[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1490862594-2712-3-git-send-email-nab@linux-iscsi.org>
Date: Thu, 30 Mar 2017 08:29:54 +0000
From: "Nicholas A. Bellinger" <nab@...ux-iscsi.org>
To: target-devel <target-devel@...r.kernel.org>
Cc: linux-scsi <linux-scsi@...r.kernel.org>,
lkml <linux-kernel@...r.kernel.org>,
Rob Millner <rlm@...erainc.com>, Chu Yuan Lin <cyl@...era.io>,
James Shen <jcs@...era.io>,
Nicholas Bellinger <nab@...ux-iscsi.org>
Subject: [PATCH 2/2] target: Avoid mappedlun symlink creation during lun shutdown
From: Nicholas Bellinger <nab@...ux-iscsi.org>
This patch closes a race between se_lun deletion during configfs
unlink in target_fabric_port_unlink() -> core_dev_del_lun()
-> core_tpg_remove_lun(), when transport_clear_lun_ref() blocks
waiting for percpu_ref RCU grace period to finish, but a new
NodeACL mappedlun is added before the RCU grace period has
completed.
This can happen in target_fabric_mappedlun_link() because it
only checks for se_lun->lun_se_dev, which is not cleared until
after transport_clear_lun_ref() percpu_ref RCU grace period
finishes.
This bug originally manifested as NULL pointer dereference
OOPsen in target_stat_scsi_att_intr_port_show_attr_dev() on
v4.1.y code, because it dereferences lun->lun_se_dev without
a explicit NULL pointer check.
In post v4.1 code with target-core RCU conversion, the code
in target_stat_scsi_att_intr_port_show_attr_dev() no longer
uses se_lun->lun_se_dev, but the same race still exists.
To address the bug, go ahead and set se_lun>lun_shutdown as
early as possible in core_tpg_remove_lun(), and ensure new
NodeACL mappedlun creation in target_fabric_mappedlun_link()
fails during se_lun shutdown.
Reported-by: James Shen <jcs@...era.io>
Cc: James Shen <jcs@...era.io>
Tested-by: James Shen <jcs@...era.io>
Signed-off-by: Nicholas Bellinger <nab@...ux-iscsi.org>
---
drivers/target/target_core_fabric_configfs.c | 5 +++++
drivers/target/target_core_tpg.c | 4 ++++
include/target/target_core_base.h | 1 +
3 files changed, 10 insertions(+)
diff --git a/drivers/target/target_core_fabric_configfs.c b/drivers/target/target_core_fabric_configfs.c
index d8a16ca..d1e6cab 100644
--- a/drivers/target/target_core_fabric_configfs.c
+++ b/drivers/target/target_core_fabric_configfs.c
@@ -92,6 +92,11 @@ static int target_fabric_mappedlun_link(
pr_err("Source se_lun->lun_se_dev does not exist\n");
return -EINVAL;
}
+ if (lun->lun_shutdown) {
+ pr_err("Unable to create mappedlun symlink because"
+ " lun->lun_shutdown=true\n");
+ return -EINVAL;
+ }
se_tpg = lun->lun_tpg;
nacl_ci = &lun_acl_ci->ci_parent->ci_group->cg_item;
diff --git a/drivers/target/target_core_tpg.c b/drivers/target/target_core_tpg.c
index 6fb1919..dfaef4d 100644
--- a/drivers/target/target_core_tpg.c
+++ b/drivers/target/target_core_tpg.c
@@ -642,6 +642,8 @@ void core_tpg_remove_lun(
*/
struct se_device *dev = rcu_dereference_raw(lun->lun_se_dev);
+ lun->lun_shutdown = true;
+
core_clear_lun_from_tpg(lun, tpg);
/*
* Wait for any active I/O references to percpu se_lun->lun_ref to
@@ -663,6 +665,8 @@ void core_tpg_remove_lun(
}
if (!(dev->se_hba->hba_flags & HBA_FLAGS_INTERNAL_USE))
hlist_del_rcu(&lun->link);
+
+ lun->lun_shutdown = false;
mutex_unlock(&tpg->tpg_lun_mutex);
percpu_ref_exit(&lun->lun_ref);
diff --git a/include/target/target_core_base.h b/include/target/target_core_base.h
index 4b784b6..2e28246 100644
--- a/include/target/target_core_base.h
+++ b/include/target/target_core_base.h
@@ -705,6 +705,7 @@ struct se_lun {
u64 unpacked_lun;
#define SE_LUN_LINK_MAGIC 0xffff7771
u32 lun_link_magic;
+ bool lun_shutdown;
bool lun_access_ro;
u32 lun_index;
--
1.9.1
Powered by blists - more mailing lists