DPDK patches and discussions
 help / color / mirror / Atom feed
* [dpdk-dev] [PATCH] net/bonding: fix add bonded device itself as its slave
@ 2021-04-15  7:09 Min Hu (Connor)
  2021-04-20  1:37 ` Ferruh Yigit
  0 siblings, 1 reply; 2+ messages in thread
From: Min Hu (Connor) @ 2021-04-15  7:09 UTC (permalink / raw)
  To: dev; +Cc: ferruh.yigit, chas3, humin29

From: Chengchang Tang <tangchengchang@huawei.com>

Adding the bond device as its own slave should be forbidden. This
will cause a recursive endless loop in many subsequent operations,
and eventually lead to coredump.

This problem was found in testpmd, the related logs are as follows:
testpmd> create bonded device 1 0
Created new bonded device net_bonding_testpmd_0 on (port 4).
testpmd> add bonding slave 4 4
Segmentation fault (core dumped)

The call stack is as follows:
0x000000000064eb90 in rte_eth_dev_info_get ()
0x00000000006df4b4 in bond_ethdev_info ()
0x000000000064eb90 in rte_eth_dev_info_get ()
0x00000000006df4b4 in bond_ethdev_info ()
0x000000000064eb90 in rte_eth_dev_info_get ()
0x0000000000564e58 in eth_dev_info_get_print_err ()
0x000000000055e8a4 in init_port_config ()
0x000000000052730c in cmd_add_bonding_slave_parsed ()
0x0000000000646f60 in cmdline_parse ()
0x0000000000645e08 in cmdline_valid_buffer ()
0x000000000064956c in rdline_char_in ()
0x0000000000645ee0 in cmdline_in ()
0x00000000006460a4 in cmdline_interact ()
0x0000000000531904 in prompt ()
0x000000000051cca8 in main ()

Fixes: 2efb58cbab6e ("bond: new link bonding library")
Cc: stable@dpdk.org

Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
---
 drivers/net/bonding/eth_bond_private.h |  2 +-
 drivers/net/bonding/rte_eth_bond_api.c | 26 +++++++++++++++++---------
 2 files changed, 18 insertions(+), 10 deletions(-)

diff --git a/drivers/net/bonding/eth_bond_private.h b/drivers/net/bonding/eth_bond_private.h
index 75fb8dc..fc179a2 100644
--- a/drivers/net/bonding/eth_bond_private.h
+++ b/drivers/net/bonding/eth_bond_private.h
@@ -212,7 +212,7 @@ int
 valid_bonded_port_id(uint16_t port_id);
 
 int
-valid_slave_port_id(uint16_t port_id, uint8_t mode);
+valid_slave_port_id(struct bond_dev_private *internals, uint16_t port_id);
 
 void
 deactivate_slave(struct rte_eth_dev *eth_dev, uint16_t port_id);
diff --git a/drivers/net/bonding/rte_eth_bond_api.c b/drivers/net/bonding/rte_eth_bond_api.c
index 17e6ff8..eb8d15d 100644
--- a/drivers/net/bonding/rte_eth_bond_api.c
+++ b/drivers/net/bonding/rte_eth_bond_api.c
@@ -56,19 +56,25 @@ check_for_master_bonded_ethdev(const struct rte_eth_dev *eth_dev)
 }
 
 int
-valid_slave_port_id(uint16_t port_id, uint8_t mode)
+valid_slave_port_id(struct bond_dev_private *internals, uint16_t slave_port_id)
 {
-	RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -1);
+	RTE_ETH_VALID_PORTID_OR_ERR_RET(slave_port_id, -1);
 
-	/* Verify that port_id refers to a non bonded port */
-	if (check_for_bonded_ethdev(&rte_eth_devices[port_id]) == 0 &&
-			mode == BONDING_MODE_8023AD) {
+	/* Verify that slave_port_id refers to a non bonded port */
+	if (check_for_bonded_ethdev(&rte_eth_devices[slave_port_id]) == 0 &&
+			internals->mode == BONDING_MODE_8023AD) {
 		RTE_BOND_LOG(ERR, "Cannot add slave to bonded device in 802.3ad"
 				" mode as slave is also a bonded device, only "
 				"physical devices can be support in this mode.");
 		return -1;
 	}
 
+	if (internals->port_id == slave_port_id) {
+		RTE_BOND_LOG(ERR,
+			"Cannot add the bonded device itself as its slave.");
+		return -1;
+	}
+
 	return 0;
 }
 
@@ -456,7 +462,7 @@ __eth_bond_slave_add_lock_free(uint16_t bonded_port_id, uint16_t slave_port_id)
 	bonded_eth_dev = &rte_eth_devices[bonded_port_id];
 	internals = bonded_eth_dev->data->dev_private;
 
-	if (valid_slave_port_id(slave_port_id, internals->mode) != 0)
+	if (valid_slave_port_id(internals, slave_port_id) != 0)
 		return -1;
 
 	slave_eth_dev = &rte_eth_devices[slave_port_id];
@@ -605,13 +611,15 @@ rte_eth_bond_slave_add(uint16_t bonded_port_id, uint16_t slave_port_id)
 
 	int retval;
 
-	/* Verify that port id's are valid bonded and slave ports */
 	if (valid_bonded_port_id(bonded_port_id) != 0)
 		return -1;
 
 	bonded_eth_dev = &rte_eth_devices[bonded_port_id];
 	internals = bonded_eth_dev->data->dev_private;
 
+	if (valid_slave_port_id(internals, slave_port_id) != 0)
+		return -1;
+
 	rte_spinlock_lock(&internals->lock);
 
 	retval = __eth_bond_slave_add_lock_free(bonded_port_id, slave_port_id);
@@ -635,7 +643,7 @@ __eth_bond_slave_remove_lock_free(uint16_t bonded_port_id,
 	bonded_eth_dev = &rte_eth_devices[bonded_port_id];
 	internals = bonded_eth_dev->data->dev_private;
 
-	if (valid_slave_port_id(slave_port_id, internals->mode) < 0)
+	if (valid_slave_port_id(internals, slave_port_id) < 0)
 		return -1;
 
 	/* first remove from active slave list */
@@ -783,7 +791,7 @@ rte_eth_bond_primary_set(uint16_t bonded_port_id, uint16_t slave_port_id)
 
 	internals = rte_eth_devices[bonded_port_id].data->dev_private;
 
-	if (valid_slave_port_id(slave_port_id, internals->mode) != 0)
+	if (valid_slave_port_id(internals, slave_port_id) != 0)
 		return -1;
 
 	internals->user_defined_primary_port = 1;
-- 
2.7.4


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [dpdk-dev] [PATCH] net/bonding: fix add bonded device itself as its slave
  2021-04-15  7:09 [dpdk-dev] [PATCH] net/bonding: fix add bonded device itself as its slave Min Hu (Connor)
@ 2021-04-20  1:37 ` Ferruh Yigit
  0 siblings, 0 replies; 2+ messages in thread
From: Ferruh Yigit @ 2021-04-20  1:37 UTC (permalink / raw)
  To: Min Hu (Connor), dev; +Cc: chas3

On 4/15/2021 8:09 AM, Min Hu (Connor) wrote:
> From: Chengchang Tang <tangchengchang@huawei.com>
> 
> Adding the bond device as its own slave should be forbidden. This
> will cause a recursive endless loop in many subsequent operations,
> and eventually lead to coredump.
> 
> This problem was found in testpmd, the related logs are as follows:
> testpmd> create bonded device 1 0
> Created new bonded device net_bonding_testpmd_0 on (port 4).
> testpmd> add bonding slave 4 4
> Segmentation fault (core dumped)
> 
> The call stack is as follows:
> 0x000000000064eb90 in rte_eth_dev_info_get ()
> 0x00000000006df4b4 in bond_ethdev_info ()
> 0x000000000064eb90 in rte_eth_dev_info_get ()
> 0x00000000006df4b4 in bond_ethdev_info ()
> 0x000000000064eb90 in rte_eth_dev_info_get ()
> 0x0000000000564e58 in eth_dev_info_get_print_err ()
> 0x000000000055e8a4 in init_port_config ()
> 0x000000000052730c in cmd_add_bonding_slave_parsed ()
> 0x0000000000646f60 in cmdline_parse ()
> 0x0000000000645e08 in cmdline_valid_buffer ()
> 0x000000000064956c in rdline_char_in ()
> 0x0000000000645ee0 in cmdline_in ()
> 0x00000000006460a4 in cmdline_interact ()
> 0x0000000000531904 in prompt ()
> 0x000000000051cca8 in main ()
> 
> Fixes: 2efb58cbab6e ("bond: new link bonding library")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
> Signed-off-by: Min Hu (Connor) <humin29@huawei.com>

Applied to dpdk-next-net/main, thanks.


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2021-04-20  1:38 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-04-15  7:09 [dpdk-dev] [PATCH] net/bonding: fix add bonded device itself as its slave Min Hu (Connor)
2021-04-20  1:37 ` Ferruh Yigit

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).