linux-kernel - [PATCH RFC v1] random: use static branch for crng

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <20220503134052.646325-1-Jason@zx2c4.com>
Date:   Tue,  3 May 2022 15:40:52 +0200
From:   "Jason A. Donenfeld" <Jason@...c4.com>
To:     linux-kernel@...r.kernel.org, linux-crypto@...r.kernel.org
Cc:     "Jason A. Donenfeld" <Jason@...c4.com>,
        Theodore Ts'o <tytso@....edu>,
        Dominik Brodowski <linux@...inikbrodowski.net>,
        Steven Rostedt <rostedt@...dmis.org>,
        Ard Biesheuvel <ardb@...nel.org>
Subject: [PATCH RFC v1] random: use static branch for crng_ready()

Since crng_ready() is only false briefly during initialization and then
forever after becomes true, we don't need to evaluate it after, making
it a prime candidate for a static branch.

One complication, however, is that it changes state in a particular call
to credit_entropy_bits(), which might be made from atomic context. So
rather than changing it then, we keep around the same state variable we
had before, but the next time it's used from non-atomic context, we
change it lazily then.

This is an RFC for now because it seems a bit complicated and fiddly,
and potentially not worth the complexity. I'm interested to hear if
people have opinions about this or if there's a better way to do it.

Cc: Theodore Ts'o <tytso@....edu>
Cc: Dominik Brodowski <linux@...inikbrodowski.net>
Cc: Steven Rostedt <rostedt@...dmis.org>
Cc: Ard Biesheuvel <ardb@...nel.org>
Signed-off-by: Jason A. Donenfeld <Jason@...c4.com>
---
 drivers/char/random.c | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/drivers/char/random.c b/drivers/char/random.c
index 845f610b6611..977093022430 100644
--- a/drivers/char/random.c
+++ b/drivers/char/random.c
@@ -75,12 +75,13 @@
  * crng_init =  0 --> Uninitialized
  *		1 --> Initialized
  *		2 --> Initialized from input_pool
+ *		3 --> Initialized from input_pool and static key set
  *
  * crng_init is protected by base_crng->lock, and only increases
- * its value (from 0->1->2).
+ * its value (from 0->1->2->3).
  */
 static int crng_init = 0;
-#define crng_ready() (likely(crng_init > 1))
+static DEFINE_STATIC_KEY_FALSE(crng_ready_static);
 /* Various types of waiters for crng_init->2 transition. */
 static DECLARE_WAIT_QUEUE_HEAD(crng_init_wait);
 static struct fasync_struct *fasync;
@@ -96,6 +97,17 @@ static int ratelimit_disable __read_mostly;
 module_param_named(ratelimit_disable, ratelimit_disable, int, 0644);
 MODULE_PARM_DESC(ratelimit_disable, "Disable random ratelimit suppression");
 
+static bool crng_ready_slowpath(void)
+{
+	if (crng_init <= 1)
+		return false;
+	if (in_atomic() || irqs_disabled() || cmpxchg(&crng_init, 2, 3) != 2)
+		return true;
+	static_branch_enable(&crng_ready_static);
+	return true;
+}
+#define crng_ready() (static_branch_likely(&crng_ready_static) || unlikely(crng_ready_slowpath()))
+
 /*
  * Returns whether or not the input pool has been seeded and thus guaranteed
  * to supply cryptographically secure random numbers. This applies to: the
-- 
2.35.1