Skip to contentSkip to conversation

bpf: Add ability to pin bpf timer to calling CPU

PATCH Applied David Vernet 3 participants 6 messages Oct 4, 2023 – Oct 9, 2023

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

v2 cover bpf: Add ability to pin bpf timer to calling CPU

BPF supports creating high resolution timers using bpf_timer_* helper functions. Currently, only the BPF_F_TIMER_ABS flag is supported, which specifies that the timeout should be interpreted as absolute time. It would also be useful to be able to pin that timer to a core. For example, if you wanted to make a subset of cores run without timer interrupts, and only have the timer be invoked on a single core.

This patch set adds support for this with a new BPF_F_TIMER_CPU_PIN flag. When specified, the HRTIMER_MODE_PINNED flag is passed to hrtimer_start().

This patch set is based off of commit 93fb2776f43e ("Merge branch 'bpf-xsk-sh-umem'").

---

v1: https://lore.kernel.org/bpf/[email protected]/

v1 -> v2 changes:
- Put declaration of soft_timer_pinned and abs_timer_pinned selftest maps on same line as abs_timer (Song)

David Vernet (2):
bpf: Add ability to pin bpf timer to calling CPU bpf/selftests: Test pinning bpf timer to a core

 include/uapi/linux/bpf.h                      |  4 ++
 kernel/bpf/helpers.c                          |  5 +-
 tools/include/uapi/linux/bpf.h                |  4 ++
 .../testing/selftests/bpf/prog_tests/timer.c  |  4 ++
 tools/testing/selftests/bpf/progs/timer.c     | 63 ++++++++++++++++++-
 5 files changed, 78 insertions(+), 2 deletions(-)
Signature · 1 line
2.41.0
#1 lore raw

PATCH v2 1/2 bpf: Add ability to pin bpf timer to calling CPU

BPF supports creating high resolution timers using bpf_timer_* helper functions. Currently, only the BPF_F_TIMER_ABS flag is supported, which specifies that the timeout should be interpreted as absolute time. It would also be useful to be able to pin that timer to a core. For example, if you wanted to make a subset of cores run without timer interrupts, and only have the timer be invoked on a single core.

This patch adds support for this with a new BPF_F_TIMER_CPU_PIN flag. When specified, the HRTIMER_MODE_PINNED flag is passed to hrtimer_start(). A subsequent patch will update selftests to validate.

---
 include/uapi/linux/bpf.h       | 4 ++++
 kernel/bpf/helpers.c           | 5 ++++-
 tools/include/uapi/linux/bpf.h | 4 ++++
 3 files changed, 12 insertions(+), 1 deletion(-)
include/uapi/linux/bpf.h +4 -0
index 70bfa997e896..a7d4a1a69f21 100644
@@ -5096,6 +5096,8 @@ union bpf_attr {
  *		**BPF_F_TIMER_ABS**
  *			Start the timer in absolute expire value instead of the
  *			default relative one.
+ *		**BPF_F_TIMER_CPU_PIN**
+ *			Timer will be pinned to the CPU of the caller.
  *
  *	Return
  *		0 on success.
@@ -7309,9 +7311,11 @@ struct bpf_core_relo {
  * Flags to control bpf_timer_start() behaviour.
  *     - BPF_F_TIMER_ABS: Timeout passed is absolute time, by default it is
  *       relative to current time.
+ *     - BPF_F_TIMER_CPU_PIN: Timer will be pinned to the CPU of the caller.
  */
 enum {
 	BPF_F_TIMER_ABS = (1ULL << 0),
+	BPF_F_TIMER_CPU_PIN = (1ULL << 1),
 };
 
 /* BPF numbers iterator state */
kernel/bpf/helpers.c +4 -1
index dd1c69ee3375..d2840dd5b00d 100644
@@ -1272,7 +1272,7 @@ BPF_CALL_3(bpf_timer_start, struct bpf_timer_kern *, timer, u64, nsecs, u64, fla
 
 	if (in_nmi())
 		return -EOPNOTSUPP;
-	if (flags > BPF_F_TIMER_ABS)
+	if (flags & ~(BPF_F_TIMER_ABS | BPF_F_TIMER_CPU_PIN))
 		return -EINVAL;
 	__bpf_spin_lock_irqsave(&timer->lock);
 	t = timer->timer;
@@ -1286,6 +1286,9 @@ BPF_CALL_3(bpf_timer_start, struct bpf_timer_kern *, timer, u64, nsecs, u64, fla
 	else
 		mode = HRTIMER_MODE_REL_SOFT;
 
+	if (flags & BPF_F_TIMER_CPU_PIN)
+		mode |= HRTIMER_MODE_PINNED;
+
 	hrtimer_start(&t->timer, ns_to_ktime(nsecs), mode);
 out:
 	__bpf_spin_unlock_irqrestore(&timer->lock);
tools/include/uapi/linux/bpf.h +4 -0
index 70bfa997e896..a7d4a1a69f21 100644
@@ -5096,6 +5096,8 @@ union bpf_attr {
  *		**BPF_F_TIMER_ABS**
  *			Start the timer in absolute expire value instead of the
  *			default relative one.
+ *		**BPF_F_TIMER_CPU_PIN**
+ *			Timer will be pinned to the CPU of the caller.
  *
  *	Return
  *		0 on success.
@@ -7309,9 +7311,11 @@ struct bpf_core_relo {
  * Flags to control bpf_timer_start() behaviour.
  *     - BPF_F_TIMER_ABS: Timeout passed is absolute time, by default it is
  *       relative to current time.
+ *     - BPF_F_TIMER_CPU_PIN: Timer will be pinned to the CPU of the caller.
  */
 enum {
 	BPF_F_TIMER_ABS = (1ULL << 0),
+	BPF_F_TIMER_CPU_PIN = (1ULL << 1),
 };
 
 /* BPF numbers iterator state */
Signature · 1 line
2.41.0
Acked-by on v2 1/2
#2 lore raw
Quoted 7 lines · David Vernet

BPF supports creating high resolution timers using bpf_timer_* helper functions. Currently, only the BPF_F_TIMER_ABS flag is supported, which specifies that the timeout should be interpreted as absolute time. It would also be useful to be able to pin that timer to a core. For example, if you wanted to make a subset of cores run without timer interrupts, and only have the timer be invoked on a single core.

Quoted 6 lines

This patch adds support for this with a new BPF_F_TIMER_CPU_PIN flag. When specified, the HRTIMER_MODE_PINNED flag is passed to hrtimer_start(). A subsequent patch will update selftests to validate.

#3 lore raw

PATCH v2 2/2 bpf/selftests: Test pinning bpf timer to a core

In reply to David Vernet

Now that we support pinning a BPF timer to the current core, we should test it with some selftests. This patch adds two new testcases to the timer suite, which verifies that a BPF timer both with and without BPF_F_TIMER_ABS, can be pinned to the calling core with BPF_F_TIMER_CPU_PIN.

---
 .../testing/selftests/bpf/prog_tests/timer.c  |  4 ++
 tools/testing/selftests/bpf/progs/timer.c     | 63 ++++++++++++++++++-
 2 files changed, 66 insertions(+), 1 deletion(-)
tools/testing/selftests/bpf/prog_tests/timer.c +4 -0
index 290c21dbe65a..d8bc838445ec 100644
@@ -14,6 +14,7 @@ static int timer(struct timer *timer_skel)
 
 	ASSERT_EQ(timer_skel->data->callback_check, 52, "callback_check1");
 	ASSERT_EQ(timer_skel->data->callback2_check, 52, "callback2_check1");
+	ASSERT_EQ(timer_skel->bss->pinned_callback_check, 0, "pinned_callback_check1");
 
 	prog_fd = bpf_program__fd(timer_skel->progs.test1);
 	err = bpf_prog_test_run_opts(prog_fd, &topts);
@@ -32,6 +33,9 @@ static int timer(struct timer *timer_skel)
 	/* check that timer_cb3() was executed twice */
 	ASSERT_EQ(timer_skel->bss->abs_data, 12, "abs_data");
 
+	/* check that timer_cb_pinned() was executed twice */
+	ASSERT_EQ(timer_skel->bss->pinned_callback_check, 2, "pinned_callback_check");
+
 	/* check that there were no errors in timer execution */
 	ASSERT_EQ(timer_skel->bss->err, 0, "err");
 
tools/testing/selftests/bpf/progs/timer.c +62 -1

78 lines not on this page · raw message

Signature · 1 line
2.41.0
Acked-by on v2 2/2
#4 lore raw

Re: [PATCH bpf-next v2 2/2] bpf/selftests: Test pinning bpf timer to a core

Quoted 8 lines · David Vernet

Now that we support pinning a BPF timer to the current core, we should test it with some selftests. This patch adds two new testcases to the timer suite, which verifies that a BPF timer both with and without BPF_F_TIMER_ABS, can be pinned to the calling core with BPF_F_TIMER_CPU_PIN.

#5 lore raw
In reply to David Vernet
This series was applied to bpf/bpf-next.git (master) by Daniel Borkmann <[email protected]>:

Hello:

This series was applied to bpf/bpf-next.git (master) by Daniel Borkmann <[email protected]>:

Quoted 8 lines · On Wed, 4 Oct 2023 11:23:37 -0500 you wrote:

BPF supports creating high resolution timers using bpf_timer_* helper functions. Currently, only the BPF_F_TIMER_ABS flag is supported, which specifies that the timeout should be interpreted as absolute time. It would also be useful to be able to pin that timer to a core. For example, if you wanted to make a subset of cores run without timer interrupts, and only have the timer be invoked on a single core.

[...]

Here is the summary with links:
- [bpf-next,v2,1/2] bpf: Add ability to pin bpf timer to calling CPU

    https://git.kernel.org/bpf/bpf-next/c/d6247ecb6c1e
  - [bpf-next,v2,2/2] bpf/selftests: Test pinning bpf timer to a core
    https://git.kernel.org/bpf/bpf-next/c/0d7ae0686075

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html