diff mbox

[v2,5/6] target/iscsi: Fix a deadlock between the XCOPY code and session shutdown

Message ID 20170510203736.12462-6-bart.vanassche@sandisk.com (mailing list archive)
State New, archived
Headers show

Commit Message

Bart Van Assche May 10, 2017, 8:37 p.m. UTC 
Move the code for parsing an XCOPY command from the context of
the iSCSI receiver thread to the context of the XCOPY workqueue.
Keep the simple XCOPY checks in the context of the iSCSI receiver
thread. Move the code for allocating and freeing struct xcopy_op
from the code that parses an XCOPY command to its caller.

This patch fixes the following deadlock:
diff mbox

Patch

======================================================
[ INFO: possible circular locking dependency detected ]
4.10.0-rc7-dbg+ #1 Not tainted
-------------------------------------------------------
rmdir/13321 is trying to acquire lock:
 (&sess->cmdsn_mutex){+.+.+.}, at: [<ffffffffa02cb47d>] iscsit_free_all_ooo_cmdsns+0x2d/0xb0 [iscsi_target_mod]

but task is already holding lock:
 (&sb->s_type->i_mutex_key#14){++++++}, at: [<ffffffff811c6e20>] vfs_rmdir+0x50/0x140

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:
-> #1 (&sb->s_type->i_mutex_key#14){++++++}:
 lock_acquire+0x71/0x90
 down_write+0x3f/0x70
 configfs_depend_item+0x3a/0xb0 [configfs]
 target_depend_item+0x13/0x20 [target_core_mod]
 target_xcopy_locate_se_dev_e4+0xdd/0x1a0 [target_core_mod]
 target_do_xcopy+0x34b/0x970 [target_core_mod]
 __target_execute_cmd+0x22/0xa0 [target_core_mod]
 target_execute_cmd+0x233/0x2c0 [target_core_mod]
 iscsit_execute_cmd+0x208/0x270 [iscsi_target_mod]
 iscsit_sequence_cmd+0x10b/0x190 [iscsi_target_mod]
 iscsit_get_rx_pdu+0x37d/0xcd0 [iscsi_target_mod]
 iscsi_target_rx_thread+0x6e/0xa0 [iscsi_target_mod]
 kthread+0x102/0x140
 ret_from_fork+0x31/0x40

-> #0 (&sess->cmdsn_mutex){+.+.+.}:
 __lock_acquire+0x10e6/0x1260
 lock_acquire+0x71/0x90
 mutex_lock_nested+0x5f/0x670
 iscsit_free_all_ooo_cmdsns+0x2d/0xb0 [iscsi_target_mod]
 iscsit_close_session+0xac/0x200 [iscsi_target_mod]
 lio_tpg_close_session+0x9f/0xb0 [iscsi_target_mod]
 target_shutdown_sessions+0xc3/0xd0 [target_core_mod]
 core_tpg_del_initiator_node_acl+0x91/0x140 [target_core_mod]
 target_fabric_nacl_base_release+0x20/0x30 [target_core_mod]
 config_item_release+0x5a/0xc0 [configfs]
 config_item_put+0x1d/0x1f [configfs]
 configfs_rmdir+0x1a6/0x300 [configfs]
 vfs_rmdir+0xb7/0x140
 do_rmdir+0x1f4/0x200
 SyS_rmdir+0x11/0x20
 entry_SYSCALL_64_fastpath+0x23/0xc6

other info that might help us debug this:

 Possible unsafe locking scenario:
       CPU0                    CPU1
       ----                    ----
  lock(&sb->s_type->i_mutex_key#14);
                               lock(&sess->cmdsn_mutex);
                               lock(&sb->s_type->i_mutex_key#14);
  lock(&sess->cmdsn_mutex);

 *** DEADLOCK ***

3 locks held by rmdir/13321:
 #0:  (sb_writers#10){.+.+.+}, at: [<ffffffff811e1aff>] mnt_want_write+0x1f/0x50
 #1:  (&default_group_class[depth - 1]#2/1){+.+.+.}, at: [<ffffffff811cc8ce>] do_rmdir+0x15e/0x200
 #2:  (&sb->s_type->i_mutex_key#14){++++++}, at: [<ffffffff811c6e20>] vfs_rmdir+0x50/0x140

stack backtrace:
CPU: 2 PID: 13321 Comm: rmdir Not tainted 4.10.0-rc7-dbg+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.0.0-prebuilt.qemu-project.org 04/01/2014
Call Trace:
 dump_stack+0x86/0xc3
 print_circular_bug+0x1c7/0x220
 __lock_acquire+0x10e6/0x1260
 lock_acquire+0x71/0x90
 mutex_lock_nested+0x5f/0x670
 iscsit_free_all_ooo_cmdsns+0x2d/0xb0 [iscsi_target_mod]
 iscsit_close_session+0xac/0x200 [iscsi_target_mod]
 lio_tpg_close_session+0x9f/0xb0 [iscsi_target_mod]
 target_shutdown_sessions+0xc3/0xd0 [target_core_mod]
 core_tpg_del_initiator_node_acl+0x91/0x140 [target_core_mod]
 target_fabric_nacl_base_release+0x20/0x30 [target_core_mod]
 config_item_release+0x5a/0xc0 [configfs]
 config_item_put+0x1d/0x1f [configfs]
 configfs_rmdir+0x1a6/0x300 [configfs]
 vfs_rmdir+0xb7/0x140
 do_rmdir+0x1f4/0x200
 SyS_rmdir+0x11/0x20
 entry_SYSCALL_64_fastpath+0x23/0xc6

Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Andy Grover <agrover@redhat.com>
Cc: David Disseldorp <ddiss@suse.de>
Cc: <stable@vger.kernel.org>
---
 drivers/target/target_core_xcopy.c | 110 +++++++++++++++++++++++--------------
 1 file changed, 69 insertions(+), 41 deletions(-)

diff --git a/drivers/target/target_core_xcopy.c b/drivers/target/target_core_xcopy.c
index cac5a20a4de0..aecb36d955f0 100644
--- a/drivers/target/target_core_xcopy.c
+++ b/drivers/target/target_core_xcopy.c
@@ -40,6 +40,8 @@ 
 
 static struct workqueue_struct *xcopy_wq = NULL;
 
+static sense_reason_t target_parse_xcopy_cmd(struct xcopy_op *xop);
+
 static int target_xcopy_gen_naa_ieee(struct se_device *dev, unsigned char *buf)
 {
 	int off = 0;
@@ -781,13 +783,24 @@  static int target_xcopy_write_destination(
 static void target_xcopy_do_work(struct work_struct *work)
 {
 	struct xcopy_op *xop = container_of(work, struct xcopy_op, xop_work);
-	struct se_device *src_dev = xop->src_dev, *dst_dev = xop->dst_dev;
 	struct se_cmd *ec_cmd = xop->xop_se_cmd;
-	sector_t src_lba = xop->src_lba, dst_lba = xop->dst_lba, end_lba;
+	struct se_device *src_dev, *dst_dev;
+	sector_t src_lba, dst_lba, end_lba;
 	unsigned int max_sectors;
-	int rc;
-	unsigned short nolb = xop->nolb, cur_nolb, max_nolb, copied_nolb = 0;
+	int rc = 0;
+	unsigned short nolb, cur_nolb, max_nolb, copied_nolb = 0;
+
+	if (target_parse_xcopy_cmd(xop) != TCM_NO_SENSE)
+		goto err_free;
 
+	if (WARN_ON_ONCE(!xop->src_dev) || WARN_ON_ONCE(!xop->dst_dev))
+		goto err_free;
+
+	src_dev = xop->src_dev;
+	dst_dev = xop->dst_dev;
+	src_lba = xop->src_lba;
+	dst_lba = xop->dst_lba;
+	nolb = xop->nolb;
 	end_lba = src_lba + nolb;
 	/*
 	 * Break up XCOPY I/O into hw_max_sectors sized I/O based on the
@@ -855,6 +868,8 @@  static void target_xcopy_do_work(struct work_struct *work)
 
 out:
 	xcopy_pt_undepend_remotedev(xop);
+
+err_free:
 	kfree(xop);
 	/*
 	 * Don't override an error scsi status if it has already been set
@@ -867,48 +882,22 @@  static void target_xcopy_do_work(struct work_struct *work)
 	target_complete_cmd(ec_cmd, ec_cmd->scsi_status);
 }
 
-sense_reason_t target_do_xcopy(struct se_cmd *se_cmd)
+/*
+ * Returns TCM_NO_SENSE upon success or a sense code != TCM_NO_SENSE if parsing
+ * fails.
+ */
+sense_reason_t target_parse_xcopy_cmd(struct xcopy_op *xop)
 {
-	struct se_device *dev = se_cmd->se_dev;
-	struct xcopy_op *xop = NULL;
+	struct se_cmd *se_cmd = xop->xop_se_cmd;
 	unsigned char *p = NULL, *seg_desc;
-	unsigned int list_id, list_id_usage, sdll, inline_dl, sa;
+	unsigned int list_id, list_id_usage, sdll, inline_dl;
 	sense_reason_t ret = TCM_INVALID_PARAMETER_LIST;
 	int rc;
 	unsigned short tdll;
 
-	if (!dev->dev_attrib.emulate_3pc) {
-		pr_err("EXTENDED_COPY operation explicitly disabled\n");
-		return TCM_UNSUPPORTED_SCSI_OPCODE;
-	}
-
-	sa = se_cmd->t_task_cdb[1] & 0x1f;
-	if (sa != 0x00) {
-		pr_err("EXTENDED_COPY(LID4) not supported\n");
-		return TCM_UNSUPPORTED_SCSI_OPCODE;
-	}
-
-	if (se_cmd->data_length == 0) {
-		target_complete_cmd(se_cmd, SAM_STAT_GOOD);
-		return TCM_NO_SENSE;
-	}
-	if (se_cmd->data_length < XCOPY_HDR_LEN) {
-		pr_err("XCOPY parameter truncation: length %u < hdr_len %u\n",
-				se_cmd->data_length, XCOPY_HDR_LEN);
-		return TCM_PARAMETER_LIST_LENGTH_ERROR;
-	}
-
-	xop = kzalloc(sizeof(struct xcopy_op), GFP_KERNEL);
-	if (!xop) {
-		pr_err("Unable to allocate xcopy_op\n");
-		return TCM_OUT_OF_RESOURCES;
-	}
-	xop->xop_se_cmd = se_cmd;
-
 	p = transport_kmap_data_sg(se_cmd);
 	if (!p) {
 		pr_err("transport_kmap_data_sg() failed in target_do_xcopy\n");
-		kfree(xop);
 		return TCM_OUT_OF_RESOURCES;
 	}
 
@@ -977,18 +966,57 @@  sense_reason_t target_do_xcopy(struct se_cmd *se_cmd)
 	pr_debug("XCOPY: Processed %d target descriptors, length: %u\n", rc,
 				rc * XCOPY_TARGET_DESC_LEN);
 	transport_kunmap_data_sg(se_cmd);
-
-	INIT_WORK(&xop->xop_work, target_xcopy_do_work);
-	queue_work(xcopy_wq, &xop->xop_work);
 	return TCM_NO_SENSE;
 
 out:
 	if (p)
 		transport_kunmap_data_sg(se_cmd);
-	kfree(xop);
 	return ret;
 }
 
+sense_reason_t target_do_xcopy(struct se_cmd *se_cmd)
+{
+	struct se_device *dev = se_cmd->se_dev;
+	struct xcopy_op *xop;
+	unsigned int sa;
+
+	if (!dev->dev_attrib.emulate_3pc) {
+		pr_err("EXTENDED_COPY operation explicitly disabled\n");
+		return TCM_UNSUPPORTED_SCSI_OPCODE;
+	}
+
+	sa = se_cmd->t_task_cdb[1] & 0x1f;
+	if (sa != 0x00) {
+		pr_err("EXTENDED_COPY(LID4) not supported\n");
+		return TCM_UNSUPPORTED_SCSI_OPCODE;
+	}
+
+	if (se_cmd->data_length == 0) {
+		target_complete_cmd(se_cmd, SAM_STAT_GOOD);
+		return TCM_NO_SENSE;
+	}
+	if (se_cmd->data_length < XCOPY_HDR_LEN) {
+		pr_err("XCOPY parameter truncation: length %u < hdr_len %u\n",
+				se_cmd->data_length, XCOPY_HDR_LEN);
+		return TCM_PARAMETER_LIST_LENGTH_ERROR;
+	}
+
+	xop = kzalloc(sizeof(struct xcopy_op), GFP_KERNEL);
+	if (!xop)
+		goto err;
+	xop->xop_se_cmd = se_cmd;
+	INIT_WORK(&xop->xop_work, target_xcopy_do_work);
+	if (WARN_ON_ONCE(!queue_work(xcopy_wq, &xop->xop_work)))
+		goto free;
+	return TCM_NO_SENSE;
+
+free:
+	kfree(xop);
+
+err:
+	return TCM_OUT_OF_RESOURCES;
+}
+
 static sense_reason_t target_rcr_operating_parameters(struct se_cmd *se_cmd)
 {
 	unsigned char *p;