[libgpiod][PATCH 1/3] tests: don't use the same chip object from different threads

[libgpiod][PATCH 1/3] tests: don't use the same chip object from different threads

[Bartosz Golaszewski]

From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

There are no thread-safety guarantees in libgpiod. Let's not reuse the
chip object created in one thread to generate info events in another but
create a second chip for that purpose instead.

Reported-by: Erik Schilling <erik.schilling@linaro.org>
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
---
 tests/tests-info-event.c | 62 +++++++++++++++++++++-------------------
 1 file changed, 33 insertions(+), 29 deletions(-)

diff --git a/tests/tests-info-event.c b/tests/tests-info-event.c
index d069811..cbd9e9e 100644
--- a/tests/tests-info-event.c
+++ b/tests/tests-info-event.c
@@ -49,43 +49,59 @@ GPIOD_TEST_CASE(event_timeout)
 }
 
 struct request_ctx {
-	struct gpiod_chip *chip;
-	struct gpiod_line_config *line_cfg;
-	struct gpiod_line_settings *settings;
+	const char *path;
 	guint offset;
 };
 
 static gpointer request_reconfigure_release_line(gpointer data)
 {
+	g_autoptr(struct_gpiod_line_settings) settings = NULL;
+	g_autoptr(struct_gpiod_line_config) line_cfg = NULL;
 	g_autoptr(struct_gpiod_line_request) request = NULL;
+	g_autoptr(struct_gpiod_chip) chip = NULL;
 	struct request_ctx *ctx = data;
 	gint ret;
 
+	chip = gpiod_chip_open(ctx->path);
+	g_assert_nonnull(chip);
+	if (g_test_failed())
+		return NULL;
+
+	line_cfg = gpiod_line_config_new();
+	g_assert_nonnull(line_cfg);
+	if (g_test_failed())
+		return NULL;
+
+	settings = gpiod_line_settings_new();
+	g_assert_nonnull(settings);
+	if (g_test_failed())
+		return NULL;
+
 	g_usleep(1000);
 
-	ret = gpiod_line_config_add_line_settings(ctx->line_cfg, &ctx->offset,
-						  1, ctx->settings);
+	ret = gpiod_line_config_add_line_settings(line_cfg, &ctx->offset,
+						  1, settings);
 	g_assert_cmpint(ret, ==, 0);
 	if (g_test_failed())
 		return NULL;
 
-	request = gpiod_chip_request_lines(ctx->chip, NULL, ctx->line_cfg);
+	request = gpiod_chip_request_lines(chip, NULL, line_cfg);
 	g_assert_nonnull(request);
 	if (g_test_failed())
 		return NULL;
 
 	g_usleep(1000);
 
-	gpiod_line_config_reset(ctx->line_cfg);
-	gpiod_line_settings_set_direction(ctx->settings,
+	gpiod_line_config_reset(line_cfg);
+	gpiod_line_settings_set_direction(settings,
 					  GPIOD_LINE_DIRECTION_OUTPUT);
-	ret = gpiod_line_config_add_line_settings(ctx->line_cfg, &ctx->offset,
-						  1, ctx->settings);
+	ret = gpiod_line_config_add_line_settings(line_cfg, &ctx->offset,
+						  1, settings);
 	g_assert_cmpint(ret, ==, 0);
 	if (g_test_failed())
 		return NULL;
 
-	ret = gpiod_line_request_reconfigure_lines(request, ctx->line_cfg);
+	ret = gpiod_line_request_reconfigure_lines(request, line_cfg);
 	g_assert_cmpint(ret, ==, 0);
 	if (g_test_failed())
 		return NULL;
@@ -106,25 +122,19 @@ GPIOD_TEST_CASE(request_reconfigure_release_events)
 	g_autoptr(struct_gpiod_info_event) request_event = NULL;
 	g_autoptr(struct_gpiod_info_event) reconfigure_event = NULL;
 	g_autoptr(struct_gpiod_info_event) release_event = NULL;
-	g_autoptr(struct_gpiod_line_config) line_cfg = NULL;
-	g_autoptr(struct_gpiod_line_settings) settings = NULL;
 	g_autoptr(GThread) thread = NULL;
 	struct gpiod_line_info *request_info, *reconfigure_info, *release_info;
 	guint64 request_ts, reconfigure_ts, release_ts;
 	struct request_ctx ctx;
+	const char *chip_path = g_gpiosim_chip_get_dev_path(sim);
 	gint ret;
 
-	chip = gpiod_test_open_chip_or_fail(g_gpiosim_chip_get_dev_path(sim));
-	line_cfg = gpiod_test_create_line_config_or_fail();
-	settings = gpiod_test_create_line_settings_or_fail();
-
+	chip = gpiod_test_open_chip_or_fail(chip_path);
 	info = gpiod_test_chip_watch_line_info_or_fail(chip, 3);
 
 	g_assert_false(gpiod_line_info_is_used(info));
 
-	ctx.chip = chip;
-	ctx.line_cfg = line_cfg;
-	ctx.settings = settings;
+	ctx.path = chip_path;
 	ctx.offset = 3;
 
 	thread = g_thread_new("request-release",
@@ -199,26 +209,20 @@ GPIOD_TEST_CASE(chip_fd_can_be_polled)
 	g_autoptr(struct_gpiod_chip) chip = NULL;
 	g_autoptr(struct_gpiod_line_info) info = NULL;
 	g_autoptr(struct_gpiod_info_event) event = NULL;
-	g_autoptr(struct_gpiod_line_config) line_cfg = NULL;
-	g_autoptr(struct_gpiod_line_settings) settings = NULL;
 	g_autoptr(GThread) thread = NULL;
+	const char *chip_path = g_gpiosim_chip_get_dev_path(sim);
 	struct gpiod_line_info *evinfo;
 	struct request_ctx ctx;
 	struct timespec ts;
 	struct pollfd pfd;
 	gint ret, fd;
 
-	chip = gpiod_test_open_chip_or_fail(g_gpiosim_chip_get_dev_path(sim));
-	settings = gpiod_test_create_line_settings_or_fail();
-	line_cfg = gpiod_test_create_line_config_or_fail();
-
+	chip = gpiod_test_open_chip_or_fail(chip_path);
 	info = gpiod_test_chip_watch_line_info_or_fail(chip, 3);
 
 	g_assert_false(gpiod_line_info_is_used(info));
 
-	ctx.chip = chip;
-	ctx.line_cfg = line_cfg;
-	ctx.settings = settings;
+	ctx.path = chip_path;
 	ctx.offset = 3;
 
 	thread = g_thread_new("request-release",
-- 
2.39.2

[libgpiod][PATCH 2/3] bindings: cxx: tests: don't use the same chip from different threads

[Bartosz Golaszewski]

From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

There are no thread-safety guarantees in libgpiod. Let's not reuse the
chip object created in one thread to generate info events in another but
create a second chip for that purpose instead.

Reported-by: Erik Schilling <erik.schilling@linaro.org>
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
---
 bindings/cxx/tests/tests-info-event.cpp | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/bindings/cxx/tests/tests-info-event.cpp b/bindings/cxx/tests/tests-info-event.cpp
index 249b1e8..21c0ef0 100644
--- a/bindings/cxx/tests/tests-info-event.cpp
+++ b/bindings/cxx/tests/tests-info-event.cpp
@@ -3,6 +3,7 @@
 
 #include <catch2/catch.hpp>
 #include <chrono>
+#include <filesystem>
 #include <gpiod.hpp>
 #include <sstream>
 #include <thread>
@@ -17,11 +18,11 @@ using event_type = ::gpiod::info_event::event_type;
 
 namespace {
 
-void request_reconfigure_release_line(::gpiod::chip& chip)
+void request_reconfigure_release_line(const ::std::filesystem::path& chip_path)
 {
 	::std::this_thread::sleep_for(::std::chrono::milliseconds(10));
 
-	auto request = chip
+	auto request = ::gpiod::chip(chip_path)
 		.prepare_request()
 		.add_line_settings(7, ::gpiod::line_settings())
 		.do_request();
@@ -48,7 +49,9 @@ TEST_CASE("Lines can be watched", "[info-event][chip]")
 		.set_num_lines(8)
 		.build();
 
-	::gpiod::chip chip(sim.dev_path());
+	const auto chip_path = sim.dev_path();
+
+	::gpiod::chip chip(chip_path);
 
 	SECTION("watch_line_info() returns line info")
 	{
@@ -74,7 +77,7 @@ TEST_CASE("Lines can be watched", "[info-event][chip]")
 
 		REQUIRE(info.direction() == direction::INPUT);
 
-		::std::thread thread(request_reconfigure_release_line, ::std::ref(chip));
+		::std::thread thread(request_reconfigure_release_line, ::std::ref(chip_path));
 
 		REQUIRE(chip.wait_info_event(::std::chrono::seconds(1)));
 		auto event = chip.read_info_event();
-- 
2.39.2

Re: [libgpiod][PATCH 2/3] bindings: cxx: tests: don't use the same chip from different threads

[Erik Schilling]

On Tue Sep 19, 2023 at 11:31 AM CEST, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>
> There are no thread-safety guarantees in libgpiod. Let's not reuse the
> chip object created in one thread to generate info events in another but
> create a second chip for that purpose instead.
>
> Reported-by: Erik Schilling <erik.schilling@linaro.org>
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

Reviewed-by: Erik Schilling <erik.schilling@linaro.org>

> ---
>  bindings/cxx/tests/tests-info-event.cpp | 11 +++++++----
>  1 file changed, 7 insertions(+), 4 deletions(-)
>
> diff --git a/bindings/cxx/tests/tests-info-event.cpp b/bindings/cxx/tests/tests-info-event.cpp
> index 249b1e8..21c0ef0 100644
> --- a/bindings/cxx/tests/tests-info-event.cpp
> +++ b/bindings/cxx/tests/tests-info-event.cpp
> @@ -3,6 +3,7 @@
>  
>  #include <catch2/catch.hpp>
>  #include <chrono>
> +#include <filesystem>
>  #include <gpiod.hpp>
>  #include <sstream>
>  #include <thread>
> @@ -17,11 +18,11 @@ using event_type = ::gpiod::info_event::event_type;
>  
>  namespace {
>  
> -void request_reconfigure_release_line(::gpiod::chip& chip)
> +void request_reconfigure_release_line(const ::std::filesystem::path& chip_path)
>  {
>  	::std::this_thread::sleep_for(::std::chrono::milliseconds(10));
>  
> -	auto request = chip
> +	auto request = ::gpiod::chip(chip_path)
>  		.prepare_request()
>  		.add_line_settings(7, ::gpiod::line_settings())
>  		.do_request();
> @@ -48,7 +49,9 @@ TEST_CASE("Lines can be watched", "[info-event][chip]")
>  		.set_num_lines(8)
>  		.build();
>  
> -	::gpiod::chip chip(sim.dev_path());
> +	const auto chip_path = sim.dev_path();
> +
> +	::gpiod::chip chip(chip_path);
>  
>  	SECTION("watch_line_info() returns line info")
>  	{
> @@ -74,7 +77,7 @@ TEST_CASE("Lines can be watched", "[info-event][chip]")
>  
>  		REQUIRE(info.direction() == direction::INPUT);
>  
> -		::std::thread thread(request_reconfigure_release_line, ::std::ref(chip));
> +		::std::thread thread(request_reconfigure_release_line, ::std::ref(chip_path));
>  
>  		REQUIRE(chip.wait_info_event(::std::chrono::seconds(1)));
>  		auto event = chip.read_info_event();

[libgpiod][PATCH 3/3] bindings: python: tests: don't use the same chip from different threads

[Bartosz Golaszewski]

From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

There are no thread-safety guarantees in libgpiod. Let's not reuse the
chip object created in one thread to generate info events in another but
use a global request function instead.

Reported-by: Erik Schilling <erik.schilling@linaro.org>
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
---
 bindings/python/tests/tests_info_event.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/bindings/python/tests/tests_info_event.py b/bindings/python/tests/tests_info_event.py
index f3926d9..6bb09d5 100644
--- a/bindings/python/tests/tests_info_event.py
+++ b/bindings/python/tests/tests_info_event.py
@@ -37,9 +37,9 @@ class InfoEventDataclassBehavior(TestCase):
                     event.line_info = 4
 
 
-def request_reconfigure_release_line(chip, offset):
+def request_reconfigure_release_line(chip_path, offset):
     time.sleep(0.1)
-    with chip.request_lines(config={offset: None}) as request:
+    with gpiod.request_lines(chip_path, config={offset: None}) as request:
         time.sleep(0.1)
         request.reconfigure_lines(
             config={offset: gpiod.LineSettings(direction=Direction.OUTPUT)}
@@ -95,7 +95,7 @@ class WatchingInfoEventWorks(TestCase):
         self.assertEqual(info.direction, Direction.INPUT)
 
         self.thread = threading.Thread(
-            target=partial(request_reconfigure_release_line, self.chip, 7)
+            target=partial(request_reconfigure_release_line, self.sim.dev_path, 7)
         )
         self.thread.start()
 
-- 
2.39.2

Re: [libgpiod][PATCH 3/3] bindings: python: tests: don't use the same chip from different threads

[Erik Schilling]

On Tue Sep 19, 2023 at 11:31 AM CEST, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>
> There are no thread-safety guarantees in libgpiod. Let's not reuse the
> chip object created in one thread to generate info events in another but
> use a global request function instead.
>
> Reported-by: Erik Schilling <erik.schilling@linaro.org>
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

Reviewed-by: Erik Schilling <erik.schilling@linaro.org>

> ---
>  bindings/python/tests/tests_info_event.py | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/bindings/python/tests/tests_info_event.py b/bindings/python/tests/tests_info_event.py
> index f3926d9..6bb09d5 100644
> --- a/bindings/python/tests/tests_info_event.py
> +++ b/bindings/python/tests/tests_info_event.py
> @@ -37,9 +37,9 @@ class InfoEventDataclassBehavior(TestCase):
>                      event.line_info = 4
>  
>  
> -def request_reconfigure_release_line(chip, offset):
> +def request_reconfigure_release_line(chip_path, offset):
>      time.sleep(0.1)
> -    with chip.request_lines(config={offset: None}) as request:
> +    with gpiod.request_lines(chip_path, config={offset: None}) as request:
>          time.sleep(0.1)
>          request.reconfigure_lines(
>              config={offset: gpiod.LineSettings(direction=Direction.OUTPUT)}
> @@ -95,7 +95,7 @@ class WatchingInfoEventWorks(TestCase):
>          self.assertEqual(info.direction, Direction.INPUT)
>  
>          self.thread = threading.Thread(
> -            target=partial(request_reconfigure_release_line, self.chip, 7)
> +            target=partial(request_reconfigure_release_line, self.sim.dev_path, 7)
>          )
>          self.thread.start()
>  

Re: [libgpiod][PATCH 1/3] tests: don't use the same chip object from different threads

[Erik Schilling]

On Tue Sep 19, 2023 at 11:31 AM CEST, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
>
> There are no thread-safety guarantees in libgpiod. Let's not reuse the
> chip object created in one thread to generate info events in another but
> create a second chip for that purpose instead.
>
> Reported-by: Erik Schilling <erik.schilling@linaro.org>
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

Reviewed-by: Erik Schilling <erik.schilling@linaro.org>

> ---
>  tests/tests-info-event.c | 62 +++++++++++++++++++++-------------------
>  1 file changed, 33 insertions(+), 29 deletions(-)
>
> diff --git a/tests/tests-info-event.c b/tests/tests-info-event.c
> index d069811..cbd9e9e 100644
> --- a/tests/tests-info-event.c
> +++ b/tests/tests-info-event.c
> @@ -49,43 +49,59 @@ GPIOD_TEST_CASE(event_timeout)
>  }
>  
>  struct request_ctx {
> -	struct gpiod_chip *chip;
> -	struct gpiod_line_config *line_cfg;
> -	struct gpiod_line_settings *settings;
> +	const char *path;
>  	guint offset;
>  };
>  
>  static gpointer request_reconfigure_release_line(gpointer data)
>  {
> +	g_autoptr(struct_gpiod_line_settings) settings = NULL;
> +	g_autoptr(struct_gpiod_line_config) line_cfg = NULL;
>  	g_autoptr(struct_gpiod_line_request) request = NULL;
> +	g_autoptr(struct_gpiod_chip) chip = NULL;
>  	struct request_ctx *ctx = data;
>  	gint ret;
>  
> +	chip = gpiod_chip_open(ctx->path);
> +	g_assert_nonnull(chip);
> +	if (g_test_failed())
> +		return NULL;
> +
> +	line_cfg = gpiod_line_config_new();
> +	g_assert_nonnull(line_cfg);
> +	if (g_test_failed())
> +		return NULL;
> +
> +	settings = gpiod_line_settings_new();
> +	g_assert_nonnull(settings);
> +	if (g_test_failed())
> +		return NULL;
> +
>  	g_usleep(1000);
>  
> -	ret = gpiod_line_config_add_line_settings(ctx->line_cfg, &ctx->offset,
> -						  1, ctx->settings);
> +	ret = gpiod_line_config_add_line_settings(line_cfg, &ctx->offset,
> +						  1, settings);
>  	g_assert_cmpint(ret, ==, 0);
>  	if (g_test_failed())
>  		return NULL;
>  
> -	request = gpiod_chip_request_lines(ctx->chip, NULL, ctx->line_cfg);
> +	request = gpiod_chip_request_lines(chip, NULL, line_cfg);
>  	g_assert_nonnull(request);
>  	if (g_test_failed())
>  		return NULL;
>  
>  	g_usleep(1000);
>  
> -	gpiod_line_config_reset(ctx->line_cfg);
> -	gpiod_line_settings_set_direction(ctx->settings,
> +	gpiod_line_config_reset(line_cfg);
> +	gpiod_line_settings_set_direction(settings,
>  					  GPIOD_LINE_DIRECTION_OUTPUT);
> -	ret = gpiod_line_config_add_line_settings(ctx->line_cfg, &ctx->offset,
> -						  1, ctx->settings);
> +	ret = gpiod_line_config_add_line_settings(line_cfg, &ctx->offset,
> +						  1, settings);
>  	g_assert_cmpint(ret, ==, 0);
>  	if (g_test_failed())
>  		return NULL;
>  
> -	ret = gpiod_line_request_reconfigure_lines(request, ctx->line_cfg);
> +	ret = gpiod_line_request_reconfigure_lines(request, line_cfg);
>  	g_assert_cmpint(ret, ==, 0);
>  	if (g_test_failed())
>  		return NULL;
> @@ -106,25 +122,19 @@ GPIOD_TEST_CASE(request_reconfigure_release_events)
>  	g_autoptr(struct_gpiod_info_event) request_event = NULL;
>  	g_autoptr(struct_gpiod_info_event) reconfigure_event = NULL;
>  	g_autoptr(struct_gpiod_info_event) release_event = NULL;
> -	g_autoptr(struct_gpiod_line_config) line_cfg = NULL;
> -	g_autoptr(struct_gpiod_line_settings) settings = NULL;
>  	g_autoptr(GThread) thread = NULL;
>  	struct gpiod_line_info *request_info, *reconfigure_info, *release_info;
>  	guint64 request_ts, reconfigure_ts, release_ts;
>  	struct request_ctx ctx;
> +	const char *chip_path = g_gpiosim_chip_get_dev_path(sim);
>  	gint ret;
>  
> -	chip = gpiod_test_open_chip_or_fail(g_gpiosim_chip_get_dev_path(sim));
> -	line_cfg = gpiod_test_create_line_config_or_fail();
> -	settings = gpiod_test_create_line_settings_or_fail();
> -
> +	chip = gpiod_test_open_chip_or_fail(chip_path);
>  	info = gpiod_test_chip_watch_line_info_or_fail(chip, 3);
>  
>  	g_assert_false(gpiod_line_info_is_used(info));
>  
> -	ctx.chip = chip;
> -	ctx.line_cfg = line_cfg;
> -	ctx.settings = settings;
> +	ctx.path = chip_path;
>  	ctx.offset = 3;
>  
>  	thread = g_thread_new("request-release",
> @@ -199,26 +209,20 @@ GPIOD_TEST_CASE(chip_fd_can_be_polled)
>  	g_autoptr(struct_gpiod_chip) chip = NULL;
>  	g_autoptr(struct_gpiod_line_info) info = NULL;
>  	g_autoptr(struct_gpiod_info_event) event = NULL;
> -	g_autoptr(struct_gpiod_line_config) line_cfg = NULL;
> -	g_autoptr(struct_gpiod_line_settings) settings = NULL;
>  	g_autoptr(GThread) thread = NULL;
> +	const char *chip_path = g_gpiosim_chip_get_dev_path(sim);
>  	struct gpiod_line_info *evinfo;
>  	struct request_ctx ctx;
>  	struct timespec ts;
>  	struct pollfd pfd;
>  	gint ret, fd;
>  
> -	chip = gpiod_test_open_chip_or_fail(g_gpiosim_chip_get_dev_path(sim));
> -	settings = gpiod_test_create_line_settings_or_fail();
> -	line_cfg = gpiod_test_create_line_config_or_fail();
> -
> +	chip = gpiod_test_open_chip_or_fail(chip_path);
>  	info = gpiod_test_chip_watch_line_info_or_fail(chip, 3);
>  
>  	g_assert_false(gpiod_line_info_is_used(info));
>  
> -	ctx.chip = chip;
> -	ctx.line_cfg = line_cfg;
> -	ctx.settings = settings;
> +	ctx.path = chip_path;
>  	ctx.offset = 3;
>  
>  	thread = g_thread_new("request-release",