Compare commits

...
7 changed files with 1150 additions and 148 deletions

No files matched your search

+340 -133
View File
@@ -11,6 +11,7 @@
#include <boost/log/trivial.hpp>
#include <algorithm>
#include <cassert>
#include <cctype>
#include <cstddef>
#include <cstdlib>
@@ -130,6 +131,20 @@ static AuditEventCategory event_category(const std::string& event_name)
return it == audit_event_categories.end() ? AuditEventCategory::None : it->second;
}
// Events the 3.12 stdlib raises with sys.audit() inside the function that performs the whole
// operation, so the raising frame bounds everything the operation does. Every other mapped event
// is a primitive raised from C, whose current frame is whatever Python code called it -- possibly
// a library loop over many targets -- so it never marks an operation. glob.glob and glob.glob/2
// are raised in iglob(), which returns its lazy generator before any directory is listed.
static const std::unordered_set<std::string> operation_root_events{
"urllib.Request", "http.client.connect", "os.walk", "os.fwalk",
"glob.glob", "glob.glob/2", "pathlib.Path.glob", "pathlib.Path.rglob",
"subprocess.Popen", "pty.spawn", "tempfile.mkstemp", "tempfile.mkdtemp",
"shutil.copyfile", "shutil.copymode", "shutil.copystat", "shutil.copytree",
"shutil.chown", "shutil.move", "shutil.rmtree", "shutil.make_archive",
"shutil.unpack_archive",
};
// True for the categories whose targets are filesystem paths, as opposed to a network
// address or a process command line -- the deny-path and allowed-root checks only make sense
// against a path.
@@ -146,6 +161,24 @@ static bool is_fs_category(AuditEventCategory category)
}
}
// The category's name in a saved event grant, e.g. "FsRead" in "FsRead:open".
static const char* category_name(AuditEventCategory category)
{
switch (category) {
case AuditEventCategory::FsRead: return "FsRead";
case AuditEventCategory::FsReadWrite: return "FsReadWrite";
case AuditEventCategory::FsCreate: return "FsCreate";
case AuditEventCategory::FsDelete: return "FsDelete";
case AuditEventCategory::Http: return "Http";
case AuditEventCategory::Socket: return "Socket";
case AuditEventCategory::ProcessCreate: return "ProcessCreate";
case AuditEventCategory::ProcessReplace: return "ProcessReplace";
case AuditEventCategory::Threading: return "Threading";
case AuditEventCategory::None: break;
}
return "None";
}
// ---------------------------------------------------------------------------
// Path safety
// ---------------------------------------------------------------------------
@@ -215,26 +248,50 @@ bool is_inside_allowed_root(const boost::filesystem::path& candidate, const boos
thread_local std::string PluginAuditManager::m_current_plugin_key = "";
thread_local std::string PluginAuditManager::m_current_capability_name = "";
thread_local std::vector<AllowedRoot> PluginAuditManager::m_scoped_allowed_roots;
thread_local std::vector<PyObject*> PluginAuditManager::m_root_frames;
thread_local int PluginAuditManager::m_releasing_frames = 0;
thread_local bool PluginAuditManager::m_audit_denial_pending = false;
thread_local bool PluginAuditManager::m_has_last_violation = false;
thread_local AuditViolation PluginAuditManager::m_last_violation;
ScopedPluginAuditContext::ScopedPluginAuditContext(const std::string& plugin_key,
const std::string& capability_name)
: m_previous_id(PluginAuditManager::instance().current_plugin())
: m_plugin_key(plugin_key)
, m_capability(capability_name)
, m_previous_id(PluginAuditManager::instance().current_plugin())
, m_previous_capability(PluginAuditManager::instance().current_capability())
, m_previous_scoped_roots(PluginAuditManager::m_scoped_allowed_roots)
, m_previous_root_frames(std::move(PluginAuditManager::m_root_frames))
{
PluginAuditManager::instance().set_current_plugin(plugin_key);
PluginAuditManager::instance().set_current_capability(capability_name);
PluginAuditManager::m_scoped_allowed_roots.clear();
PluginAuditManager::m_root_frames.clear();
}
ScopedPluginAuditContext::~ScopedPluginAuditContext()
{
PluginAuditManager::instance().set_current_plugin(m_previous_id);
PluginAuditManager::instance().set_current_capability(m_previous_capability);
PluginAuditManager& mgr = PluginAuditManager::instance();
// Let this scope's root frames go while it is still the current context, and as its own
// plugin: the trampoline's denial path clears the key before the scope ends, and a finalizer
// must never be judged as the outer plugin, or bypass the audit when there is none. A
// finalizer may record a new root frame, so repeat until none is held.
if (!PluginAuditManager::m_root_frames.empty()) {
assert(PyGILState_Check());
mgr.set_current_plugin(m_plugin_key);
mgr.set_current_capability(m_capability);
while (!PluginAuditManager::m_root_frames.empty()) {
std::vector<PyObject*> frames = std::move(PluginAuditManager::m_root_frames);
PluginAuditManager::m_root_frames.clear();
mgr.release_frames(frames);
}
}
mgr.set_current_plugin(m_previous_id);
mgr.set_current_capability(m_previous_capability);
PluginAuditManager::m_scoped_allowed_roots = std::move(m_previous_scoped_roots);
PluginAuditManager::m_root_frames = std::move(m_previous_root_frames);
}
// ---------------------------------------------------------------------------
@@ -575,34 +632,6 @@ bool PluginAuditManager::last_violation(AuditViolation& violation) const
return true;
}
bool PluginAuditManager::has_approved_ancestor(const std::string& plugin_key,
const std::vector<std::string>& call_site_ids) const
{
if (plugin_key.empty() || call_site_ids.empty())
return false;
std::lock_guard<std::mutex> lock(m_mutex);
auto it = m_approved_call_sites.find(plugin_key);
if (it == m_approved_call_sites.end())
return false;
for (const auto& id : call_site_ids)
if (it->second.count(id))
return true;
return false;
}
void PluginAuditManager::record_approved_call_sites(const std::string& plugin_key,
const std::vector<std::string>& call_site_ids)
{
if (plugin_key.empty() || call_site_ids.empty())
return;
std::lock_guard<std::mutex> lock(m_mutex);
auto& approved = m_approved_call_sites[plugin_key];
approved.insert(call_site_ids.begin(), call_site_ids.end());
}
// ---------------------------------------------------------------------------
// The C-level audit hook
// ---------------------------------------------------------------------------
@@ -663,40 +692,88 @@ std::string python_str(PyObject* object)
return result;
}
std::vector<std::string> call_site_identities(const std::string& plugin_root)
{
std::vector<std::string> ids;
if (plugin_root.empty())
return ids;
// The interpreter's stdlib folders, and the site-packages folders inside them that hold
// third-party code, both resolved. Set by install_hook(); like the verdict cache below, only
// touched while the GIL is held.
static std::vector<boost::filesystem::path> s_stdlib_dirs;
static std::vector<boost::filesystem::path> s_site_package_dirs;
static std::unordered_map<std::string, bool> s_stdlib_file_verdicts; // co_filename -> is stdlib
void collect_stdlib_dirs()
{
s_stdlib_dirs.clear();
s_site_package_dirs.clear();
s_stdlib_file_verdicts.clear();
PyObject* sysconfig = PyImport_ImportModule("sysconfig");
PyObject* paths = sysconfig ? PyObject_CallMethod(sysconfig, "get_paths", nullptr) : nullptr;
if (paths && PyDict_Check(paths)) {
const auto collect = [paths](const char* key, std::vector<boost::filesystem::path>& out) {
const std::string dir = python_unicode(PyDict_GetItemString(paths, key)); // borrowed
if (dir.empty())
return;
boost::system::error_code ec;
const boost::filesystem::path resolved = boost::filesystem::weakly_canonical(dir, ec);
out.push_back(ec ? boost::filesystem::path(dir) : resolved);
};
collect("stdlib", s_stdlib_dirs);
collect("platstdlib", s_stdlib_dirs);
collect("purelib", s_site_package_dirs);
collect("platlib", s_site_package_dirs);
}
Py_XDECREF(paths);
Py_XDECREF(sysconfig);
if (PyErr_Occurred()) {
PyErr_Clear();
BOOST_LOG_TRIVIAL(error) << "[AUDIT] Failed to read the stdlib folders from sysconfig";
}
}
bool is_stdlib_frame(PyObject* frame)
{
PyCodeObject* code = PyFrame_GetCode(reinterpret_cast<PyFrameObject*>(frame)); // new reference
const std::string filename = python_unicode(code->co_filename);
Py_DECREF(code);
return PluginAuditManager::is_stdlib_file(filename);
}
// The frames on this thread's Python stack, innermost first. The pointers stay valid while the
// caller runs, because a running activation keeps its frame object alive.
std::vector<PyObject*> stack_frames()
{
std::vector<PyObject*> frames;
PyFrameObject* frame = PyEval_GetFrame(); // borrowed reference
Py_XINCREF(frame); // normalize to an owned reference for the loop below
while (frame) {
PyCodeObject* code = PyFrame_GetCode(frame); // new reference
const std::string filename = python_unicode(reinterpret_cast<PyObject*>(code->co_filename));
const bool is_plugin_frame = !filename.empty() && boost::algorithm::starts_with(filename, plugin_root);
std::string id;
if (!is_plugin_frame && !filename.empty()) {
const std::string funcname = python_unicode(reinterpret_cast<PyObject*>(code->co_name));
id = filename + ":" + funcname + ":" + std::to_string(code->co_firstlineno);
}
Py_DECREF(code);
frames.push_back(reinterpret_cast<PyObject*>(frame));
PyFrameObject* back = PyFrame_GetBack(frame); // new reference, or nullptr at the top of the stack
Py_DECREF(frame);
frame = back;
if (is_plugin_frame)
break;
if (!id.empty())
ids.push_back(std::move(id));
}
return frames;
}
Py_XDECREF(frame); // only holds a reference here if the loop exited via break
// Moves the held root frames whose operation has returned from `held` to `dropped`. A suspended
// generator's frame (os.walk between two items) is off the stack, but its operation resumes, so
// it stays held until the scope ends.
void take_returned_frames(std::vector<PyObject*>& held, std::vector<PyObject*>& dropped)
{
if (held.empty())
return;
return ids;
const std::vector<PyObject*> stack = stack_frames();
const auto running = [&stack](PyObject* frame) {
if (std::find(stack.begin(), stack.end(), frame) != stack.end())
return true;
// NULL once the generator has finished or been deleted: its frame then owns its own data.
PyObject* generator = PyFrame_GetGenerator(reinterpret_cast<PyFrameObject*>(frame)); // new reference
const bool suspended = generator != nullptr;
Py_XDECREF(generator);
return suspended;
};
const auto returned = std::stable_partition(held.begin(), held.end(), running);
dropped.insert(dropped.end(), returned, held.end());
held.erase(returned, held.end());
}
static const std::unordered_set<std::string> two_path_fs_events{
@@ -717,12 +794,15 @@ static const std::unordered_map<std::string, std::vector<Py_ssize_t>> audit_targ
{"socket.getnameinfo", {0}},
{"socket.getservbyname", {0}},
{"socket.getservbyport", {0}},
{"socket.sendto", {1}},
{"socket.sendmsg", {1}},
{"os.system", {0}},
{"subprocess.Popen", {1, 0}},
{"os.posix_spawn", {1, 0}},
{"os.spawn", {2, 1}},
{"os.startfile", {0}},
{"os.startfile/2", {0}},
{"pty.spawn", {0}},
{"_winapi.CreateProcess", {1, 0}},
{"_posixsubprocess.fork_exec", {0}},
@@ -767,7 +847,11 @@ std::vector<std::string> audit_targets(const std::string& event_name, AuditEvent
const auto it = audit_target_arg_indices.find(event_name);
if (it != audit_target_arg_indices.end()) {
for (const Py_ssize_t index : it->second) {
std::string value = python_str(tuple_item(args, index));
// None is an argument left out (sendmsg without an address), not a target named "None".
PyObject* const item = tuple_item(args, index);
if (item == Py_None)
continue;
std::string value = python_str(item);
if (!value.empty()) {
targets.push_back(std::move(value));
break;
@@ -778,8 +862,14 @@ std::vector<std::string> audit_targets(const std::string& event_name, AuditEvent
return targets;
}
std::vector<std::string>* permission_list_for(AuditEventCategory category, PluginPermissions& permissions)
// The saved list that grants an event: its category's target list, or the event-grant list for
// an event with no target. Kept apart, so that no grant doubles as a target of the same name (an
// os.fork grant is not an os.exec of a command named "os.fork").
std::vector<std::string>* permission_list_for(AuditEventCategory category, bool by_event, PluginPermissions& permissions)
{
if (by_event)
return &permissions.events;
switch (category) {
case AuditEventCategory::FsRead: return &permissions.fs_read;
case AuditEventCategory::FsReadWrite: return &permissions.fs_readwrite;
@@ -792,6 +882,13 @@ std::vector<std::string>* permission_list_for(AuditEventCategory category, Plugi
}
}
// An event with no target is granted by its effective category and name, e.g. "FsRead:open"
// for a read of a file descriptor, which does not grant a write of one ("FsReadWrite:open").
std::string event_grant_key(AuditEventCategory category, const std::string& event_name)
{
return std::string(category_name(category)) + ":" + event_name;
}
bool has_permission(const std::vector<std::string>& granted, const std::string& target)
{
return std::find(granted.begin(), granted.end(), target) != granted.end();
@@ -902,22 +999,25 @@ bool prompt_for_targets(AuditEventCategory category, const std::string& plugin_n
// Records a grant in the plugin's sidecar so it is not asked again. Reads the install state
// freshly because the async prompt outlives the caller's stack copy of it.
void persist_grant(const std::string& plugin_key, AuditEventCategory category, const std::vector<std::string>& targets)
void persist_grant(const std::string& plugin_key, AuditEventCategory category, bool by_event, const std::vector<std::string>& keys)
{
PluginInstallState state;
if (!PluginManager::instance().get_install_state(plugin_key, state))
return;
std::vector<std::string>* permission_list = permission_list_for(category, state.permissions);
std::vector<std::string>* permission_list = permission_list_for(category, by_event, state.permissions);
if (!permission_list)
return;
for (const auto& target : targets)
persist_permission(plugin_key, state, *permission_list, target);
for (const auto& key : keys)
persist_permission(plugin_key, state, *permission_list, key);
}
// Requests permission for an audited event, returning true when it is already granted or the user
// approves an inline prompt.
static PluginAuditManager::PermissionPrompt s_permission_prompt_for_testing;
// Asks the user to grant an audited event, returning true when they approve an inline prompt.
// `keys` are what a Yes saves: the unresolved targets, or the event-grant key when the event has
// no target (by_event), which the prompt does not show as a target.
//
// An audited event can fire on a thread the UI thread may itself be blocked waiting on: the
// SlicingPipeline hook runs on the slicing worker thread (see PluginHooks.cpp), and
@@ -929,56 +1029,205 @@ void persist_grant(const std::string& plugin_key, AuditEventCategory category, c
// prompt); the grant is persisted once the user accepts, so a later attempt succeeds without
// re-prompting.
bool request_permission(AuditEventCategory category, const std::string& plugin_key, const std::string& plugin_name,
const std::string& event_name, const std::vector<std::string>& unresolved)
const std::string& event_name, bool by_event, const std::vector<std::string>& keys)
{
const std::vector<std::string> shown = by_event ? std::vector<std::string>{} : keys;
if (s_permission_prompt_for_testing)
return s_permission_prompt_for_testing(event_name, shown);
if (wxTheApp == nullptr || GUI::wxGetApp().is_closing())
return false;
if (wxIsMainThread())
return prompt_for_targets(category, plugin_name, event_name, unresolved);
return prompt_for_targets(category, plugin_name, event_name, shown);
GUI::wxGetApp().CallAfter([category, plugin_key, plugin_name, event_name, unresolved]() {
GUI::wxGetApp().CallAfter([category, plugin_key, plugin_name, event_name, by_event, keys, shown]() {
if (wxTheApp == nullptr || GUI::wxGetApp().is_closing())
return;
if (prompt_for_targets(category, plugin_name, event_name, unresolved))
persist_grant(plugin_key, category, unresolved);
if (prompt_for_targets(category, plugin_name, event_name, shown))
persist_grant(plugin_key, category, by_event, keys);
});
return false;
}
// Decides an event that no allowed root or running operation covers: a saved grant allows it,
// otherwise the user is asked and a Yes is saved.
int decide_audited_event(PluginAuditManager& mgr,
PluginInstallState& state,
const std::string& plugin_key,
const std::string& plugin_name,
const std::string& event_name,
AuditEventCategory category,
const std::vector<std::string>& targets,
std::vector<std::string>* permission_list,
const std::vector<std::string>& call_site_ids)
const std::vector<std::string>& targets)
{
std::vector<std::string> unresolved = targets;
const bool by_event = targets.empty();
const std::vector<std::string> keys = by_event ? std::vector<std::string>{event_grant_key(category, event_name)} : targets;
std::vector<std::string>* permission_list = permission_list_for(category, by_event, state.permissions);
std::vector<std::string> unresolved = keys;
if (permission_list) {
unresolved.clear();
for (const auto& target : targets)
if (!has_permission(*permission_list, target))
unresolved.push_back(target);
if (!targets.empty() && unresolved.empty())
for (const auto& key : keys)
if (!has_permission(*permission_list, key))
unresolved.push_back(key);
if (unresolved.empty())
return 0;
}
if (!request_permission(category, plugin_key, plugin_name, event_name, unresolved))
// Letting a root frame go runs finalizers inside a scope destructor or another event's hook,
// where no dialog may open: what they do must already be granted.
if (mgr.releasing_frames())
return report_denied(mgr, event_name, {false, "audit permission required"});
if (!request_permission(category, plugin_key, plugin_name, event_name, by_event, unresolved))
return report_denied(mgr, event_name, {false, "audit permission required"});
if (permission_list)
for (const auto& target : unresolved)
persist_permission(plugin_key, state, *permission_list, target);
mgr.record_approved_call_sites(plugin_key, call_site_ids);
for (const auto& key : unresolved)
persist_permission(plugin_key, state, *permission_list, key);
return 0;
}
// Judges an audited event raised while a plugin context is active.
int audit_plugin_event(PluginAuditManager& mgr, const std::string& event_name, AuditEventCategory event_type, PyObject* args)
{
const bool fs_category = is_fs_category(event_type);
const std::vector<std::string> targets = audit_targets(event_name, event_type, args);
// A denied path (secrets, certificates, config files -- see is_denied_path) is an
// unconditional block: checked before the running-operation check below, so no granted
// operation can launder access to one, and before the allowed-root shortcut further down,
// so an allowed root (e.g. the read-only resources folder) cannot make a denied path
// underneath it reachable.
if (fs_category) {
for (const auto& target : targets) {
if (mgr.is_denied_path(boost::filesystem::path(target)))
return report_denied(mgr, event_name, {false, "denied path"}, target);
}
}
// A nested event of an operation already granted, e.g. socket.connect inside an approved
// urlopen(), is not asked again. urllib.Request is always judged on its own URL, so a
// redirect or a nested urlopen() inside a running request is not covered by it.
if (event_name != "urllib.Request" && mgr.covered_by_root_frame())
return 0;
// A filesystem target that resolves entirely inside a pre-determined allowed root -- the
// plugin system's own data_dir() tree (which holds each plugin's storage folder and the
// installed/system profile cache), the read-only bundled resources folder, or a per-call
// scoped root such as the current G-code folder -- is part of the plugin system's normal
// workflow and does not need a prompt.
if (fs_category && !targets.empty()) {
const bool is_write = event_type != AuditEventCategory::FsRead;
const bool all_inside_allowed_root =
std::all_of(targets.begin(), targets.end(), [&](const std::string& target) {
return mgr.check_path_access(boost::filesystem::path(target), is_write).allowed;
});
if (all_inside_allowed_root)
return 0;
}
PluginInstallState state;
const bool have_install_state = PluginManager::instance().get_install_state(mgr.current_plugin(), state);
if (!have_install_state) {
BOOST_LOG_TRIVIAL(warning) << __FUNCTION__ << " Failed to get install state for " << mgr.current_plugin();
}
const std::string plugin_name = state.plugin_name.empty() ? mgr.current_plugin() : state.plugin_name;
const int result = decide_audited_event(mgr, state, mgr.current_plugin(), plugin_name, event_name, event_type, targets);
// A granted operation covers its nested events while it runs. A targetless grant names no
// operation, so it covers nothing else.
if (result == 0 && !targets.empty() && operation_root_events.count(event_name))
mgr.record_root_frame();
return result;
}
} // namespace PluginAuditDetail
void PluginAuditManager::record_root_frame()
{
PyObject* frame = reinterpret_cast<PyObject*>(PyEval_GetFrame()); // borrowed reference
if (!frame || !PluginAuditDetail::is_stdlib_frame(frame))
return;
if (std::find(m_root_frames.begin(), m_root_frames.end(), frame) != m_root_frames.end())
return; // glob.glob and glob.glob/2 share a frame
Py_INCREF(frame);
m_root_frames.push_back(frame);
}
bool PluginAuditManager::covered_by_root_frame() const
{
if (m_root_frames.empty())
return false;
// Only stdlib frames may sit between the event and the operation: plugin or library code
// the operation calls back into (an os.walk onerror, a urllib handler) acts on its own.
for (PyObject* frame : PluginAuditDetail::stack_frames()) {
if (std::find(m_root_frames.begin(), m_root_frames.end(), frame) != m_root_frames.end())
return true;
if (!PluginAuditDetail::is_stdlib_frame(frame))
return false;
}
return false;
}
bool PluginAuditManager::releasing_frames() const { return m_releasing_frames > 0; }
bool PluginAuditManager::is_stdlib_file(const std::string& filename)
{
// Frozen stdlib modules (os, posixpath, codecs, ...) carry a pseudo filename.
if (boost::algorithm::starts_with(filename, "<frozen "))
return true;
if (filename.empty() || filename.front() == '<')
return false;
// Both sides are resolved: co_filename follows the Python home as it was given, which may
// be a symlink.
const auto [it, inserted] = PluginAuditDetail::s_stdlib_file_verdicts.try_emplace(filename, false);
if (inserted) {
const boost::filesystem::path file(filename);
const auto inside_any = [&file](const std::vector<boost::filesystem::path>& dirs) {
return std::any_of(dirs.begin(), dirs.end(),
[&file](const boost::filesystem::path& dir) { return is_inside_allowed_root(file, dir); });
};
it->second = inside_any(PluginAuditDetail::s_stdlib_dirs) && !inside_any(PluginAuditDetail::s_site_package_dirs);
}
return it->second;
}
void PluginAuditManager::release_frames(std::vector<PyObject*>& frames)
{
if (frames.empty())
return;
// A finalizer's own denial must neither replace the PermissionError of the event being
// decided nor leave a stale denial behind for the next call.
PyObject* const error = PyErr_GetRaisedException();
const bool denial_pending = m_audit_denial_pending;
const bool has_violation = m_has_last_violation;
const AuditViolation violation = m_last_violation;
std::vector<PyObject*> released;
released.swap(frames);
++m_releasing_frames;
for (PyObject* frame : released)
Py_DECREF(frame);
--m_releasing_frames;
m_audit_denial_pending = denial_pending;
m_has_last_violation = has_violation;
m_last_violation = violation;
PyErr_SetRaisedException(error);
}
void PluginAuditManager::set_permission_prompt_for_testing(PermissionPrompt prompt)
{
PluginAuditDetail::s_permission_prompt_for_testing = std::move(prompt);
}
int PluginAuditManager::audit_hook(const char* event, PyObject* args, void* user_data)
{
auto* mgr = static_cast<PluginAuditManager*>(user_data);
@@ -1006,62 +1255,20 @@ int PluginAuditManager::audit_hook(const char* event, PyObject* args, void* user
if (event_type == AuditEventCategory::None)
return 0;
const bool fs_category = is_fs_category(event_type);
const std::vector<std::string> targets = PluginAuditDetail::audit_targets(event_name, event_type, args);
// A denied path (secrets, certificates, config files -- see is_denied_path) is an
// unconditional block: checked before the ancestor-cascade check below, so a cascade
// approval recorded for an unrelated action can never launder access to one, and before
// the allowed-root shortcut further down, so an allowed root (e.g. the read-only resources
// folder) cannot make a denied path underneath it reachable.
if (fs_category) {
for (const auto& target : targets) {
if (mgr->is_denied_path(boost::filesystem::path(target)))
return PluginAuditDetail::report_denied(*mgr, event_name, {false, "denied path"}, target);
}
}
PluginDescriptor plugin_descriptor;
PluginManager::instance().try_get_plugin_descriptor(mgr->current_plugin(), plugin_descriptor);
// Some plugins call other audited events, e.g. urlib.request will call socket.connect. So this is so that if urlib.request
// was already approved, socket.connect won't trigger another permission dialog.
const std::vector<std::string> call_site_ids =
PluginAuditDetail::call_site_identities(plugin_descriptor.plugin_root);
if (mgr->has_approved_ancestor(mgr->current_plugin(), call_site_ids))
return 0;
// A filesystem target that resolves entirely inside a pre-determined allowed root -- the
// plugin system's own data_dir() tree (which holds each plugin's storage folder and the
// installed/system profile cache), the read-only bundled resources folder, or a per-call
// scoped root such as the current G-code folder -- is part of the plugin system's normal
// workflow and does not need a prompt.
if (fs_category && !targets.empty()) {
const bool is_write = event_type != AuditEventCategory::FsRead;
const bool all_inside_allowed_root =
std::all_of(targets.begin(), targets.end(), [&](const std::string& target) {
return mgr->check_path_access(boost::filesystem::path(target), is_write).allowed;
});
if (all_inside_allowed_root)
return 0;
}
PluginInstallState state;
const bool have_install_state = PluginManager::instance().get_install_state(mgr->current_plugin(), state);
if (!have_install_state) {
BOOST_LOG_TRIVIAL(warning) << __FUNCTION__ << " Failed to get install state for " << mgr->current_plugin();
}
const std::string plugin_name = state.plugin_name.empty() ? mgr->current_plugin() : state.plugin_name;
std::vector<std::string>* permission_list = PluginAuditDetail::permission_list_for(event_type, state.permissions);
return PluginAuditDetail::decide_audited_event(*mgr, state, mgr->current_plugin(), plugin_name, event_name,
event_type, targets, permission_list, call_site_ids);
// Root frames whose operation has returned are let go once this event is decided, so that a
// finalizer they run re-enters a hook whose held frames are already consistent.
std::vector<PyObject*> returned_frames;
PluginAuditDetail::take_returned_frames(m_root_frames, returned_frames);
const int result = PluginAuditDetail::audit_plugin_event(*mgr, event_name, event_type, args);
mgr->release_frames(returned_frames);
return result;
}
void PluginAuditManager::install_hook()
{
// Tells the stdlib's own frames apart, for the operation-root frames (see record_root_frame).
PluginAuditDetail::collect_stdlib_dirs();
// data_dir() is the primary globally-allowed root during plugin execution: read+write. It
// covers the plugin system's own workflow needs -- each plugin's storage folder
// (data_dir()/orca_plugins) and the installed/system profile cache (data_dir()/system) --
+34 -13
View File
@@ -4,10 +4,9 @@
// Via pybind11 so this file requests the same python3xx.lib as everything else.
#include <boost/filesystem/path.hpp>
#include <pybind11/conduit/wrap_include_python_h.h>
#include <functional>
#include <mutex>
#include <string>
#include <unordered_map>
#include <unordered_set>
#include <vector>
#include <boost/filesystem.hpp>
@@ -144,15 +143,26 @@ public:
void clear_last_violation();
bool last_violation(AuditViolation& violation) const;
// --- call-site cascade cache ---
// A single plugin action often fires several nested CPython audit events as it passes
// through stdlib layers (urllib.request calling http.client calling socket, for example).
// Once the user approves one event, every stdlib frame still on the stack for that call
// is recorded here by (filename, function, first line) identity. A later event whose own
// ancestor chain still contains one of those frames is the same logical action seen from
// a deeper layer, so it is auto-approved instead of prompting again.
bool has_approved_ancestor(const std::string& plugin_key, const std::vector<std::string>& call_site_ids) const;
void record_approved_call_sites(const std::string& plugin_key, const std::vector<std::string>& call_site_ids);
// --- operation-root frames ---
// One stdlib operation fires several nested audit events as it passes through lower layers:
// urlopen() raises urllib.Request, then http.client and socket events. When an operation's
// own root event is granted (by a saved grant or the user), the stdlib frame that raised it
// is held, and a later event whose stack reaches that frame through stdlib frames only is
// part of the same operation, so it is allowed without a prompt of its own. Plugin or
// library code the operation calls back into is judged on its own. A frame is let go once it
// returns, a suspended generator's (os.walk) when the audit scope ends. Letting a frame go may
// run finalizers of objects its locals held; they are judged as the plugin and never prompt.
void record_root_frame();
bool covered_by_root_frame() const;
bool releasing_frames() const;
// True when code from `filename` (a code object's co_filename) is the interpreter's own
// stdlib: the only code a root frame's coverage passes through. Valid after install_hook().
static bool is_stdlib_file(const std::string& filename);
// Answers permission prompts in place of the dialog, for tests. Production never sets it.
using PermissionPrompt = std::function<bool(const std::string& event_name, const std::vector<std::string>& targets)>;
static void set_permission_prompt_for_testing(PermissionPrompt prompt);
bool verbose_events = true;
@@ -163,24 +173,32 @@ private:
static int audit_hook(const char* event, PyObject* args, void* user_data);
// Drops the references in `frames`, keeping the caller's error and denial state intact.
void release_frames(std::vector<PyObject*>& frames);
static thread_local std::string m_current_plugin_key;
static thread_local std::string m_current_capability_name;
static thread_local std::vector<AllowedRoot> m_scoped_allowed_roots;
static thread_local std::vector<PyObject*> m_root_frames; // strong references
static thread_local int m_releasing_frames;
static thread_local bool m_audit_denial_pending;
static thread_local bool m_has_last_violation;
static thread_local AuditViolation m_last_violation;
// mutable: is_denied_filename() and has_approved_ancestor() are const queries that must lock.
// mutable: is_denied_filename() is a const query that must lock.
mutable std::mutex m_mutex;
std::vector<AllowedRoot> m_global_allowed_roots;
std::vector<std::string> m_denied_filenames;
std::vector<std::string> m_denied_path_keywords;
std::unordered_map<std::string, std::unordered_set<std::string>> m_approved_call_sites; // plugin_key -> call-site ids
};
// RAII guard that sets the current plugin key and capability name, restoring the previous
// pair on scope exit. `capability_name` may be empty for calls that are not scoped to a
// single capability.
//
// Each scope owns the root frames recorded inside it and lets them go, as its own plugin, before
// restoring the outer context. That needs the GIL whenever a frame is still held, so construct
// the scope after the GIL guard, to be destroyed before it.
class ScopedPluginAuditContext
{
public:
@@ -194,9 +212,12 @@ public:
ScopedPluginAuditContext& operator=(const ScopedPluginAuditContext&) = delete;
private:
std::string m_plugin_key;
std::string m_capability;
std::string m_previous_id;
std::string m_previous_capability;
std::vector<AllowedRoot> m_previous_scoped_roots;
std::vector<PyObject*> m_previous_root_frames;
};
} // namespace Slic3r
+2
View File
@@ -808,6 +808,7 @@ bool read_install_state(const boost::filesystem::path& plugin_dir, PluginInstall
read_string_list("network_socket", parsed.permissions.network_socket);
read_string_list("process", parsed.permissions.process);
read_string_list("threading", parsed.permissions.threading);
read_string_list("events", parsed.permissions.events);
}
if (state.contains("enabled") && state["enabled"].is_boolean())
@@ -852,6 +853,7 @@ bool write_install_state(const boost::filesystem::path& plugin_dir, const Plugin
{"network_socket", state.permissions.network_socket},
{"process", state.permissions.process},
{"threading", state.permissions.threading},
{"events", state.permissions.events},
};
nlohmann::json capabilities = nlohmann::json::array();
+2
View File
@@ -104,6 +104,8 @@ struct PluginPermissions
std::vector<std::string> network_socket;
std::vector<std::string> process;
std::vector<std::string> threading;
// Grants for audit events that expose no target, keyed "<category>:<event>" (e.g. "Socket:socket.__new__").
std::vector<std::string> events;
};
struct PluginInstallState {
+1 -1
View File
@@ -102,9 +102,9 @@ public:
// mistake), both fall back to the base's empty object rather than writing `"cap_config": null`.
nlohmann::json get_default_config() const override
{
pybind11::gil_scoped_acquire gil;
ORCA_PY_AUDIT_SCOPE();
try {
pybind11::gil_scoped_acquire gil;
pybind11::function override = pybind11::get_override(static_cast<const Base*>(this), "get_default_config");
if (!override)
return Base::get_default_config();
@@ -282,11 +282,11 @@ public:
int request_bind_ticket(std::string* ticket) override
{
try {
ORCA_PY_AUDIT_SCOPE();
::Slic3r::PluginCapabilityInterface::RefCounter _orca_ref_counter(*this);
::Slic3r::PythonGILState gil;
if (!gil)
throw std::runtime_error("Python interpreter is shutting down");
ORCA_PY_AUDIT_SCOPE();
pybind11::function override =
pybind11::get_override(static_cast<const PrinterAgentPluginCapability*>(this), "request_bind_ticket");
if (!override)
+770
View File
@@ -1,22 +1,48 @@
#include <boost/filesystem/path.hpp>
#include <boost/filesystem/operations.hpp>
#include <boost/system/error_code.hpp>
#include <catch2/catch_all.hpp>
#include <libslic3r/Utils.hpp>
#include <libslic3r/libslic3r.h> // GCODEVIEWER_APP_KEY, SLIC3R_APP_KEY (via libslic3r_version.h)
#include <slic3r/plugin/PluginAuditManager.hpp>
#include <slic3r/plugin/PluginFsUtils.hpp>
#include <slic3r/plugin/PluginManager.hpp>
#include <slic3r/plugin/PythonInterpreter.hpp>
#include <slic3r/plugin/PythonPluginBridge.hpp>
#include <slic3r/plugin/PythonPluginInterface.hpp>
#include <slic3r/Utils/IPrinterAgent.hpp>
#include <slic3r/Utils/OrcaCloudServiceAgent.hpp> // secret_constants::USER_SECRET_FILENAME
#include <catch2/catch_message.hpp>
#include <catch2/catch_test_macros.hpp>
#include <catch2/matchers/catch_matchers.hpp>
#include <catch2/matchers/catch_matchers_string.hpp>
#include "libslic3r_version.h"
#include "plugin_test_utils.hpp"
#include "test_utils.hpp"
#include <boost/filesystem.hpp>
#include <nlohmann/json.hpp>
#include <pybind11/cast.h>
#include <pybind11/embed.h>
#include <pybind11/eval.h>
#include <pybind11/gil.h>
#include <pybind11/pybind11.h>
#include <pybind11/pytypes.h>
#include <pybind11/stl.h>
#include <fstream>
#include <functional>
#include <ios>
#include <memory>
#include <string>
#include <utility>
#include <vector>
using namespace Slic3r;
namespace fs = boost::filesystem;
namespace py = pybind11;
namespace {
@@ -349,3 +375,747 @@ TEST_CASE("Plugin audit a scoped root can also be registered read-only", "[audit
CHECK_FALSE(mgr.check_open((readonly_dir / "vendor.json").string(), "w").allowed);
}
}
// ---------------------------------------------------------------------------
// The audit hook, run in the embedded interpreter
// ---------------------------------------------------------------------------
//
// Each TEST_CASE runs in its own process (catch_discover_tests), so each brings the plugin system
// up with the audit hook installed, once, and shuts it down at the end. The tests loop over their
// cases instead of using SECTIONs, which would restart the interpreter for every section.
namespace {
// Discovered, never loaded: the hook only needs the plugin's root and its sidecar.
const char* const AUDIT_PLUGIN_SOURCE = R"PY(# /// script
# requires-python = ">=3.12"
#
# [tool.orcaslicer.plugin]
# name = "Audit Plugin"
# description = "Plugin audit hook fixture"
# author = "OrcaSlicer"
# version = "1.0"
# type = "script"
# ///
)PY";
// A third-party library, installed outside the stdlib and outside the plugin.
const char* const LIBRARY_SOURCE = R"PY(
import socket
import urllib.request
def read_all(paths):
for path in paths:
with open(path) as f:
f.read()
def connect_from_host(port):
socket.gethostname()
socket.create_connection(("127.0.0.1", port)).close()
def fetch(url):
with urllib.request.urlopen(url) as response:
return response.read()
def fetch_each(urls):
for url in urls:
yield fetch(url)
class ReadingHandler(urllib.request.BaseHandler):
"""Reads a file while each response is processed."""
def __init__(self, path):
self.path = path
self.outcome = None
def http_response(self, request, response):
try:
open(self.path).close()
self.outcome = "ok"
except PermissionError:
self.outcome = "denied"
return response
)PY";
// Test-side Python, run outside any plugin context: files outside every allowed root, a local
// HTTP server, the library, and every module the tests use imported up front, so that no import
// happens inside a plugin context. Code from exec() runs as "<string>", which is not stdlib, so
// attempt() and fetch() stand in for plugin code.
const char* const AUDIT_SETUP = R"PY(
import http.client, http.server, os, pathlib, shutil, socket, subprocess, sys, tempfile, threading
import urllib.request, zipfile
if os.name != "nt":
import pty
APPROVED = os.path.join(OUTSIDE, "approved.txt")
PRIVATE = os.path.join(OUTSIDE, "private.txt")
MISSING = os.path.join(OUTSIDE, "missing")
TREE = os.path.join(OUTSIDE, "tree")
TRUE_COMMAND = ["cmd", "/c", "exit 0"] if os.name == "nt" else ["true"]
for path in (APPROVED, PRIVATE):
with open(path, "w") as f:
f.write("x")
for sub in ("one", "two"):
os.makedirs(os.path.join(TREE, sub))
with open(os.path.join(TREE, sub, "f.txt"), "w") as f:
f.write("x")
LIBRARY = os.path.join(OUTSIDE, "library")
os.makedirs(LIBRARY)
with open(os.path.join(LIBRARY, "netlib.py"), "w") as f:
f.write(LIBRARY_SOURCE)
sys.path.insert(0, LIBRARY)
import netlib
class Handler(http.server.BaseHTTPRequestHandler):
def do_GET(self):
if self.path == "/redirect":
self.send_response(302)
self.send_header("Location", URL_B)
body = b""
else:
self.send_response(200)
body = b"ok"
self.send_header("Content-Length", str(len(body)))
self.end_headers()
self.wfile.write(body)
def log_message(self, *args):
pass
server = http.server.ThreadingHTTPServer(("127.0.0.1", 0), Handler)
threading.Thread(target=server.serve_forever, daemon=True).start()
PORT = server.server_address[1]
BASE = "http://127.0.0.1:%d" % PORT
URL_A, URL_B, URL_REDIRECT = BASE + "/a", BASE + "/b", BASE + "/redirect"
urllib.request.install_opener(urllib.request.build_opener(urllib.request.ProxyHandler({})))
def fetch(url):
with urllib.request.urlopen(url) as response:
return response.read()
fetch(URL_A) # urllib imports some modules on its first request
def attempt(action):
try:
action()
return "ok"
except PermissionError:
return "denied"
def spawn_checked(argv):
parent = os.getpid()
try:
return pty.spawn(argv)
finally:
if os.getpid() != parent:
os._exit(1) # a child whose exec was refused must not go on running the tests
finalizer_outcomes = []
class TrackedRequest(urllib.request.Request):
def __del__(self):
finalizer_outcomes.append(attempt(lambda: open(PRIVATE).close()))
)PY";
struct Prompt
{
std::string plugin;
std::string event;
std::vector<std::string> targets;
};
// One line per prompt, "event: target ...", so a failed check shows every prompt.
std::string describe(const std::vector<Prompt>& prompts)
{
std::string text;
for (const Prompt& prompt : prompts) {
text += prompt.event + ":";
for (const std::string& target : prompt.targets)
text += " " + target;
text += "\n";
}
return text;
}
// The plugin system with the audit hook installed, and the permission dialog replaced by a
// recorder that answers with `approve`. Declare it first in a test, so it is destroyed last.
struct AuditHarness
{
// Before initialize(): the interpreter creates {data_dir}/python/packages and {data_dir}/log.
// data_dir() is a global allowed root from then on, so the test files live in `outside`.
ScopedDataDir data{"plugin-audit-hook"};
bool initialized = PluginManager::instance().initialize();
ScopedTemporaryDir outside{"orca-plugin-audit-outside"};
std::vector<Prompt> prompts;
std::function<bool(const Prompt&)> approve = [](const Prompt&) { return false; };
AuditHarness()
{
PluginAuditManager::set_permission_prompt_for_testing(
[this](const std::string& event, const std::vector<std::string>& targets) {
prompts.push_back({PluginAuditManager::instance().current_plugin(), event, targets});
return approve(prompts.back());
});
}
~AuditHarness()
{
PluginAuditManager::set_permission_prompt_for_testing(nullptr);
PluginManager::instance().shutdown();
PythonInterpreter::instance().shutdown();
}
// Call before taking the GIL: discovery is not Python work.
void add_plugin(const std::string& key)
{
const fs::path root = data.plugins_dir() / key;
fs::create_directories(root);
std::ofstream((root / (key + ".py")).string(), std::ios::binary) << AUDIT_PLUGIN_SOURCE;
save_permissions(key, {});
PluginManager::instance().discover_plugins(/*async=*/false, /*clear=*/true);
}
// Grants as if saved in an earlier session.
void save_permissions(const std::string& key, const PluginPermissions& permissions)
{
PluginInstallState state;
state.installed_from = "local";
state.installed_version = "1.0";
state.plugin_name = key;
state.permissions = permissions;
REQUIRE(write_install_state(data.plugins_dir() / key, state));
}
PluginPermissions saved_permissions(const std::string& key) const
{
PluginInstallState state;
read_install_state(data.plugins_dir() / key, state);
return state.permissions;
}
};
// The test-side Python globals. Construct and destroy it with the GIL held; it stops the HTTP
// server before the interpreter shuts down.
struct PythonSide
{
py::dict globals;
explicit PythonSide(const AuditHarness& harness)
{
globals["OUTSIDE"] = harness.outside.path().string();
globals["LIBRARY_SOURCE"] = LIBRARY_SOURCE;
exec(AUDIT_SETUP);
}
// A Python error fails the test here, as text: an error_already_set left to propagate would
// outlive the interpreter the harness shuts down.
void exec(const char* code)
{
std::string error;
try {
py::exec(code, globals);
} catch (py::error_already_set& raised) {
error = raised.what();
}
if (!error.empty())
FAIL(error);
}
~PythonSide()
{
try {
globals["server"].attr("shutdown")();
globals["server"].attr("server_close")();
} catch (...) {
}
}
std::string str(const char* name) const { return py::str(globals[name]).cast<std::string>(); }
std::string eval(const char* expression) const { return py::str(py::eval(expression, globals)).cast<std::string>(); }
std::vector<std::string> strings(const char* name) const { return globals[name].cast<std::vector<std::string>>(); }
// Runs `code` as `plugin_key`, as a trampoline call does: inside the caller's GIL, in an
// audit scope of its own.
void run_as(const std::string& plugin_key, const char* code)
{
ScopedPluginAuditContext audit(plugin_key);
exec(code);
}
};
PluginPermissions http_grants(std::vector<std::string> urls)
{
PluginPermissions permissions;
permissions.network_http = std::move(urls);
return permissions;
}
const char* const PLUGIN = "AuditPlugin";
} // namespace
TEST_CASE("A saved request grant covers the request's own connection in a later session", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin(PLUGIN);
py::gil_scoped_acquire gil;
PythonSide python(harness);
harness.save_permissions(PLUGIN, http_grants({python.str("URL_A")}));
python.run_as(PLUGIN, "outcome = attempt(lambda: fetch(URL_A))");
CHECK(python.str("outcome") == "ok");
CHECK(describe(harness.prompts).empty());
}
TEST_CASE("A saved request grant does not cover another URL or a redirect", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin(PLUGIN);
py::gil_scoped_acquire gil;
PythonSide python(harness);
harness.save_permissions(PLUGIN, http_grants({python.str("URL_A"), python.str("URL_REDIRECT")}));
python.run_as(PLUGIN, R"PY(
outcomes = [attempt(lambda: fetch(url)) for url in (URL_A, URL_B, URL_REDIRECT)]
outcomes += [attempt(lambda: netlib.fetch(url)) for url in (URL_A, URL_B)]
outcomes.append(attempt(lambda: list(netlib.fetch_each([URL_A, URL_B]))))
)PY");
// The redirect target is asked for inside the approved redirecting request, and a library's
// wrapper or generator asks for the second URL as plain code does.
CHECK(python.strings("outcomes") == std::vector<std::string>{"ok", "denied", "denied", "ok", "denied", "denied"});
const std::string url_b = "urllib.Request: " + python.str("URL_B") + "\n";
CHECK(describe(harness.prompts) == url_b + url_b + url_b + url_b);
}
TEST_CASE("Code an approved operation calls back into is judged on its own", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin(PLUGIN);
py::gil_scoped_acquire gil;
PythonSide python(harness);
python.exec(R"PY(
CALLBACK_COPY = os.path.join(OUTSIDE, "callback-copy")
handler = netlib.ReadingHandler(PRIVATE)
handler_opener = urllib.request.build_opener(urllib.request.ProxyHandler({}), handler)
walk_outcomes = []
copy_outcomes = []
def on_walk_error(error):
walk_outcomes.append(attempt(lambda: open(PRIVATE).close()))
def copy_file(source, destination):
copy_outcomes.append(attempt(lambda: open(PRIVATE).close()))
)PY");
PluginPermissions permissions = http_grants({python.str("URL_A")});
permissions.fs_read = {python.str("APPROVED"), python.str("MISSING")};
permissions.fs_readwrite = {python.str("TREE"), python.str("CALLBACK_COPY")};
harness.save_permissions(PLUGIN, permissions);
// A library handler that urllib calls while the approved request runs, the plugin's own
// os.walk error callback and copytree copy function (called once per file), and a library
// loop over a saved file and an unsaved one.
python.run_as(PLUGIN, R"PY(
request_outcome = attempt(lambda: handler_opener.open(URL_A).read())
for _ in os.walk(MISSING, onerror=on_walk_error):
pass
shutil.copytree(TREE, CALLBACK_COPY, copy_function=copy_file)
loop_outcome = attempt(lambda: netlib.read_all([APPROVED, PRIVATE]))
)PY");
CHECK(python.str("request_outcome") == "ok");
CHECK(python.eval("handler.outcome") == "denied");
CHECK(python.strings("walk_outcomes") == std::vector<std::string>{"denied"});
CHECK(python.strings("copy_outcomes") == std::vector<std::string>{"denied", "denied"});
CHECK(python.str("loop_outcome") == "denied");
const std::string private_file = "open: " + python.str("PRIVATE") + "\n";
CHECK(describe(harness.prompts) == private_file + private_file + private_file + private_file + private_file);
}
TEST_CASE("A saved directory walk covers each directory it lists", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin(PLUGIN);
py::gil_scoped_acquire gil;
PythonSide python(harness);
PluginPermissions permissions;
permissions.fs_read = {python.str("TREE"), python.str("APPROVED")};
harness.save_permissions(PLUGIN, permissions);
// The walk lists the tree's two subdirectories on later resumes, after the plugin's own
// audited reads in between. Abandoned walks leave frames whose generator is gone, which the
// next audited event lets go.
python.run_as(PLUGIN, R"PY(
walked = []
for root, dirs, files in os.walk(TREE):
walked.append(root)
open(APPROVED).close()
globbed = list(pathlib.Path(TREE).glob("**/*"))
abandoned = os.walk(TREE)
next(abandoned)
del abandoned
open(APPROVED).close()
for _ in pathlib.Path(TREE).glob("*"):
break
open(APPROVED).close()
)PY");
CHECK(py::len(python.globals["walked"]) == 3);
CHECK(py::len(python.globals["globbed"]) == 4);
CHECK(describe(harness.prompts).empty());
}
TEST_CASE("A freshly approved operation covers its nested events", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin(PLUGIN);
py::gil_scoped_acquire gil;
PythonSide python(harness);
struct Case
{
const char* prompts;
const char* code;
};
// Each operation names targets no earlier one saved, so each is asked once. FsCreate and
// FsDelete grants are never saved, so those operations are asked every time. Only the
// expected prompts are approved, so a nested event left uncovered fails the operation.
std::vector<Case> cases = {
{"shutil.copytree", "shutil.copytree(TREE, os.path.join(OUTSIDE, 'copy'))"},
{"shutil.copyfile", "shutil.copyfile(APPROVED, os.path.join(OUTSIDE, 'copied.txt'))"},
{"shutil.move", "shutil.move(os.path.join(OUTSIDE, 'copied.txt'), os.path.join(OUTSIDE, 'moved.txt'))"},
{"shutil.copystat", "shutil.copystat(os.path.join(TREE, 'one', 'f.txt'), PRIVATE)"},
{"shutil.make_archive", "shutil.make_archive(os.path.join(OUTSIDE, 'archive'), 'zip', root_dir=TREE)"},
{"shutil.unpack_archive", "shutil.unpack_archive(os.path.join(OUTSIDE, 'archive.zip'), os.path.join(OUTSIDE, 'unpacked'))"},
{"pathlib.Path.glob", "list(pathlib.Path(TREE).glob('**/*'))"},
{"subprocess.Popen", "subprocess.run(TRUE_COMMAND, check=True)"},
{"tempfile.mkdtemp", "tempfile.mkdtemp(dir=OUTSIDE)"},
{"tempfile.mkstemp", "os.close(tempfile.mkstemp(dir=OUTSIDE)[0])"},
{"shutil.rmtree", "shutil.rmtree(os.path.join(OUTSIDE, 'copy'))"},
// send() runs after connect() has returned, so its own grant is asked once.
{"http.client.connect\nhttp.client.send",
"connection = http.client.HTTPConnection('127.0.0.1', PORT)\n"
"connection.request('GET', '/a')\n"
"connection.getresponse().read()\n"
"connection.close()"},
};
#ifndef _WIN32
cases.push_back({"os.fwalk", "list(os.fwalk(os.path.join(OUTSIDE, 'unpacked')))"});
// The forked child's exec is covered by the spawn too: had it prompted, the child's own
// denial would end it with status 1.
cases.push_back({"pty.spawn", "spawn_status = spawn_checked(TRUE_COMMAND + ['spawned'])"});
#endif
for (const Case& operation : cases) {
INFO(operation.code);
const std::string expected = std::string(operation.prompts) + "\n";
harness.approve = [&expected](const Prompt& prompt) {
return expected.find(prompt.event + "\n") != std::string::npos;
};
harness.prompts.clear();
python.run_as(PLUGIN, operation.code);
std::string asked;
for (const Prompt& prompt : harness.prompts)
asked += (asked.empty() ? "" : "\n") + prompt.event;
CHECK(asked == operation.prompts);
}
#ifndef _WIN32
CHECK(python.eval("spawn_status") == "0");
#endif
}
TEST_CASE("An event with no target is granted by name and remembered", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin(PLUGIN);
harness.approve = [](const Prompt& prompt) { return prompt.event != "socket.sendto"; };
py::gil_scoped_acquire gil;
PythonSide python(harness);
python.run_as(PLUGIN, R"PY(
socket.socket().close()
udp = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
sendto_outcome = attempt(lambda: udp.sendto(b"x", ("127.0.0.1", 9)))
udp.close()
)PY");
// The socket is asked for once, without a target to show; the datagram's address is a target.
CHECK(python.str("sendto_outcome") == "denied");
CHECK(describe(harness.prompts) == "socket.__new__:\nsocket.sendto: ('127.0.0.1', 9)\n");
CHECK(harness.saved_permissions(PLUGIN).events == std::vector<std::string>{"Socket:socket.__new__"});
CHECK(harness.saved_permissions(PLUGIN).network_socket.empty());
harness.prompts.clear();
python.run_as(PLUGIN, "socket.socket().close()");
CHECK(describe(harness.prompts).empty());
// A grant for an event with no target covers nothing else: a library's approved
// gethostname() does not carry the connection it makes next.
harness.approve = [](const Prompt& prompt) { return prompt.event == "socket.gethostname"; };
python.run_as(PLUGIN, "connect_outcome = attempt(lambda: netlib.connect_from_host(PORT))");
CHECK(python.str("connect_outcome") == "denied");
CHECK(describe(harness.prompts) == "socket.gethostname:\nsocket.getaddrinfo: 127.0.0.1\n");
harness.prompts.clear();
harness.approve = [](const Prompt&) { return true; };
#ifndef _WIN32
// sendmsg() without an address names no target either.
python.run_as(PLUGIN, R"PY(
left, right = socket.socketpair()
left.sendmsg([b"x"])
left.close()
right.close()
)PY");
CHECK(describe(harness.prompts) == "socket.sendmsg:\n");
CHECK(harness.saved_permissions(PLUGIN).events ==
std::vector<std::string>{"Socket:socket.__new__", "Socket:socket.gethostname", "Socket:socket.sendmsg"});
#endif
}
TEST_CASE("An event grant does not stand in for another access or a target of the same name", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin(PLUGIN);
harness.approve = [asked = 0](const Prompt&) mutable { return asked++ == 0; };
py::gil_scoped_acquire gil;
PythonSide python(harness);
python.exec(R"PY(
READ_FD = os.open(APPROVED, os.O_RDONLY)
WRITE_FD = os.open(os.path.join(OUTSIDE, "scratch.txt"), os.O_WRONLY | os.O_CREAT)
)PY");
// A file descriptor has no path to show or save: a granted read of one is not a write.
python.run_as(PLUGIN, R"PY(
fd_outcomes = [attempt(lambda: open(READ_FD, "r", closefd=False).close()),
attempt(lambda: open(WRITE_FD, "w", closefd=False).close())]
)PY");
CHECK(python.strings("fd_outcomes") == std::vector<std::string>{"ok", "denied"});
CHECK(describe(harness.prompts) == "open:\nopen:\n");
CHECK(harness.saved_permissions(PLUGIN).events == std::vector<std::string>{"FsRead:open"});
// A saved os.fork event grant is not an os.system command named "os.fork".
PluginPermissions permissions;
permissions.events = {"FsRead:open", "ProcessCreate:os.fork"};
harness.save_permissions(PLUGIN, permissions);
harness.prompts.clear();
python.run_as(PLUGIN, R"PY(
reread_outcome = attempt(lambda: open(READ_FD, "r", closefd=False).close())
system_outcome = attempt(lambda: os.system("os.fork"))
os.close(READ_FD)
os.close(WRITE_FD)
)PY");
CHECK(python.str("reread_outcome") == "ok");
CHECK(python.str("system_outcome") == "denied");
REQUIRE(harness.prompts.size() == 1);
CHECK(harness.prompts.front().event == "os.system");
CHECK(harness.saved_permissions(PLUGIN).process.empty());
}
TEST_CASE("A finalizer run when a root frame is let go is judged as the plugin without a prompt", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin(PLUGIN);
py::gil_scoped_acquire gil;
PythonSide python(harness);
PluginPermissions permissions = http_grants({python.str("URL_A")});
permissions.fs_read = {python.str("APPROVED")};
harness.save_permissions(PLUGIN, permissions);
PluginAuditManager& mgr = PluginAuditManager::instance();
// Once the request returns, only its held frame keeps the request object alive, and the
// request's finalizer reads a file the plugin was never granted.
python.run_as(PLUGIN, "fetch(TrackedRequest(URL_A))");
CHECK(python.strings("finalizer_outcomes") == std::vector<std::string>{"denied"}); // at scope end
CHECK(describe(harness.prompts).empty());
CHECK_FALSE(mgr.audit_denial_pending());
python.exec("finalizer_outcomes.clear()");
python.run_as(PLUGIN, R"PY(
fetch(TrackedRequest(URL_A))
open(APPROVED).close() # the next event lets the returned request's frame go
outcomes_before_scope_end = list(finalizer_outcomes)
)PY");
CHECK(python.strings("outcomes_before_scope_end") == std::vector<std::string>{"denied"});
CHECK(describe(harness.prompts).empty());
// The finalizer's denial during another event's decision leaves that event's own denial intact.
python.exec("finalizer_outcomes.clear()");
python.run_as(PLUGIN, R"PY(
fetch(TrackedRequest(URL_A))
try:
os.listdir(OUTSIDE)
outer_error = ""
except PermissionError as error:
outer_error = str(error)
)PY");
CHECK(python.strings("finalizer_outcomes") == std::vector<std::string>{"denied"});
CHECK_THAT(python.str("outer_error"), Catch::Matchers::ContainsSubstring("os.listdir"));
CHECK(describe(harness.prompts) == "os.listdir: " + python.str("OUTSIDE") + "\n");
AuditViolation violation;
REQUIRE(mgr.last_violation(violation));
CHECK(violation.event_name == "os.listdir");
CHECK(mgr.audit_denial_pending());
}
TEST_CASE("A nested audit scope does not see the outer scope's running operations", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.add_plugin("PluginA");
harness.add_plugin("PluginB");
py::gil_scoped_acquire gil;
PythonSide python(harness);
PluginPermissions permissions;
permissions.fs_read = {python.str("TREE")};
harness.save_permissions("PluginA", permissions);
{
// Two walks plugin A started and suspended after listing the tree itself.
ScopedPluginAuditContext plugin_a("PluginA");
python.exec("first = os.walk(TREE)\nnext(first)\nsecond = os.walk(TREE)\nnext(second)");
{
// Plugin B resuming one of them is asked for each subdirectory, as plugin B.
ScopedPluginAuditContext plugin_b("PluginB");
python.exec("listed_by_b = len(list(first))");
}
// Back in plugin A's scope, its walks are covered again.
python.exec("listed_by_a = len(list(second))");
}
CHECK(python.globals["listed_by_b"].cast<int>() == 0);
CHECK(python.globals["listed_by_a"].cast<int>() == 2);
REQUIRE(harness.prompts.size() == 2);
for (const Prompt& prompt : harness.prompts) {
CHECK(prompt.plugin == "PluginB");
CHECK(prompt.event == "os.scandir");
}
}
TEST_CASE("Root frames held when a capability call ends are let go with the GIL, as the plugin", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
harness.approve = [](const Prompt& prompt) { return prompt.event == "os.walk"; };
py::gil_scoped_acquire gil;
PythonSide python(harness);
(void) PythonPluginBridge::instance(); // force the embedded module registration into the binary
python.globals["orca"] = py::module_::import("orca");
// walk_and_drop() leaves a returned walk's frame held by the audit scope alone, with a marker
// in its locals; the frame is let go when the capability call ends, which runs the marker's
// finalizer.
python.exec(R"PY(
class Marker:
def __init__(self, action=None):
self.action = action
def __del__(self):
finalized.append(self.action() if self.action else "ran")
finalized = []
def walk_and_drop(marker):
walker = os.walk(TREE, onerror=marker)
next(walker)
del walker
class ConfigCapability(orca.PythonPluginBase):
def get_name(self):
return "config_cap"
def get_default_config(self):
walk_and_drop(Marker())
return {"finalized_during_call": len(finalized)}
def has_config_ui(self):
walker = os.walk(TREE, onerror=Marker(lambda: attempt(lambda: open(PRIVATE).close())))
next(walker)
attempt(lambda: os.listdir(OUTSIDE)) # denied: the trampoline clears the plugin key
del walker # the walk's frame is now held by the audit scope alone
raise RuntimeError("the capability fails after a denial")
class TicketAgent(orca.printer_agent.PrinterAgentBase):
def get_name(self):
return "ticket_agent"
def request_bind_ticket(self):
walk_and_drop(Marker())
return (0, "ticket")
)PY");
py::object config_instance = python.globals["ConfigCapability"]();
auto config_cap = config_instance.cast<std::shared_ptr<PluginCapabilityInterface>>();
config_cap->set_audit_plugin_key("GilPlugin");
config_cap->set_resolved_identity("config_cap", PluginCapabilityType::Script);
py::object agent_instance = python.globals["TicketAgent"]();
auto agent_cap = agent_instance.cast<std::shared_ptr<PluginCapabilityInterface>>();
agent_cap->set_audit_plugin_key("GilPlugin");
agent_cap->set_resolved_identity("ticket_agent", PluginCapabilityType::PrinterConnection);
auto agent = std::dynamic_pointer_cast<IPrinterAgent>(agent_cap);
REQUIRE(agent);
// A host call arrives without the GIL; both calls take it before their audit scope.
nlohmann::json config;
int ticket_result = -1;
std::string ticket;
{
py::gil_scoped_release released;
config = config_cap->get_default_config();
ticket_result = agent->request_bind_ticket(&ticket);
}
CHECK(config == nlohmann::json{{"finalized_during_call", 0}});
CHECK(ticket_result == 0);
CHECK(ticket == "ticket");
CHECK(python.strings("finalized") == std::vector<std::string>{"ran", "ran"});
// On the trampoline's denial path the frame is let go as the plugin, though the ambient key
// was cleared: the finalizer's read is audited and denied without a prompt.
python.exec("finalized.clear()");
harness.prompts.clear();
CHECK_THROWS_AS(config_cap->has_config_ui(), py::error_already_set);
CHECK(python.strings("finalized") == std::vector<std::string>{"denied"});
CHECK(describe(harness.prompts) == "os.walk: " + python.str("TREE") + "\nos.listdir: " + python.str("OUTSIDE") + "\n");
}
TEST_CASE("Stdlib code is recognized through a symlinked Python home", "[audit][Python]")
{
AuditHarness harness;
if (!harness.initialized)
SKIP("Bundled Python interpreter unavailable: " + PythonInterpreter::instance().last_error());
py::gil_scoped_acquire gil;
const py::dict paths = py::module_::import("sysconfig").attr("get_paths")();
const fs::path stdlib = py::str(paths["stdlib"]).cast<std::string>();
const fs::path purelib = py::str(paths["purelib"]).cast<std::string>();
const fs::path link = harness.outside.path() / "linked-stdlib";
boost::system::error_code ec;
fs::create_directory_symlink(stdlib, link, ec);
if (ec)
SKIP("Cannot create a directory symlink here: " + ec.message());
CHECK(PluginAuditManager::is_stdlib_file((stdlib / "os.py").string()));
CHECK(PluginAuditManager::is_stdlib_file((link / "urllib" / "request.py").string()));
CHECK(PluginAuditManager::is_stdlib_file("<frozen os>"));
CHECK_FALSE(PluginAuditManager::is_stdlib_file((purelib / "requests" / "api.py").string()));
CHECK_FALSE(PluginAuditManager::is_stdlib_file((link / "site-packages" / "requests" / "api.py").string()));
CHECK_FALSE(PluginAuditManager::is_stdlib_file((harness.outside.path() / "plugin.py").string()));
CHECK_FALSE(PluginAuditManager::is_stdlib_file("<string>"));
}