public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Kumar Kartikeya Dwivedi <memxor@gmail.com>
To: bpf@vger.kernel.org, linux-kernel@vger.kernel.org
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Waiman Long <llong@redhat.com>,
	Alexei Starovoitov <ast@kernel.org>,
	Andrii Nakryiko <andrii@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Martin KaFai Lau <martin.lau@kernel.org>,
	Eduard Zingerman <eddyz87@gmail.com>,
	"Paul E. McKenney" <paulmck@kernel.org>,
	Tejun Heo <tj@kernel.org>, Barret Rhoden <brho@google.com>,
	Josh Don <joshdon@google.com>, Dohyun Kim <dohyunkim@google.com>,
	kernel-team@meta.com
Subject: [PATCH bpf-next v1 12/22] rqspinlock: Add basic support for CONFIG_PARAVIRT
Date: Tue,  7 Jan 2025 05:59:54 -0800	[thread overview]
Message-ID: <20250107140004.2732830-13-memxor@gmail.com> (raw)
In-Reply-To: <20250107140004.2732830-1-memxor@gmail.com>

We ripped out PV and virtualization related bits from rqspinlock in an
earlier commit, however, a fair lock performs poorly within a virtual
machine when the lock holder is preempted. As such, retain the
virt_spin_lock fallback to test and set lock, but with timeout and
deadlock detection.

We don't integrate support for CONFIG_PARAVIRT_SPINLOCKS yet, as that
requires more involved algorithmic changes and introduces more
complexity. It can be done when the need arises in the future.

Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
---
 arch/x86/include/asm/rqspinlock.h | 20 ++++++++++++++++
 include/asm-generic/rqspinlock.h  |  7 ++++++
 kernel/locking/rqspinlock.c       | 38 +++++++++++++++++++++++++++++++
 3 files changed, 65 insertions(+)
 create mode 100644 arch/x86/include/asm/rqspinlock.h

diff --git a/arch/x86/include/asm/rqspinlock.h b/arch/x86/include/asm/rqspinlock.h
new file mode 100644
index 000000000000..ecfb7dfe6370
--- /dev/null
+++ b/arch/x86/include/asm/rqspinlock.h
@@ -0,0 +1,20 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_X86_RQSPINLOCK_H
+#define _ASM_X86_RQSPINLOCK_H
+
+#include <asm/paravirt.h>
+
+#ifdef CONFIG_PARAVIRT
+DECLARE_STATIC_KEY_FALSE(virt_spin_lock_key);
+
+#define resilient_virt_spin_lock_enabled resilient_virt_spin_lock_enabled
+static __always_inline bool resilient_virt_spin_lock_enabled(void)
+{
+       return static_branch_likely(&virt_spin_lock_key);
+}
+
+#endif /* CONFIG_PARAVIRT */
+
+#include <asm-generic/rqspinlock.h>
+
+#endif /* _ASM_X86_RQSPINLOCK_H */
diff --git a/include/asm-generic/rqspinlock.h b/include/asm-generic/rqspinlock.h
index c7e33ccc57a6..dc436ab01471 100644
--- a/include/asm-generic/rqspinlock.h
+++ b/include/asm-generic/rqspinlock.h
@@ -17,6 +17,13 @@ struct qspinlock;
 
 extern int resilient_queued_spin_lock_slowpath(struct qspinlock *lock, u32 val, u64 timeout);
 
+#ifndef resilient_virt_spin_lock_enabled
+static __always_inline bool resilient_virt_spin_lock_enabled(void)
+{
+	return false;
+}
+#endif
+
 /*
  * Default timeout for waiting loops is 0.5 seconds
  */
diff --git a/kernel/locking/rqspinlock.c b/kernel/locking/rqspinlock.c
index b7c86127d288..e397f91ebcf6 100644
--- a/kernel/locking/rqspinlock.c
+++ b/kernel/locking/rqspinlock.c
@@ -247,6 +247,41 @@ static noinline int check_timeout(struct qspinlock *lock, u32 mask,
  */
 #define RES_RESET_TIMEOUT(ts) ({ (ts).timeout_end = 0; })
 
+#ifdef CONFIG_PARAVIRT
+
+static inline int resilient_virt_spin_lock(struct qspinlock *lock, struct rqspinlock_timeout *ts)
+{
+	int val, ret = 0;
+
+	RES_RESET_TIMEOUT(*ts);
+	grab_held_lock_entry(lock);
+retry:
+	val = atomic_read(&lock->val);
+
+	if (val || !atomic_try_cmpxchg(&lock->val, &val, _Q_LOCKED_VAL)) {
+		if (RES_CHECK_TIMEOUT(*ts, ret, ~0u)) {
+			lockevent_inc(rqspinlock_lock_timeout);
+			goto timeout;
+		}
+		cpu_relax();
+		goto retry;
+	}
+
+	return 0;
+timeout:
+	release_held_lock_entry();
+	return ret;
+}
+
+#else
+
+static __always_inline int resilient_virt_spin_lock(struct qspinlock *lock, struct rqspinlock_timeout *ts)
+{
+	return 0;
+}
+
+#endif /* CONFIG_PARAVIRT */
+
 /*
  * Per-CPU queue node structures; we can never have more than 4 nested
  * contexts: task, softirq, hardirq, nmi.
@@ -287,6 +322,9 @@ int __lockfunc resilient_queued_spin_lock_slowpath(struct qspinlock *lock, u32 v
 
 	RES_INIT_TIMEOUT(ts, timeout);
 
+	if (resilient_virt_spin_lock_enabled())
+		return resilient_virt_spin_lock(lock, &ts);
+
 	/*
 	 * Wait for in-progress pending->locked hand-overs with a bounded
 	 * number of spins so that we guarantee forward progress.
-- 
2.43.5


  parent reply	other threads:[~2025-01-07 14:00 UTC|newest]

Thread overview: 63+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-01-07 13:59 [PATCH bpf-next v1 00/22] Resilient Queued Spin Lock Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 01/22] locking: Move MCS struct definition to public header Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 02/22] locking: Move common qspinlock helpers to a private header Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 03/22] locking: Allow obtaining result of arch_mcs_spin_lock_contended Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 04/22] locking: Copy out qspinlock.c to rqspinlock.c Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 05/22] rqspinlock: Add rqspinlock.h header Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 06/22] rqspinlock: Drop PV and virtualization support Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 07/22] rqspinlock: Add support for timeouts Kumar Kartikeya Dwivedi
2025-01-07 14:50   ` Peter Zijlstra
2025-01-07 17:14     ` Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 08/22] rqspinlock: Protect pending bit owners from stalls Kumar Kartikeya Dwivedi
2025-01-07 14:51   ` Peter Zijlstra
2025-01-07 17:14     ` Kumar Kartikeya Dwivedi
2025-01-07 19:17       ` Peter Zijlstra
2025-01-07 19:22         ` Peter Zijlstra
2025-01-07 19:54           ` Kumar Kartikeya Dwivedi
2025-01-08  2:19   ` Waiman Long
2025-01-08 20:13     ` Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 09/22] rqspinlock: Protect waiters in queue " Kumar Kartikeya Dwivedi
2025-01-08  3:38   ` Waiman Long
2025-01-08 20:42     ` Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 10/22] rqspinlock: Protect waiters in trylock fallback " Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 11/22] rqspinlock: Add deadlock detection and recovery Kumar Kartikeya Dwivedi
2025-01-08 16:06   ` Waiman Long
2025-01-08 20:19     ` Kumar Kartikeya Dwivedi
2025-01-09  0:32       ` Waiman Long
2025-01-07 13:59 ` Kumar Kartikeya Dwivedi [this message]
2025-01-08 16:27   ` [PATCH bpf-next v1 12/22] rqspinlock: Add basic support for CONFIG_PARAVIRT Waiman Long
2025-01-08 20:32     ` Kumar Kartikeya Dwivedi
2025-01-09  0:48       ` Waiman Long
2025-01-09  2:42         ` Alexei Starovoitov
2025-01-09  2:58           ` Waiman Long
2025-01-09  3:37             ` Alexei Starovoitov
2025-01-09  3:46               ` Waiman Long
2025-01-09  3:53                 ` Alexei Starovoitov
2025-01-09  3:58                   ` Waiman Long
2025-01-07 13:59 ` [PATCH bpf-next v1 13/22] rqspinlock: Add helper to print a splat on timeout or deadlock Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 14/22] rqspinlock: Add macros for rqspinlock usage Kumar Kartikeya Dwivedi
2025-01-08 16:55   ` Waiman Long
2025-01-08 20:41     ` Kumar Kartikeya Dwivedi
2025-01-09  1:11       ` Waiman Long
2025-01-09  3:30         ` Alexei Starovoitov
2025-01-09  4:09           ` Waiman Long
2025-01-07 13:59 ` [PATCH bpf-next v1 15/22] rqspinlock: Add locktorture support Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 16/22] rqspinlock: Add entry to Makefile, MAINTAINERS Kumar Kartikeya Dwivedi
2025-01-07 13:59 ` [PATCH bpf-next v1 17/22] bpf: Convert hashtab.c to rqspinlock Kumar Kartikeya Dwivedi
2025-01-07 14:00 ` [PATCH bpf-next v1 18/22] bpf: Convert percpu_freelist.c " Kumar Kartikeya Dwivedi
2025-01-07 14:00 ` [PATCH bpf-next v1 19/22] bpf: Convert lpm_trie.c " Kumar Kartikeya Dwivedi
2025-01-07 14:00 ` [PATCH bpf-next v1 20/22] bpf: Introduce rqspinlock kfuncs Kumar Kartikeya Dwivedi
2025-01-08 10:23   ` kernel test robot
2025-01-08 10:23   ` kernel test robot
2025-01-08 10:44   ` kernel test robot
2025-01-07 14:00 ` [PATCH bpf-next v1 21/22] bpf: Implement verifier support for rqspinlock Kumar Kartikeya Dwivedi
2025-01-07 14:00 ` [PATCH bpf-next v1 22/22] selftests/bpf: Add tests " Kumar Kartikeya Dwivedi
2025-01-07 23:54 ` [PATCH bpf-next v1 00/22] Resilient Queued Spin Lock Linus Torvalds
2025-01-08  9:18   ` Peter Zijlstra
2025-01-08 20:12     ` Kumar Kartikeya Dwivedi
2025-01-08 20:30       ` Linus Torvalds
2025-01-08 21:06         ` Kumar Kartikeya Dwivedi
2025-01-08 21:30         ` Paul E. McKenney
2025-01-09 13:59       ` Waiman Long
2025-01-09 21:13         ` Kumar Kartikeya Dwivedi
2025-01-09 21:18           ` Waiman Long

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20250107140004.2732830-13-memxor@gmail.com \
    --to=memxor@gmail.com \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=brho@google.com \
    --cc=daniel@iogearbox.net \
    --cc=dohyunkim@google.com \
    --cc=eddyz87@gmail.com \
    --cc=joshdon@google.com \
    --cc=kernel-team@meta.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=llong@redhat.com \
    --cc=martin.lau@kernel.org \
    --cc=paulmck@kernel.org \
    --cc=peterz@infradead.org \
    --cc=tj@kernel.org \
    --cc=torvalds@linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox