[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20220204155142.56419-1-Jason@zx2c4.com>
Date: Fri, 4 Feb 2022 16:51:42 +0100
From: "Jason A. Donenfeld" <Jason@...c4.com>
To: Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
Andy Lutomirski <luto@...capital.net>,
Boqun Feng <boqun.feng@...il.com>,
Will Deacon <will@...nel.org>,
Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>,
Waiman Long <longman@...hat.com>,
Sultan Alsawaf <sultan@...neltoast.com>,
"Theodore Ts'o" <tytso@....edu>, Andy Lutomirski <luto@...nel.org>,
Jonathan Neuschäfer <j.neuschaefer@....net>,
LKML <linux-kernel@...r.kernel.org>,
Thomas Gleixner <tglx@...utronix.de>
Cc: "Jason A. Donenfeld" <Jason@...c4.com>
Subject: [PATCH v3] random: remove batched entropy locking
Rather than use spinlocks to protect batched entropy, we can instead
disable interrupts locally, since we're dealing with per-cpu data, and
manage resets with a basic generation counter. At the same time, we
can't quite do this on PREEMPT_RT, where we still want spinlocks-as-
mutexes semantics. So we use a local_lock_t, which provides the right
behavior for each. Because this is a per-cpu lock, that generation
counter is still doing the necessary CPU-to-CPU communication.
This should improve performance a bit. It will also fix the linked splat
that Jonathan received with a PROVE_RAW_LOCK_NESTING=y.
Note that Sebastian has pointed out a few other areas where
using spinlock_t in an IRQ context is potentially problematic for
PREEMPT_RT. This patch handles one of those cases, and we'll hopefully
have additional patches for other cases.
Suggested-by: Andy Lutomirski <luto@...nel.org>
Reported-by: Jonathan Neuschäfer <j.neuschaefer@....net>
Tested-by: Jonathan Neuschäfer <j.neuschaefer@....net>
Link: https://lore.kernel.org/lkml/YfMa0QgsjCVdRAvJ@latitude/
Cc: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Signed-off-by: Jason A. Donenfeld <Jason@...c4.com>
---
drivers/char/random.c | 55 ++++++++++++++++++++++---------------------
1 file changed, 28 insertions(+), 27 deletions(-)
diff --git a/drivers/char/random.c b/drivers/char/random.c
index 455615ac169a..3e54b90a3ff8 100644
--- a/drivers/char/random.c
+++ b/drivers/char/random.c
@@ -1732,13 +1732,16 @@ struct ctl_table random_table[] = {
};
#endif /* CONFIG_SYSCTL */
+static atomic_t batch_generation = ATOMIC_INIT(0);
+
struct batched_entropy {
union {
u64 entropy_u64[CHACHA_BLOCK_SIZE / sizeof(u64)];
u32 entropy_u32[CHACHA_BLOCK_SIZE / sizeof(u32)];
};
+ local_lock_t lock;
unsigned int position;
- spinlock_t batch_lock;
+ int generation;
};
/*
@@ -1750,7 +1753,7 @@ struct batched_entropy {
* point prior.
*/
static DEFINE_PER_CPU(struct batched_entropy, batched_entropy_u64) = {
- .batch_lock = __SPIN_LOCK_UNLOCKED(batched_entropy_u64.lock),
+ .lock = INIT_LOCAL_LOCK(batched_entropy_u64.lock)
};
u64 get_random_u64(void)
@@ -1759,41 +1762,54 @@ u64 get_random_u64(void)
unsigned long flags;
struct batched_entropy *batch;
static void *previous;
+ int next_gen;
warn_unseeded_randomness(&previous);
- batch = raw_cpu_ptr(&batched_entropy_u64);
- spin_lock_irqsave(&batch->batch_lock, flags);
- if (batch->position % ARRAY_SIZE(batch->entropy_u64) == 0) {
+ batch = this_cpu_ptr(&batched_entropy_u64);
+ local_lock_irqsave(&batch->lock, flags);
+
+ next_gen = atomic_read(&batch_generation);
+ if (batch->position % ARRAY_SIZE(batch->entropy_u64) == 0 ||
+ next_gen != batch->generation) {
extract_crng((u8 *)batch->entropy_u64);
batch->position = 0;
+ batch->generation = next_gen;
}
+
ret = batch->entropy_u64[batch->position++];
- spin_unlock_irqrestore(&batch->batch_lock, flags);
+ local_unlock_irqrestore(&batch->lock, flags);
return ret;
}
EXPORT_SYMBOL(get_random_u64);
static DEFINE_PER_CPU(struct batched_entropy, batched_entropy_u32) = {
- .batch_lock = __SPIN_LOCK_UNLOCKED(batched_entropy_u32.lock),
+ .lock = INIT_LOCAL_LOCK(batched_entropy_u32.lock)
};
+
u32 get_random_u32(void)
{
u32 ret;
unsigned long flags;
struct batched_entropy *batch;
static void *previous;
+ int next_gen;
warn_unseeded_randomness(&previous);
- batch = raw_cpu_ptr(&batched_entropy_u32);
- spin_lock_irqsave(&batch->batch_lock, flags);
- if (batch->position % ARRAY_SIZE(batch->entropy_u32) == 0) {
+ batch = this_cpu_ptr(&batched_entropy_u32);
+ local_lock_irqsave(&batch->lock, flags);
+
+ next_gen = atomic_read(&batch_generation);
+ if (batch->position % ARRAY_SIZE(batch->entropy_u32) == 0 ||
+ next_gen != batch->generation) {
extract_crng((u8 *)batch->entropy_u32);
batch->position = 0;
+ batch->generation = next_gen;
}
+
ret = batch->entropy_u32[batch->position++];
- spin_unlock_irqrestore(&batch->batch_lock, flags);
+ local_unlock_irqrestore(&batch->lock, flags);
return ret;
}
EXPORT_SYMBOL(get_random_u32);
@@ -1804,22 +1820,7 @@ EXPORT_SYMBOL(get_random_u32);
* next usage. */
static void invalidate_batched_entropy(void)
{
- int cpu;
- unsigned long flags;
-
- for_each_possible_cpu(cpu) {
- struct batched_entropy *batched_entropy;
-
- batched_entropy = per_cpu_ptr(&batched_entropy_u32, cpu);
- spin_lock_irqsave(&batched_entropy->batch_lock, flags);
- batched_entropy->position = 0;
- spin_unlock(&batched_entropy->batch_lock);
-
- batched_entropy = per_cpu_ptr(&batched_entropy_u64, cpu);
- spin_lock(&batched_entropy->batch_lock);
- batched_entropy->position = 0;
- spin_unlock_irqrestore(&batched_entropy->batch_lock, flags);
- }
+ atomic_inc(&batch_generation);
}
/**
--
2.35.0
Powered by blists - more mailing lists