The patch below does not apply to the 5.18-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From b954ebba296bb2eb2e38322f17aaa6426934bd7e Mon Sep 17 00:00:00 2001
From: Damien Le Moal <damien.lemoal(a)opensource.wdc.com>
Date: Tue, 12 Apr 2022 20:52:35 +0900
Subject: [PATCH] zonefs: Clear inode information flags on inode creation
Ensure that the i_flags field of struct zonefs_inode_info is cleared to
0 when initializing a zone file inode, avoiding seeing the flag
ZONEFS_ZONE_OPEN being incorrectly set.
Fixes: b5c00e975779 ("zonefs: open/close zone on file open/close")
Cc: <stable(a)vger.kernel.org>
Signed-off-by: Damien Le Moal <damien.lemoal(a)opensource.wdc.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn(a)wdc.com>
Reviewed-by: Chaitanya Kulkarni <kch(a)nvidia.com>
Reviewed-by: Hans Holmberg <hans.holmberg(a)wdc.com>
diff --git a/fs/zonefs/super.c b/fs/zonefs/super.c
index 3614c7834007..75d8dabe0807 100644
--- a/fs/zonefs/super.c
+++ b/fs/zonefs/super.c
@@ -1142,6 +1142,7 @@ static struct inode *zonefs_alloc_inode(struct super_block *sb)
inode_init_once(&zi->i_vnode);
mutex_init(&zi->i_truncate_mutex);
zi->i_wr_refcnt = 0;
+ zi->i_flags = 0;
return &zi->i_vnode;
}
The patch below does not apply to the 5.18-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From 19139539207934aef6335bdef09c9e4bd70d1808 Mon Sep 17 00:00:00 2001
From: Damien Le Moal <damien.lemoal(a)opensource.wdc.com>
Date: Tue, 12 Apr 2022 17:41:37 +0900
Subject: [PATCH] zonefs: Fix management of open zones
The mount option "explicit_open" manages the device open zone
resources to ensure that if an application opens a sequential file for
writing, the file zone can always be written by explicitly opening
the zone and accounting for that state with the s_open_zones counter.
However, if some zones are already open when mounting, the device open
zone resource usage status will be larger than the initial s_open_zones
value of 0. Ensure that this inconsistency does not happen by closing
any sequential zone that is open when mounting.
Furthermore, with ZNS drives, closing an explicitly open zone that has
not been written will change the zone state to "closed", that is, the
zone will remain in an active state. Since this can then cause failures
of explicit open operations on other zones if the drive active zone
resources are exceeded, we need to make sure that the zone is not
active anymore by resetting it instead of closing it. To address this,
zonefs_zone_mgmt() is modified to change a REQ_OP_ZONE_CLOSE request
into a REQ_OP_ZONE_RESET for sequential zones that have not been
written.
Fixes: b5c00e975779 ("zonefs: open/close zone on file open/close")
Cc: <stable(a)vger.kernel.org>
Signed-off-by: Damien Le Moal <damien.lemoal(a)opensource.wdc.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn(a)wdc.com>
Reviewed-by: Hans Holmberg <hans.holmberg(a)wdc.com>
diff --git a/fs/zonefs/super.c b/fs/zonefs/super.c
index 75d8dabe0807..e20e7c841489 100644
--- a/fs/zonefs/super.c
+++ b/fs/zonefs/super.c
@@ -35,6 +35,17 @@ static inline int zonefs_zone_mgmt(struct inode *inode,
lockdep_assert_held(&zi->i_truncate_mutex);
+ /*
+ * With ZNS drives, closing an explicitly open zone that has not been
+ * written will change the zone state to "closed", that is, the zone
+ * will remain active. Since this can then cause failure of explicit
+ * open operation on other zones if the drive active zone resources
+ * are exceeded, make sure that the zone does not remain active by
+ * resetting it.
+ */
+ if (op == REQ_OP_ZONE_CLOSE && !zi->i_wpoffset)
+ op = REQ_OP_ZONE_RESET;
+
trace_zonefs_zone_mgmt(inode, op);
ret = blkdev_zone_mgmt(inode->i_sb->s_bdev, op, zi->i_zsector,
zi->i_zone_size >> SECTOR_SHIFT, GFP_NOFS);
@@ -1294,12 +1305,13 @@ static void zonefs_init_dir_inode(struct inode *parent, struct inode *inode,
inc_nlink(parent);
}
-static void zonefs_init_file_inode(struct inode *inode, struct blk_zone *zone,
- enum zonefs_ztype type)
+static int zonefs_init_file_inode(struct inode *inode, struct blk_zone *zone,
+ enum zonefs_ztype type)
{
struct super_block *sb = inode->i_sb;
struct zonefs_sb_info *sbi = ZONEFS_SB(sb);
struct zonefs_inode_info *zi = ZONEFS_I(inode);
+ int ret = 0;
inode->i_ino = zone->start >> sbi->s_zone_sectors_shift;
inode->i_mode = S_IFREG | sbi->s_perm;
@@ -1324,6 +1336,22 @@ static void zonefs_init_file_inode(struct inode *inode, struct blk_zone *zone,
sb->s_maxbytes = max(zi->i_max_size, sb->s_maxbytes);
sbi->s_blocks += zi->i_max_size >> sb->s_blocksize_bits;
sbi->s_used_blocks += zi->i_wpoffset >> sb->s_blocksize_bits;
+
+ /*
+ * For sequential zones, make sure that any open zone is closed first
+ * to ensure that the initial number of open zones is 0, in sync with
+ * the open zone accounting done when the mount option
+ * ZONEFS_MNTOPT_EXPLICIT_OPEN is used.
+ */
+ if (type == ZONEFS_ZTYPE_SEQ &&
+ (zone->cond == BLK_ZONE_COND_IMP_OPEN ||
+ zone->cond == BLK_ZONE_COND_EXP_OPEN)) {
+ mutex_lock(&zi->i_truncate_mutex);
+ ret = zonefs_zone_mgmt(inode, REQ_OP_ZONE_CLOSE);
+ mutex_unlock(&zi->i_truncate_mutex);
+ }
+
+ return ret;
}
static struct dentry *zonefs_create_inode(struct dentry *parent,
@@ -1333,6 +1361,7 @@ static struct dentry *zonefs_create_inode(struct dentry *parent,
struct inode *dir = d_inode(parent);
struct dentry *dentry;
struct inode *inode;
+ int ret;
dentry = d_alloc_name(parent, name);
if (!dentry)
@@ -1343,10 +1372,16 @@ static struct dentry *zonefs_create_inode(struct dentry *parent,
goto dput;
inode->i_ctime = inode->i_mtime = inode->i_atime = dir->i_ctime;
- if (zone)
- zonefs_init_file_inode(inode, zone, type);
- else
+ if (zone) {
+ ret = zonefs_init_file_inode(inode, zone, type);
+ if (ret) {
+ iput(inode);
+ goto dput;
+ }
+ } else {
zonefs_init_dir_inode(dir, inode, type);
+ }
+
d_add(dentry, inode);
dir->i_size++;
From: Pablo Neira Ayuso <pablo(a)netfilter.org>
commit 520778042ccca019f3ffa136dd0ca565c486cedd upstream.
Since 3e135cd499bf ("netfilter: nft_dynset: dynamic stateful expression
instantiation"), it is possible to attach stateful expressions to set
elements.
cd5125d8f518 ("netfilter: nf_tables: split set destruction in deactivate
and destroy phase") introduces conditional destruction on the object to
accomodate transaction semantics.
nft_expr_init() calls expr->ops->init() first, then check for
NFT_STATEFUL_EXPR, this stills allows to initialize a non-stateful
lookup expressions which points to a set, which might lead to UAF since
the set is not properly detached from the set->binding for this case.
Anyway, this combination is non-sense from nf_tables perspective.
This patch fixes this problem by checking for NFT_STATEFUL_EXPR before
expr->ops->init() is called.
The reporter provides a KASAN splat and a poc reproducer (similar to
those autogenerated by syzbot to report use-after-free errors). It is
unknown to me if they are using syzbot or if they use similar automated
tool to locate the bug that they are reporting.
For the record, this is the KASAN splat.
[ 85.431824] ==================================================================
[ 85.432901] BUG: KASAN: use-after-free in nf_tables_bind_set+0x81b/0xa20
[ 85.433825] Write of size 8 at addr ffff8880286f0e98 by task poc/776
[ 85.434756]
[ 85.434999] CPU: 1 PID: 776 Comm: poc Tainted: G W 5.18.0+ #2
[ 85.436023] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-2 04/01/2014
Fixes: 0b2d8a7b638b ("netfilter: nf_tables: add helper functions for expression handling")
Reported-and-tested-by: Aaron Adams <edg-e(a)nccgroup.com>
Signed-off-by: Pablo Neira Ayuso <pablo(a)netfilter.org>
[Ajay: Regenerated the patch for v4.14.y]
Signed-off-by: Ajay Kaher <akaher(a)vmware.com>
---
net/netfilter/nf_tables_api.c | 16 ++++++++++------
net/netfilter/nft_dynset.c | 3 ---
2 files changed, 10 insertions(+), 9 deletions(-)
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 7c95314..28fc44e 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -1952,23 +1952,27 @@ struct nft_expr *nft_expr_init(const struct nft_ctx *ctx,
err = nf_tables_expr_parse(ctx, nla, &info);
if (err < 0)
- goto err1;
+ goto err_expr_parse;
+
+ err = -EOPNOTSUPP;
+ if (!(info.ops->type->flags & NFT_EXPR_STATEFUL))
+ goto err_expr_stateful;
err = -ENOMEM;
expr = kzalloc(info.ops->size, GFP_KERNEL);
if (expr == NULL)
- goto err2;
+ goto err_expr_stateful;
err = nf_tables_newexpr(ctx, &info, expr);
if (err < 0)
- goto err3;
+ goto err_expr_new;
return expr;
-err3:
+err_expr_new:
kfree(expr);
-err2:
+err_expr_stateful:
module_put(info.ops->type->owner);
-err1:
+err_expr_parse:
return ERR_PTR(err);
}
diff --git a/net/netfilter/nft_dynset.c b/net/netfilter/nft_dynset.c
index 278d765..f8688f9 100644
--- a/net/netfilter/nft_dynset.c
+++ b/net/netfilter/nft_dynset.c
@@ -191,9 +191,6 @@ static int nft_dynset_init(const struct nft_ctx *ctx,
if (IS_ERR(priv->expr))
return PTR_ERR(priv->expr);
- err = -EOPNOTSUPP;
- if (!(priv->expr->ops->type->flags & NFT_EXPR_STATEFUL))
- goto err1;
} else if (set->flags & NFT_SET_EVAL)
return -EINVAL;
--
2.7.4