From: Eric Biggers ebiggers@google.com
crypto_shash_descsize() returns the size of the shash_desc context needed to compute the hash, not the size of the hash itself.
crypto_shash_digestsize() would be correct, or alternatively using c->hash_len and c->hmac_desc_len which already store the correct values. But actually it's simpler to just use stack arrays, so do that instead.
Fixes: 49525e5eecca ("ubifs: Add helper functions for authentication support") Fixes: da8ef65f9573 ("ubifs: Authenticate replayed journal") Cc: stable@vger.kernel.org # v4.20+ Cc: Sascha Hauer s.hauer@pengutronix.de Signed-off-by: Eric Biggers ebiggers@google.com --- fs/ubifs/auth.c | 17 ++++------------- fs/ubifs/replay.c | 13 ++----------- 2 files changed, 6 insertions(+), 24 deletions(-)
diff --git a/fs/ubifs/auth.c b/fs/ubifs/auth.c index 8cdbd53d780ca7..f985a3fbbb36a1 100644 --- a/fs/ubifs/auth.c +++ b/fs/ubifs/auth.c @@ -79,13 +79,9 @@ int ubifs_prepare_auth_node(struct ubifs_info *c, void *node, struct shash_desc *inhash) { struct ubifs_auth_node *auth = node; - u8 *hash; + u8 hash[UBIFS_HASH_ARR_SZ]; int err;
- hash = kmalloc(crypto_shash_descsize(c->hash_tfm), GFP_NOFS); - if (!hash) - return -ENOMEM; - { SHASH_DESC_ON_STACK(hash_desc, c->hash_tfm);
@@ -94,21 +90,16 @@ int ubifs_prepare_auth_node(struct ubifs_info *c, void *node,
err = crypto_shash_final(hash_desc, hash); if (err) - goto out; + return err; }
err = ubifs_hash_calc_hmac(c, hash, auth->hmac); if (err) - goto out; + return err;
auth->ch.node_type = UBIFS_AUTH_NODE; ubifs_prepare_node(c, auth, ubifs_auth_node_sz(c), 0); - - err = 0; -out: - kfree(hash); - - return err; + return 0; }
static struct shash_desc *ubifs_get_desc(const struct ubifs_info *c, diff --git a/fs/ubifs/replay.c b/fs/ubifs/replay.c index b28ac4dfb4070a..01fcf79750472b 100644 --- a/fs/ubifs/replay.c +++ b/fs/ubifs/replay.c @@ -601,18 +601,12 @@ static int authenticate_sleb(struct ubifs_info *c, struct ubifs_scan_leb *sleb, struct ubifs_scan_node *snod; int n_nodes = 0; int err; - u8 *hash, *hmac; + u8 hash[UBIFS_HASH_ARR_SZ]; + u8 hmac[UBIFS_HMAC_ARR_SZ];
if (!ubifs_authenticated(c)) return sleb->nodes_cnt;
- hash = kmalloc(crypto_shash_descsize(c->hash_tfm), GFP_NOFS); - hmac = kmalloc(c->hmac_desc_len, GFP_NOFS); - if (!hash || !hmac) { - err = -ENOMEM; - goto out; - } - list_for_each_entry(snod, &sleb->nodes, list) {
n_nodes++; @@ -662,9 +656,6 @@ static int authenticate_sleb(struct ubifs_info *c, struct ubifs_scan_leb *sleb, err = 0; } out: - kfree(hash); - kfree(hmac); - return err ? err : n_nodes - n_not_auth; }
On Fri, May 01, 2020 at 10:59:45PM -0700, Eric Biggers wrote:
From: Eric Biggers ebiggers@google.com
crypto_shash_descsize() returns the size of the shash_desc context needed to compute the hash, not the size of the hash itself.
crypto_shash_digestsize() would be correct, or alternatively using c->hash_len and c->hmac_desc_len which already store the correct values. But actually it's simpler to just use stack arrays, so do that instead.
Fixes: 49525e5eecca ("ubifs: Add helper functions for authentication support") Fixes: da8ef65f9573 ("ubifs: Authenticate replayed journal") Cc: stable@vger.kernel.org # v4.20+ Cc: Sascha Hauer s.hauer@pengutronix.de Signed-off-by: Eric Biggers ebiggers@google.com
Looks better that way, thanks.
Acked-by: Sascha Hauer s.hauer@pengutronix.de
Sascha
On Mon, May 04, 2020 at 09:16:44AM +0200, Sascha Hauer wrote:
On Fri, May 01, 2020 at 10:59:45PM -0700, Eric Biggers wrote:
From: Eric Biggers ebiggers@google.com
crypto_shash_descsize() returns the size of the shash_desc context needed to compute the hash, not the size of the hash itself.
crypto_shash_digestsize() would be correct, or alternatively using c->hash_len and c->hmac_desc_len which already store the correct values. But actually it's simpler to just use stack arrays, so do that instead.
Fixes: 49525e5eecca ("ubifs: Add helper functions for authentication support") Fixes: da8ef65f9573 ("ubifs: Authenticate replayed journal") Cc: stable@vger.kernel.org # v4.20+ Cc: Sascha Hauer s.hauer@pengutronix.de Signed-off-by: Eric Biggers ebiggers@google.com
Looks better that way, thanks.
Acked-by: Sascha Hauer s.hauer@pengutronix.de
Richard, could you take this through the ubifs tree for 5.8?
- Eric
----- Ursprüngliche Mail -----
Von: "Eric Biggers" ebiggers@kernel.org An: "Sascha Hauer" s.hauer@pengutronix.de, "richard" richard@nod.at CC: "linux-mtd" linux-mtd@lists.infradead.org, "Linux Crypto Mailing List" linux-crypto@vger.kernel.org, "stable" stable@vger.kernel.org Gesendet: Freitag, 15. Mai 2020 21:17:04 Betreff: Re: [PATCH] ubifs: fix wrong use of crypto_shash_descsize()
On Mon, May 04, 2020 at 09:16:44AM +0200, Sascha Hauer wrote:
On Fri, May 01, 2020 at 10:59:45PM -0700, Eric Biggers wrote:
From: Eric Biggers ebiggers@google.com
crypto_shash_descsize() returns the size of the shash_desc context needed to compute the hash, not the size of the hash itself.
crypto_shash_digestsize() would be correct, or alternatively using c->hash_len and c->hmac_desc_len which already store the correct values. But actually it's simpler to just use stack arrays, so do that instead.
Fixes: 49525e5eecca ("ubifs: Add helper functions for authentication support") Fixes: da8ef65f9573 ("ubifs: Authenticate replayed journal") Cc: stable@vger.kernel.org # v4.20+ Cc: Sascha Hauer s.hauer@pengutronix.de Signed-off-by: Eric Biggers ebiggers@google.com
Looks better that way, thanks.
Acked-by: Sascha Hauer s.hauer@pengutronix.de
Richard, could you take this through the ubifs tree for 5.8?
Sure. I actually will send a PR with various MTD related fixes for 5.7.
Thanks, //richard
On Fri, May 15, 2020 at 10:50 PM Richard Weinberger richard@nod.at wrote:
----- Ursprüngliche Mail -----
Von: "Eric Biggers" ebiggers@kernel.org An: "Sascha Hauer" s.hauer@pengutronix.de, "richard" richard@nod.at CC: "linux-mtd" linux-mtd@lists.infradead.org, "Linux Crypto Mailing List" linux-crypto@vger.kernel.org, "stable" stable@vger.kernel.org Gesendet: Freitag, 15. Mai 2020 21:17:04 Betreff: Re: [PATCH] ubifs: fix wrong use of crypto_shash_descsize()
On Mon, May 04, 2020 at 09:16:44AM +0200, Sascha Hauer wrote:
On Fri, May 01, 2020 at 10:59:45PM -0700, Eric Biggers wrote:
From: Eric Biggers ebiggers@google.com
crypto_shash_descsize() returns the size of the shash_desc context needed to compute the hash, not the size of the hash itself.
crypto_shash_digestsize() would be correct, or alternatively using c->hash_len and c->hmac_desc_len which already store the correct values. But actually it's simpler to just use stack arrays, so do that instead.
Fixes: 49525e5eecca ("ubifs: Add helper functions for authentication support") Fixes: da8ef65f9573 ("ubifs: Authenticate replayed journal") Cc: stable@vger.kernel.org # v4.20+ Cc: Sascha Hauer s.hauer@pengutronix.de Signed-off-by: Eric Biggers ebiggers@google.com
Looks better that way, thanks.
Acked-by: Sascha Hauer s.hauer@pengutronix.de
Richard, could you take this through the ubifs tree for 5.8?
Sure. I actually will send a PR with various MTD related fixes for 5.7.
Applied. Thanks for fixing!
linux-stable-mirror@lists.linaro.org