All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH net-next v2] selftest: epoll_busy_poll: epoll busy poll tests
@ 2024-05-06 20:53 Joe Damato
  2024-05-07  1:12 ` Jakub Kicinski
  0 siblings, 1 reply; 4+ messages in thread
From: Joe Damato @ 2024-05-06 20:53 UTC (permalink / raw)
  To: linux-kselftest, netdev, linux-kernel
  Cc: nalramli, Joe Damato, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Shuah Khan

Add a simple test for the epoll busy poll ioctls, using the kernel
selftest harness.

This test ensures that the ioctls have the expected return codes and
that the kernel properly gets and sets epoll busy poll parameters.

The test can be expanded in the future to do real busy polling (provided
another machine to act as the client is available).

Signed-off-by: Joe Damato <jdamato@fastly.com>
---
 tools/testing/selftests/net/.gitignore        |   1 +
 tools/testing/selftests/net/Makefile          |   2 +-
 tools/testing/selftests/net/epoll_busy_poll.c | 271 ++++++++++++++++++
 3 files changed, 273 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/net/epoll_busy_poll.c

diff --git a/tools/testing/selftests/net/.gitignore b/tools/testing/selftests/net/.gitignore
index d996a0ab0765..777cfd027076 100644
--- a/tools/testing/selftests/net/.gitignore
+++ b/tools/testing/selftests/net/.gitignore
@@ -5,6 +5,7 @@ bind_wildcard
 csum
 cmsg_sender
 diag_uid
+epoll_busy_poll
 fin_ack_lat
 gro
 hwtstamp_config
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile
index 5befca249452..b0b893009867 100644
--- a/tools/testing/selftests/net/Makefile
+++ b/tools/testing/selftests/net/Makefile
@@ -67,7 +67,7 @@ TEST_GEN_FILES += ipsec
 TEST_GEN_FILES += ioam6_parser
 TEST_GEN_FILES += gro
 TEST_GEN_PROGS = reuseport_bpf reuseport_bpf_cpu reuseport_bpf_numa
-TEST_GEN_PROGS += reuseport_dualstack reuseaddr_conflict tls tun tap
+TEST_GEN_PROGS += reuseport_dualstack reuseaddr_conflict tls tun tap epoll_busy_poll
 TEST_GEN_FILES += toeplitz
 TEST_GEN_FILES += cmsg_sender
 TEST_GEN_FILES += stress_reuseport_listen
diff --git a/tools/testing/selftests/net/epoll_busy_poll.c b/tools/testing/selftests/net/epoll_busy_poll.c
new file mode 100644
index 000000000000..166fabc6cc7e
--- /dev/null
+++ b/tools/testing/selftests/net/epoll_busy_poll.c
@@ -0,0 +1,271 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+/* Basic per-epoll context busy poll test.
+ *
+ * Only tests the ioctls, but should be expanded to test two connected hosts in
+ * the future
+ */
+
+#define _GNU_SOURCE
+
+#include <error.h>
+#include <errno.h>
+#include <inttypes.h>
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#include <sys/epoll.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+
+#include "../kselftest_harness.h"
+
+/* if the headers haven't been updated, we need to define some things */
+#if !defined(EPOLL_IOC_TYPE)
+struct epoll_params {
+	uint32_t busy_poll_usecs;
+	uint16_t busy_poll_budget;
+	uint8_t prefer_busy_poll;
+
+	/* pad the struct to a multiple of 64bits */
+	uint8_t __pad;
+};
+
+#define EPOLL_IOC_TYPE 0x8A
+#define EPIOCSPARAMS _IOW(EPOLL_IOC_TYPE, 0x01, struct epoll_params)
+#define EPIOCGPARAMS _IOR(EPOLL_IOC_TYPE, 0x02, struct epoll_params)
+#endif
+
+FIXTURE(invalid_fd)
+{
+	int invalid_fd;
+	struct epoll_params params;
+};
+
+FIXTURE_SETUP(invalid_fd)
+{
+	int ret;
+
+	ret = socket(AF_UNIX, SOCK_DGRAM, 0);
+	EXPECT_NE(-1, ret)
+		TH_LOG("error creating unix socket");
+
+	self->invalid_fd = ret;
+}
+
+FIXTURE_TEARDOWN(invalid_fd)
+{
+	int ret;
+
+	ret = close(self->invalid_fd);
+	EXPECT_EQ(0, ret);
+}
+
+TEST_F(invalid_fd, test_invalid_fd)
+{
+	int ret;
+
+	ret = ioctl(self->invalid_fd, EPIOCGPARAMS, &self->params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("EPIOCGPARAMS on invalid epoll FD should error");
+
+	EXPECT_EQ(ENOTTY, errno)
+		TH_LOG("EPIOCGPARAMS on invalid epoll FD should set errno to ENOTTY");
+
+	memset(&self->params, 0, sizeof(struct epoll_params));
+
+	ret = ioctl(self->invalid_fd, EPIOCSPARAMS, &self->params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("EPIOCSPARAMS on invalid epoll FD should error");
+
+	EXPECT_EQ(ENOTTY, errno)
+		TH_LOG("EPIOCSPARAMS on invalid epoll FD should set errno to ENOTTY");
+}
+
+FIXTURE(epoll_busy_poll)
+{
+	int fd;
+	struct epoll_params params;
+	struct epoll_params *invalid_params;
+};
+
+FIXTURE_SETUP(epoll_busy_poll)
+{
+	int ret;
+
+	ret = epoll_create1(0);
+	EXPECT_NE(-1, ret)
+		TH_LOG("epoll_create1 failed?");
+
+	self->fd = ret;
+}
+
+FIXTURE_TEARDOWN(epoll_busy_poll)
+{
+	int ret;
+
+	ret = close(self->fd);
+	EXPECT_EQ(0, ret);
+}
+
+TEST_F(epoll_busy_poll, test_get_params)
+{
+	/* begin by getting the epoll params from the kernel
+	 *
+	 * the default should be default and all fields should be zero'd by the
+	 * kernel, so set params fields to garbage to test this.
+	 */
+	int ret = 0;
+
+	self->params.busy_poll_usecs = 0xff;
+	self->params.busy_poll_budget = 0xff;
+	self->params.prefer_busy_poll = 1;
+	self->params.__pad = 0xf;
+
+	ret = ioctl(self->fd, EPIOCGPARAMS, &self->params);
+	EXPECT_EQ(0, ret)
+		TH_LOG("ioctl EPIOCGPARAMS should succeed");
+
+	EXPECT_EQ(0, self->params.busy_poll_usecs)
+		TH_LOG("EPIOCGPARAMS busy_poll_usecs should have been 0");
+
+	EXPECT_EQ(0, self->params.busy_poll_budget)
+		TH_LOG("EPIOCGPARAMS busy_poll_budget should have been 0");
+
+	EXPECT_EQ(0, self->params.prefer_busy_poll)
+		TH_LOG("EPIOCGPARAMS prefer_busy_poll should have been 0");
+
+	EXPECT_EQ(0, self->params.__pad)
+		TH_LOG("EPIOCGPARAMS __pad should have been 0");
+
+	self->invalid_params = (struct epoll_params *)0xdeadbeef;
+	ret = ioctl(self->fd, EPIOCGPARAMS, self->invalid_params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("EPIOCGPARAMS should error with invalid params");
+
+	EXPECT_EQ(EFAULT, errno)
+		TH_LOG("EPIOCGPARAMS with invalid params should set errno to EFAULT");
+}
+
+TEST_F(epoll_busy_poll, test_set_invalid)
+{
+	int ret;
+
+	memset(&self->params, 0, sizeof(struct epoll_params));
+
+	self->params.__pad = 1;
+
+	ret = ioctl(self->fd, EPIOCSPARAMS, &self->params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("EPIOCSPARAMS non-zero __pad should error");
+
+	EXPECT_EQ(EINVAL, errno)
+		TH_LOG("EPIOCSPARAMS non-zero __pad errno should be EINVAL");
+
+	self->params.__pad = 0;
+	self->params.busy_poll_usecs = (unsigned int)INT_MAX + 1;
+
+	ret = ioctl(self->fd, EPIOCSPARAMS, &self->params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("EPIOCSPARAMS should error busy_poll_usecs > S32_MAX");
+
+	EXPECT_EQ(EINVAL, errno)
+		TH_LOG("EPIOCSPARAMS busy_poll_usecs > S32_MAX errno should be EINVAL");
+
+	self->params.__pad = 0;
+	self->params.busy_poll_usecs = 32;
+	self->params.prefer_busy_poll = 2;
+
+	ret = ioctl(self->fd, EPIOCSPARAMS, &self->params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("EPIOCSPARAMS should error prefer_busy_poll > 1");
+
+	EXPECT_EQ(EINVAL, errno)
+		TH_LOG("EPIOCSPARAMS prefer_busy_poll > 1 errno should be EINVAL");
+
+	self->params.__pad = 0;
+	self->params.busy_poll_usecs = 32;
+	self->params.prefer_busy_poll = 1;
+
+	/* set budget well above kernel's NAPI_POLL_WEIGHT of 64 */
+	self->params.busy_poll_budget = 65535;
+
+	ret = ioctl(self->fd, EPIOCSPARAMS, &self->params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("EPIOCSPARAMS should error busy_poll_budget > NAPI_POLL_WEIGHT");
+
+	EXPECT_EQ(EPERM, errno)
+		TH_LOG("EPIOCSPARAMS errno should be EPERM busy_poll_budget > NAPI_POLL_WEIGHT");
+
+	self->invalid_params = (struct epoll_params *)0xdeadbeef;
+	ret = ioctl(self->fd, EPIOCSPARAMS, self->invalid_params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("EPIOCSPARAMS should error when epoll_params is invalid");
+
+	EXPECT_EQ(EFAULT, errno)
+		TH_LOG("EPIOCSPARAMS should set errno to EFAULT when epoll_params is invalid");
+}
+
+TEST_F(epoll_busy_poll, test_set_and_get_valid)
+{
+	int ret;
+
+	memset(&self->params, 0, sizeof(struct epoll_params));
+
+	self->params.busy_poll_usecs = 25;
+	self->params.busy_poll_budget = 16;
+	self->params.prefer_busy_poll = 1;
+
+	ret = ioctl(self->fd, EPIOCSPARAMS, &self->params);
+
+	EXPECT_EQ(0, ret)
+		TH_LOG("EPIOCSPARAMS with valid params should not error");
+
+	/* check that the kernel returns the same values back */
+
+	memset(&self->params, 0, sizeof(struct epoll_params));
+
+	ret = ioctl(self->fd, EPIOCGPARAMS, &self->params);
+
+	EXPECT_EQ(0, ret)
+		TH_LOG("EPIOCGPARAMS should not error");
+
+	EXPECT_EQ(25, self->params.busy_poll_usecs)
+		TH_LOG("params.busy_poll_usecs incorrect");
+
+	EXPECT_EQ(16, self->params.busy_poll_budget)
+		TH_LOG("params.busy_poll_budget incorrect");
+
+	EXPECT_EQ(1, self->params.prefer_busy_poll)
+		TH_LOG("params.prefer_busy_poll incorrect");
+
+	EXPECT_EQ(0, self->params.__pad)
+		TH_LOG("params.__pad was not 0");
+}
+
+TEST_F(epoll_busy_poll, test_invalid_ioctl)
+{
+	int invalid_ioctl = EPIOCGPARAMS + 10;
+	int ret;
+
+	ret = ioctl(self->fd, invalid_ioctl, &self->params);
+
+	EXPECT_EQ(-1, ret)
+		TH_LOG("invalid ioctl should return error");
+
+	EXPECT_EQ(EINVAL, errno)
+		TH_LOG("invalid ioctl should set errno to EINVAL");
+}
+
+TEST_HARNESS_MAIN
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH net-next v2] selftest: epoll_busy_poll: epoll busy poll tests
  2024-05-06 20:53 [PATCH net-next v2] selftest: epoll_busy_poll: epoll busy poll tests Joe Damato
@ 2024-05-07  1:12 ` Jakub Kicinski
  2024-05-07  1:40   ` Joe Damato
  0 siblings, 1 reply; 4+ messages in thread
From: Jakub Kicinski @ 2024-05-07  1:12 UTC (permalink / raw)
  To: Joe Damato
  Cc: linux-kselftest, netdev, linux-kernel, nalramli, David S. Miller,
	Eric Dumazet, Paolo Abeni, Shuah Khan

On Mon,  6 May 2024 20:53:22 +0000 Joe Damato wrote:
> Add a simple test for the epoll busy poll ioctls, using the kernel
> selftest harness.
> 
> This test ensures that the ioctls have the expected return codes and
> that the kernel properly gets and sets epoll busy poll parameters.
> 
> The test can be expanded in the future to do real busy polling (provided
> another machine to act as the client is available).

Hm, we get:

# timeout set to 3600
# selftests: net: epoll_busy_poll
# TAP version 13
# 1..5
# # Starting 5 tests from 2 test cases.
# #  RUN           invalid_fd.test_invalid_fd ...
# #            OK  invalid_fd.test_invalid_fd
# ok 1 invalid_fd.test_invalid_fd
# #  RUN           epoll_busy_poll.test_get_params ...
# #            OK  epoll_busy_poll.test_get_params
# ok 2 epoll_busy_poll.test_get_params
# #  RUN           epoll_busy_poll.test_set_invalid ...
# # epoll_busy_poll.c:204:test_set_invalid:Expected -1 (-1) == ret (0)
# # epoll_busy_poll.c:205:test_set_invalid:EPIOCSPARAMS should error busy_poll_budget > NAPI_POLL_WEIGHT
# # epoll_busy_poll.c:207:test_set_invalid:Expected EPERM (1) == errno (22)
# # epoll_busy_poll.c:208:test_set_invalid:EPIOCSPARAMS errno should be EPERM busy_poll_budget > NAPI_POLL_WEIGHT
# # test_set_invalid: Test failed
# #          FAIL  epoll_busy_poll.test_set_invalid
# not ok 3 epoll_busy_poll.test_set_invalid
# #  RUN           epoll_busy_poll.test_set_and_get_valid ...
# #            OK  epoll_busy_poll.test_set_and_get_valid
# ok 4 epoll_busy_poll.test_set_and_get_valid
# #  RUN           epoll_busy_poll.test_invalid_ioctl ...
# #            OK  epoll_busy_poll.test_invalid_ioctl
# ok 5 epoll_busy_poll.test_invalid_ioctl

https://netdev-3.bots.linux.dev/vmksft-net/results/584001/98-epoll-busy-poll/stdout

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH net-next v2] selftest: epoll_busy_poll: epoll busy poll tests
  2024-05-07  1:12 ` Jakub Kicinski
@ 2024-05-07  1:40   ` Joe Damato
  2024-05-07  2:02     ` Jakub Kicinski
  0 siblings, 1 reply; 4+ messages in thread
From: Joe Damato @ 2024-05-07  1:40 UTC (permalink / raw)
  To: Jakub Kicinski
  Cc: linux-kselftest, netdev, linux-kernel, nalramli, David S. Miller,
	Eric Dumazet, Paolo Abeni, Shuah Khan

On Mon, May 06, 2024 at 06:12:54PM -0700, Jakub Kicinski wrote:
> On Mon,  6 May 2024 20:53:22 +0000 Joe Damato wrote:
> > Add a simple test for the epoll busy poll ioctls, using the kernel
> > selftest harness.
> > 
> > This test ensures that the ioctls have the expected return codes and
> > that the kernel properly gets and sets epoll busy poll parameters.
> > 
> > The test can be expanded in the future to do real busy polling (provided
> > another machine to act as the client is available).
> 
> Hm, we get:
> 
> # timeout set to 3600
> # selftests: net: epoll_busy_poll
> # TAP version 13
> # 1..5
> # # Starting 5 tests from 2 test cases.
> # #  RUN           invalid_fd.test_invalid_fd ...
> # #            OK  invalid_fd.test_invalid_fd
> # ok 1 invalid_fd.test_invalid_fd
> # #  RUN           epoll_busy_poll.test_get_params ...
> # #            OK  epoll_busy_poll.test_get_params
> # ok 2 epoll_busy_poll.test_get_params
> # #  RUN           epoll_busy_poll.test_set_invalid ...
> # # epoll_busy_poll.c:204:test_set_invalid:Expected -1 (-1) == ret (0)
> # # epoll_busy_poll.c:205:test_set_invalid:EPIOCSPARAMS should error busy_poll_budget > NAPI_POLL_WEIGHT
> # # epoll_busy_poll.c:207:test_set_invalid:Expected EPERM (1) == errno (22)
> # # epoll_busy_poll.c:208:test_set_invalid:EPIOCSPARAMS errno should be EPERM busy_poll_budget > NAPI_POLL_WEIGHT
> # # test_set_invalid: Test failed
> # #          FAIL  epoll_busy_poll.test_set_invalid
> # not ok 3 epoll_busy_poll.test_set_invalid
> # #  RUN           epoll_busy_poll.test_set_and_get_valid ...
> # #            OK  epoll_busy_poll.test_set_and_get_valid
> # ok 4 epoll_busy_poll.test_set_and_get_valid
> # #  RUN           epoll_busy_poll.test_invalid_ioctl ...
> # #            OK  epoll_busy_poll.test_invalid_ioctl
> # ok 5 epoll_busy_poll.test_invalid_ioctl
> 
> https://netdev-3.bots.linux.dev/vmksft-net/results/584001/98-epoll-busy-poll/stdout

Ah, sorry -- this is because I had assumed the test would run without
CAP_NET_ADMIN, but since:

  epoll_busy_poll.c:204:test_set_invalid:Expected -1 (-1) == ret (0)

succeeds (ret = 0), clearly I am mistaken. Sorry about that.

I think I'll spin up a v3 and I'll add a test with and without
CAP_NET_ADMIN to check both cases, which would probably be better anyway.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH net-next v2] selftest: epoll_busy_poll: epoll busy poll tests
  2024-05-07  1:40   ` Joe Damato
@ 2024-05-07  2:02     ` Jakub Kicinski
  0 siblings, 0 replies; 4+ messages in thread
From: Jakub Kicinski @ 2024-05-07  2:02 UTC (permalink / raw)
  To: Joe Damato
  Cc: linux-kselftest, netdev, linux-kernel, nalramli, David S. Miller,
	Eric Dumazet, Paolo Abeni, Shuah Khan

On Mon, 6 May 2024 18:40:00 -0700 Joe Damato wrote:
> Ah, sorry -- this is because I had assumed the test would run without
> CAP_NET_ADMIN, but since:
> 
>   epoll_busy_poll.c:204:test_set_invalid:Expected -1 (-1) == ret (0)
> 
> succeeds (ret = 0), clearly I am mistaken. Sorry about that.
> 
> I think I'll spin up a v3 and I'll add a test with and without
> CAP_NET_ADMIN to check both cases, which would probably be better anyway.

FWIW the tests run a in separate process from the harness, so it may 
be possible to drop privileges inside the test, without affecting other
test cases. But I've never done it myself, so not sure how easy it is
to do in practice..

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2024-05-07  2:02 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-05-06 20:53 [PATCH net-next v2] selftest: epoll_busy_poll: epoll busy poll tests Joe Damato
2024-05-07  1:12 ` Jakub Kicinski
2024-05-07  1:40   ` Joe Damato
2024-05-07  2:02     ` Jakub Kicinski

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.