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-prev] [thread-next>] [day] [month] [year] [list]
Date:   Thu, 01 Jun 2017 16:40:55 +0100
From:   Ben Hutchings <ben@...adent.org.uk>
To:     linux-kernel@...r.kernel.org, stable@...r.kernel.org
CC:     akpm@...ux-foundation.org,
        "Fabio Estevam" <fabio.estevam@...escale.com>,
        "Marek Vasut" <marex@...x.de>,
        "Herbert Xu" <herbert@...dor.apana.org.au>,
        "Tom Lendacky" <thomas.lendacky@....com>,
        "David S. Miller" <davem@...emloft.net>,
        "Shawn Guo" <shawn.guo@...aro.org>
Subject: [PATCH 3.2 085/101] crypto: hash - Fix the pointer voodoo in
 unaligned ahash

3.2.89-rc1 review patch.  If anyone has any objections, please let me know.

------------------

From: Marek Vasut <marex@...x.de>

commit ab6bf4e5e5e4298e8649e635bee25542cccbfd97 upstream.

Add documentation for the pointer voodoo that is happening in crypto/ahash.c
in ahash_op_unaligned(). This code is quite confusing, so add a beefy chunk
of documentation.

Moreover, make sure the mangled request is completely restored after finishing
this unaligned operation. This means restoring all of .result, .base.data
and .base.complete .

Also, remove the crypto_completion_t complete = ... line present in the
ahash_op_unaligned_done() function. This type actually declares a function
pointer, which is very confusing.

Finally, yet very important nonetheless, make sure the req->priv is free()'d
only after the original request is restored in ahash_op_unaligned_done().
The req->priv data must not be free()'d before that in ahash_op_unaligned_finish(),
since we would be accessing previously free()'d data in ahash_op_unaligned_done()
and cause corruption.

Signed-off-by: Marek Vasut <marex@...x.de>
Cc: David S. Miller <davem@...emloft.net>
Cc: Fabio Estevam <fabio.estevam@...escale.com>
Cc: Herbert Xu <herbert@...dor.apana.org.au>
Cc: Shawn Guo <shawn.guo@...aro.org>
Cc: Tom Lendacky <thomas.lendacky@....com>
Signed-off-by: Herbert Xu <herbert@...dor.apana.org.au>
Signed-off-by: Ben Hutchings <ben@...adent.org.uk>
---
 crypto/ahash.c | 56 +++++++++++++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 49 insertions(+), 7 deletions(-)

--- a/crypto/ahash.c
+++ b/crypto/ahash.c
@@ -202,22 +202,34 @@ static void ahash_op_unaligned_finish(st
 		memcpy(priv->result, req->result,
 		       crypto_ahash_digestsize(crypto_ahash_reqtfm(req)));
 
+	/* Restore the original crypto request. */
+	req->result = priv->result;
+	req->base.complete = priv->complete;
+	req->base.data = priv->data;
+	req->priv = NULL;
+
+	/* Free the req->priv.priv from the ADJUSTED request. */
 	kzfree(priv);
 }
 
 static void ahash_op_unaligned_done(struct crypto_async_request *req, int err)
 {
 	struct ahash_request *areq = req->data;
-	struct ahash_request_priv *priv = areq->priv;
-	crypto_completion_t complete = priv->complete;
-	void *data = priv->data;
 
-	ahash_op_unaligned_finish(areq, err);
+	/*
+	 * Restore the original request, see ahash_op_unaligned() for what
+	 * goes where.
+	 *
+	 * The "struct ahash_request *req" here is in fact the "req.base"
+	 * from the ADJUSTED request from ahash_op_unaligned(), thus as it
+	 * is a pointer to self, it is also the ADJUSTED "req" .
+	 */
 
-	areq->base.complete = complete;
-	areq->base.data = data;
+	/* First copy areq->result into areq->priv.result */
+	ahash_op_unaligned_finish(areq, err);
 
-	complete(&areq->base, err);
+	/* Complete the ORIGINAL request. */
+	areq->base.complete(&areq->base, err);
 }
 
 static int ahash_op_unaligned(struct ahash_request *req,
@@ -235,9 +247,39 @@ static int ahash_op_unaligned(struct aha
 	if (!priv)
 		return -ENOMEM;
 
+	/*
+	 * WARNING: Voodoo programming below!
+	 *
+	 * The code below is obscure and hard to understand, thus explanation
+	 * is necessary. See include/crypto/hash.h and include/linux/crypto.h
+	 * to understand the layout of structures used here!
+	 *
+	 * The code here will replace portions of the ORIGINAL request with
+	 * pointers to new code and buffers so the hashing operation can store
+	 * the result in aligned buffer. We will call the modified request
+	 * an ADJUSTED request.
+	 *
+	 * The newly mangled request will look as such:
+	 *
+	 * req {
+	 *   .result        = ADJUSTED[new aligned buffer]
+	 *   .base.complete = ADJUSTED[pointer to completion function]
+	 *   .base.data     = ADJUSTED[*req (pointer to self)]
+	 *   .priv          = ADJUSTED[new priv] {
+	 *           .result   = ORIGINAL(result)
+	 *           .complete = ORIGINAL(base.complete)
+	 *           .data     = ORIGINAL(base.data)
+	 *   }
+	 */
+
 	priv->result = req->result;
 	priv->complete = req->base.complete;
 	priv->data = req->base.data;
+	/*
+	 * WARNING: We do not backup req->priv here! The req->priv
+	 *          is for internal use of the Crypto API and the
+	 *          user must _NOT_ _EVER_ depend on it's content!
+	 */
 
 	req->result = PTR_ALIGN((u8 *)priv->ubuf, alignmask + 1);
 	req->base.complete = ahash_op_unaligned_done;

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ