All of lore.kernel.org
 help / color / mirror / Atom feed
From: <gregkh@linuxfoundation.org>
To: idryomov@gmail.com, gregkh@linuxfoundation.org,
	l.stach@pengutronix.de, sage@redhat.com
Cc: <stable@vger.kernel.org>, <stable-commits@vger.kernel.org>
Subject: Patch "libceph: stop allocating a new cipher on every crypto request" has been added to the 4.9-stable tree
Date: Mon, 23 Jan 2017 17:06:21 +0100	[thread overview]
Message-ID: <148518758122388@kroah.com> (raw)


This is a note to let you know that I've just added the patch titled

    libceph: stop allocating a new cipher on every crypto request

to the 4.9-stable tree which can be found at:
    http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
     libceph-stop-allocating-a-new-cipher-on-every-crypto-request.patch
and it can be found in the queue-4.9 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@vger.kernel.org> know about it.


>From 7af3ea189a9a13f090de51c97f676215dabc1205 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Fri, 2 Dec 2016 16:35:08 +0100
Subject: libceph: stop allocating a new cipher on every crypto request

From: Ilya Dryomov <idryomov@gmail.com>

commit 7af3ea189a9a13f090de51c97f676215dabc1205 upstream.

This is useless and more importantly not allowed on the writeback path,
because crypto_alloc_skcipher() allocates memory with GFP_KERNEL, which
can recurse back into the filesystem:

    kworker/9:3     D ffff92303f318180     0 20732      2 0x00000080
    Workqueue: ceph-msgr ceph_con_workfn [libceph]
     ffff923035dd4480 ffff923038f8a0c0 0000000000000001 000000009eb27318
     ffff92269eb28000 ffff92269eb27338 ffff923036b145ac ffff923035dd4480
     00000000ffffffff ffff923036b145b0 ffffffff951eb4e1 ffff923036b145a8
    Call Trace:
     [<ffffffff951eb4e1>] ? schedule+0x31/0x80
     [<ffffffff951eb77a>] ? schedule_preempt_disabled+0xa/0x10
     [<ffffffff951ed1f4>] ? __mutex_lock_slowpath+0xb4/0x130
     [<ffffffff951ed28b>] ? mutex_lock+0x1b/0x30
     [<ffffffffc0a974b3>] ? xfs_reclaim_inodes_ag+0x233/0x2d0 [xfs]
     [<ffffffff94d92ba5>] ? move_active_pages_to_lru+0x125/0x270
     [<ffffffff94f2b985>] ? radix_tree_gang_lookup_tag+0xc5/0x1c0
     [<ffffffff94dad0f3>] ? __list_lru_walk_one.isra.3+0x33/0x120
     [<ffffffffc0a98331>] ? xfs_reclaim_inodes_nr+0x31/0x40 [xfs]
     [<ffffffff94e05bfe>] ? super_cache_scan+0x17e/0x190
     [<ffffffff94d919f3>] ? shrink_slab.part.38+0x1e3/0x3d0
     [<ffffffff94d9616a>] ? shrink_node+0x10a/0x320
     [<ffffffff94d96474>] ? do_try_to_free_pages+0xf4/0x350
     [<ffffffff94d967ba>] ? try_to_free_pages+0xea/0x1b0
     [<ffffffff94d863bd>] ? __alloc_pages_nodemask+0x61d/0xe60
     [<ffffffff94ddf42d>] ? cache_grow_begin+0x9d/0x560
     [<ffffffff94ddfb88>] ? fallback_alloc+0x148/0x1c0
     [<ffffffff94ed84e7>] ? __crypto_alloc_tfm+0x37/0x130
     [<ffffffff94de09db>] ? __kmalloc+0x1eb/0x580
     [<ffffffffc09fe2db>] ? crush_choose_firstn+0x3eb/0x470 [libceph]
     [<ffffffff94ed84e7>] ? __crypto_alloc_tfm+0x37/0x130
     [<ffffffff94ed9c19>] ? crypto_spawn_tfm+0x39/0x60
     [<ffffffffc08b30a3>] ? crypto_cbc_init_tfm+0x23/0x40 [cbc]
     [<ffffffff94ed857c>] ? __crypto_alloc_tfm+0xcc/0x130
     [<ffffffff94edcc23>] ? crypto_skcipher_init_tfm+0x113/0x180
     [<ffffffff94ed7cc3>] ? crypto_create_tfm+0x43/0xb0
     [<ffffffff94ed83b0>] ? crypto_larval_lookup+0x150/0x150
     [<ffffffff94ed7da2>] ? crypto_alloc_tfm+0x72/0x120
     [<ffffffffc0a01dd7>] ? ceph_aes_encrypt2+0x67/0x400 [libceph]
     [<ffffffffc09fd264>] ? ceph_pg_to_up_acting_osds+0x84/0x5b0 [libceph]
     [<ffffffff950d40a0>] ? release_sock+0x40/0x90
     [<ffffffff95139f94>] ? tcp_recvmsg+0x4b4/0xae0
     [<ffffffffc0a02714>] ? ceph_encrypt2+0x54/0xc0 [libceph]
     [<ffffffffc0a02b4d>] ? ceph_x_encrypt+0x5d/0x90 [libceph]
     [<ffffffffc0a02bdf>] ? calcu_signature+0x5f/0x90 [libceph]
     [<ffffffffc0a02ef5>] ? ceph_x_sign_message+0x35/0x50 [libceph]
     [<ffffffffc09e948c>] ? prepare_write_message_footer+0x5c/0xa0 [libceph]
     [<ffffffffc09ecd18>] ? ceph_con_workfn+0x2258/0x2dd0 [libceph]
     [<ffffffffc09e9903>] ? queue_con_delay+0x33/0xd0 [libceph]
     [<ffffffffc09f68ed>] ? __submit_request+0x20d/0x2f0 [libceph]
     [<ffffffffc09f6ef8>] ? ceph_osdc_start_request+0x28/0x30 [libceph]
     [<ffffffffc0b52603>] ? rbd_queue_workfn+0x2f3/0x350 [rbd]
     [<ffffffff94c94ec0>] ? process_one_work+0x160/0x410
     [<ffffffff94c951bd>] ? worker_thread+0x4d/0x480
     [<ffffffff94c95170>] ? process_one_work+0x410/0x410
     [<ffffffff94c9af8d>] ? kthread+0xcd/0xf0
     [<ffffffff951efb2f>] ? ret_from_fork+0x1f/0x40
     [<ffffffff94c9aec0>] ? kthread_create_on_node+0x190/0x190

Allocating the cipher along with the key fixes the issue - as long the
key doesn't change, a single cipher context can be used concurrently in
multiple requests.

We still can't take that GFP_KERNEL allocation though.  Both
ceph_crypto_key_clone() and ceph_crypto_key_decode() are called from
GFP_NOFS context, so resort to memalloc_noio_{save,restore}() here.

Reported-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Sage Weil <sage@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 net/ceph/crypto.c |   85 ++++++++++++++++++++++++++++++++++++++----------------
 net/ceph/crypto.h |    1 
 2 files changed, 61 insertions(+), 25 deletions(-)

--- a/net/ceph/crypto.c
+++ b/net/ceph/crypto.c
@@ -13,14 +13,60 @@
 #include <linux/ceph/decode.h>
 #include "crypto.h"
 
+/*
+ * Set ->key and ->tfm.  The rest of the key should be filled in before
+ * this function is called.
+ */
+static int set_secret(struct ceph_crypto_key *key, void *buf)
+{
+	unsigned int noio_flag;
+	int ret;
+
+	key->key = NULL;
+	key->tfm = NULL;
+
+	switch (key->type) {
+	case CEPH_CRYPTO_NONE:
+		return 0; /* nothing to do */
+	case CEPH_CRYPTO_AES:
+		break;
+	default:
+		return -ENOTSUPP;
+	}
+
+	WARN_ON(!key->len);
+	key->key = kmemdup(buf, key->len, GFP_NOIO);
+	if (!key->key) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	/* crypto_alloc_skcipher() allocates with GFP_KERNEL */
+	noio_flag = memalloc_noio_save();
+	key->tfm = crypto_alloc_skcipher("cbc(aes)", 0, CRYPTO_ALG_ASYNC);
+	memalloc_noio_restore(noio_flag);
+	if (IS_ERR(key->tfm)) {
+		ret = PTR_ERR(key->tfm);
+		key->tfm = NULL;
+		goto fail;
+	}
+
+	ret = crypto_skcipher_setkey(key->tfm, key->key, key->len);
+	if (ret)
+		goto fail;
+
+	return 0;
+
+fail:
+	ceph_crypto_key_destroy(key);
+	return ret;
+}
+
 int ceph_crypto_key_clone(struct ceph_crypto_key *dst,
 			  const struct ceph_crypto_key *src)
 {
 	memcpy(dst, src, sizeof(struct ceph_crypto_key));
-	dst->key = kmemdup(src->key, src->len, GFP_NOFS);
-	if (!dst->key)
-		return -ENOMEM;
-	return 0;
+	return set_secret(dst, src->key);
 }
 
 int ceph_crypto_key_encode(struct ceph_crypto_key *key, void **p, void *end)
@@ -37,16 +83,16 @@ int ceph_crypto_key_encode(struct ceph_c
 
 int ceph_crypto_key_decode(struct ceph_crypto_key *key, void **p, void *end)
 {
+	int ret;
+
 	ceph_decode_need(p, end, 2*sizeof(u16) + sizeof(key->created), bad);
 	key->type = ceph_decode_16(p);
 	ceph_decode_copy(p, &key->created, sizeof(key->created));
 	key->len = ceph_decode_16(p);
 	ceph_decode_need(p, end, key->len, bad);
-	key->key = kmalloc(key->len, GFP_NOFS);
-	if (!key->key)
-		return -ENOMEM;
-	ceph_decode_copy(p, key->key, key->len);
-	return 0;
+	ret = set_secret(key, *p);
+	*p += key->len;
+	return ret;
 
 bad:
 	dout("failed to decode crypto key\n");
@@ -85,14 +131,11 @@ void ceph_crypto_key_destroy(struct ceph
 	if (key) {
 		kfree(key->key);
 		key->key = NULL;
+		crypto_free_skcipher(key->tfm);
+		key->tfm = NULL;
 	}
 }
 
-static struct crypto_skcipher *ceph_crypto_alloc_cipher(void)
-{
-	return crypto_alloc_skcipher("cbc(aes)", 0, CRYPTO_ALG_ASYNC);
-}
-
 static const u8 *aes_iv = (u8 *)CEPH_AES_IV;
 
 /*
@@ -168,8 +211,7 @@ static void teardown_sgtable(struct sg_t
 static int ceph_aes_crypt(const struct ceph_crypto_key *key, bool encrypt,
 			  void *buf, int buf_len, int in_len, int *pout_len)
 {
-	struct crypto_skcipher *tfm = ceph_crypto_alloc_cipher();
-	SKCIPHER_REQUEST_ON_STACK(req, tfm);
+	SKCIPHER_REQUEST_ON_STACK(req, key->tfm);
 	struct sg_table sgt;
 	struct scatterlist prealloc_sg;
 	char iv[AES_BLOCK_SIZE] __aligned(8);
@@ -177,20 +219,15 @@ static int ceph_aes_crypt(const struct c
 	int crypt_len = encrypt ? in_len + pad_byte : in_len;
 	int ret;
 
-	if (IS_ERR(tfm))
-		return PTR_ERR(tfm);
-
 	WARN_ON(crypt_len > buf_len);
 	if (encrypt)
 		memset(buf + in_len, pad_byte, pad_byte);
 	ret = setup_sgtable(&sgt, &prealloc_sg, buf, crypt_len);
 	if (ret)
-		goto out_tfm;
+		return ret;
 
-	crypto_skcipher_setkey((void *)tfm, key->key, key->len);
 	memcpy(iv, aes_iv, AES_BLOCK_SIZE);
-
-	skcipher_request_set_tfm(req, tfm);
+	skcipher_request_set_tfm(req, key->tfm);
 	skcipher_request_set_callback(req, 0, NULL, NULL);
 	skcipher_request_set_crypt(req, sgt.sgl, sgt.sgl, crypt_len, iv);
 
@@ -232,8 +269,6 @@ static int ceph_aes_crypt(const struct c
 
 out_sgt:
 	teardown_sgtable(&sgt);
-out_tfm:
-	crypto_free_skcipher(tfm);
 	return ret;
 }
 
--- a/net/ceph/crypto.h
+++ b/net/ceph/crypto.h
@@ -12,6 +12,7 @@ struct ceph_crypto_key {
 	struct ceph_timespec created;
 	int len;
 	void *key;
+	struct crypto_skcipher *tfm;
 };
 
 int ceph_crypto_key_clone(struct ceph_crypto_key *dst,


Patches currently in stable-queue which might be from idryomov@gmail.com are

queue-4.9/libceph-switch-ceph_x_decrypt-to-ceph_crypt.patch
queue-4.9/libceph-introduce-ceph_x_encrypt_offset.patch
queue-4.9/ceph-fix-endianness-of-getattr-mask-in-ceph_d_revalidate.patch
queue-4.9/libceph-rename-and-align-ceph_x_authorizer-reply_buf.patch
queue-4.9/libceph-ceph_x_encrypt_buflen-takes-in_len.patch
queue-4.9/ceph-fix-endianness-bug-in-frag_tree_split_cmp.patch
queue-4.9/libceph-uninline-ceph_crypto_key_destroy.patch
queue-4.9/libceph-introduce-ceph_crypt-for-in-place-en-decryption.patch
queue-4.9/libceph-make-sure-ceph_aes_crypt-iv-is-aligned.patch
queue-4.9/ceph-fix-ceph_get_caps-interruption.patch
queue-4.9/libceph-old_key-in-process_one_ticket-is-redundant.patch
queue-4.9/libceph-tweak-calcu_signature-a-little.patch
queue-4.9/libceph-remove-now-unused-ceph_-en-de-crypt-functions.patch
queue-4.9/libceph-stop-allocating-a-new-cipher-on-every-crypto-request.patch
queue-4.9/ceph-fix-bad-endianness-handling-in-parse_reply_info_extra.patch
queue-4.9/libceph-switch-ceph_x_encrypt-to-ceph_crypt.patch

                 reply	other threads:[~2017-01-23 16:06 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=148518758122388@kroah.com \
    --to=gregkh@linuxfoundation.org \
    --cc=idryomov@gmail.com \
    --cc=l.stach@pengutronix.de \
    --cc=sage@redhat.com \
    --cc=stable-commits@vger.kernel.org \
    --cc=stable@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.