_plasma.py *** Error in 
> `/home/antoine/miniconda3/envs/pyarrow/bin/python': double free or corruption 
> (!prev): 0x01699520 ***
> [...]
> Current thread 0x7fe7e8570700 (most recent call first):
>   File "/home/antoine/arrow/python/pyarrow/tests/test_plasma.py", line 211 in 
> test_connection_failure_raises_exception
> [...]
> {code}
> Here is the C backtrace under gdb:
> {code}
> #0  0x769d0428 in __GI_raise (sig=sig@entry=6) at 
> ../sysdeps/unix/sysv/linux/raise.c:54
> #1  0x769d202a in __GI_abort () at abort.c:89
> #2  0x76a127ea in __libc_message (do_abort=do_abort@entry=2, 
> fmt=fmt@entry=0x76b2bed8 "*** Error in `%s': %s: 0x%s ***\n")
> at ../sysdeps/posix/libc_fatal.c:175
> #3  0x76a1b37a in malloc_printerr (ar_ptr=, 
> ptr=, str=0x76b2c008 "double free or corruption (!prev)", 
> action=3)
> at malloc.c:5006
> #4  _int_free (av=, p=, have_lock=0) at 
> malloc.c:3867
> #5  0x76a1f53c in __GI___libc_free (mem=) at 
> malloc.c:2968
> #6  0x7fffbdfcc504 in std::_Sp_counted_ptr (__gnu_cxx::_Lock_policy)2>::_M_dispose (this=0x9defb0)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:373
> #7  0x7fffbdfc903c in 
> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release (this=0x9defb0) 
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:149
> #8  0x7fffbdfc82b9 in 
> std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count 
> (this=0x7fffc1214510, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:666
> #9  0x7fffbdfc8276 in std::__shared_ptr (__gnu_cxx::_Lock_policy)2>::~__shared_ptr (this=0x7fffc1214508, 
> __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:914
> #10 0x7fffbdfc8fc4 in std::shared_ptr::~shared_ptr 
> (this=0x7fffc1214508, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr.h:93
> #11 0x7fffbdfc8fde in 
> __Pyx_call_destructor (x=...)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:281
> #12 0x7fffbdfbc317 in __pyx_tp_dealloc_7pyarrow_6plasma_PlasmaClient 
> (o=0x7fffc12144f0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:10383
> #13 0x7fffbdfb8986 in __pyx_pf_7pyarrow_6plasma_2connect (__pyx_self=0x0, 
> __pyx_v_store_socket_name=0x7fffbc922c48, 
> __pyx_v_manager_socket_name=0x77fa0ab0, __pyx_v_release_delay=0, 
> __pyx_v_num_retries=1)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:9147
> #14 0x7fffbdfb7dec in __pyx_pw_7pyarrow_6plasma_3connect (__pyx_self=0x0, 
> __pyx_args=0x7fffbc4d9688, __pyx_kwds=0x0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:8978
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (ARROW-2489) [Plasma] test_plasma.py crashes

2018-04-21 Thread Robert Nishihara (JIRA)

[ 
https://issues.apache.org/jira/browse/ARROW-2489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16446980#comment-16446980
 ] 

Robert Nishihara commented on ARROW-2489:
-

Oh wow, nice job tracking that down. I'm all for the PIMPL solution to 
Arrow-2448, but why would PIMPL solve this issue?

> [Plasma] test_plasma.py crashes
> ---
>
> Key: ARROW-2489
> URL: https://issues.apache.org/jira/browse/ARROW-2489
> Project: Apache Arrow
>  Issue Type: Bug
>  Components: GPU, Plasma (C++), Python
>Reporter: Antoine Pitrou
>Priority: Major
>
> This is new here:
> {code}$ py.test   --tb=native pyarrow/tests/test_plasma.py 
> = test 
> session starts 
> ==
> platform linux -- Python 3.6.5, pytest-3.3.2, py-1.5.2, pluggy-0.6.0
> rootdir: /home/antoine/arrow/python, inifile: setup.cfg
> plugins: xdist-1.22.0, timeout-1.2.1, repeat-0.4.1, forked-0.2, 
> faulthandler-1.3.1
> collected 23 items
>   
>
> pyarrow/tests/test_plasma.py *** Error in 
> `/home/antoine/miniconda3/envs/pyarrow/bin/python': double free or corruption 
> (!prev): 0x01699520 ***
> [...]
> Current thread 0x7fe7e8570700 (most recent call first):
>   File "/home/antoine/arrow/python/pyarrow/tests/test_plasma.py", line 211 in 
> test_connection_failure_raises_exception
> [...]
> {code}
> Here is the C backtrace under gdb:
> {code}
> #0  0x769d0428 in __GI_raise (sig=sig@entry=6) at 
> ../sysdeps/unix/sysv/linux/raise.c:54
> #1  0x769d202a in __GI_abort () at abort.c:89
> #2  0x76a127ea in __libc_message (do_abort=do_abort@entry=2, 
> fmt=fmt@entry=0x76b2bed8 "*** Error in `%s': %s: 0x%s ***\n")
> at ../sysdeps/posix/libc_fatal.c:175
> #3  0x76a1b37a in malloc_printerr (ar_ptr=, 
> ptr=, str=0x76b2c008 "double free or corruption (!prev)", 
> action=3)
> at malloc.c:5006
> #4  _int_free (av=, p=, have_lock=0) at 
> malloc.c:3867
> #5  0x76a1f53c in __GI___libc_free (mem=) at 
> malloc.c:2968
> #6  0x7fffbdfcc504 in std::_Sp_counted_ptr (__gnu_cxx::_Lock_policy)2>::_M_dispose (this=0x9defb0)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:373
> #7  0x7fffbdfc903c in 
> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release (this=0x9defb0) 
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:149
> #8  0x7fffbdfc82b9 in 
> std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count 
> (this=0x7fffc1214510, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:666
> #9  0x7fffbdfc8276 in std::__shared_ptr (__gnu_cxx::_Lock_policy)2>::~__shared_ptr (this=0x7fffc1214508, 
> __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:914
> #10 0x7fffbdfc8fc4 in std::shared_ptr::~shared_ptr 
> (this=0x7fffc1214508, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr.h:93
> #11 0x7fffbdfc8fde in 
> __Pyx_call_destructor (x=...)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:281
> #12 0x7fffbdfbc317 in __pyx_tp_dealloc_7pyarrow_6plasma_PlasmaClient 
> (o=0x7fffc12144f0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:10383
> #13 0x7fffbdfb8986 in __pyx_pf_7pyarrow_6plasma_2connect (__pyx_self=0x0, 
> __pyx_v_store_socket_name=0x7fffbc922c48, 
> __pyx_v_manager_socket_name=0x77fa0ab0, __pyx_v_release_delay=0, 
> __pyx_v_num_retries=1)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:9147
> #14 0x7fffbdfb7dec in __pyx_pw_7pyarrow_6plasma_3connect (__pyx_self=0x0, 
> __pyx_args=0x7fffbc4d9688, __pyx_kwds=0x0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:8978
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (ARROW-2489) [Plasma] test_plasma.py crashes

2018-04-21 Thread Antoine Pitrou (JIRA)

[ 
https://issues.apache.org/jira/browse/ARROW-2489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16446943#comment-16446943
 ] 

Antoine Pitrou commented on ARROW-2489:
---

Ok, it's an ABI issue.

When you compile Plasma with GPU support enabled, there's an additional pointer 
field at the end of {{PlasmaClient}}. However, pyarrow's {{setup.py}} doesn't 
enable {{PLASMA_GPU}} (AFAICT, it has no easy way of deciding how to enable 
it). So the Cython extension allocates a {{PlasmaClient}} with 8 missing bytes 
at the end, and calls the {{PlasmaClient}} constructor which writes past the 
allocated memory. Depending on the exact structure size, this may garble some 
safety tags used by the libc's allocator, hence the error and the crash.

One solution is to migrate {{PlasmaClient}} to use a pimpl, which could also 
help fixing ARROW-2448.

> [Plasma] test_plasma.py crashes
> ---
>
> Key: ARROW-2489
> URL: https://issues.apache.org/jira/browse/ARROW-2489
> Project: Apache Arrow
>  Issue Type: Bug
>  Components: GPU, Plasma (C++), Python
>Reporter: Antoine Pitrou
>Priority: Major
>
> This is new here:
> {code}$ py.test   --tb=native pyarrow/tests/test_plasma.py 
> = test 
> session starts 
> ==
> platform linux -- Python 3.6.5, pytest-3.3.2, py-1.5.2, pluggy-0.6.0
> rootdir: /home/antoine/arrow/python, inifile: setup.cfg
> plugins: xdist-1.22.0, timeout-1.2.1, repeat-0.4.1, forked-0.2, 
> faulthandler-1.3.1
> collected 23 items
>   
>
> pyarrow/tests/test_plasma.py *** Error in 
> `/home/antoine/miniconda3/envs/pyarrow/bin/python': double free or corruption 
> (!prev): 0x01699520 ***
> [...]
> Current thread 0x7fe7e8570700 (most recent call first):
>   File "/home/antoine/arrow/python/pyarrow/tests/test_plasma.py", line 211 in 
> test_connection_failure_raises_exception
> [...]
> {code}
> Here is the C backtrace under gdb:
> {code}
> #0  0x769d0428 in __GI_raise (sig=sig@entry=6) at 
> ../sysdeps/unix/sysv/linux/raise.c:54
> #1  0x769d202a in __GI_abort () at abort.c:89
> #2  0x76a127ea in __libc_message (do_abort=do_abort@entry=2, 
> fmt=fmt@entry=0x76b2bed8 "*** Error in `%s': %s: 0x%s ***\n")
> at ../sysdeps/posix/libc_fatal.c:175
> #3  0x76a1b37a in malloc_printerr (ar_ptr=, 
> ptr=, str=0x76b2c008 "double free or corruption (!prev)", 
> action=3)
> at malloc.c:5006
> #4  _int_free (av=, p=, have_lock=0) at 
> malloc.c:3867
> #5  0x76a1f53c in __GI___libc_free (mem=) at 
> malloc.c:2968
> #6  0x7fffbdfcc504 in std::_Sp_counted_ptr (__gnu_cxx::_Lock_policy)2>::_M_dispose (this=0x9defb0)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:373
> #7  0x7fffbdfc903c in 
> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release (this=0x9defb0) 
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:149
> #8  0x7fffbdfc82b9 in 
> std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count 
> (this=0x7fffc1214510, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:666
> #9  0x7fffbdfc8276 in std::__shared_ptr (__gnu_cxx::_Lock_policy)2>::~__shared_ptr (this=0x7fffc1214508, 
> __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:914
> #10 0x7fffbdfc8fc4 in std::shared_ptr::~shared_ptr 
> (this=0x7fffc1214508, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr.h:93
> #11 0x7fffbdfc8fde in 
> __Pyx_call_destructor (x=...)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:281
> #12 0x7fffbdfbc317 in __pyx_tp_dealloc_7pyarrow_6plasma_PlasmaClient 
> (o=0x7fffc12144f0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:10383
> #13 0x7fffbdfb8986 in __pyx_pf_7pyarrow_6plasma_2connect (__pyx_self=0x0, 
> __pyx_v_store_socket_name=0x7fffbc922c48, 
> __pyx_v_manager_socket_name=0x77fa0ab0, __pyx_v_release_delay=0, 
> __pyx_v_num_retries=1)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:9147
> #14 0x7fffbdfb7dec in __pyx_pw_7pyarrow_6plasma_3connect (__pyx_self=0x0, 
> __pyx_args=0x7fffbc4d9688, __pyx_kwds=0x0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:8978
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (ARROW-2489) [Plasma] test_plasma.py crashes

2018-04-21 Thread Antoine Pitrou (JIRA)

[ 
https://issues.apache.org/jira/browse/ARROW-2489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16446916#comment-16446916
 ] 

Antoine Pitrou commented on ARROW-2489:
---

This seems somehow related to GPU support. If I disable the following line in 
the plasma client constructor:
{code:cpp}
CudaDeviceManager::GetInstance(_);
{code}
then the crash goes away.

> [Plasma] test_plasma.py crashes
> ---
>
> Key: ARROW-2489
> URL: https://issues.apache.org/jira/browse/ARROW-2489
> Project: Apache Arrow
>  Issue Type: Bug
>  Components: Plasma (C++)
>Reporter: Antoine Pitrou
>Priority: Major
>
> This is new here:
> {code}$ py.test   --tb=native pyarrow/tests/test_plasma.py 
> = test 
> session starts 
> ==
> platform linux -- Python 3.6.5, pytest-3.3.2, py-1.5.2, pluggy-0.6.0
> rootdir: /home/antoine/arrow/python, inifile: setup.cfg
> plugins: xdist-1.22.0, timeout-1.2.1, repeat-0.4.1, forked-0.2, 
> faulthandler-1.3.1
> collected 23 items
>   
>
> pyarrow/tests/test_plasma.py *** Error in 
> `/home/antoine/miniconda3/envs/pyarrow/bin/python': double free or corruption 
> (!prev): 0x01699520 ***
> [...]
> Current thread 0x7fe7e8570700 (most recent call first):
>   File "/home/antoine/arrow/python/pyarrow/tests/test_plasma.py", line 211 in 
> test_connection_failure_raises_exception
> [...]
> {code}
> Here is the C backtrace under gdb:
> {code}
> #0  0x769d0428 in __GI_raise (sig=sig@entry=6) at 
> ../sysdeps/unix/sysv/linux/raise.c:54
> #1  0x769d202a in __GI_abort () at abort.c:89
> #2  0x76a127ea in __libc_message (do_abort=do_abort@entry=2, 
> fmt=fmt@entry=0x76b2bed8 "*** Error in `%s': %s: 0x%s ***\n")
> at ../sysdeps/posix/libc_fatal.c:175
> #3  0x76a1b37a in malloc_printerr (ar_ptr=, 
> ptr=, str=0x76b2c008 "double free or corruption (!prev)", 
> action=3)
> at malloc.c:5006
> #4  _int_free (av=, p=, have_lock=0) at 
> malloc.c:3867
> #5  0x76a1f53c in __GI___libc_free (mem=) at 
> malloc.c:2968
> #6  0x7fffbdfcc504 in std::_Sp_counted_ptr (__gnu_cxx::_Lock_policy)2>::_M_dispose (this=0x9defb0)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:373
> #7  0x7fffbdfc903c in 
> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release (this=0x9defb0) 
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:149
> #8  0x7fffbdfc82b9 in 
> std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count 
> (this=0x7fffc1214510, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:666
> #9  0x7fffbdfc8276 in std::__shared_ptr (__gnu_cxx::_Lock_policy)2>::~__shared_ptr (this=0x7fffc1214508, 
> __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:914
> #10 0x7fffbdfc8fc4 in std::shared_ptr::~shared_ptr 
> (this=0x7fffc1214508, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr.h:93
> #11 0x7fffbdfc8fde in 
> __Pyx_call_destructor (x=...)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:281
> #12 0x7fffbdfbc317 in __pyx_tp_dealloc_7pyarrow_6plasma_PlasmaClient 
> (o=0x7fffc12144f0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:10383
> #13 0x7fffbdfb8986 in __pyx_pf_7pyarrow_6plasma_2connect (__pyx_self=0x0, 
> __pyx_v_store_socket_name=0x7fffbc922c48, 
> __pyx_v_manager_socket_name=0x77fa0ab0, __pyx_v_release_delay=0, 
> __pyx_v_num_retries=1)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:9147
> #14 0x7fffbdfb7dec in __pyx_pw_7pyarrow_6plasma_3connect (__pyx_self=0x0, 
> __pyx_args=0x7fffbc4d9688, __pyx_kwds=0x0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:8978
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (ARROW-2489) [Plasma] test_plasma.py crashes

2018-04-21 Thread Antoine Pitrou (JIRA)

[ 
https://issues.apache.org/jira/browse/ARROW-2489?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16446902#comment-16446902
 ] 

Antoine Pitrou commented on ARROW-2489:
---

Bisecting tells me the following commit introduced the issue:

{code}
commit 7eeca3ac823d5dda05f018661b453354cb782406
Author: Philipp Moritz 
Date:   Thu Apr 19 12:25:27 2018 -0700

ARROW-2458: [Plasma] Use one thread pool per PlasmaClient

This removes the global `thread pool_` variable from the PlasmaClient.

cc @robertnishihara @pitrou

Author: Philipp Moritz 

Closes #1893 from pcmoritz/fix-plasma-client and squashes the following 
commits:

742b542  fix linting
b1130bd  remove global variable from PlasmaClient
{code}

[~pcmoritz]

> [Plasma] test_plasma.py crashes
> ---
>
> Key: ARROW-2489
> URL: https://issues.apache.org/jira/browse/ARROW-2489
> Project: Apache Arrow
>  Issue Type: Bug
>  Components: Plasma (C++)
>Reporter: Antoine Pitrou
>Priority: Major
>
> This is new here:
> {code}$ py.test   --tb=native pyarrow/tests/test_plasma.py 
> = test 
> session starts 
> ==
> platform linux -- Python 3.6.5, pytest-3.3.2, py-1.5.2, pluggy-0.6.0
> rootdir: /home/antoine/arrow/python, inifile: setup.cfg
> plugins: xdist-1.22.0, timeout-1.2.1, repeat-0.4.1, forked-0.2, 
> faulthandler-1.3.1
> collected 23 items
>   
>
> pyarrow/tests/test_plasma.py *** Error in 
> `/home/antoine/miniconda3/envs/pyarrow/bin/python': double free or corruption 
> (!prev): 0x01699520 ***
> [...]
> Current thread 0x7fe7e8570700 (most recent call first):
>   File "/home/antoine/arrow/python/pyarrow/tests/test_plasma.py", line 211 in 
> test_connection_failure_raises_exception
> [...]
> {code}
> Here is the C backtrace under gdb:
> {code}
> #0  0x769d0428 in __GI_raise (sig=sig@entry=6) at 
> ../sysdeps/unix/sysv/linux/raise.c:54
> #1  0x769d202a in __GI_abort () at abort.c:89
> #2  0x76a127ea in __libc_message (do_abort=do_abort@entry=2, 
> fmt=fmt@entry=0x76b2bed8 "*** Error in `%s': %s: 0x%s ***\n")
> at ../sysdeps/posix/libc_fatal.c:175
> #3  0x76a1b37a in malloc_printerr (ar_ptr=, 
> ptr=, str=0x76b2c008 "double free or corruption (!prev)", 
> action=3)
> at malloc.c:5006
> #4  _int_free (av=, p=, have_lock=0) at 
> malloc.c:3867
> #5  0x76a1f53c in __GI___libc_free (mem=) at 
> malloc.c:2968
> #6  0x7fffbdfcc504 in std::_Sp_counted_ptr (__gnu_cxx::_Lock_policy)2>::_M_dispose (this=0x9defb0)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:373
> #7  0x7fffbdfc903c in 
> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release (this=0x9defb0) 
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:149
> #8  0x7fffbdfc82b9 in 
> std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count 
> (this=0x7fffc1214510, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:666
> #9  0x7fffbdfc8276 in std::__shared_ptr (__gnu_cxx::_Lock_policy)2>::~__shared_ptr (this=0x7fffc1214508, 
> __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr_base.h:914
> #10 0x7fffbdfc8fc4 in std::shared_ptr::~shared_ptr 
> (this=0x7fffc1214508, __in_chrg=)
> at /usr/include/c++/4.9/bits/shared_ptr.h:93
> #11 0x7fffbdfc8fde in 
> __Pyx_call_destructor (x=...)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:281
> #12 0x7fffbdfbc317 in __pyx_tp_dealloc_7pyarrow_6plasma_PlasmaClient 
> (o=0x7fffc12144f0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:10383
> #13 0x7fffbdfb8986 in __pyx_pf_7pyarrow_6plasma_2connect (__pyx_self=0x0, 
> __pyx_v_store_socket_name=0x7fffbc922c48, 
> __pyx_v_manager_socket_name=0x77fa0ab0, __pyx_v_release_delay=0, 
> __pyx_v_num_retries=1)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:9147
> #14 0x7fffbdfb7dec in __pyx_pw_7pyarrow_6plasma_3connect (__pyx_self=0x0, 
> __pyx_args=0x7fffbc4d9688, __pyx_kwds=0x0)
> at /home/antoine/arrow/python/build/temp.linux-x86_64-3.6/plasma.cxx:8978
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)