* [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup
2019-05-04 9:29 [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Qi Zhang
@ 2019-05-04 9:29 ` Qi Zhang
2019-05-04 9:29 ` [dpdk-dev] [PATCH 1/3] net/i40e: fix invalid Tx threshold setup Qi Zhang
` (3 subsequent siblings)
4 siblings, 0 replies; 12+ messages in thread
From: Qi Zhang @ 2019-05-04 9:29 UTC (permalink / raw)
To: beilei.xing, wenzhuo.lu, qiming.yang, konstantin.ananyev; +Cc: dev, Qi Zhang
When tx_free_thresh + tx_rs_thresh > nb_desc, it is possible
that an outdated DD status be checked as tx_next_dd, then segment fault
happen due to free a NULL mbuf pointer.
The issue usually happens with an aggresive tx_free_thresh, for example:
./testpmd -c 0x3 -n 4 -- -i --rxq=16 --txq=16 --rxd=1024 --txd=1024 --txfreet=1020
The patchset fix this issue on i40e, ixgbe and ice.
Qi Zhang (3):
net/i40e: fix invalid Tx threshold setup
net/ice: fix invalid Tx threshold setup
net/ixgbe: fix invalid Tx threshold setup
drivers/net/i40e/i40e_rxtx.c | 19 +++++++++++++++++--
drivers/net/ice/ice_rxtx.c | 21 ++++++++++++++++++---
drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++--
3 files changed, 52 insertions(+), 7 deletions(-)
--
2.13.6
^ permalink raw reply [flat|nested] 12+ messages in thread* [dpdk-dev] [PATCH 1/3] net/i40e: fix invalid Tx threshold setup
2019-05-04 9:29 [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Qi Zhang
2019-05-04 9:29 ` Qi Zhang
@ 2019-05-04 9:29 ` Qi Zhang
2019-05-04 9:29 ` Qi Zhang
2019-05-04 9:29 ` [dpdk-dev] [PATCH 2/3] net/ice: " Qi Zhang
` (2 subsequent siblings)
4 siblings, 1 reply; 12+ messages in thread
From: Qi Zhang @ 2019-05-04 9:29 UTC (permalink / raw)
To: beilei.xing, wenzhuo.lu, qiming.yang, konstantin.ananyev
Cc: dev, Qi Zhang, stable
Tx desc's DD status is not cleaned by NIC automatically after packets
have been transmitted until software refill a new packet during next
loop. So when tx_free_thresh + tx_rs_thresh > nb_desc, it is possible
that an outdated DD status be checked as tx_next_dd, then segment fault
happen due to free a NULL mbuf pointer.
Then patch fixes this issue by
1. try to adapt tx_rs_thresh to an aggresive tx_free_thresh.
2. queue setup fail when tx_free_thresh + tx_rs_thresh > nb_desc
Fixes: 4861cde46116 ("i40e: new poll mode driver")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
---
drivers/net/i40e/i40e_rxtx.c | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c
index 1489552da..4640a9c66 100644
--- a/drivers/net/i40e/i40e_rxtx.c
+++ b/drivers/net/i40e/i40e_rxtx.c
@@ -2169,15 +2169,30 @@ i40e_dev_tx_queue_setup(struct rte_eth_dev *dev,
* - tx_rs_thresh must be a divisor of the ring size.
* - tx_free_thresh must be greater than 0.
* - tx_free_thresh must be less than the size of the ring minus 3.
+ * - tx_free_thresh + tx_rs_thresh must not exceed nb_desc.
*
* One descriptor in the TX ring is used as a sentinel to avoid a H/W
* race condition, hence the maximum threshold constraints. When set
* to zero use default values.
*/
- tx_rs_thresh = (uint16_t)((tx_conf->tx_rs_thresh) ?
- tx_conf->tx_rs_thresh : DEFAULT_TX_RS_THRESH);
tx_free_thresh = (uint16_t)((tx_conf->tx_free_thresh) ?
tx_conf->tx_free_thresh : DEFAULT_TX_FREE_THRESH);
+ /* force tx_rs_thresh to adapt an aggresive tx_free_thresh */
+ tx_rs_thresh = (DEFAULT_TX_RS_THRESH + tx_free_thresh > nb_desc) ?
+ nb_desc - tx_free_thresh : DEFAULT_TX_RS_THRESH;
+ if (tx_conf->tx_rs_thresh > 0)
+ tx_rs_thresh = tx_conf->tx_rs_thresh;
+ if (tx_rs_thresh + tx_free_thresh > nb_desc) {
+ PMD_INIT_LOG(ERR, "tx_rs_thresh + tx_free_thresh must not "
+ "exceed nb_desc. (tx_rs_thresh=%u "
+ "tx_free_thresh=%u nb_desc=%u port=%d queue=%d)",
+ (unsigned int)tx_rs_thresh,
+ (unsigned int)tx_free_thresh,
+ (unsigned int)nb_desc,
+ (int)dev->data->port_id,
+ (int)queue_idx);
+ return I40E_ERR_PARAM;
+ }
if (tx_rs_thresh >= (nb_desc - 2)) {
PMD_INIT_LOG(ERR, "tx_rs_thresh must be less than the "
"number of TX descriptors minus 2. "
--
2.13.6
^ permalink raw reply [flat|nested] 12+ messages in thread* [dpdk-dev] [PATCH 1/3] net/i40e: fix invalid Tx threshold setup
2019-05-04 9:29 ` [dpdk-dev] [PATCH 1/3] net/i40e: fix invalid Tx threshold setup Qi Zhang
@ 2019-05-04 9:29 ` Qi Zhang
0 siblings, 0 replies; 12+ messages in thread
From: Qi Zhang @ 2019-05-04 9:29 UTC (permalink / raw)
To: beilei.xing, wenzhuo.lu, qiming.yang, konstantin.ananyev
Cc: dev, Qi Zhang, stable
Tx desc's DD status is not cleaned by NIC automatically after packets
have been transmitted until software refill a new packet during next
loop. So when tx_free_thresh + tx_rs_thresh > nb_desc, it is possible
that an outdated DD status be checked as tx_next_dd, then segment fault
happen due to free a NULL mbuf pointer.
Then patch fixes this issue by
1. try to adapt tx_rs_thresh to an aggresive tx_free_thresh.
2. queue setup fail when tx_free_thresh + tx_rs_thresh > nb_desc
Fixes: 4861cde46116 ("i40e: new poll mode driver")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
---
drivers/net/i40e/i40e_rxtx.c | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c
index 1489552da..4640a9c66 100644
--- a/drivers/net/i40e/i40e_rxtx.c
+++ b/drivers/net/i40e/i40e_rxtx.c
@@ -2169,15 +2169,30 @@ i40e_dev_tx_queue_setup(struct rte_eth_dev *dev,
* - tx_rs_thresh must be a divisor of the ring size.
* - tx_free_thresh must be greater than 0.
* - tx_free_thresh must be less than the size of the ring minus 3.
+ * - tx_free_thresh + tx_rs_thresh must not exceed nb_desc.
*
* One descriptor in the TX ring is used as a sentinel to avoid a H/W
* race condition, hence the maximum threshold constraints. When set
* to zero use default values.
*/
- tx_rs_thresh = (uint16_t)((tx_conf->tx_rs_thresh) ?
- tx_conf->tx_rs_thresh : DEFAULT_TX_RS_THRESH);
tx_free_thresh = (uint16_t)((tx_conf->tx_free_thresh) ?
tx_conf->tx_free_thresh : DEFAULT_TX_FREE_THRESH);
+ /* force tx_rs_thresh to adapt an aggresive tx_free_thresh */
+ tx_rs_thresh = (DEFAULT_TX_RS_THRESH + tx_free_thresh > nb_desc) ?
+ nb_desc - tx_free_thresh : DEFAULT_TX_RS_THRESH;
+ if (tx_conf->tx_rs_thresh > 0)
+ tx_rs_thresh = tx_conf->tx_rs_thresh;
+ if (tx_rs_thresh + tx_free_thresh > nb_desc) {
+ PMD_INIT_LOG(ERR, "tx_rs_thresh + tx_free_thresh must not "
+ "exceed nb_desc. (tx_rs_thresh=%u "
+ "tx_free_thresh=%u nb_desc=%u port=%d queue=%d)",
+ (unsigned int)tx_rs_thresh,
+ (unsigned int)tx_free_thresh,
+ (unsigned int)nb_desc,
+ (int)dev->data->port_id,
+ (int)queue_idx);
+ return I40E_ERR_PARAM;
+ }
if (tx_rs_thresh >= (nb_desc - 2)) {
PMD_INIT_LOG(ERR, "tx_rs_thresh must be less than the "
"number of TX descriptors minus 2. "
--
2.13.6
^ permalink raw reply [flat|nested] 12+ messages in thread
* [dpdk-dev] [PATCH 2/3] net/ice: fix invalid Tx threshold setup
2019-05-04 9:29 [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Qi Zhang
2019-05-04 9:29 ` Qi Zhang
2019-05-04 9:29 ` [dpdk-dev] [PATCH 1/3] net/i40e: fix invalid Tx threshold setup Qi Zhang
@ 2019-05-04 9:29 ` Qi Zhang
2019-05-04 9:29 ` Qi Zhang
2019-05-04 9:29 ` [dpdk-dev] [PATCH 3/3] net/ixgbe: " Qi Zhang
2019-05-15 6:36 ` [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Xing, Beilei
4 siblings, 1 reply; 12+ messages in thread
From: Qi Zhang @ 2019-05-04 9:29 UTC (permalink / raw)
To: beilei.xing, wenzhuo.lu, qiming.yang, konstantin.ananyev
Cc: dev, Qi Zhang, stable
Tx desc's DD status is not cleaned by NIC automatically after packets
have been transmitted until software refill a new packet during next
loop. So when tx_free_thresh + tx_rs_thresh > nb_desc, it is possible
that an outdated DD status be checked as tx_next_dd, then segment fault
happen due to free a NULL mbuf pointer.
Then patch fixes this issue by
1. try to adapt tx_rs_thresh to an aggresive tx_free_thresh.
2. queue setup fail when tx_free_thresh + tx_rs_thresh > nb_desc
Fixes: 50370662b727 ("net/ice: support device and queue ops")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
---
drivers/net/ice/ice_rxtx.c | 21 ++++++++++++++++++---
1 file changed, 18 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ice/ice_rxtx.c b/drivers/net/ice/ice_rxtx.c
index ace766b1d..620a5ea2b 100644
--- a/drivers/net/ice/ice_rxtx.c
+++ b/drivers/net/ice/ice_rxtx.c
@@ -764,17 +764,32 @@ ice_tx_queue_setup(struct rte_eth_dev *dev,
* - tx_rs_thresh must be a divisor of the ring size.
* - tx_free_thresh must be greater than 0.
* - tx_free_thresh must be less than the size of the ring minus 3.
+ * - tx_free_thresh + tx_rs_thresh must not exceed nb_desc.
*
* One descriptor in the TX ring is used as a sentinel to avoid a H/W
* race condition, hence the maximum threshold constraints. When set
* to zero use default values.
*/
- tx_rs_thresh = (uint16_t)(tx_conf->tx_rs_thresh ?
- tx_conf->tx_rs_thresh :
- ICE_DEFAULT_TX_RSBIT_THRESH);
tx_free_thresh = (uint16_t)(tx_conf->tx_free_thresh ?
tx_conf->tx_free_thresh :
ICE_DEFAULT_TX_FREE_THRESH);
+ /* force tx_rs_thresh to adapt an aggresive tx_free_thresh */
+ tx_rs_thresh =
+ (ICE_DEFAULT_TX_RSBIT_THRESH + tx_free_thresh > nb_desc) ?
+ nb_desc - tx_free_thresh : ICE_DEFAULT_TX_RSBIT_THRESH;
+ if (tx_conf->tx_rs_thresh)
+ tx_rs_thresh = tx_conf->tx_rs_thresh;
+ if (tx_rs_thresh + tx_free_thresh > nb_desc) {
+ PMD_INIT_LOG(ERR, "tx_rs_thresh + tx_free_thresh must not "
+ "exceed nb_desc. (tx_rs_thresh=%u "
+ "tx_free_thresh=%u nb_desc=%u port = %d queue=%d)",
+ (unsigned int)tx_rs_thresh,
+ (unsigned int)tx_free_thresh,
+ (unsigned int)nb_desc,
+ (int)dev->data->port_id,
+ (int)queue_idx);
+ return -EINVAL;
+ }
if (tx_rs_thresh >= (nb_desc - 2)) {
PMD_INIT_LOG(ERR, "tx_rs_thresh must be less than the "
"number of TX descriptors minus 2. "
--
2.13.6
^ permalink raw reply [flat|nested] 12+ messages in thread* [dpdk-dev] [PATCH 2/3] net/ice: fix invalid Tx threshold setup
2019-05-04 9:29 ` [dpdk-dev] [PATCH 2/3] net/ice: " Qi Zhang
@ 2019-05-04 9:29 ` Qi Zhang
0 siblings, 0 replies; 12+ messages in thread
From: Qi Zhang @ 2019-05-04 9:29 UTC (permalink / raw)
To: beilei.xing, wenzhuo.lu, qiming.yang, konstantin.ananyev
Cc: dev, Qi Zhang, stable
Tx desc's DD status is not cleaned by NIC automatically after packets
have been transmitted until software refill a new packet during next
loop. So when tx_free_thresh + tx_rs_thresh > nb_desc, it is possible
that an outdated DD status be checked as tx_next_dd, then segment fault
happen due to free a NULL mbuf pointer.
Then patch fixes this issue by
1. try to adapt tx_rs_thresh to an aggresive tx_free_thresh.
2. queue setup fail when tx_free_thresh + tx_rs_thresh > nb_desc
Fixes: 50370662b727 ("net/ice: support device and queue ops")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
---
drivers/net/ice/ice_rxtx.c | 21 ++++++++++++++++++---
1 file changed, 18 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ice/ice_rxtx.c b/drivers/net/ice/ice_rxtx.c
index ace766b1d..620a5ea2b 100644
--- a/drivers/net/ice/ice_rxtx.c
+++ b/drivers/net/ice/ice_rxtx.c
@@ -764,17 +764,32 @@ ice_tx_queue_setup(struct rte_eth_dev *dev,
* - tx_rs_thresh must be a divisor of the ring size.
* - tx_free_thresh must be greater than 0.
* - tx_free_thresh must be less than the size of the ring minus 3.
+ * - tx_free_thresh + tx_rs_thresh must not exceed nb_desc.
*
* One descriptor in the TX ring is used as a sentinel to avoid a H/W
* race condition, hence the maximum threshold constraints. When set
* to zero use default values.
*/
- tx_rs_thresh = (uint16_t)(tx_conf->tx_rs_thresh ?
- tx_conf->tx_rs_thresh :
- ICE_DEFAULT_TX_RSBIT_THRESH);
tx_free_thresh = (uint16_t)(tx_conf->tx_free_thresh ?
tx_conf->tx_free_thresh :
ICE_DEFAULT_TX_FREE_THRESH);
+ /* force tx_rs_thresh to adapt an aggresive tx_free_thresh */
+ tx_rs_thresh =
+ (ICE_DEFAULT_TX_RSBIT_THRESH + tx_free_thresh > nb_desc) ?
+ nb_desc - tx_free_thresh : ICE_DEFAULT_TX_RSBIT_THRESH;
+ if (tx_conf->tx_rs_thresh)
+ tx_rs_thresh = tx_conf->tx_rs_thresh;
+ if (tx_rs_thresh + tx_free_thresh > nb_desc) {
+ PMD_INIT_LOG(ERR, "tx_rs_thresh + tx_free_thresh must not "
+ "exceed nb_desc. (tx_rs_thresh=%u "
+ "tx_free_thresh=%u nb_desc=%u port = %d queue=%d)",
+ (unsigned int)tx_rs_thresh,
+ (unsigned int)tx_free_thresh,
+ (unsigned int)nb_desc,
+ (int)dev->data->port_id,
+ (int)queue_idx);
+ return -EINVAL;
+ }
if (tx_rs_thresh >= (nb_desc - 2)) {
PMD_INIT_LOG(ERR, "tx_rs_thresh must be less than the "
"number of TX descriptors minus 2. "
--
2.13.6
^ permalink raw reply [flat|nested] 12+ messages in thread
* [dpdk-dev] [PATCH 3/3] net/ixgbe: fix invalid Tx threshold setup
2019-05-04 9:29 [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Qi Zhang
` (2 preceding siblings ...)
2019-05-04 9:29 ` [dpdk-dev] [PATCH 2/3] net/ice: " Qi Zhang
@ 2019-05-04 9:29 ` Qi Zhang
2019-05-04 9:29 ` Qi Zhang
2019-05-15 6:36 ` [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Xing, Beilei
4 siblings, 1 reply; 12+ messages in thread
From: Qi Zhang @ 2019-05-04 9:29 UTC (permalink / raw)
To: beilei.xing, wenzhuo.lu, qiming.yang, konstantin.ananyev
Cc: dev, Qi Zhang, stable
Tx desc's DD status is not cleaned by NIC automatically after packets
have been transmitted until software refill a new packet during next
loop. So when tx_free_thresh + tx_rs_thresh > nb_desc, it is possible
that an outdated DD status be checked as tx_next_dd, then segment fault
happen due to free a NULL mbuf pointer.
Then patch fixes this issue by
1. try to adapt tx_rs_thresh to an aggresive tx_free_thresh.
2. queue setup fail when tx_free_thresh + tx_rs_thresh > nb_desc
Fixes: af75078fece3 ("first public release")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
---
drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ixgbe/ixgbe_rxtx.c b/drivers/net/ixgbe/ixgbe_rxtx.c
index 1fbc754ae..3072bc1b5 100644
--- a/drivers/net/ixgbe/ixgbe_rxtx.c
+++ b/drivers/net/ixgbe/ixgbe_rxtx.c
@@ -2496,14 +2496,29 @@ ixgbe_dev_tx_queue_setup(struct rte_eth_dev *dev,
* tx_rs_thresh must be a divisor of the ring size.
* tx_free_thresh must be greater than 0.
* tx_free_thresh must be less than the size of the ring minus 3.
+ * tx_free_thresh + tx_rs_thresh must not exceed nb_desc.
* One descriptor in the TX ring is used as a sentinel to avoid a
* H/W race condition, hence the maximum threshold constraints.
* When set to zero use default values.
*/
- tx_rs_thresh = (uint16_t)((tx_conf->tx_rs_thresh) ?
- tx_conf->tx_rs_thresh : DEFAULT_TX_RS_THRESH);
tx_free_thresh = (uint16_t)((tx_conf->tx_free_thresh) ?
tx_conf->tx_free_thresh : DEFAULT_TX_FREE_THRESH);
+ /* force tx_rs_thresh to adapt an aggresive tx_free_thresh */
+ tx_rs_thresh = (DEFAULT_TX_RS_THRESH + tx_free_thresh > nb_desc) ?
+ nb_desc - tx_free_thresh : DEFAULT_TX_RS_THRESH;
+ if (tx_conf->tx_rs_thresh > 0)
+ tx_rs_thresh = tx_conf->tx_rs_thresh;
+ if (tx_rs_thresh + tx_free_thresh > nb_desc) {
+ PMD_INIT_LOG(ERR, "tx_rs_thresh + tx_free_thresh must not "
+ "exceed nb_desc. (tx_rs_thresh=%u "
+ "tx_free_thresh=%u nb_desc=%u port = %d queue=%d)",
+ (unsigned int)tx_rs_thresh,
+ (unsigned int)tx_free_thresh,
+ (unsigned int)nb_desc,
+ (int)dev->data->port_id,
+ (int)queue_idx);
+ return -(EINVAL);
+ }
if (tx_rs_thresh >= (nb_desc - 2)) {
PMD_INIT_LOG(ERR, "tx_rs_thresh must be less than the number "
"of TX descriptors minus 2. (tx_rs_thresh=%u "
--
2.13.6
^ permalink raw reply [flat|nested] 12+ messages in thread* [dpdk-dev] [PATCH 3/3] net/ixgbe: fix invalid Tx threshold setup
2019-05-04 9:29 ` [dpdk-dev] [PATCH 3/3] net/ixgbe: " Qi Zhang
@ 2019-05-04 9:29 ` Qi Zhang
0 siblings, 0 replies; 12+ messages in thread
From: Qi Zhang @ 2019-05-04 9:29 UTC (permalink / raw)
To: beilei.xing, wenzhuo.lu, qiming.yang, konstantin.ananyev
Cc: dev, Qi Zhang, stable
Tx desc's DD status is not cleaned by NIC automatically after packets
have been transmitted until software refill a new packet during next
loop. So when tx_free_thresh + tx_rs_thresh > nb_desc, it is possible
that an outdated DD status be checked as tx_next_dd, then segment fault
happen due to free a NULL mbuf pointer.
Then patch fixes this issue by
1. try to adapt tx_rs_thresh to an aggresive tx_free_thresh.
2. queue setup fail when tx_free_thresh + tx_rs_thresh > nb_desc
Fixes: af75078fece3 ("first public release")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
---
drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ixgbe/ixgbe_rxtx.c b/drivers/net/ixgbe/ixgbe_rxtx.c
index 1fbc754ae..3072bc1b5 100644
--- a/drivers/net/ixgbe/ixgbe_rxtx.c
+++ b/drivers/net/ixgbe/ixgbe_rxtx.c
@@ -2496,14 +2496,29 @@ ixgbe_dev_tx_queue_setup(struct rte_eth_dev *dev,
* tx_rs_thresh must be a divisor of the ring size.
* tx_free_thresh must be greater than 0.
* tx_free_thresh must be less than the size of the ring minus 3.
+ * tx_free_thresh + tx_rs_thresh must not exceed nb_desc.
* One descriptor in the TX ring is used as a sentinel to avoid a
* H/W race condition, hence the maximum threshold constraints.
* When set to zero use default values.
*/
- tx_rs_thresh = (uint16_t)((tx_conf->tx_rs_thresh) ?
- tx_conf->tx_rs_thresh : DEFAULT_TX_RS_THRESH);
tx_free_thresh = (uint16_t)((tx_conf->tx_free_thresh) ?
tx_conf->tx_free_thresh : DEFAULT_TX_FREE_THRESH);
+ /* force tx_rs_thresh to adapt an aggresive tx_free_thresh */
+ tx_rs_thresh = (DEFAULT_TX_RS_THRESH + tx_free_thresh > nb_desc) ?
+ nb_desc - tx_free_thresh : DEFAULT_TX_RS_THRESH;
+ if (tx_conf->tx_rs_thresh > 0)
+ tx_rs_thresh = tx_conf->tx_rs_thresh;
+ if (tx_rs_thresh + tx_free_thresh > nb_desc) {
+ PMD_INIT_LOG(ERR, "tx_rs_thresh + tx_free_thresh must not "
+ "exceed nb_desc. (tx_rs_thresh=%u "
+ "tx_free_thresh=%u nb_desc=%u port = %d queue=%d)",
+ (unsigned int)tx_rs_thresh,
+ (unsigned int)tx_free_thresh,
+ (unsigned int)nb_desc,
+ (int)dev->data->port_id,
+ (int)queue_idx);
+ return -(EINVAL);
+ }
if (tx_rs_thresh >= (nb_desc - 2)) {
PMD_INIT_LOG(ERR, "tx_rs_thresh must be less than the number "
"of TX descriptors minus 2. (tx_rs_thresh=%u "
--
2.13.6
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup
2019-05-04 9:29 [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Qi Zhang
` (3 preceding siblings ...)
2019-05-04 9:29 ` [dpdk-dev] [PATCH 3/3] net/ixgbe: " Qi Zhang
@ 2019-05-15 6:36 ` Xing, Beilei
2019-05-15 6:36 ` Xing, Beilei
2019-05-15 7:09 ` Zhang, Qi Z
4 siblings, 2 replies; 12+ messages in thread
From: Xing, Beilei @ 2019-05-15 6:36 UTC (permalink / raw)
To: Zhang, Qi Z, Lu, Wenzhuo, Yang, Qiming, Ananyev, Konstantin; +Cc: dev
> -----Original Message-----
> From: Zhang, Qi Z
> Sent: Saturday, May 4, 2019 5:30 PM
> To: Xing, Beilei <beilei.xing@intel.com>; Lu, Wenzhuo
> <wenzhuo.lu@intel.com>; Yang, Qiming <qiming.yang@intel.com>; Ananyev,
> Konstantin <konstantin.ananyev@intel.com>
> Cc: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>
> Subject: [PATCH 0/3] fix invalid Tx threshhold setup
>
> When tx_free_thresh + tx_rs_thresh > nb_desc, it is possible that an
> outdated DD status be checked as tx_next_dd, then segment fault happen
> due to free a NULL mbuf pointer.
>
> The issue usually happens with an aggresive tx_free_thresh, for example:
>
> ./testpmd -c 0x3 -n 4 -- -i --rxq=16 --txq=16 --rxd=1024 --txd=1024 --
> txfreet=1020
>
> The patchset fix this issue on i40e, ixgbe and ice.
>
> Qi Zhang (3):
> net/i40e: fix invalid Tx threshold setup
> net/ice: fix invalid Tx threshold setup
> net/ixgbe: fix invalid Tx threshold setup
>
> drivers/net/i40e/i40e_rxtx.c | 19 +++++++++++++++++--
> drivers/net/ice/ice_rxtx.c | 21 ++++++++++++++++++---
> drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++--
> 3 files changed, 52 insertions(+), 7 deletions(-)
>
> --
> 2.13.6
Acked-by: Beilei Xing <beilei.xing@intel.com>
^ permalink raw reply [flat|nested] 12+ messages in thread* Re: [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup
2019-05-15 6:36 ` [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Xing, Beilei
@ 2019-05-15 6:36 ` Xing, Beilei
2019-05-15 7:09 ` Zhang, Qi Z
1 sibling, 0 replies; 12+ messages in thread
From: Xing, Beilei @ 2019-05-15 6:36 UTC (permalink / raw)
To: Zhang, Qi Z, Lu, Wenzhuo, Yang, Qiming, Ananyev, Konstantin; +Cc: dev
> -----Original Message-----
> From: Zhang, Qi Z
> Sent: Saturday, May 4, 2019 5:30 PM
> To: Xing, Beilei <beilei.xing@intel.com>; Lu, Wenzhuo
> <wenzhuo.lu@intel.com>; Yang, Qiming <qiming.yang@intel.com>; Ananyev,
> Konstantin <konstantin.ananyev@intel.com>
> Cc: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>
> Subject: [PATCH 0/3] fix invalid Tx threshhold setup
>
> When tx_free_thresh + tx_rs_thresh > nb_desc, it is possible that an
> outdated DD status be checked as tx_next_dd, then segment fault happen
> due to free a NULL mbuf pointer.
>
> The issue usually happens with an aggresive tx_free_thresh, for example:
>
> ./testpmd -c 0x3 -n 4 -- -i --rxq=16 --txq=16 --rxd=1024 --txd=1024 --
> txfreet=1020
>
> The patchset fix this issue on i40e, ixgbe and ice.
>
> Qi Zhang (3):
> net/i40e: fix invalid Tx threshold setup
> net/ice: fix invalid Tx threshold setup
> net/ixgbe: fix invalid Tx threshold setup
>
> drivers/net/i40e/i40e_rxtx.c | 19 +++++++++++++++++--
> drivers/net/ice/ice_rxtx.c | 21 ++++++++++++++++++---
> drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++--
> 3 files changed, 52 insertions(+), 7 deletions(-)
>
> --
> 2.13.6
Acked-by: Beilei Xing <beilei.xing@intel.com>
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup
2019-05-15 6:36 ` [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup Xing, Beilei
2019-05-15 6:36 ` Xing, Beilei
@ 2019-05-15 7:09 ` Zhang, Qi Z
2019-05-15 7:09 ` Zhang, Qi Z
1 sibling, 1 reply; 12+ messages in thread
From: Zhang, Qi Z @ 2019-05-15 7:09 UTC (permalink / raw)
To: Xing, Beilei, Lu, Wenzhuo, Yang, Qiming, Ananyev, Konstantin; +Cc: dev
> -----Original Message-----
> From: Xing, Beilei
> Sent: Wednesday, May 15, 2019 2:36 PM
> To: Zhang, Qi Z <qi.z.zhang@intel.com>; Lu, Wenzhuo
> <wenzhuo.lu@intel.com>; Yang, Qiming <qiming.yang@intel.com>; Ananyev,
> Konstantin <konstantin.ananyev@intel.com>
> Cc: dev@dpdk.org
> Subject: RE: [PATCH 0/3] fix invalid Tx threshhold setup
>
>
>
> > -----Original Message-----
> > From: Zhang, Qi Z
> > Sent: Saturday, May 4, 2019 5:30 PM
> > To: Xing, Beilei <beilei.xing@intel.com>; Lu, Wenzhuo
> > <wenzhuo.lu@intel.com>; Yang, Qiming <qiming.yang@intel.com>; Ananyev,
> > Konstantin <konstantin.ananyev@intel.com>
> > Cc: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>
> > Subject: [PATCH 0/3] fix invalid Tx threshhold setup
> >
> > When tx_free_thresh + tx_rs_thresh > nb_desc, it is possible that an
> > outdated DD status be checked as tx_next_dd, then segment fault happen
> > due to free a NULL mbuf pointer.
> >
> > The issue usually happens with an aggresive tx_free_thresh, for example:
> >
> > ./testpmd -c 0x3 -n 4 -- -i --rxq=16 --txq=16 --rxd=1024 --txd=1024 --
> > txfreet=1020
> >
> > The patchset fix this issue on i40e, ixgbe and ice.
> >
> > Qi Zhang (3):
> > net/i40e: fix invalid Tx threshold setup
> > net/ice: fix invalid Tx threshold setup
> > net/ixgbe: fix invalid Tx threshold setup
> >
> > drivers/net/i40e/i40e_rxtx.c | 19 +++++++++++++++++--
> > drivers/net/ice/ice_rxtx.c | 21 ++++++++++++++++++---
> > drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++--
> > 3 files changed, 52 insertions(+), 7 deletions(-)
> >
> > --
> > 2.13.6
>
> Acked-by: Beilei Xing <beilei.xing@intel.com>
Applied to dpdk-next-net-intel.
Thanks
Qi
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [dpdk-dev] [PATCH 0/3] fix invalid Tx threshhold setup
2019-05-15 7:09 ` Zhang, Qi Z
@ 2019-05-15 7:09 ` Zhang, Qi Z
0 siblings, 0 replies; 12+ messages in thread
From: Zhang, Qi Z @ 2019-05-15 7:09 UTC (permalink / raw)
To: Xing, Beilei, Lu, Wenzhuo, Yang, Qiming, Ananyev, Konstantin; +Cc: dev
> -----Original Message-----
> From: Xing, Beilei
> Sent: Wednesday, May 15, 2019 2:36 PM
> To: Zhang, Qi Z <qi.z.zhang@intel.com>; Lu, Wenzhuo
> <wenzhuo.lu@intel.com>; Yang, Qiming <qiming.yang@intel.com>; Ananyev,
> Konstantin <konstantin.ananyev@intel.com>
> Cc: dev@dpdk.org
> Subject: RE: [PATCH 0/3] fix invalid Tx threshhold setup
>
>
>
> > -----Original Message-----
> > From: Zhang, Qi Z
> > Sent: Saturday, May 4, 2019 5:30 PM
> > To: Xing, Beilei <beilei.xing@intel.com>; Lu, Wenzhuo
> > <wenzhuo.lu@intel.com>; Yang, Qiming <qiming.yang@intel.com>; Ananyev,
> > Konstantin <konstantin.ananyev@intel.com>
> > Cc: dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>
> > Subject: [PATCH 0/3] fix invalid Tx threshhold setup
> >
> > When tx_free_thresh + tx_rs_thresh > nb_desc, it is possible that an
> > outdated DD status be checked as tx_next_dd, then segment fault happen
> > due to free a NULL mbuf pointer.
> >
> > The issue usually happens with an aggresive tx_free_thresh, for example:
> >
> > ./testpmd -c 0x3 -n 4 -- -i --rxq=16 --txq=16 --rxd=1024 --txd=1024 --
> > txfreet=1020
> >
> > The patchset fix this issue on i40e, ixgbe and ice.
> >
> > Qi Zhang (3):
> > net/i40e: fix invalid Tx threshold setup
> > net/ice: fix invalid Tx threshold setup
> > net/ixgbe: fix invalid Tx threshold setup
> >
> > drivers/net/i40e/i40e_rxtx.c | 19 +++++++++++++++++--
> > drivers/net/ice/ice_rxtx.c | 21 ++++++++++++++++++---
> > drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++--
> > 3 files changed, 52 insertions(+), 7 deletions(-)
> >
> > --
> > 2.13.6
>
> Acked-by: Beilei Xing <beilei.xing@intel.com>
Applied to dpdk-next-net-intel.
Thanks
Qi
^ permalink raw reply [flat|nested] 12+ messages in thread