FreeChainXenon/binutils-gdb
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
binutils-gdb/gdb/python/py-micmd.c

740 lines
23 KiB
C
Raw Normal View History

gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
/* MI Command Set for GDB, the GNU debugger.
Copyright (C) 2019-2022 Free Software Foundation, Inc.
This file is part of GDB.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>. */
/* GDB/MI commands implemented in Python. */
#include "defs.h"
#include "python-internal.h"
#include "arch-utils.h"
#include "charset.h"
#include "language.h"
#include "mi/mi-cmds.h"
#include "mi/mi-parse.h"
#include "cli/cli-cmds.h"
#include <string>
/* Debugging of Python MI commands. */
static bool pymicmd_debug;
/* Implementation of "show debug py-micmd". */
static void
show_pymicmd_debug (struct ui_file *file, int from_tty,
struct cmd_list_element *c, const char *value)
{
Unify gdb printf functions Now that filtered and unfiltered output can be treated identically, we can unify the printf family of functions. This is done under the name "gdb_printf". Most of this patch was written by script.
2022-01-02 11:46:15 -07:00
gdb_printf (file, _("Python MI command debugging is %s.\n"), value);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
}
/* Print a "py-micmd" debug statement. */
#define pymicmd_debug_printf(fmt, ...) \
debug_prefixed_printf_cond (pymicmd_debug, "py-micmd", fmt, ##__VA_ARGS__)
/* Print a "py-micmd" enter/exit debug statements. */
#define PYMICMD_SCOPED_DEBUG_ENTER_EXIT \
scoped_debug_enter_exit (pymicmd_debug, "py-micmd")
struct mi_command_py;
/* Representation of a Python gdb.MICommand object. */
struct micmdpy_object
{
PyObject_HEAD
/* The object representing this command in the MI command table. This
pointer can be nullptr if the command is not currently installed into
the MI command table (see gdb.MICommand.installed property). */
struct mi_command_py *mi_command;
/* The string representing the name of this command, without the leading
dash. This string is never nullptr once the Python object has been
initialised.
The memory for this string was allocated with malloc, and needs to be
deallocated with free when the Python object is deallocated.
When the MI_COMMAND field is not nullptr, then the mi_command_py
object's name will point back to this string. */
char *mi_command_name;
};
/* The MI command implemented in Python. */
struct mi_command_py : public mi_command
{
/* Constructs a new mi_command_py object. NAME is command name without
leading dash. OBJECT is a reference to a Python object implementing
the command. This object must inherit from gdb.MICommand and must
implement the invoke method. */
mi_command_py (const char *name, micmdpy_object *object)
: mi_command (name, nullptr),
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
m_pyobj (gdbpy_ref<micmdpy_object>::new_reference (object))
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
{
pymicmd_debug_printf ("this = %p", this);
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
m_pyobj->mi_command = this;
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
}
~mi_command_py ()
{
/* The Python object representing a MI command contains a pointer back
to this c++ object. We can safely set this pointer back to nullptr
now, to indicate the Python object no longer references a valid c++
object.
However, the Python object also holds the storage for our name
string. We can't clear that here as our parent's destructor might
still want to reference that string. Instead we rely on the Python
object deallocator to free that memory, and reset the pointer. */
m_pyobj->mi_command = nullptr;
pymicmd_debug_printf ("this = %p", this);
};
/* Validate that CMD_OBJ, a non-nullptr pointer, is installed into the MI
command table correctly. This function looks up the command in the MI
command table and checks that the object we get back references
CMD_OBJ. This function is only intended for calling within a
gdb_assert. This function performs many assertions internally, and
then always returns true. */
static void validate_installation (micmdpy_object *cmd_obj);
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* Update M_PYOBJ to NEW_PYOBJ. The pointer from M_PYOBJ that points
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
back to this object is swapped with the pointer in NEW_PYOBJ, which
must be nullptr, so that NEW_PYOBJ now points back to this object.
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
Additionally our parent's name string is stored in M_PYOBJ, so we
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
swap the name string with NEW_PYOBJ.
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
Before this call M_PYOBJ is the Python object representing this MI
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
command object. After this call has completed, NEW_PYOBJ now
represents this MI command object. */
void swap_python_object (micmdpy_object *new_pyobj)
{
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* Current object has a backlink, new object doesn't have a backlink. */
gdb_assert (m_pyobj->mi_command != nullptr);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
gdb_assert (new_pyobj->mi_command == nullptr);
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* Clear the current M_PYOBJ's backlink, set NEW_PYOBJ's backlink. */
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
std::swap (new_pyobj->mi_command, m_pyobj->mi_command);
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* Both object have names. */
gdb_assert (m_pyobj->mi_command_name != nullptr);
gdb_assert (new_pyobj->mi_command_name != nullptr);
/* mi_command::m_name is the string owned by the current object. */
gdb_assert (m_pyobj->mi_command_name == this->name ());
/* The name in mi_command::m_name is owned by the current object. Rather
than changing the value of mi_command::m_name (which is not accessible
from here) to point to the name owned by the new object, swap the names
of the two objects, since we know they are identical strings. */
gdb_assert (strcmp (new_pyobj->mi_command_name,
m_pyobj->mi_command_name) == 0);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
std::swap (new_pyobj->mi_command_name, m_pyobj->mi_command_name);
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* Take a reference to the new object, drop the reference to the current
object. */
m_pyobj = gdbpy_ref<micmdpy_object>::new_reference (new_pyobj);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
}
/* Called when the MI command is invoked. */
gdb/mi: consistently notify user when GDB/MI client uses -thread-select GDB notifies users about user selected thread changes somewhat inconsistently as mentioned on gdb-patches mailing list here: https://sourceware.org/pipermail/gdb-patches/2022-February/185989.html Consider GDB debugging a multi-threaded inferior with both CLI and GDB/MI interfaces connected to separate terminals. Assuming inferior is stopped and thread 1 is selected, when a thread 2 is selected using '-thread-select 2' command on GDB/MI terminal: -thread-select 2 ^done,new-thread-id="2",frame={level="0",addr="0x00005555555551cd",func="child_sub_function",args=[],file="/home/jv/Projects/gdb/users_jv_patches/gdb/testsuite/gdb.mi/user-selected-context-sync.c",fullname="/home/uuu/gdb/gdb/testsuite/gdb.mi/user-selected-context-sync.c",line="30",arch="i386:x86-64"} (gdb) and on CLI terminal we get the notification (as expected): [Switching to thread 2 (Thread 0x7ffff7daa640 (LWP 389659))] #0 child_sub_function () at /home/uuu/gdb/gdb/testsuite/gdb.mi/user-selected-context-sync.c:30 30 volatile int dummy = 0; However, now that thread 2 is selected, if thread 1 is selected using 'thread-select --thread 1 1' command on GDB/MI terminal terminal: -thread-select --thread 1 1 ^done,new-thread-id="1",frame={level="0",addr="0x0000555555555294",func="main",args=[],file="/home/jv/Projects/gdb/users_jv_patches/gdb/testsuite/gdb.mi/user-selected-context-sync.c",fullname="/home/jv/Projects/gdb/users_jv_patches/gdb/testsuite/gdb.mi/user-selected-context-sync.c",line="66",arch="i386:x86-64"} (gdb) but no notification is printed on CLI terminal, despite the fact that user selected thread has changed. The problem is that when `-thread-select --thread 1 1` is executed then thread is switched to thread 1 before mi_cmd_thread_select () is called, therefore the condition "inferior_ptid != previous_ptid" there does not hold. To address this problem, we have to move notification logic up to mi_cmd_execute () where --thread option is processed and notify user selected contents observers there if context changes. However, this in itself breaks GDB/MI because it would cause context notification to be sent on MI channel. This is because by the time we notify, MI notification suppression is already restored (done in mi_command::invoke(). Therefore we had to lift notification suppression logic also up to mi_cmd_execute (). This change in made distinction between mi_command::invoke() and mi_command::do_invoke() unnecessary as all mi_command::invoke() did (after the change) was to call do_invoke(). So this patches removes do_invoke() and moves the command execution logic directly to invoke(). With this change, all gdb.mi tests pass, tested on x86_64-linux. Co-authored-by: Andrew Burgess <aburgess@redhat.com> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=20631
2022-03-16 15:08:22 +00:00
virtual void invoke(struct mi_parse *parse) const override;
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
private:
/* The Python object representing this MI command. */
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
gdbpy_ref<micmdpy_object> m_pyobj;
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
};
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
using mi_command_py_up = std::unique_ptr<mi_command_py>;
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
extern PyTypeObject micmdpy_object_type
CPYCHECKER_TYPE_OBJECT_FOR_TYPEDEF ("micmdpy_object");
/* Holds a Python object containing the string 'invoke'. */
static PyObject *invoke_cst;
/* Convert KEY_OBJ into a string that can be used as a field name in MI
output. KEY_OBJ must be a Python string object, and must only contain
characters suitable for use as an MI field name.
If KEY_OBJ is not a string, or if KEY_OBJ contains invalid characters,
then an error is thrown. Otherwise, KEY_OBJ is converted to a string
and returned. */
static gdb::unique_xmalloc_ptr<char>
py_object_to_mi_key (PyObject *key_obj)
{
/* The key must be a string. */
gdb/python: remove Python 2/3 compatibility macros New in this version: - Rebase on master, fix a few more issues that appeared. python-internal.h contains a number of macros that helped make the code work with both Python 2 and 3. Remove them and adjust the code to use the Python 3 functions. Change-Id: I99a3d80067fb2d65de4f69f6473ba6ffd16efb2d
2022-03-21 10:07:41 -04:00
if (!PyUnicode_Check (key_obj))
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
{
gdbpy_ref<> key_repr (PyObject_Repr (key_obj));
gdb::unique_xmalloc_ptr<char> key_repr_string;
if (key_repr != nullptr)
key_repr_string = python_string_to_target_string (key_repr.get ());
if (key_repr_string == nullptr)
gdbpy_handle_exception ();
gdbpy_error (_("non-string object used as key: %s"),
key_repr_string.get ());
}
gdb::unique_xmalloc_ptr<char> key_string
= python_string_to_target_string (key_obj);
if (key_string == nullptr)
gdbpy_handle_exception ();
/* Predicate function, returns true if NAME is a valid field name for use
in MI result output, otherwise, returns false. */
auto is_valid_key_name = [] (const char *name) -> bool
{
gdb_assert (name != nullptr);
if (*name == '\0' || !isalpha (*name))
return false;
for (; *name != '\0'; ++name)
if (!isalnum (*name) && *name != '_' && *name != '-')
return false;
return true;
};
if (!is_valid_key_name (key_string.get ()))
{
if (*key_string.get () == '\0')
gdbpy_error (_("Invalid empty key in MI result"));
else
gdbpy_error (_("Invalid key in MI result: %s"), key_string.get ());
}
return key_string;
}
/* Serialize RESULT and print it in MI format to the current_uiout.
FIELD_NAME is used as the name of this result field.
RESULT can be a dictionary, a sequence, an iterator, or an object that
can be converted to a string, these are converted to the matching MI
output format (dictionaries as tuples, sequences and iterators as lists,
and strings as named fields).
If anything goes wrong while formatting the output then an error is
thrown.
This function is the recursive inner core of serialize_mi_result, and
should only be called from that function. */
static void
serialize_mi_result_1 (PyObject *result, const char *field_name)
{
struct ui_out *uiout = current_uiout;
if (PyDict_Check (result))
{
PyObject *key, *value;
Py_ssize_t pos = 0;
ui_out_emit_tuple tuple_emitter (uiout, field_name);
while (PyDict_Next (result, &pos, &key, &value))
{
gdb::unique_xmalloc_ptr<char> key_string
(py_object_to_mi_key (key));
serialize_mi_result_1 (value, key_string.get ());
}
}
gdb/python: remove Python 2/3 compatibility macros New in this version: - Rebase on master, fix a few more issues that appeared. python-internal.h contains a number of macros that helped make the code work with both Python 2 and 3. Remove them and adjust the code to use the Python 3 functions. Change-Id: I99a3d80067fb2d65de4f69f6473ba6ffd16efb2d
2022-03-21 10:07:41 -04:00
else if (PySequence_Check (result) && !PyUnicode_Check (result))
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
{
ui_out_emit_list list_emitter (uiout, field_name);
Py_ssize_t len = PySequence_Size (result);
if (len == -1)
gdbpy_handle_exception ();
for (Py_ssize_t i = 0; i < len; ++i)
{
gdbpy_ref<> item (PySequence_ITEM (result, i));
if (item == nullptr)
gdbpy_handle_exception ();
serialize_mi_result_1 (item.get (), nullptr);
}
}
else if (PyIter_Check (result))
{
gdbpy_ref<> item;
ui_out_emit_list list_emitter (uiout, field_name);
while (true)
{
item.reset (PyIter_Next (result));
if (item == nullptr)
{
if (PyErr_Occurred () != nullptr)
gdbpy_handle_exception ();
break;
}
serialize_mi_result_1 (item.get (), nullptr);
}
}
else
{
gdb::unique_xmalloc_ptr<char> string (gdbpy_obj_to_string (result));
if (string == nullptr)
gdbpy_handle_exception ();
uiout->field_string (field_name, string.get ());
}
}
/* Serialize RESULT and print it in MI format to the current_uiout.
This function handles the top-level result initially returned from the
invoke method of the Python command implementation. At the top-level
the result must be a dictionary. The values within this dictionary can
be a wider range of types. Handling the values of the top-level
dictionary is done by serialize_mi_result_1, see that function for more
details.
If anything goes wrong while parsing and printing the MI output then an
error is thrown. */
static void
serialize_mi_result (PyObject *result)
{
/* At the top-level, the result must be a dictionary. */
if (!PyDict_Check (result))
gdbpy_error (_("Result from invoke must be a dictionary"));
PyObject *key, *value;
Py_ssize_t pos = 0;
while (PyDict_Next (result, &pos, &key, &value))
{
gdb::unique_xmalloc_ptr<char> key_string
(py_object_to_mi_key (key));
serialize_mi_result_1 (value, key_string.get ());
}
}
/* Called when the MI command is invoked. PARSE contains the parsed
command line arguments from the user. */
void
gdb/mi: consistently notify user when GDB/MI client uses -thread-select GDB notifies users about user selected thread changes somewhat inconsistently as mentioned on gdb-patches mailing list here: https://sourceware.org/pipermail/gdb-patches/2022-February/185989.html Consider GDB debugging a multi-threaded inferior with both CLI and GDB/MI interfaces connected to separate terminals. Assuming inferior is stopped and thread 1 is selected, when a thread 2 is selected using '-thread-select 2' command on GDB/MI terminal: -thread-select 2 ^done,new-thread-id="2",frame={level="0",addr="0x00005555555551cd",func="child_sub_function",args=[],file="/home/jv/Projects/gdb/users_jv_patches/gdb/testsuite/gdb.mi/user-selected-context-sync.c",fullname="/home/uuu/gdb/gdb/testsuite/gdb.mi/user-selected-context-sync.c",line="30",arch="i386:x86-64"} (gdb) and on CLI terminal we get the notification (as expected): [Switching to thread 2 (Thread 0x7ffff7daa640 (LWP 389659))] #0 child_sub_function () at /home/uuu/gdb/gdb/testsuite/gdb.mi/user-selected-context-sync.c:30 30 volatile int dummy = 0; However, now that thread 2 is selected, if thread 1 is selected using 'thread-select --thread 1 1' command on GDB/MI terminal terminal: -thread-select --thread 1 1 ^done,new-thread-id="1",frame={level="0",addr="0x0000555555555294",func="main",args=[],file="/home/jv/Projects/gdb/users_jv_patches/gdb/testsuite/gdb.mi/user-selected-context-sync.c",fullname="/home/jv/Projects/gdb/users_jv_patches/gdb/testsuite/gdb.mi/user-selected-context-sync.c",line="66",arch="i386:x86-64"} (gdb) but no notification is printed on CLI terminal, despite the fact that user selected thread has changed. The problem is that when `-thread-select --thread 1 1` is executed then thread is switched to thread 1 before mi_cmd_thread_select () is called, therefore the condition "inferior_ptid != previous_ptid" there does not hold. To address this problem, we have to move notification logic up to mi_cmd_execute () where --thread option is processed and notify user selected contents observers there if context changes. However, this in itself breaks GDB/MI because it would cause context notification to be sent on MI channel. This is because by the time we notify, MI notification suppression is already restored (done in mi_command::invoke(). Therefore we had to lift notification suppression logic also up to mi_cmd_execute (). This change in made distinction between mi_command::invoke() and mi_command::do_invoke() unnecessary as all mi_command::invoke() did (after the change) was to call do_invoke(). So this patches removes do_invoke() and moves the command execution logic directly to invoke(). With this change, all gdb.mi tests pass, tested on x86_64-linux. Co-authored-by: Andrew Burgess <aburgess@redhat.com> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=20631
2022-03-16 15:08:22 +00:00
mi_command_py::invoke (struct mi_parse *parse) const
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
{
PYMICMD_SCOPED_DEBUG_ENTER_EXIT;
pymicmd_debug_printf ("this = %p, name = %s", this, name ());
mi_parse_argv (parse->args, parse);
if (parse->argv == nullptr)
error (_("Problem parsing arguments: %s %s"), parse->command, parse->args);
gdbpy_enter enter_py;
/* Place all the arguments into a list which we pass as a single argument
to the MI command's invoke method. */
gdbpy_ref<> argobj (PyList_New (parse->argc));
if (argobj == nullptr)
gdbpy_handle_exception ();
for (int i = 0; i < parse->argc; ++i)
{
gdbpy_ref<> str (PyUnicode_Decode (parse->argv[i],
strlen (parse->argv[i]),
host_charset (), nullptr));
if (PyList_SetItem (argobj.get (), i, str.release ()) < 0)
gdbpy_handle_exception ();
}
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
gdb_assert (this->m_pyobj != nullptr);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
gdb_assert (PyErr_Occurred () == nullptr);
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
gdbpy_ref<> result
(PyObject_CallMethodObjArgs ((PyObject *) this->m_pyobj.get (), invoke_cst,
argobj.get (), nullptr));
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
if (result == nullptr)
gdbpy_handle_exception ();
if (result != Py_None)
serialize_mi_result (result.get ());
}
/* See declaration above. */
void
mi_command_py::validate_installation (micmdpy_object *cmd_obj)
{
gdb_assert (cmd_obj != nullptr);
mi_command_py *cmd = cmd_obj->mi_command;
gdb_assert (cmd != nullptr);
const char *name = cmd_obj->mi_command_name;
gdb_assert (name != nullptr);
gdb_assert (name == cmd->name ());
mi_command *mi_cmd = mi_cmd_lookup (name);
gdb_assert (mi_cmd == cmd);
gdb_assert (cmd->m_pyobj == cmd_obj);
}
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* Return CMD as an mi_command_py if it is a Python MI command, else
nullptr. */
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
static mi_command_py *
as_mi_command_py (mi_command *cmd)
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
{
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
return dynamic_cast<mi_command_py *> (cmd);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
}
/* Uninstall OBJ, making the MI command represented by OBJ unavailable for
use by the user. On success 0 is returned, otherwise -1 is returned
and a Python exception will be set. */
static int
micmdpy_uninstall_command (micmdpy_object *obj)
{
PYMICMD_SCOPED_DEBUG_ENTER_EXIT;
gdb_assert (obj->mi_command != nullptr);
gdb_assert (obj->mi_command_name != nullptr);
pymicmd_debug_printf ("name = %s", obj->mi_command_name);
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* Remove the command from the internal MI table of commands. This will
cause the mi_command_py object to be deleted, which will clear the
backlink in OBJ. */
bool removed = remove_mi_cmd_entry (obj->mi_command->name ());
gdb_assert (removed);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
gdb_assert (obj->mi_command == nullptr);
return 0;
}
/* Install OBJ as a usable MI command. Return 0 on success, and -1 on
error, in which case, a Python error will have been set.
After successful completion the command name associated with OBJ will
be installed in the MI command table (so it can be found if the user
enters that command name), additionally, OBJ will have been added to
the gdb._mi_commands dictionary (using the command name as its key),
this will ensure that OBJ remains live even if the user gives up all
references. */
static int
micmdpy_install_command (micmdpy_object *obj)
{
PYMICMD_SCOPED_DEBUG_ENTER_EXIT;
gdb_assert (obj->mi_command == nullptr);
gdb_assert (obj->mi_command_name != nullptr);
pymicmd_debug_printf ("name = %s", obj->mi_command_name);
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* Look up this command name in MI_COMMANDS, a command with this name may
already exist. */
mi_command *cmd = mi_cmd_lookup (obj->mi_command_name);
mi_command_py *cmd_py = as_mi_command_py (cmd);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
if (cmd != nullptr && cmd_py == nullptr)
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
{
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* There is already an MI command registered with that name, and it's not
a Python one. Forbid replacing a non-Python MI command. */
PyErr_SetString (PyExc_RuntimeError,
_("unable to add command, name is already in use"));
return -1;
}
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
if (cmd_py != nullptr)
{
/* There is already a Python MI command registered with that name, swap
in the new gdb.MICommand implementation. */
cmd_py->swap_python_object (obj);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
}
else
{
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* There's no MI command registered with that name at all, create one. */
mi_command_py_up mi_cmd (new mi_command_py (obj->mi_command_name, obj));
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
/* Add the command to the gdb internal MI command table. */
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
bool result = insert_mi_cmd_entry (std::move (mi_cmd));
gdb_assert (result);
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
}
return 0;
}
/* Implement gdb.MICommand.__init__. The init method takes the name of
the MI command as the first argument, which must be a string, starting
with a single dash. */
static int
micmdpy_init (PyObject *self, PyObject *args, PyObject *kwargs)
{
PYMICMD_SCOPED_DEBUG_ENTER_EXIT;
micmdpy_object *cmd = (micmdpy_object *) self;
static const char *keywords[] = { "name", nullptr };
const char *name;
if (!gdb_PyArg_ParseTupleAndKeywords (args, kwargs, "s", keywords,
&name))
return -1;
/* Validate command name */
const int name_len = strlen (name);
if (name_len == 0)
{
PyErr_SetString (PyExc_ValueError, _("MI command name is empty."));
return -1;
}
else if ((name_len < 2) || (name[0] != '-') || !isalnum (name[1]))
{
PyErr_SetString (PyExc_ValueError,
_("MI command name does not start with '-'"
" followed by at least one letter or digit."));
return -1;
}
else
{
for (int i = 2; i < name_len; i++)
{
if (!isalnum (name[i]) && name[i] != '-')
{
PyErr_Format
(PyExc_ValueError,
_("MI command name contains invalid character: %c."),
name[i]);
return -1;
}
}
/* Skip over the leading dash. For the rest of this function the
dash is not important. */
++name;
}
/* If this object already has a name set, then this object has been
initialized before. We handle this case a little differently. */
if (cmd->mi_command_name != nullptr)
{
/* First, we don't allow the user to change the MI command name.
Supporting this would be tricky as we would need to delete the
mi_command_py from the MI command table, however, the user might
be trying to perform this reinitialization from within the very
command we're about to delete... it all gets very messy.
So, for now at least, we don't allow this. This doesn't seem like
an excessive restriction. */
if (strcmp (cmd->mi_command_name, name) != 0)
{
PyErr_SetString
(PyExc_ValueError,
_("can't reinitialize object with a different command name"));
return -1;
}
/* If there's already an object registered with the MI command table,
then we're done. That object must be a mi_command_py, which
should reference back to this micmdpy_object. */
if (cmd->mi_command != nullptr)
{
mi_command_py::validate_installation (cmd);
return 0;
}
}
else
cmd->mi_command_name = xstrdup (name);
/* Now we can install this mi_command_py in the MI command table. */
return micmdpy_install_command (cmd);
}
/* Called when a gdb.MICommand object is deallocated. */
static void
micmdpy_dealloc (PyObject *obj)
{
PYMICMD_SCOPED_DEBUG_ENTER_EXIT;
micmdpy_object *cmd = (micmdpy_object *) obj;
/* If the Python object failed to initialize, then the name field might
be nullptr. */
pymicmd_debug_printf ("obj = %p, name = %s", cmd,
(cmd->mi_command_name == nullptr
? "(null)" : cmd->mi_command_name));
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
/* As the mi_command_py object holds a reference to the micmdpy_object,
the only way the dealloc function can be called is if the mi_command_py
object has been deleted, in which case the following assert will
hold. */
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
gdb_assert (cmd->mi_command == nullptr);
/* Free the memory that holds the command name. */
xfree (cmd->mi_command_name);
cmd->mi_command_name = nullptr;
/* Finally, free the memory for this Python object. */
Py_TYPE (obj)->tp_free (obj);
}
/* Python initialization for the MI commands components. */
int
gdbpy_initialize_micommands ()
{
micmdpy_object_type.tp_new = PyType_GenericNew;
if (PyType_Ready (&micmdpy_object_type) < 0)
return -1;
if (gdb_pymodule_addobject (gdb_module, "MICommand",
(PyObject *) &micmdpy_object_type)
< 0)
return -1;
gdb/python: remove Python 2/3 compatibility macros New in this version: - Rebase on master, fix a few more issues that appeared. python-internal.h contains a number of macros that helped make the code work with both Python 2 and 3. Remove them and adjust the code to use the Python 3 functions. Change-Id: I99a3d80067fb2d65de4f69f6473ba6ffd16efb2d
2022-03-21 10:07:41 -04:00
invoke_cst = PyUnicode_FromString ("invoke");
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
if (invoke_cst == nullptr)
return -1;
return 0;
}
gdb/python: remove gdb._mi_commands dict The motivation for this patch is the fact that py-micmd.c doesn't build with Python 2, due to PyDict_GetItemWithError being a Python 3-only function: CXX python/py-micmd.o /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c: In function ‘int micmdpy_uninstall_command(micmdpy_object*)’: /home/smarchi/src/binutils-gdb/gdb/python/py-micmd.c:430:20: error: ‘PyDict_GetItemWithError’ was not declared in this scope; did you mean ‘PyDict_GetItemString’? 430 | PyObject *curr = PyDict_GetItemWithError (mi_cmd_dict.get (), | ^~~~~~~~~~~~~~~~~~~~~~~ | PyDict_GetItemString A first solution to fix this would be to try to replace PyDict_GetItemWithError equivalent Python 2 code. But I looked at why we are doing this in the first place: it is to maintain the `gdb._mi_commands` Python dictionary that we use as a `name -> gdb.MICommand object` map. Since the `gdb._mi_commands` dictionary is never actually used in Python, it seems like a lot of trouble to use a Python object for this. My first idea was to replace it with a C++ map (std::unordered_map<std::string, gdbpy_ref<micmdpy_object>>). While implementing this, I realized we don't really need this map at all. The mi_command_py objects registered in the main MI command table can own their backing micmdpy_object (that's a gdb.MICommand, but seen from the C++ code). To know whether an mi_command is an mi_command_py, we can use a dynamic cast. Since there's one less data structure to maintain, there are less chances of messing things up. - Change mi_command_py::m_pyobj to a gdbpy_ref, the mi_command_py is now what keeps the MICommand alive. - Set micmdpy_object::mi_command in the constructor of mi_command_py. If mi_command_py manages setting/clearing that field in swap_python_object, I think it makes sense that it also takes care of setting it initially. - Move a bunch of checks from micmdpy_install_command to swap_python_object, and make them gdb_asserts. - In micmdpy_install_command, start by doing an mi_cmd_lookup. This is needed to know whether there's a Python MI command already registered with that name. But we can already tell if there's a non-Python command registered with that name. Return an error if that happens, rather than waiting for insert_mi_cmd_entry to fail. Change the error message to "name is already in use" rather than "may already be in use", since it's more precise. I asked Andrew about the original intent of using a Python dictionary object to hold the command objects. The reason was to make sure the objects get destroyed when the Python runtime gets finalized, not later. Holding the objects in global C++ data structures and not doing anything more means that the held Python objects will be decref'd after the Python interpreter has been finalized. That's not desirable. I tried it and it indeed segfaults. Handle this by adding a gdbpy_finalize_micommands function called in finalize_python. This is the mirror of gdbpy_initialize_micommands called in do_start_initialization. In there, delete all Python MI commands. I think it makes sense to do it this way: if it was somehow possible to unload Python support from GDB in the middle of a session we'd want to unregister any Python MI command. Otherwise, these MI commands would be backed with a stale PyObject or simply nothing. Delete tests that were related to `gdb._mi_commands`. Co-Authored-By: Andrew Burgess <aburgess@redhat.com> Change-Id: I060d5ebc7a096c67487998a8a4ca1e8e56f12cd3
2022-03-18 15:55:26 -04:00
void
gdbpy_finalize_micommands ()
{
/* mi_command_py objects hold references to micmdpy_object objects. They must
be dropped before the Python interpreter is finalized. Do so by removing
those MI command entries, thus deleting the mi_command_py objects. */
remove_mi_cmd_entries ([] (mi_command *cmd)
{
return as_mi_command_py (cmd) != nullptr;
});
}
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
/* Get the gdb.MICommand.name attribute, returns a string, the name of this
MI command. */
static PyObject *
micmdpy_get_name (PyObject *self, void *closure)
{
struct micmdpy_object *micmd_obj = (struct micmdpy_object *) self;
gdb_assert (micmd_obj->mi_command_name != nullptr);
std::string name_str = string_printf ("-%s", micmd_obj->mi_command_name);
gdb/python: remove Python 2/3 compatibility macros New in this version: - Rebase on master, fix a few more issues that appeared. python-internal.h contains a number of macros that helped make the code work with both Python 2 and 3. Remove them and adjust the code to use the Python 3 functions. Change-Id: I99a3d80067fb2d65de4f69f6473ba6ffd16efb2d
2022-03-21 10:07:41 -04:00
return PyUnicode_FromString (name_str.c_str ());
gdb/python/mi: create MI commands using python This commit allows a user to create custom MI commands using Python similarly to what is possible for Python CLI commands. A new subclass of mi_command is defined for Python MI commands, mi_command_py. A new file, gdb/python/py-micmd.c contains the logic for Python MI commands. This commit is based on work linked too from this mailing list thread: https://sourceware.org/pipermail/gdb/2021-November/049774.html Which has also been previously posted to the mailing list here: https://sourceware.org/pipermail/gdb-patches/2019-May/158010.html And was recently reposted here: https://sourceware.org/pipermail/gdb-patches/2022-January/185190.html The version in this patch takes some core code from the previously posted patches, but also has some significant differences, especially after the feedback given here: https://sourceware.org/pipermail/gdb-patches/2022-February/185767.html A new MI command can be implemented in Python like this: class echo_args(gdb.MICommand): def invoke(self, args): return { 'args': args } echo_args("-echo-args") The 'args' parameter (to the invoke method) is a list containing (almost) all command line arguments passed to the MI command (--thread and --frame are handled before the Python code is called, and removed from the args list). This list can be empty if the MI command was passed no arguments. When used within gdb the above command produced output like this: (gdb) -echo-args a b c ^done,args=["a","b","c"] (gdb) The 'invoke' method of the new command must return a dictionary. The keys of this dictionary are then used as the field names in the mi command output (e.g. 'args' in the above). The values of the result returned by invoke can be dictionaries, lists, iterators, or an object that can be converted to a string. These are processed recursively to create the mi output. And so, this is valid: class new_command(gdb.MICommand): def invoke(self,args): return { 'result_one': { 'abc': 123, 'def': 'Hello' }, 'result_two': [ { 'a': 1, 'b': 2 }, { 'c': 3, 'd': 4 } ] } Which produces output like: (gdb) -new-command ^done,result_one={abc="123",def="Hello"},result_two=[{a="1",b="2"},{c="3",d="4"}] (gdb) I have required that the fields names used in mi result output must match the regexp: "^[a-zA-Z][-_a-zA-Z0-9]*$" (without the quotes). This restriction was never written down anywhere before, but seems sensible to me, and we can always loosen this rule later if it proves to be a problem. Much harder to try and add a restriction later, once people are already using the API. What follows are some details about how this implementation differs from the original patch that was posted to the mailing list. In this patch, I have changed how the lifetime of the Python gdb.MICommand objects is managed. In the original patch, these object were kept alive by an owned reference within the mi_command_py object. As such, the Python object would not be deleted until the mi_command_py object itself was deleted. This caused a problem, the mi_command_py were held in the global mi command table (in mi/mi-cmds.c), which, as a global, was not cleared until program shutdown. By this point the Python interpreter has already been shutdown. Attempting to delete the mi_command_py object at this point was causing GDB to try and invoke Python code after finalising the Python interpreter, and we would crash. To work around this problem, the original patch added code in python/python.c that would search the mi command table, and delete the mi_command_py objects before the Python environment was finalised. In contrast, in this patch, I have added a new global dictionary to the gdb module, gdb._mi_commands. We already have several such global data stores related to pretty printers, and frame unwinders. The MICommand objects are placed into the new gdb.mi_commands dictionary, and it is this reference that keeps the objects alive. When GDB's Python interpreter is shut down gdb._mi_commands is deleted, and any MICommand objects within it are deleted at this point. This change avoids having to make the mi_cmd_table global, and walk over it from within GDB's python related code. This patch handles command redefinition entirely within GDB's python code, though this does impose one small restriction which is not present in the original code (detailed below), I don't think this is a big issue. However, the original patch relied on being able to finish executing the mi_command::do_invoke member function after the mi_command object had been deleted. Though continuing to execute a member function after an object is deleted is well defined, it is also (IMHO) risky, its too easy for someone to later add a use of the object without realising that the object might sometimes, have been deleted. The new patch avoids this issue. The one restriction that is added to avoid this, is that an MICommand object can't be reinitialised with a different command name, so: (gdb) python cmd = MyMICommand("-abc") (gdb) python cmd.__init__("-def") can't reinitialize object with a different command name This feels like a pretty weird edge case, and I'm happy to live with this restriction. I have also changed how the memory is managed for the command name. In the most recently posted patch series, the command name is moved into a subclass of mi_command, the python mi_command_py, which inherits from mi_command is then free to use a smart pointer to manage the memory for the name. In this patch, I leave the mi_command class unchanged, and instead hold the memory for the name within the Python object, as the lifetime of the Python object always exceeds the c++ object stored in the mi_cmd_table. This adds a little more complexity in py-micmd.c, but leaves the mi_command class nice and simple. Next, this patch adds some extra functionality, there's a MICommand.name read-only attribute containing the name of the command, and a read-write MICommand.installed attribute that can be used to install (make the command available for use) and uninstall (remove the command from the mi_cmd_table so it can't be used) the command. This attribute will be automatically updated if a second command replaces an earlier command. This patch adds additional error handling, and makes more use the gdbpy_handle_exception function. Co-Authored-By: Jan Vrany <jan.vrany@labware.com>
2020-06-23 14:45:38 +01:00
}
/* Get the gdb.MICommand.installed property. Returns true if this MI
command is installed into the MI command table, otherwise returns
false. */
static PyObject *
micmdpy_get_installed (PyObject *self, void *closure)
{
struct micmdpy_object *micmd_obj = (struct micmdpy_object *) self;
if (micmd_obj->mi_command == nullptr)
Py_RETURN_FALSE;
Py_RETURN_TRUE;
}
/* Set the gdb.MICommand.installed property. The property can be set to
either true or false. Setting the property to true will cause the
command to be installed into the MI command table (if it isn't
already), while setting this property to false will cause the command
to be removed from the MI command table (if it is present). */
static int
micmdpy_set_installed (PyObject *self, PyObject *newvalue, void *closure)
{
struct micmdpy_object *micmd_obj = (struct micmdpy_object *) self;
bool installed_p = PyObject_IsTrue (newvalue);
if (installed_p == (micmd_obj->mi_command != nullptr))
return 0;
if (installed_p)
return micmdpy_install_command (micmd_obj);
else
return micmdpy_uninstall_command (micmd_obj);
}
/* The gdb.MICommand properties. */
static gdb_PyGetSetDef micmdpy_object_getset[] = {
{ "name", micmdpy_get_name, nullptr, "The command's name.", nullptr },
{ "installed", micmdpy_get_installed, micmdpy_set_installed,
"Is this command installed for use.", nullptr },
{ nullptr } /* Sentinel. */
};
/* The gdb.MICommand descriptor. */
PyTypeObject micmdpy_object_type = {
PyVarObject_HEAD_INIT (nullptr, 0) "gdb.MICommand", /*tp_name */
sizeof (micmdpy_object), /*tp_basicsize */
0, /*tp_itemsize */
micmdpy_dealloc, /*tp_dealloc */
0, /*tp_print */
0, /*tp_getattr */
0, /*tp_setattr */
0, /*tp_compare */
0, /*tp_repr */
0, /*tp_as_number */
0, /*tp_as_sequence */
0, /*tp_as_mapping */
0, /*tp_hash */
0, /*tp_call */
0, /*tp_str */
0, /*tp_getattro */
0, /*tp_setattro */
0, /*tp_as_buffer */
Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags */
"GDB mi-command object", /* tp_doc */
0, /* tp_traverse */
0, /* tp_clear */
0, /* tp_richcompare */
0, /* tp_weaklistoffset */
0, /* tp_iter */
0, /* tp_iternext */
0, /* tp_methods */
0, /* tp_members */
micmdpy_object_getset, /* tp_getset */
0, /* tp_base */
0, /* tp_dict */
0, /* tp_descr_get */
0, /* tp_descr_set */
0, /* tp_dictoffset */
micmdpy_init, /* tp_init */
0, /* tp_alloc */
};
void _initialize_py_micmd ();
void
_initialize_py_micmd ()
{
add_setshow_boolean_cmd
("py-micmd", class_maintenance, &pymicmd_debug,
_("Set Python micmd debugging."),
_("Show Python micmd debugging."),
_("When on, Python micmd debugging is enabled."),
nullptr,
show_pymicmd_debug,
&setdebuglist, &showdebuglist);
}
Reference in a new issue Copy permalink