cx-oracle-users

[cx-oracle-users] Segmentation fault in Python 3 - Cursor - chained exceptions

[Boris D. <bor...@gm...>]

Hi all,

The following test program gives a segmentation fault:

C:\>python
Python 3.3.3 (v3.3.3:c3896275c0f6, Nov 18 2013, 21:19:30) [MSC v.1600 64
bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import cx_Oracle
>>> c=cx_Oracle.connect(<connect string>).cursor()
>>> try:
...     c.execute("select")
... except:
...     try:
...             c.execute("select")
...     except:
...             pass
...

C:\>

Access Violation
Faulting application name: python.exe
Fault offset: 0x000000000000cacd


Platform: 64bit, Win7, Oracle Client 11g (11.2.0.1.0), python 3.3.3,
cx_Oracle 5.1.2



At a guess, the problem is because in the function Cursor_SetErrorOffset()
/Cursor.c/:

static void Cursor_SetErrorOffset(
    udt_Cursor *self)                   // cursor to get the error offset
from
{
    PyObject *type, *value, *traceback;
    udt_Error *error;

    PyErr_Fetch(&type, &value, &traceback);
    if (type == g_DatabaseErrorException) {
        error = (udt_Error*) value;
        OCIAttrGet(self->handle, OCI_HTYPE_STMT, &error->offset, 0,
                OCI_ATTR_PARSE_ERROR_OFFSET,
self->environment->errorHandle);
    }
    PyErr_Restore(type, value, traceback);
}

OCIAttrGet() call attempts to write a value to the address
/&error->offset/,
therefore after the end of an allocated memory block of the borrowed
pointer /PyObject *value/

typedef struct {
    PyObject_HEAD
    sb4 code;
    ub4 offset;
    PyObject *message;
    const char *context;
} udt_Error;


Commenting it out /* OCIAttrGet(self->handle, OCI_HTYPE_STMT,
&error->offset, 0, OCI_ATTR_PARSE_ERROR_OFFSET,
self->environment->errorHandle);*/
fix this crash!


Regards,
Boris

Re: [cx-oracle-users] Segmentation fault in Python 3 - Cursor - chained exceptions

[Anthony T. <ant...@gm...>]

Thanks for the simple program to demonstrate the problem. :-) I have just
pushed a change to the source that resolves this issue. Apparently in
Python 3 the first time the exception is referenced it is not normalized
and so the error object I am creating is immediately accessible. The second
level of exception results in a normalized exception which means that it
can't be accessed in the same way. Fortunately the normalization can be
forced so that I can access the error object the same way each time. A
subtle and rare bug so I appreciate the pointer!

Anthony


On Tue, May 13, 2014 at 1:15 AM, Boris Dzuba <bor...@gm...> wrote:

> Hi all,
>
> The following test program gives a segmentation fault:
>
> C:\>python
> Python 3.3.3 (v3.3.3:c3896275c0f6, Nov 18 2013, 21:19:30) [MSC v.1600 64
> bit (AMD64)] on win32
> Type "help", "copyright", "credits" or "license" for more information.
> >>> import cx_Oracle
> >>> c=cx_Oracle.connect(<connect string>).cursor()
> >>> try:
> ...     c.execute("select")
> ... except:
> ...     try:
> ...             c.execute("select")
> ...     except:
> ...             pass
> ...
>
> C:\>
>
> Access Violation
> Faulting application name: python.exe
> Fault offset: 0x000000000000cacd
>
>
> Platform: 64bit, Win7, Oracle Client 11g (11.2.0.1.0), python 3.3.3,
> cx_Oracle 5.1.2
>
>
>
> At a guess, the problem is because in the function Cursor_SetErrorOffset()
> /Cursor.c/:
>
> static void Cursor_SetErrorOffset(
>     udt_Cursor *self)                   // cursor to get the error offset
> from
> {
>     PyObject *type, *value, *traceback;
>     udt_Error *error;
>
>     PyErr_Fetch(&type, &value, &traceback);
>     if (type == g_DatabaseErrorException) {
>         error = (udt_Error*) value;
>         OCIAttrGet(self->handle, OCI_HTYPE_STMT, &error->offset, 0,
>                 OCI_ATTR_PARSE_ERROR_OFFSET,
> self->environment->errorHandle);
>     }
>     PyErr_Restore(type, value, traceback);
> }
>
> OCIAttrGet() call attempts to write a value to the address
> /&error->offset/,
> therefore after the end of an allocated memory block of the borrowed
> pointer /PyObject *value/
>
> typedef struct {
>     PyObject_HEAD
>     sb4 code;
>     ub4 offset;
>     PyObject *message;
>     const char *context;
> } udt_Error;
>
>
> Commenting it out /* OCIAttrGet(self->handle, OCI_HTYPE_STMT,
> &error->offset, 0, OCI_ATTR_PARSE_ERROR_OFFSET,
> self->environment->errorHandle);*/
> fix this crash!
>
>
> Regards,
> Boris
>
>
> ------------------------------------------------------------------------------
> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
> Instantly run your Selenium tests across 300+ browser/OS combos.
> Get unparalleled scalability from the best Selenium testing platform
> available
> Simple to use. Nothing to install. Get started now for free."
> http://p.sf.net/sfu/SauceLabs
> _______________________________________________
> cx-oracle-users mailing list
> cx-...@li...
> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>
>

Re: [cx-oracle-users] Segmentation fault in Python 3 - Cursor - chained exceptions

[Boris D. <bor...@gm...>]

Thank you for the quick response. Do you already have any plans, when will
new version be released ?


2014-05-13 23:17 GMT+02:00 Anthony Tuininga <ant...@gm...>:

> Thanks for the simple program to demonstrate the problem. :-) I have just
> pushed a change to the source that resolves this issue. Apparently in
> Python 3 the first time the exception is referenced it is not normalized
> and so the error object I am creating is immediately accessible. The second
> level of exception results in a normalized exception which means that it
> can't be accessed in the same way. Fortunately the normalization can be
> forced so that I can access the error object the same way each time. A
> subtle and rare bug so I appreciate the pointer!
>
> Anthony
>
>
> On Tue, May 13, 2014 at 1:15 AM, Boris Dzuba <bor...@gm...>wrote:
>
>> Hi all,
>>
>> The following test program gives a segmentation fault:
>>
>> C:\>python
>> Python 3.3.3 (v3.3.3:c3896275c0f6, Nov 18 2013, 21:19:30) [MSC v.1600 64
>> bit (AMD64)] on win32
>> Type "help", "copyright", "credits" or "license" for more information.
>> >>> import cx_Oracle
>> >>> c=cx_Oracle.connect(<connect string>).cursor()
>> >>> try:
>> ...     c.execute("select")
>> ... except:
>> ...     try:
>> ...             c.execute("select")
>> ...     except:
>> ...             pass
>> ...
>>
>> C:\>
>>
>> Access Violation
>> Faulting application name: python.exe
>> Fault offset: 0x000000000000cacd
>>
>>
>> Platform: 64bit, Win7, Oracle Client 11g (11.2.0.1.0), python 3.3.3,
>> cx_Oracle 5.1.2
>>
>>
>>
>> At a guess, the problem is because in the function
>> Cursor_SetErrorOffset() /Cursor.c/:
>>
>> static void Cursor_SetErrorOffset(
>>     udt_Cursor *self)                   // cursor to get the error offset
>> from
>> {
>>     PyObject *type, *value, *traceback;
>>     udt_Error *error;
>>
>>     PyErr_Fetch(&type, &value, &traceback);
>>     if (type == g_DatabaseErrorException) {
>>         error = (udt_Error*) value;
>>         OCIAttrGet(self->handle, OCI_HTYPE_STMT, &error->offset, 0,
>>                 OCI_ATTR_PARSE_ERROR_OFFSET,
>> self->environment->errorHandle);
>>     }
>>     PyErr_Restore(type, value, traceback);
>> }
>>
>> OCIAttrGet() call attempts to write a value to the address
>> /&error->offset/,
>> therefore after the end of an allocated memory block of the borrowed
>> pointer /PyObject *value/
>>
>> typedef struct {
>>     PyObject_HEAD
>>     sb4 code;
>>     ub4 offset;
>>     PyObject *message;
>>     const char *context;
>> } udt_Error;
>>
>>
>> Commenting it out /* OCIAttrGet(self->handle, OCI_HTYPE_STMT,
>> &error->offset, 0, OCI_ATTR_PARSE_ERROR_OFFSET,
>> self->environment->errorHandle);*/
>> fix this crash!
>>
>>
>> Regards,
>> Boris
>>
>>
>> ------------------------------------------------------------------------------
>> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
>> Instantly run your Selenium tests across 300+ browser/OS combos.
>> Get unparalleled scalability from the best Selenium testing platform
>> available
>> Simple to use. Nothing to install. Get started now for free."
>> http://p.sf.net/sfu/SauceLabs
>> _______________________________________________
>> cx-oracle-users mailing list
>> cx-...@li...
>> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>>
>>
>
>
> ------------------------------------------------------------------------------
> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
> Instantly run your Selenium tests across 300+ browser/OS combos.
> Get unparalleled scalability from the best Selenium testing platform
> available
> Simple to use. Nothing to install. Get started now for free."
> http://p.sf.net/sfu/SauceLabs
> _______________________________________________
> cx-oracle-users mailing list
> cx-...@li...
> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>
>

Re: [cx-oracle-users] Segmentation fault in Python 3 - Cursor - chained exceptions

[Anthony T. <ant...@gm...>]

I am hoping this weekend to make a new release. It has been far too long! I
need binaries for Python 3.4 and Oracle 12c at the least. :-)


On Wed, May 14, 2014 at 6:28 AM, Boris Dzuba <bor...@gm...> wrote:

> Thank you for the quick response. Do you already have any plans, when will
> new version be released ?
>
>
> 2014-05-13 23:17 GMT+02:00 Anthony Tuininga <ant...@gm...>:
>
>> Thanks for the simple program to demonstrate the problem. :-) I have just
>> pushed a change to the source that resolves this issue. Apparently in
>> Python 3 the first time the exception is referenced it is not normalized
>> and so the error object I am creating is immediately accessible. The second
>> level of exception results in a normalized exception which means that it
>> can't be accessed in the same way. Fortunately the normalization can be
>> forced so that I can access the error object the same way each time. A
>> subtle and rare bug so I appreciate the pointer!
>>
>> Anthony
>>
>>
>> On Tue, May 13, 2014 at 1:15 AM, Boris Dzuba <bor...@gm...>wrote:
>>
>>> Hi all,
>>>
>>> The following test program gives a segmentation fault:
>>>
>>> C:\>python
>>> Python 3.3.3 (v3.3.3:c3896275c0f6, Nov 18 2013, 21:19:30) [MSC v.1600 64
>>> bit (AMD64)] on win32
>>> Type "help", "copyright", "credits" or "license" for more information.
>>> >>> import cx_Oracle
>>> >>> c=cx_Oracle.connect(<connect string>).cursor()
>>> >>> try:
>>> ...     c.execute("select")
>>> ... except:
>>> ...     try:
>>> ...             c.execute("select")
>>> ...     except:
>>> ...             pass
>>> ...
>>>
>>> C:\>
>>>
>>> Access Violation
>>> Faulting application name: python.exe
>>> Fault offset: 0x000000000000cacd
>>>
>>>
>>> Platform: 64bit, Win7, Oracle Client 11g (11.2.0.1.0), python 3.3.3,
>>> cx_Oracle 5.1.2
>>>
>>>
>>>
>>> At a guess, the problem is because in the function
>>> Cursor_SetErrorOffset() /Cursor.c/:
>>>
>>> static void Cursor_SetErrorOffset(
>>>     udt_Cursor *self)                   // cursor to get the error
>>> offset from
>>> {
>>>     PyObject *type, *value, *traceback;
>>>     udt_Error *error;
>>>
>>>     PyErr_Fetch(&type, &value, &traceback);
>>>     if (type == g_DatabaseErrorException) {
>>>         error = (udt_Error*) value;
>>>         OCIAttrGet(self->handle, OCI_HTYPE_STMT, &error->offset, 0,
>>>                 OCI_ATTR_PARSE_ERROR_OFFSET,
>>> self->environment->errorHandle);
>>>     }
>>>     PyErr_Restore(type, value, traceback);
>>> }
>>>
>>> OCIAttrGet() call attempts to write a value to the address
>>> /&error->offset/,
>>> therefore after the end of an allocated memory block of the borrowed
>>> pointer /PyObject *value/
>>>
>>> typedef struct {
>>>     PyObject_HEAD
>>>     sb4 code;
>>>     ub4 offset;
>>>     PyObject *message;
>>>     const char *context;
>>> } udt_Error;
>>>
>>>
>>> Commenting it out /* OCIAttrGet(self->handle, OCI_HTYPE_STMT,
>>> &error->offset, 0, OCI_ATTR_PARSE_ERROR_OFFSET,
>>> self->environment->errorHandle);*/
>>> fix this crash!
>>>
>>>
>>> Regards,
>>> Boris
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
>>> Instantly run your Selenium tests across 300+ browser/OS combos.
>>> Get unparalleled scalability from the best Selenium testing platform
>>> available
>>> Simple to use. Nothing to install. Get started now for free."
>>> http://p.sf.net/sfu/SauceLabs
>>> _______________________________________________
>>> cx-oracle-users mailing list
>>> cx-...@li...
>>> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>>>
>>>
>>
>>
>> ------------------------------------------------------------------------------
>> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
>> Instantly run your Selenium tests across 300+ browser/OS combos.
>> Get unparalleled scalability from the best Selenium testing platform
>> available
>> Simple to use. Nothing to install. Get started now for free."
>> http://p.sf.net/sfu/SauceLabs
>> _______________________________________________
>> cx-oracle-users mailing list
>> cx-...@li...
>> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>>
>>
>
>
> ------------------------------------------------------------------------------
> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
> Instantly run your Selenium tests across 300+ browser/OS combos.
> Get unparalleled scalability from the best Selenium testing platform
> available
> Simple to use. Nothing to install. Get started now for free."
> http://p.sf.net/sfu/SauceLabs
> _______________________________________________
> cx-oracle-users mailing list
> cx-...@li...
> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>
>

Re: [cx-oracle-users] Segmentation fault in Python 3 - Cursor - chained exceptions

[Krishna M. IV <kri...@or...>]

Hi Anthony,
   does it require a new binary for Oracle 12c? The existing one should 
work with 12c client also, right? Unless you are adding some 12c 
specific features...

thanks,
krishna

On 5/14/2014 7:45 PM, Anthony Tuininga wrote:
> I am hoping this weekend to make a new release. It has been far too 
> long! I need binaries for Python 3.4 and Oracle 12c at the least. :-)
>
>
> On Wed, May 14, 2014 at 6:28 AM, Boris Dzuba <bor...@gm... 
> <mailto:bor...@gm...>> wrote:
>
>     Thank you for the quick response. Do you already have any plans,
>     when will new version be released ?
>
>
>     2014-05-13 23:17 GMT+02:00 Anthony Tuininga
>     <ant...@gm... <mailto:ant...@gm...>>:
>
>         Thanks for the simple program to demonstrate the problem. :-)
>         I have just pushed a change to the source that resolves this
>         issue. Apparently in Python 3 the first time the exception is
>         referenced it is not normalized and so the error object I am
>         creating is immediately accessible. The second level of
>         exception results in a normalized exception which means that
>         it can't be accessed in the same way. Fortunately the
>         normalization can be forced so that I can access the error
>         object the same way each time. A subtle and rare bug so I
>         appreciate the pointer!
>
>         Anthony
>
>
>         On Tue, May 13, 2014 at 1:15 AM, Boris Dzuba
>         <bor...@gm... <mailto:bor...@gm...>> wrote:
>
>             Hi all,
>
>             The following test program gives a segmentation fault:
>
>             C:\>python
>             Python 3.3.3 (v3.3.3:c3896275c0f6, Nov 18 2013, 21:19:30)
>             [MSC v.1600 64 bit (AMD64)] on win32
>             Type "help", "copyright", "credits" or "license" for more
>             information.
>             >>> import cx_Oracle
>             >>> c=cx_Oracle.connect(<connect string>).cursor()
>             >>> try:
>             ...     c.execute("select")
>             ... except:
>             ...     try:
>             ...             c.execute("select")
>             ...     except:
>             ...             pass
>             ...
>
>             C:\>
>
>             Access Violation
>             Faulting application name: python.exe
>             Fault offset: 0x000000000000cacd
>
>
>             Platform: 64bit, Win7, Oracle Client 11g (11.2.0.1.0),
>             python 3.3.3, cx_Oracle 5.1.2
>
>
>
>             At a guess, the problem is because in the function
>             Cursor_SetErrorOffset() /Cursor.c/:
>
>             static void Cursor_SetErrorOffset(
>                 udt_Cursor *self) // cursor to get the error offset from
>             {
>                 PyObject *type, *value, *traceback;
>                 udt_Error *error;
>
>                 PyErr_Fetch(&type, &value, &traceback);
>                 if (type == g_DatabaseErrorException) {
>                     error = (udt_Error*) value;
>                     OCIAttrGet(self->handle, OCI_HTYPE_STMT,
>             &error->offset, 0,
>             OCI_ATTR_PARSE_ERROR_OFFSET, self->environment->errorHandle);
>                 }
>                 PyErr_Restore(type, value, traceback);
>             }
>             OCIAttrGet() call attempts to write a value to the address
>             /&error->offset/,
>             therefore after the end of an allocated memory block of
>             the borrowed pointer /PyObject *value/
>
>             typedef struct {
>                 PyObject_HEAD
>                 sb4 code;
>                 ub4 offset;
>                 PyObject *message;
>                 const char *context;
>             } udt_Error;
>
>
>             Commenting it out /* OCIAttrGet(self->handle,
>             OCI_HTYPE_STMT, &error->offset, 0,
>             OCI_ATTR_PARSE_ERROR_OFFSET,
>             self->environment->errorHandle);*/
>             fix this crash!
>
>
>             Regards,
>             Boris
>
>             ------------------------------------------------------------------------------
>             "Accelerate Dev Cycles with Automated Cross-Browser
>             Testing - For FREE
>             Instantly run your Selenium tests across 300+ browser/OS
>             combos.
>             Get unparalleled scalability from the best Selenium
>             testing platform available
>             Simple to use. Nothing to install. Get started now for free."
>             http://p.sf.net/sfu/SauceLabs
>             _______________________________________________
>             cx-oracle-users mailing list
>             cx-...@li...
>             <mailto:cx-...@li...>
>             https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>
>
>
>         ------------------------------------------------------------------------------
>         "Accelerate Dev Cycles with Automated Cross-Browser Testing -
>         For FREE
>         Instantly run your Selenium tests across 300+ browser/OS combos.
>         Get unparalleled scalability from the best Selenium testing
>         platform available
>         Simple to use. Nothing to install. Get started now for free."
>         http://p.sf.net/sfu/SauceLabs
>         _______________________________________________
>         cx-oracle-users mailing list
>         cx-...@li...
>         <mailto:cx-...@li...>
>         https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>
>
>
>     ------------------------------------------------------------------------------
>     "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
>     Instantly run your Selenium tests across 300+ browser/OS combos.
>     Get unparalleled scalability from the best Selenium testing
>     platform available
>     Simple to use. Nothing to install. Get started now for free."
>     http://p.sf.net/sfu/SauceLabs
>     _______________________________________________
>     cx-oracle-users mailing list
>     cx-...@li...
>     <mailto:cx-...@li...>
>     https://lists.sourceforge.net/lists/listinfo/cx-oracle-users
>
>
>
>
> ------------------------------------------------------------------------------
> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
> Instantly run your Selenium tests across 300+ browser/OS combos.
> Get unparalleled scalability from the best Selenium testing platform available
> Simple to use. Nothing to install. Get started now for free."
> http://p.sf.net/sfu/SauceLabs
>
>
> _______________________________________________
> cx-oracle-users mailing list
> cx-...@li...
> https://lists.sourceforge.net/lists/listinfo/cx-oracle-users