[PATCH] scsi: iscsi_tcp: Fix use-after-free when do get_host_param

25 views
Skip to first unread message

Wu Bo

unread,
Mar 20, 2021, 4:50:25 AM3/20/21
to ldu...@suse.com, cle...@redhat.com, je...@linux.ibm.com, martin....@oracle.com, mich...@cs.wisc.edu, James.B...@suse.de, linux...@vger.kernel.org, open-...@googlegroups.com, linux-...@vger.kernel.org, linfe...@huawei.com, wub...@huawei.com, haowe...@huawei.com
When logout of iscsi session, to do destroy session process,
tcp_sw_host->session is not set to NULL.
Get host parameters access to tcp_sw_host->session at the same time,
but the session has been released at this time.

[29844.848044] sd 2:0:0:1: [sdj] Synchronizing SCSI cache
[29844.923745] scsi 2:0:0:1: alua: Detached
[29844.927840] ==================================================================
[29844.927861] BUG: KASAN: use-after-free in iscsi_sw_tcp_host_get_param+0xf4/0x218 [iscsi_tcp]
[29844.927864] Read of size 8 at addr ffff80002c0b8f68 by task iscsiadm/523945
[29844.927871] CPU: 1 PID: 523945 Comm: iscsiadm Kdump: loaded Not tainted 4.19.90.kasan.aarch64
[29844.927873] Hardware name: QEMU KVM Virtual Machine, BIOS 0.0.0 02/06/2015
[29844.927875] Call trace:
[29844.927884] dump_backtrace+0x0/0x270
[29844.927886] show_stack+0x24/0x30
[29844.927895] dump_stack+0xc4/0x120
[29844.927902] print_address_description+0x68/0x278
[29844.927904] kasan_report+0x20c/0x338
[29844.927906] __asan_load8+0x88/0xb0
[29844.927910] iscsi_sw_tcp_host_get_param+0xf4/0x218 [iscsi_tcp]
[29844.927932] show_host_param_ISCSI_HOST_PARAM_IPADDRESS+0x84/0xa0 [scsi_transport_iscsi]
[29844.927938] dev_attr_show+0x48/0x90
[29844.927943] sysfs_kf_seq_show+0x100/0x1e0
[29844.927946] kernfs_seq_show+0x88/0xa0
[29844.927949] seq_read+0x164/0x748
[29844.927951] kernfs_fop_read+0x204/0x308
[29844.927956] __vfs_read+0xd4/0x2d8
[29844.927958] vfs_read+0xa8/0x198
[29844.927960] ksys_read+0xd0/0x180
[29844.927962] __arm64_sys_read+0x4c/0x60
[29844.927966] el0_svc_common+0xa8/0x230
[29844.927969] el0_svc_handler+0xdc/0x138
[29844.927971] el0_svc+0x10/0x218
[29844.928063] Freed by task 53358:
[29844.928066] __kasan_slab_free+0x120/0x228
[29844.928068] kasan_slab_free+0x10/0x18
[29844.928069] kfree+0x98/0x278
[29844.928083] iscsi_session_release+0x84/0xa0 [scsi_transport_iscsi]
[29844.928085] device_release+0x4c/0x100
[29844.928089] kobject_put+0xc4/0x288
[29844.928091] put_device+0x24/0x30
[29844.928105] iscsi_free_session+0x60/0x70 [scsi_transport_iscsi]
[29844.928112] iscsi_session_teardown+0x134/0x158 [libiscsi]
[29844.928116] iscsi_sw_tcp_session_destroy+0x7c/0xd8 [iscsi_tcp]
[29844.928129] iscsi_if_rx+0x1538/0x1f00 [scsi_transport_iscsi]
[29844.928131] netlink_unicast+0x338/0x3c8
[29844.928133] netlink_sendmsg+0x51c/0x588
[29844.928135] sock_sendmsg+0x74/0x98
[29844.928137] ___sys_sendmsg+0x434/0x470
[29844.928139] __sys_sendmsg+0xd4/0x148
[29844.928141] __arm64_sys_sendmsg+0x50/0x60
[29844.928143] el0_svc_common+0xa8/0x230
[29844.928146] el0_svc_handler+0xdc/0x138
[29844.928147] el0_svc+0x10/0x218
[29844.928148]
[29844.928150] The buggy address belongs to the object at ffff80002c0b8880#012 which belongs to the cache kmalloc-2048 of size 2048
[29844.928153] The buggy address is located 1768 bytes inside of#012 2048-byte region [ffff80002c0b8880, ffff80002c0b9080)
[29844.928154] The buggy address belongs to the page:
[29844.928158] page:ffff7e0000b02e00 count:1 mapcount:0 mapping:ffff8000d8402600 index:0x0 compound_mapcount: 0
[29844.928902] flags: 0x7fffe0000008100(slab|head)
[29844.929215] raw: 07fffe0000008100 ffff7e0003535e08 ffff7e00024a9408 ffff8000d8402600
[29844.929217] raw: 0000000000000000 00000000000f000f 00000001ffffffff 0000000000000000
[29844.929219] page dumped because: kasan: bad access detected
[29844.929219]
[29844.929221] Memory state around the buggy address:
[29844.929223] ffff80002c0b8e00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[29844.929225] ffff80002c0b8e80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[29844.929227] >ffff80002c0b8f00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[29844.929228] ^
[29844.929230] ffff80002c0b8f80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[29844.929232] ffff80002c0b9000: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[29844.929232] ==================================================================
[29844.929234] Disabling lock debugging due to kernel taint
[29844.969534] scsi host2: iSCSI Initiator over TCP/IP

Fixes: a79af8a64d39 ("[SCSI] iscsi_tcp: use iscsi_conn_get_addr_param libiscsi function")
Signed-off-by: Wu Bo <wub...@huawei.com>
Signed-off-by: WenChao Hao <haowe...@huawei.com>
---
drivers/scsi/iscsi_tcp.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/drivers/scsi/iscsi_tcp.c b/drivers/scsi/iscsi_tcp.c
index dd33ce0..98d782d 100644
--- a/drivers/scsi/iscsi_tcp.c
+++ b/drivers/scsi/iscsi_tcp.c
@@ -901,10 +901,12 @@ static void iscsi_sw_tcp_session_destroy(struct iscsi_cls_session *cls_session)
{
struct Scsi_Host *shost = iscsi_session_to_shost(cls_session);
struct iscsi_session *session = cls_session->dd_data;
+ struct iscsi_sw_tcp_host *tcp_sw_host = iscsi_host_priv(shost);

if (WARN_ON_ONCE(session->leadconn))
return;

+ tcp_sw_host->session = NULL;
iscsi_tcp_r2tpool_free(cls_session->dd_data);
iscsi_session_teardown(cls_session);

--
1.8.3.1

Wu Bo

unread,
Mar 21, 2021, 1:15:38 AM3/21/21
to ldu...@suse.com, cle...@redhat.com, je...@linux.ibm.com, martin....@oracle.com, mich...@cs.wisc.edu, James.B...@suse.de, linux...@vger.kernel.org, open-...@googlegroups.com, linux-...@vger.kernel.org, linfe...@huawei.com, haowe...@huawei.com
Please ignore this patch, it cannot completely fix the use-after-free
issuss.


Wu Bo

unread,
Mar 21, 2021, 2:29:31 AM3/21/21
to ldu...@suse.com, cle...@redhat.com, je...@linux.ibm.com, martin....@oracle.com, mich...@cs.wisc.edu, James.B...@suse.de, linux...@vger.kernel.org, open-...@googlegroups.com, linux-...@vger.kernel.org, linfe...@huawei.com, wub...@huawei.com, haowe...@huawei.com
From: Wu Bo <wub...@huawei.com>

iscsid(cpu1): Logout of iscsi session, will do destroy session,
tcp_sw_host->session is not set to NULL before release the iscsi session.
in the iscsi_sw_tcp_session_destroy().

iscsadm(cpu2): Get host parameters access to tcp_sw_host->session in the
iscsi_sw_tcp_host_get_param(), tcp_sw_host->session is not NULL,
but pointed to a freed space.

Add ihost->lock and kref to protect the session,
between get host parameters and destroy iscsi session.
drivers/scsi/iscsi_tcp.c | 23 +++++++++++++++++++++--
1 file changed, 21 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/iscsi_tcp.c b/drivers/scsi/iscsi_tcp.c
index 93ce990..579aa80 100644
--- a/drivers/scsi/iscsi_tcp.c
+++ b/drivers/scsi/iscsi_tcp.c
@@ -783,22 +783,32 @@ static int iscsi_sw_tcp_host_get_param(struct Scsi_Host *shost,
enum iscsi_host_param param, char *buf)
{
struct iscsi_sw_tcp_host *tcp_sw_host = iscsi_host_priv(shost);
- struct iscsi_session *session = tcp_sw_host->session;
+ struct iscsi_session *session;
+ struct iscsi_host *ihost = shost_priv(shost);
struct iscsi_conn *conn;
struct iscsi_tcp_conn *tcp_conn;
struct iscsi_sw_tcp_conn *tcp_sw_conn;
struct sockaddr_in6 addr;
+ unsigned long flags;
int rc;

switch (param) {
case ISCSI_HOST_PARAM_IPADDRESS:
- if (!session)
+ spin_lock_irqsave(&ihost->lock, flags);
+ session = tcp_sw_host->session;
+ if (!session) {
+ spin_unlock_irqrestore(&ihost->lock, flags);
return -ENOTCONN;
+ }
+
+ get_device(&(session->cls_session->dev));
+ spin_unlock_irqrestore(&ihost->lock, flags);

spin_lock_bh(&session->frwd_lock);
conn = session->leadconn;
if (!conn) {
spin_unlock_bh(&session->frwd_lock);
+ put_device(&(session->cls_session->dev));
return -ENOTCONN;
}
tcp_conn = conn->dd_data;
@@ -806,12 +816,14 @@ static int iscsi_sw_tcp_host_get_param(struct Scsi_Host *shost,
tcp_sw_conn = tcp_conn->dd_data;
if (!tcp_sw_conn->sock) {
spin_unlock_bh(&session->frwd_lock);
+ put_device(&(session->cls_session->dev));
return -ENOTCONN;
}

rc = kernel_getsockname(tcp_sw_conn->sock,
(struct sockaddr *)&addr);
spin_unlock_bh(&session->frwd_lock);
+ put_device(&(session->cls_session->dev));
if (rc < 0)
return rc;

@@ -901,10 +913,17 @@ static void iscsi_sw_tcp_session_destroy(struct iscsi_cls_session *cls_session)
{
struct Scsi_Host *shost = iscsi_session_to_shost(cls_session);
struct iscsi_session *session = cls_session->dd_data;
+ struct iscsi_sw_tcp_host *tcp_sw_host = iscsi_host_priv(shost);
+ struct iscsi_host *ihost = shost_priv(shost);
+ unsigned long flags;

if (WARN_ON_ONCE(session->leadconn))
return;

+ spin_lock_irqsave(&ihost->lock, flags);
+ tcp_sw_host->session = NULL;
+ spin_unlock_irqrestore(&ihost->lock, flags);
+

michael....@oracle.com

unread,
Mar 21, 2021, 4:21:29 PM3/21/21
to Wu Bo, ldu...@suse.com, cle...@redhat.com, je...@linux.ibm.com, martin....@oracle.com, mich...@cs.wisc.edu, James.B...@suse.de, linux...@vger.kernel.org, open-...@googlegroups.com, linux-...@vger.kernel.org, linfe...@huawei.com, haowe...@huawei.com
We are tearing down the structs in the wrong order. I think sysfs removal
functions will wait for users accessing the object, so we can do:

1. remove session from sysfs (iscsi_remove_session)
2. remove host from syfs (iscsi_host_remove)

At this point we userspace is not accessing the host and session structs so
we can start to tear them down.

3. free session: iscsi_tcp_r2tpool_free, a modified iscsi_session_teardown
that only does iscsi_free_session instead of iscsi_destroy_session.
4. free host (iscsi_host_free).

Before the device_del function waited for userspace to release refcounts
for sysfs accesses we could have also moved some of thise to a release function.

Wu Bo

unread,
Mar 22, 2021, 9:19:26 PM3/22/21
to open-...@googlegroups.com, michael....@oracle.com, ldu...@suse.com, cle...@redhat.com, je...@linux.ibm.com, martin....@oracle.com, mich...@cs.wisc.edu, James.B...@suse.de, linux...@vger.kernel.org, linux-...@vger.kernel.org, linfe...@huawei.com, haowe...@huawei.com
Hi,

Thank you for your suggestion, I will try to modify tearing down the
structs order and send the V3 version after testing.

Thanks.


Reply all
Reply to author
Forward
0 new messages