Many architectures' switch_mm() (e.g. arm64) do not have an smp_mb()
which the core scheduler code has depended upon since commit:
commit 223baf9d17f25 ("sched: Fix performance regression introduced by mm_cid")
If switch_mm() doesn't call smp_mb(), sched_mm_cid_remote_clear() can
unset the activly used cid when it fails to observe active task after it
sets lazy_put.
The *is* a memory barrier between storing to rq->curr and _return to
userspace_ (as required by membarrier), but the rseq mm_cid has stricter
requirements: the barrier needs to be issued between store to rq->curr
and switch_mm_cid(), which happens earlier than:
- spin_unlock(),
- switch_to().
So it's fine when the architecture switch_mm happens to have that barrier
already, but less so when the architecture only provides the full barrier
in switch_to() or spin_unlock().
It is a bug in the rseq switch_mm_cid() implementation. All architectures
that don't have memory barriers in switch_mm(), but rather have the full
barrier either in finish_lock_switch() or switch_to() have them too late
for the needs of switch_mm_cid().
Introduce a new smp_mb__after_switch_mm(), defined as smp_mb() in the
generic barrier.h header, and use it in switch_mm_cid() for scheduler
transitions where switch_mm() is expected to provide a memory barrier.
Architectures can override smp_mb__after_switch_mm() if their
switch_mm() implementation provides an implicit memory barrier.
Override it with a no-op on x86 which implicitly provide this memory
barrier by writing to CR3.
Reported-by: levi.yun <yeoreum.yun(a)arm.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com>
Fixes: 223baf9d17f2 ("sched: Fix performance regression introduced by mm_cid")
Cc: <stable(a)vger.kernel.org> # 6.4.x
Cc: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com>
Cc: Catalin Marinas <catalin.marinas(a)arm.com>
Cc: Mark Rutland <mark.rutland(a)arm.com>
Cc: Will Deacon <will(a)kernel.org>
Cc: Peter Zijlstra <peterz(a)infradead.org>
Cc: Aaron Lu <aaron.lu(a)intel.com>
---
arch/x86/include/asm/barrier.h | 3 +++
include/asm-generic/barrier.h | 8 ++++++++
kernel/sched/sched.h | 19 +++++++++++++------
3 files changed, 24 insertions(+), 6 deletions(-)
diff --git a/arch/x86/include/asm/barrier.h b/arch/x86/include/asm/barrier.h
index 35389b2af88e..0d5e54201eb2 100644
--- a/arch/x86/include/asm/barrier.h
+++ b/arch/x86/include/asm/barrier.h
@@ -79,6 +79,9 @@ do { \
#define __smp_mb__before_atomic() do { } while (0)
#define __smp_mb__after_atomic() do { } while (0)
+/* Writing to CR3 provides a full memory barrier in switch_mm(). */
+#define smp_mb__after_switch_mm() do { } while (0)
+
#include <asm-generic/barrier.h>
/*
diff --git a/include/asm-generic/barrier.h b/include/asm-generic/barrier.h
index 961f4d88f9ef..5a6c94d7a598 100644
--- a/include/asm-generic/barrier.h
+++ b/include/asm-generic/barrier.h
@@ -296,5 +296,13 @@ do { \
#define io_stop_wc() do { } while (0)
#endif
+/*
+ * Architectures that guarantee an implicit smp_mb() in switch_mm()
+ * can override smp_mb__after_switch_mm.
+ */
+#ifndef smp_mb__after_switch_mm
+#define smp_mb__after_switch_mm() smp_mb()
+#endif
+
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_GENERIC_BARRIER_H */
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 2e5a95486a42..638ebd355912 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -79,6 +79,8 @@
# include <asm/paravirt_api_clock.h>
#endif
+#include <asm/barrier.h>
+
#include "cpupri.h"
#include "cpudeadline.h"
@@ -3481,13 +3483,18 @@ static inline void switch_mm_cid(struct rq *rq,
* between rq->curr store and load of {prev,next}->mm->pcpu_cid[cpu].
* Provide it here.
*/
- if (!prev->mm) // from kernel
+ if (!prev->mm) { // from kernel
smp_mb();
- /*
- * user -> user transition guarantees a memory barrier through
- * switch_mm() when current->mm changes. If current->mm is
- * unchanged, no barrier is needed.
- */
+ } else { // from user
+ /*
+ * user -> user transition relies on an implicit the
+ * memory barrier in switch_mm() when current->mm
+ * changes. If the architecture switch_mm() does not
+ * have an implicit memory barrier, it is emitted here.
+ * If current->mm is unchanged, no barrier is needed.
+ */
+ smp_mb__after_switch_mm();
+ }
}
if (prev->mm_cid_active) {
mm_cid_snapshot_time(rq, prev->mm);
--
2.39.2
From: "Steven Rostedt (Google)" <rostedt(a)goodmis.org>
When the trace_pipe_raw file is closed, there should be no new readers on
the file descriptor. This is mostly handled with the waking and wait_index
fields of the iterator. But there's still a slight race.
CPU 0 CPU 1
----- -----
wait_index++;
index = wait_index;
ring_buffer_wake_waiters();
wait_on_pipe()
ring_buffer_wait();
The ring_buffer_wait() will miss the wakeup from CPU 1. The problem is
that the ring_buffer_wait() needs the logic of:
prepare_to_wait();
if (!condition)
schedule();
Where the missing condition check is the iter->wait_index update.
Have the ring_buffer_wait() take a conditional callback function and a
data parameter that can be used within the wait_event_interruptible() of
the ring_buffer_wait() function.
In wait_on_pipe(), pass a condition function that will check if the
wait_index has been updated, if it has, it will return true to break out
of the wait_event_interruptible() loop.
Create a new field "closed" in the trace_iterator and set it in the
.flush() callback before calling ring_buffer_wake_waiters().
This will keep any new readers from waiting on a closed file descriptor.
Have the wait_on_pipe() condition callback also check the closed field.
Change the wait_index field of the trace_iterator to atomic_t. There's no
reason it needs to be 'long' and making it atomic and using
atomic_read_acquire() and atomic_fetch_inc_release() will provide the
necessary memory barriers.
Add a "woken" flag to tracing_buffers_splice_read() to exit the loop after
one more try to fetch data. That is, if it waited for data and something
woke it up, it should try to collect any new data and then exit back to
user space.
Link: https://lore.kernel.org/linux-trace-kernel/CAHk-=wgsNgewHFxZAJiAQznwPMqEtQm…
Link: https://lore.kernel.org/linux-trace-kernel/20240312121703.557950713@goodmis…
Cc: stable(a)vger.kernel.org
Cc: Masami Hiramatsu <mhiramat(a)kernel.org>
Cc: Mark Rutland <mark.rutland(a)arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com>
Cc: Andrew Morton <akpm(a)linux-foundation.org>
Cc: Linus Torvalds <torvalds(a)linux-foundation.org>
Cc: linke li <lilinke99(a)qq.com>
Cc: Rabin Vincent <rabin(a)rab.in>
Fixes: f3ddb74ad0790 ("tracing: Wake up ring buffer waiters on closing of the file")
Signed-off-by: Steven Rostedt (Google) <rostedt(a)goodmis.org>
---
include/linux/ring_buffer.h | 3 ++-
include/linux/trace_events.h | 5 ++++-
kernel/trace/ring_buffer.c | 13 ++++++-----
kernel/trace/trace.c | 43 ++++++++++++++++++++++++++----------
4 files changed, 45 insertions(+), 19 deletions(-)
diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h
index 338a33db1577..dc5ae4e96aee 100644
--- a/include/linux/ring_buffer.h
+++ b/include/linux/ring_buffer.h
@@ -99,7 +99,8 @@ __ring_buffer_alloc(unsigned long size, unsigned flags, struct lock_class_key *k
})
typedef bool (*ring_buffer_cond_fn)(void *data);
-int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full);
+int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full,
+ ring_buffer_cond_fn cond, void *data);
__poll_t ring_buffer_poll_wait(struct trace_buffer *buffer, int cpu,
struct file *filp, poll_table *poll_table, int full);
void ring_buffer_wake_waiters(struct trace_buffer *buffer, int cpu);
diff --git a/include/linux/trace_events.h b/include/linux/trace_events.h
index d68ff9b1247f..fc6d0af56bb1 100644
--- a/include/linux/trace_events.h
+++ b/include/linux/trace_events.h
@@ -103,13 +103,16 @@ struct trace_iterator {
unsigned int temp_size;
char *fmt; /* modified format holder */
unsigned int fmt_size;
- long wait_index;
+ atomic_t wait_index;
/* trace_seq for __print_flags() and __print_symbolic() etc. */
struct trace_seq tmp_seq;
cpumask_var_t started;
+ /* Set when the file is closed to prevent new waiters */
+ bool closed;
+
/* it's true when current open file is snapshot */
bool snapshot;
diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index f4c34b7c7e1e..350607cce869 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -902,23 +902,26 @@ static bool rb_wait_once(void *data)
* @buffer: buffer to wait on
* @cpu: the cpu buffer to wait on
* @full: wait until the percentage of pages are available, if @cpu != RING_BUFFER_ALL_CPUS
+ * @cond: condition function to break out of wait (NULL to run once)
+ * @data: the data to pass to @cond.
*
* If @cpu == RING_BUFFER_ALL_CPUS then the task will wake up as soon
* as data is added to any of the @buffer's cpu buffers. Otherwise
* it will wait for data to be added to a specific cpu buffer.
*/
-int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
+int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full,
+ ring_buffer_cond_fn cond, void *data)
{
struct ring_buffer_per_cpu *cpu_buffer;
struct wait_queue_head *waitq;
- ring_buffer_cond_fn cond;
struct rb_irq_work *rbwork;
- void *data;
long once = 0;
int ret = 0;
- cond = rb_wait_once;
- data = &once;
+ if (!cond) {
+ cond = rb_wait_once;
+ data = &once;
+ }
/*
* Depending on what the caller is waiting for, either any
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index c9c898307348..d390fea3a6a5 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -1955,15 +1955,36 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu)
#endif /* CONFIG_TRACER_MAX_TRACE */
+struct pipe_wait {
+ struct trace_iterator *iter;
+ int wait_index;
+};
+
+static bool wait_pipe_cond(void *data)
+{
+ struct pipe_wait *pwait = data;
+ struct trace_iterator *iter = pwait->iter;
+
+ if (atomic_read_acquire(&iter->wait_index) != pwait->wait_index)
+ return true;
+
+ return iter->closed;
+}
+
static int wait_on_pipe(struct trace_iterator *iter, int full)
{
+ struct pipe_wait pwait;
int ret;
/* Iterators are static, they should be filled or empty */
if (trace_buffer_iter(iter, iter->cpu_file))
return 0;
- ret = ring_buffer_wait(iter->array_buffer->buffer, iter->cpu_file, full);
+ pwait.wait_index = atomic_read_acquire(&iter->wait_index);
+ pwait.iter = iter;
+
+ ret = ring_buffer_wait(iter->array_buffer->buffer, iter->cpu_file, full,
+ wait_pipe_cond, &pwait);
#ifdef CONFIG_TRACER_MAX_TRACE
/*
@@ -8398,9 +8419,9 @@ static int tracing_buffers_flush(struct file *file, fl_owner_t id)
struct ftrace_buffer_info *info = file->private_data;
struct trace_iterator *iter = &info->iter;
- iter->wait_index++;
+ iter->closed = true;
/* Make sure the waiters see the new wait_index */
- smp_wmb();
+ (void)atomic_fetch_inc_release(&iter->wait_index);
ring_buffer_wake_waiters(iter->array_buffer->buffer, iter->cpu_file);
@@ -8500,6 +8521,7 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
.spd_release = buffer_spd_release,
};
struct buffer_ref *ref;
+ bool woken = false;
int page_size;
int entries, i;
ssize_t ret = 0;
@@ -8573,17 +8595,17 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
/* did we read anything? */
if (!spd.nr_pages) {
- long wait_index;
if (ret)
goto out;
+ if (woken)
+ goto out;
+
ret = -EAGAIN;
if ((file->f_flags & O_NONBLOCK) || (flags & SPLICE_F_NONBLOCK))
goto out;
- wait_index = READ_ONCE(iter->wait_index);
-
ret = wait_on_pipe(iter, iter->snapshot ? 0 : iter->tr->buffer_percent);
if (ret)
goto out;
@@ -8592,10 +8614,8 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
if (!tracer_tracing_is_on(iter->tr))
goto out;
- /* Make sure we see the new wait_index */
- smp_rmb();
- if (wait_index != iter->wait_index)
- goto out;
+ /* Iterate one more time to collect any new data then exit */
+ woken = true;
goto again;
}
@@ -8618,9 +8638,8 @@ static long tracing_buffers_ioctl(struct file *file, unsigned int cmd, unsigned
mutex_lock(&trace_types_lock);
- iter->wait_index++;
/* Make sure the waiters see the new wait_index */
- smp_wmb();
+ (void)atomic_fetch_inc_release(&iter->wait_index);
ring_buffer_wake_waiters(iter->array_buffer->buffer, iter->cpu_file);
--
2.43.0
From: "Steven Rostedt (Google)" <rostedt(a)goodmis.org>
Convert ring_buffer_wait() over to wait_event_interruptible(). The default
condition is to execute the wait loop inside __wait_event() just once.
This does not change the ring_buffer_wait() prototype yet, but
restructures the code so that it can take a "cond" and "data" parameter
and will call wait_event_interruptible() with a helper function as the
condition.
The helper function (rb_wait_cond) takes the cond function and data
parameters. It will first check if the buffer hit the watermark defined by
the "full" parameter and then call the passed in condition parameter. If
either are true, it returns true.
If rb_wait_cond() does not return true, it will set the appropriate
"waiters_pending" flag and returns false.
Link: https://lore.kernel.org/linux-trace-kernel/CAHk-=wgsNgewHFxZAJiAQznwPMqEtQm…
Link: https://lore.kernel.org/linux-trace-kernel/20240312121703.399598519@goodmis…
Cc: stable(a)vger.kernel.org
Cc: Masami Hiramatsu <mhiramat(a)kernel.org>
Cc: Mark Rutland <mark.rutland(a)arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com>
Cc: Andrew Morton <akpm(a)linux-foundation.org>
Cc: Linus Torvalds <torvalds(a)linux-foundation.org>
Cc: linke li <lilinke99(a)qq.com>
Cc: Rabin Vincent <rabin(a)rab.in>
Fixes: f3ddb74ad0790 ("tracing: Wake up ring buffer waiters on closing of the file")
Signed-off-by: Steven Rostedt (Google) <rostedt(a)goodmis.org>
---
include/linux/ring_buffer.h | 1 +
kernel/trace/ring_buffer.c | 116 +++++++++++++++++++++---------------
2 files changed, 69 insertions(+), 48 deletions(-)
diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h
index fa802db216f9..338a33db1577 100644
--- a/include/linux/ring_buffer.h
+++ b/include/linux/ring_buffer.h
@@ -98,6 +98,7 @@ __ring_buffer_alloc(unsigned long size, unsigned flags, struct lock_class_key *k
__ring_buffer_alloc((size), (flags), &__key); \
})
+typedef bool (*ring_buffer_cond_fn)(void *data);
int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full);
__poll_t ring_buffer_poll_wait(struct trace_buffer *buffer, int cpu,
struct file *filp, poll_table *poll_table, int full);
diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 8c3730a88662..f4c34b7c7e1e 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -843,43 +843,15 @@ static bool rb_watermark_hit(struct trace_buffer *buffer, int cpu, int full)
return ret;
}
-/**
- * ring_buffer_wait - wait for input to the ring buffer
- * @buffer: buffer to wait on
- * @cpu: the cpu buffer to wait on
- * @full: wait until the percentage of pages are available, if @cpu != RING_BUFFER_ALL_CPUS
- *
- * If @cpu == RING_BUFFER_ALL_CPUS then the task will wake up as soon
- * as data is added to any of the @buffer's cpu buffers. Otherwise
- * it will wait for data to be added to a specific cpu buffer.
- */
-int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
+static inline bool
+rb_wait_cond(struct rb_irq_work *rbwork, struct trace_buffer *buffer,
+ int cpu, int full, ring_buffer_cond_fn cond, void *data)
{
- struct ring_buffer_per_cpu *cpu_buffer;
- DEFINE_WAIT(wait);
- struct rb_irq_work *work;
- int ret = 0;
-
- /*
- * Depending on what the caller is waiting for, either any
- * data in any cpu buffer, or a specific buffer, put the
- * caller on the appropriate wait queue.
- */
- if (cpu == RING_BUFFER_ALL_CPUS) {
- work = &buffer->irq_work;
- /* Full only makes sense on per cpu reads */
- full = 0;
- } else {
- if (!cpumask_test_cpu(cpu, buffer->cpumask))
- return -ENODEV;
- cpu_buffer = buffer->buffers[cpu];
- work = &cpu_buffer->irq_work;
- }
+ if (rb_watermark_hit(buffer, cpu, full))
+ return true;
- if (full)
- prepare_to_wait(&work->full_waiters, &wait, TASK_INTERRUPTIBLE);
- else
- prepare_to_wait(&work->waiters, &wait, TASK_INTERRUPTIBLE);
+ if (cond(data))
+ return true;
/*
* The events can happen in critical sections where
@@ -902,27 +874,75 @@ int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
* a task has been queued. It's OK for spurious wake ups.
*/
if (full)
- work->full_waiters_pending = true;
+ rbwork->full_waiters_pending = true;
else
- work->waiters_pending = true;
+ rbwork->waiters_pending = true;
- if (rb_watermark_hit(buffer, cpu, full))
- goto out;
+ return false;
+}
- if (signal_pending(current)) {
- ret = -EINTR;
- goto out;
+/*
+ * The default wait condition for ring_buffer_wait() is to just to exit the
+ * wait loop the first time it is woken up.
+ */
+static bool rb_wait_once(void *data)
+{
+ long *once = data;
+
+ /* wait_event() actually calls this twice before scheduling*/
+ if (*once > 1)
+ return true;
+
+ (*once)++;
+ return false;
+}
+
+/**
+ * ring_buffer_wait - wait for input to the ring buffer
+ * @buffer: buffer to wait on
+ * @cpu: the cpu buffer to wait on
+ * @full: wait until the percentage of pages are available, if @cpu != RING_BUFFER_ALL_CPUS
+ *
+ * If @cpu == RING_BUFFER_ALL_CPUS then the task will wake up as soon
+ * as data is added to any of the @buffer's cpu buffers. Otherwise
+ * it will wait for data to be added to a specific cpu buffer.
+ */
+int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
+{
+ struct ring_buffer_per_cpu *cpu_buffer;
+ struct wait_queue_head *waitq;
+ ring_buffer_cond_fn cond;
+ struct rb_irq_work *rbwork;
+ void *data;
+ long once = 0;
+ int ret = 0;
+
+ cond = rb_wait_once;
+ data = &once;
+
+ /*
+ * Depending on what the caller is waiting for, either any
+ * data in any cpu buffer, or a specific buffer, put the
+ * caller on the appropriate wait queue.
+ */
+ if (cpu == RING_BUFFER_ALL_CPUS) {
+ rbwork = &buffer->irq_work;
+ /* Full only makes sense on per cpu reads */
+ full = 0;
+ } else {
+ if (!cpumask_test_cpu(cpu, buffer->cpumask))
+ return -ENODEV;
+ cpu_buffer = buffer->buffers[cpu];
+ rbwork = &cpu_buffer->irq_work;
}
- schedule();
- out:
if (full)
- finish_wait(&work->full_waiters, &wait);
+ waitq = &rbwork->full_waiters;
else
- finish_wait(&work->waiters, &wait);
+ waitq = &rbwork->waiters;
- if (!ret && !rb_watermark_hit(buffer, cpu, full) && signal_pending(current))
- ret = -EINTR;
+ ret = wait_event_interruptible((*waitq),
+ rb_wait_cond(rbwork, buffer, cpu, full, cond, data));
return ret;
}
--
2.43.0
From: "Steven Rostedt (Google)" <rostedt(a)goodmis.org>
If a reader of the ring buffer is doing a poll, and waiting for the ring
buffer to hit a specific watermark, there could be a case where it gets
into an infinite ping-pong loop.
The poll code has:
rbwork->full_waiters_pending = true;
if (!cpu_buffer->shortest_full ||
cpu_buffer->shortest_full > full)
cpu_buffer->shortest_full = full;
The writer will see full_waiters_pending and check if the ring buffer is
filled over the percentage of the shortest_full value. If it is, it calls
an irq_work to wake up all the waiters.
But the code could get into a circular loop:
CPU 0 CPU 1
----- -----
[ Poll ]
[ shortest_full = 0 ]
rbwork->full_waiters_pending = true;
if (rbwork->full_waiters_pending &&
[ buffer percent ] > shortest_full) {
rbwork->wakeup_full = true;
[ queue_irqwork ]
cpu_buffer->shortest_full = full;
[ IRQ work ]
if (rbwork->wakeup_full) {
cpu_buffer->shortest_full = 0;
wakeup poll waiters;
[woken]
if ([ buffer percent ] > full)
break;
rbwork->full_waiters_pending = true;
if (rbwork->full_waiters_pending &&
[ buffer percent ] > shortest_full) {
rbwork->wakeup_full = true;
[ queue_irqwork ]
cpu_buffer->shortest_full = full;
[ IRQ work ]
if (rbwork->wakeup_full) {
cpu_buffer->shortest_full = 0;
wakeup poll waiters;
[woken]
[ Wash, rinse, repeat! ]
In the poll, the shortest_full needs to be set before the
full_pending_waiters, as once that is set, the writer will compare the
current shortest_full (which is incorrect) to decide to call the irq_work,
which will reset the shortest_full (expecting the readers to update it).
Also move the setting of full_waiters_pending after the check if the ring
buffer has the required percentage filled. There's no reason to tell the
writer to wake up waiters if there are no waiters.
Link: https://lore.kernel.org/linux-trace-kernel/20240312131952.630922155@goodmis…
Cc: stable(a)vger.kernel.org
Cc: Mark Rutland <mark.rutland(a)arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com>
Cc: Andrew Morton <akpm(a)linux-foundation.org>
Fixes: 42fb0a1e84ff5 ("tracing/ring-buffer: Have polling block on watermark")
Reviewed-by: Masami Hiramatsu (Google) <mhiramat(a)kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt(a)goodmis.org>
---
kernel/trace/ring_buffer.c | 27 ++++++++++++++++++++-------
1 file changed, 20 insertions(+), 7 deletions(-)
diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 6ffbccb9bcf0..99fdda29ce4e 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -965,16 +965,32 @@ __poll_t ring_buffer_poll_wait(struct trace_buffer *buffer, int cpu,
poll_wait(filp, &rbwork->full_waiters, poll_table);
raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
- rbwork->full_waiters_pending = true;
if (!cpu_buffer->shortest_full ||
cpu_buffer->shortest_full > full)
cpu_buffer->shortest_full = full;
raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
- } else {
- poll_wait(filp, &rbwork->waiters, poll_table);
- rbwork->waiters_pending = true;
+ if (full_hit(buffer, cpu, full))
+ return EPOLLIN | EPOLLRDNORM;
+ /*
+ * Only allow full_waiters_pending update to be seen after
+ * the shortest_full is set. If the writer sees the
+ * full_waiters_pending flag set, it will compare the
+ * amount in the ring buffer to shortest_full. If the amount
+ * in the ring buffer is greater than the shortest_full
+ * percent, it will call the irq_work handler to wake up
+ * this list. The irq_handler will reset shortest_full
+ * back to zero. That's done under the reader_lock, but
+ * the below smp_mb() makes sure that the update to
+ * full_waiters_pending doesn't leak up into the above.
+ */
+ smp_mb();
+ rbwork->full_waiters_pending = true;
+ return 0;
}
+ poll_wait(filp, &rbwork->waiters, poll_table);
+ rbwork->waiters_pending = true;
+
/*
* There's a tight race between setting the waiters_pending and
* checking if the ring buffer is empty. Once the waiters_pending bit
@@ -990,9 +1006,6 @@ __poll_t ring_buffer_poll_wait(struct trace_buffer *buffer, int cpu,
*/
smp_mb();
- if (full)
- return full_hit(buffer, cpu, full) ? EPOLLIN | EPOLLRDNORM : 0;
-
if ((cpu == RING_BUFFER_ALL_CPUS && !ring_buffer_empty(buffer)) ||
(cpu != RING_BUFFER_ALL_CPUS && !ring_buffer_empty_cpu(buffer, cpu)))
return EPOLLIN | EPOLLRDNORM;
--
2.43.0
From: "Steven Rostedt (Google)" <rostedt(a)goodmis.org>
The rb_watermark_hit() checks if the amount of data in the ring buffer is
above the percentage level passed in by the "full" variable. If it is, it
returns true.
But it also sets the "shortest_full" field of the cpu_buffer that informs
writers that it needs to call the irq_work if the amount of data on the
ring buffer is above the requested amount.
The rb_watermark_hit() always sets the shortest_full even if the amount in
the ring buffer is what it wants. As it is not going to wait, because it
has what it wants, there's no reason to set shortest_full.
Link: https://lore.kernel.org/linux-trace-kernel/20240312115641.6aa8ba08@gandalf.…
Cc: stable(a)vger.kernel.org
Cc: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com>
Fixes: 42fb0a1e84ff5 ("tracing/ring-buffer: Have polling block on watermark")
Reviewed-by: Masami Hiramatsu (Google) <mhiramat(a)kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt(a)goodmis.org>
---
kernel/trace/ring_buffer.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index aa332ace108b..6ffbccb9bcf0 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -834,9 +834,10 @@ static bool rb_watermark_hit(struct trace_buffer *buffer, int cpu, int full)
pagebusy = cpu_buffer->reader_page == cpu_buffer->commit_page;
ret = !pagebusy && full_hit(buffer, cpu, full);
- if (!cpu_buffer->shortest_full ||
- cpu_buffer->shortest_full > full)
- cpu_buffer->shortest_full = full;
+ if (!ret && (!cpu_buffer->shortest_full ||
+ cpu_buffer->shortest_full > full)) {
+ cpu_buffer->shortest_full = full;
+ }
raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
}
return ret;
--
2.43.0