Received-SPF: pass (google.com: domain of linux-crypto-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
From:   Sandy Harris <sandyinchina@gmail.com>
Date:   Sun, 30 Jan 2022 17:56:04 +0800
Message-ID: <CACXcFmnPumpkfLLzzjqkBmxwtpMa0izNj3LOtf2ycTugAKAUwQ@mail.gmail.com>
Subject: [PATCH] random.c Remove locking in extract_buf()
To:     Linux Crypto Mailing List <linux-crypto@vger.kernel.org>,
        "Ted Ts'o" <tytso@mit.edu>, "Jason A. Donenfeld" <Jason@zx2c4.com>,
        m@ib.tc, Herbert Xu <herbert@gondor.apana.org.au>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

This function does not need to lock the input pool
during the hash since that only reads the pool &
we do not care if a write makes the hash result
indeterminate. "That's not a bug; it's a feature."

Removing the unnecessary lock prevents it from
delaying other threads or interrupts which write
to the input pool. Such delays are a bug.

We do need to lock the input pool when writing
to it. Changing __mix_pool_bytes() to plain
mix_pool_bytes() accomplishes that.

We do not need a lock for *out, the only other
place where this function writes. That points to
an array declared local in the calling function.
---
 drivers/char/random.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/drivers/char/random.c b/drivers/char/random.c
index 68613f0b6887..9dbf7c8c68dd 100644
--- a/drivers/char/random.c
+++ b/drivers/char/random.c
@@ -1355,7 +1355,6 @@ static void extract_buf(u8 *out)
     }

     /* Generate a hash across the pool */
-    spin_lock_irqsave(&input_pool.lock, flags);
     blake2s_update(&state, (const u8 *)input_pool_data, POOL_BYTES);
     blake2s_final(&state, hash); /* final zeros out state */

@@ -1368,8 +1367,7 @@ static void extract_buf(u8 *out)
      * brute-forcing the feedback as hard as brute-forcing the
      * hash.
      */
-    __mix_pool_bytes(hash, sizeof(hash));
-    spin_unlock_irqrestore(&input_pool.lock, flags);
+    mix_pool_bytes(hash, sizeof(hash));

     /* Note that EXTRACT_SIZE is half of hash size here, because above
      * we've dumped the full length back into mixer. By reducing the
-- 
Signed-off-by: Sandy Harris <sandyinchina@gmail.com>