From: Zhong Jinghua <[email protected]>
A use-after-free problem like below:
BUG: KASAN: use-after-free in scsi_target_reap+0x6c/0x70
Workqueue: scsi_wq_1 __iscsi_unbind_session [scsi_transport_iscsi]
Call trace:
dump_backtrace+0x0/0x320
show_stack+0x24/0x30
dump_stack+0xdc/0x128
print_address_description+0x68/0x278
kasan_report+0x1e4/0x308
__asan_report_load4_noabort+0x30/0x40
scsi_target_reap+0x6c/0x70
scsi_remove_target+0x430/0x640
__iscsi_unbind_session+0x164/0x268 [scsi_transport_iscsi]
process_one_work+0x67c/0x1350
worker_thread+0x370/0xf90
kthread+0x2a4/0x320
ret_from_fork+0x10/0x18
The problem is caused by a concurrency scenario:
T0: delete target
// echo 1 > /sys/devices/platform/host1/session1/target1:0:0/1:0:0:1/delete
T1: logout
// iscsiadm -m node --logout
T0 T1
sdev_store_delete
scsi_remove_device
device_remove_file
__scsi_remove_device
__iscsi_unbind_session
scsi_remove_target
spin_lock_irqsave
list_for_each_entry
scsi_target_reap
// starget->reap_ref 1 -> 0
kref_get(&starget->reap_ref);
// warn use-after-free.
spin_unlock_irqrestore
scsi_target_reap_ref_release
scsi_target_destroy
... // delete starget
scsi_target_reap
// UAF
When T0 reduces the reference count to 0, but has not been released,
T1 can still enter list_for_each_entry, and then kref_get reports UAF.
Fix it by using kref_get_unless_zero() to check for a reference count of
0.
Signed-off-by: Zhong Jinghua <[email protected]>
---
v2: commit message: "starget->reaf" -> "starget->reap_ref"
comment: "If it is reduced to 0, it means that other processes are releasing it and there is no need to delete it again"
->
"If the reference count is already zero, skip this target is safe because scsi_target_destroy() will wait until the
host lock has been released before freeing starget."
Resend: use .huaweicloud mailbox to send
drivers/scsi/scsi_sysfs.c | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/drivers/scsi/scsi_sysfs.c b/drivers/scsi/scsi_sysfs.c
index e7893835b99a..12e8ed6d55cb 100644
--- a/drivers/scsi/scsi_sysfs.c
+++ b/drivers/scsi/scsi_sysfs.c
@@ -1561,7 +1561,16 @@ void scsi_remove_target(struct device *dev)
starget->state == STARGET_CREATED_REMOVE)
continue;
if (starget->dev.parent == dev || &starget->dev == dev) {
- kref_get(&starget->reap_ref);
+
+ /*
+ * If the reference count is already zero, skip this
+ * target is safe because scsi_target_destroy()
+ * will wait until the host lock has been released
+ * before freeing starget.
+ */
+ if (!kref_get_unless_zero(&starget->reap_ref))
+ continue;
+
if (starget->state == STARGET_CREATED)
starget->state = STARGET_CREATED_REMOVE;
else
--
2.31.1
On 3/6/23 04:16, Zhong Jinghua wrote:
> + /*
> + * If the reference count is already zero, skip this
> + * target is safe because scsi_target_destroy()
> + * will wait until the host lock has been released
> + * before freeing starget.
> + */
The above comment has grammatical issues and is confusing. I think the
comment that I suggested was much better than the above.
Bart.
Thanks for you adevice.
Jinghua
在 2023/3/8 6:35, Bart Van Assche 写道:
> On 3/6/23 04:16, Zhong Jinghua wrote:
>> + /*
>> + * If the reference count is already zero, skip this
>> + * target is safe because scsi_target_destroy()
>> + * will wait until the host lock has been released
>> + * before freeing starget.
>> + */
>
> The above comment has grammatical issues and is confusing. I think the
> comment that I suggested was much better than the above.
>
> Bart.
>