* [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
@ 2015-09-16 21:51 Pablo de Lara
2015-09-21 22:40 ` Stephen Hemminger
2015-09-23 11:49 ` Ananyev, Konstantin
0 siblings, 2 replies; 7+ messages in thread
From: Pablo de Lara @ 2015-09-16 21:51 UTC (permalink / raw)
To: dev
Following the same approach taken with dev_started field
in rte_eth_dev_data structure, this patch adds two new fields
in it, rx_queue_state and tx_queue_state arrays, which track
which queues have been started and which not.
This is important to avoid trying to start/stop twice a queue,
which will result in undefined behaviour
(which may cause RX/TX disruption).
Mind that only the PMDs which have queue_start/stop functions
have been changed to update this field, as the functions will
check the queue state before switching it.
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
---
Changes in v2:
- Added macros to define the different queue states
drivers/net/cxgbe/cxgbe_ethdev.c | 29 +++++++++++++++++++++++++----
drivers/net/enic/enic_ethdev.c | 6 ++++++
drivers/net/fm10k/fm10k_ethdev.c | 4 ++++
drivers/net/i40e/i40e_ethdev_vf.c | 6 ++++++
drivers/net/i40e/i40e_rxtx.c | 7 ++++++-
drivers/net/ixgbe/ixgbe_rxtx.c | 4 ++++
lib/librte_ether/rte_ethdev.c | 28 ++++++++++++++++++++++++++++
lib/librte_ether/rte_ethdev.h | 10 ++++++++++
8 files changed, 89 insertions(+), 5 deletions(-)
diff --git a/drivers/net/cxgbe/cxgbe_ethdev.c b/drivers/net/cxgbe/cxgbe_ethdev.c
index 478051a..305959e 100644
--- a/drivers/net/cxgbe/cxgbe_ethdev.c
+++ b/drivers/net/cxgbe/cxgbe_ethdev.c
@@ -368,23 +368,33 @@ static int cxgbe_dev_configure(struct rte_eth_dev *eth_dev)
static int cxgbe_dev_tx_queue_start(struct rte_eth_dev *eth_dev,
uint16_t tx_queue_id)
{
+ int ret;
struct sge_eth_txq *txq = (struct sge_eth_txq *)
(eth_dev->data->tx_queues[tx_queue_id]);
dev_debug(NULL, "%s: tx_queue_id = %d\n", __func__, tx_queue_id);
- return t4_sge_eth_txq_start(txq);
+ ret = t4_sge_eth_txq_start(txq);
+ if (ret == 0)
+ eth_dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
+
+ return ret;
}
static int cxgbe_dev_tx_queue_stop(struct rte_eth_dev *eth_dev,
uint16_t tx_queue_id)
{
+ int ret;
struct sge_eth_txq *txq = (struct sge_eth_txq *)
(eth_dev->data->tx_queues[tx_queue_id]);
dev_debug(NULL, "%s: tx_queue_id = %d\n", __func__, tx_queue_id);
- return t4_sge_eth_txq_stop(txq);
+ ret = t4_sge_eth_txq_stop(txq);
+ if (ret == 0)
+ eth_dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
+
+ return ret;
}
static int cxgbe_dev_tx_queue_setup(struct rte_eth_dev *eth_dev,
@@ -460,6 +470,7 @@ static void cxgbe_dev_tx_queue_release(void *q)
static int cxgbe_dev_rx_queue_start(struct rte_eth_dev *eth_dev,
uint16_t rx_queue_id)
{
+ int ret;
struct port_info *pi = (struct port_info *)(eth_dev->data->dev_private);
struct adapter *adap = pi->adapter;
struct sge_rspq *q;
@@ -468,12 +479,18 @@ static int cxgbe_dev_rx_queue_start(struct rte_eth_dev *eth_dev,
__func__, pi->port_id, rx_queue_id);
q = eth_dev->data->rx_queues[rx_queue_id];
- return t4_sge_eth_rxq_start(adap, q);
+
+ ret = t4_sge_eth_rxq_start(adap, q);
+ if (ret == 0)
+ eth_dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
+
+ return ret;
}
static int cxgbe_dev_rx_queue_stop(struct rte_eth_dev *eth_dev,
uint16_t rx_queue_id)
{
+ int ret;
struct port_info *pi = (struct port_info *)(eth_dev->data->dev_private);
struct adapter *adap = pi->adapter;
struct sge_rspq *q;
@@ -482,7 +499,11 @@ static int cxgbe_dev_rx_queue_stop(struct rte_eth_dev *eth_dev,
__func__, pi->port_id, rx_queue_id);
q = eth_dev->data->rx_queues[rx_queue_id];
- return t4_sge_eth_rxq_stop(adap, q);
+ ret = t4_sge_eth_rxq_stop(adap, q);
+ if (ret == 0)
+ eth_dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
+
+ return ret;
}
static int cxgbe_dev_rx_queue_setup(struct rte_eth_dev *eth_dev,
diff --git a/drivers/net/enic/enic_ethdev.c b/drivers/net/enic/enic_ethdev.c
index 8280cea..56b22f1 100644
--- a/drivers/net/enic/enic_ethdev.c
+++ b/drivers/net/enic/enic_ethdev.c
@@ -194,6 +194,7 @@ static int enicpmd_dev_tx_queue_start(struct rte_eth_dev *eth_dev,
ENICPMD_FUNC_TRACE();
enic_start_wq(enic, queue_idx);
+ eth_dev->data->tx_queue_state[queue_idx] = RTE_ETH_QUEUE_STATE_STARTED;
return 0;
}
@@ -209,6 +210,8 @@ static int enicpmd_dev_tx_queue_stop(struct rte_eth_dev *eth_dev,
ret = enic_stop_wq(enic, queue_idx);
if (ret)
dev_err(enic, "error in stopping wq %d\n", queue_idx);
+ else
+ eth_dev->data->tx_queue_state[queue_idx] = RTE_ETH_QUEUE_STATE_STOPPED;
return ret;
}
@@ -221,6 +224,7 @@ static int enicpmd_dev_rx_queue_start(struct rte_eth_dev *eth_dev,
ENICPMD_FUNC_TRACE();
enic_start_rq(enic, queue_idx);
+ eth_dev->data->rx_queue_state[queue_idx] = RTE_ETH_QUEUE_STATE_STARTED;
return 0;
}
@@ -236,6 +240,8 @@ static int enicpmd_dev_rx_queue_stop(struct rte_eth_dev *eth_dev,
ret = enic_stop_rq(enic, queue_idx);
if (ret)
dev_err(enic, "error in stopping rq %d\n", queue_idx);
+ else
+ eth_dev->data->rx_queue_state[queue_idx] = RTE_ETH_QUEUE_STATE_STOPPED;
return ret;
}
diff --git a/drivers/net/fm10k/fm10k_ethdev.c b/drivers/net/fm10k/fm10k_ethdev.c
index a69c990..b7e7b5a 100644
--- a/drivers/net/fm10k/fm10k_ethdev.c
+++ b/drivers/net/fm10k/fm10k_ethdev.c
@@ -524,6 +524,7 @@ fm10k_dev_rx_queue_start(struct rte_eth_dev *dev, uint16_t rx_queue_id)
*/
FM10K_WRITE_REG(hw, FM10K_RDH(rx_queue_id), 0);
FM10K_WRITE_REG(hw, FM10K_RDT(rx_queue_id), rxq->nb_desc - 1);
+ dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
}
return err;
@@ -542,6 +543,7 @@ fm10k_dev_rx_queue_stop(struct rte_eth_dev *dev, uint16_t rx_queue_id)
/* Free mbuf and clean HW ring */
rx_queue_clean(dev->data->rx_queues[rx_queue_id]);
+ dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
}
return 0;
@@ -569,6 +571,7 @@ fm10k_dev_tx_queue_start(struct rte_eth_dev *dev, uint16_t tx_queue_id)
FM10K_WRITE_REG(hw, FM10K_TXDCTL(tx_queue_id),
FM10K_TXDCTL_ENABLE | txdctl);
FM10K_WRITE_FLUSH(hw);
+ dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
} else
err = -1;
@@ -585,6 +588,7 @@ fm10k_dev_tx_queue_stop(struct rte_eth_dev *dev, uint16_t tx_queue_id)
if (tx_queue_id < dev->data->nb_tx_queues) {
tx_queue_disable(hw, tx_queue_id);
tx_queue_clean(dev->data->tx_queues[tx_queue_id]);
+ dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
}
return 0;
diff --git a/drivers/net/i40e/i40e_ethdev_vf.c b/drivers/net/i40e/i40e_ethdev_vf.c
index b694400..0f07a77 100644
--- a/drivers/net/i40e/i40e_ethdev_vf.c
+++ b/drivers/net/i40e/i40e_ethdev_vf.c
@@ -1373,6 +1373,8 @@ i40evf_dev_rx_queue_start(struct rte_eth_dev *dev, uint16_t rx_queue_id)
if (err)
PMD_DRV_LOG(ERR, "Failed to switch RX queue %u on",
rx_queue_id);
+ else
+ dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
}
return err;
@@ -1397,6 +1399,7 @@ i40evf_dev_rx_queue_stop(struct rte_eth_dev *dev, uint16_t rx_queue_id)
i40e_rx_queue_release_mbufs(rxq);
i40e_reset_rx_queue(rxq);
+ dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
}
return 0;
@@ -1417,6 +1420,8 @@ i40evf_dev_tx_queue_start(struct rte_eth_dev *dev, uint16_t tx_queue_id)
if (err)
PMD_DRV_LOG(ERR, "Failed to switch TX queue %u on",
tx_queue_id);
+ else
+ dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
}
return err;
@@ -1441,6 +1446,7 @@ i40evf_dev_tx_queue_stop(struct rte_eth_dev *dev, uint16_t tx_queue_id)
i40e_tx_queue_release_mbufs(txq);
i40e_reset_tx_queue(txq);
+ dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
}
return 0;
diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c
index fd656d5..dc297b1 100644
--- a/drivers/net/i40e/i40e_rxtx.c
+++ b/drivers/net/i40e/i40e_rxtx.c
@@ -2009,7 +2009,8 @@ i40e_dev_rx_queue_start(struct rte_eth_dev *dev, uint16_t rx_queue_id)
i40e_rx_queue_release_mbufs(rxq);
i40e_reset_rx_queue(rxq);
- }
+ } else
+ dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
}
return err;
@@ -2038,6 +2039,7 @@ i40e_dev_rx_queue_stop(struct rte_eth_dev *dev, uint16_t rx_queue_id)
}
i40e_rx_queue_release_mbufs(rxq);
i40e_reset_rx_queue(rxq);
+ dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
}
return 0;
@@ -2063,6 +2065,8 @@ i40e_dev_tx_queue_start(struct rte_eth_dev *dev, uint16_t tx_queue_id)
if (err)
PMD_DRV_LOG(ERR, "Failed to switch TX queue %u on",
tx_queue_id);
+ else
+ dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
}
return err;
@@ -2092,6 +2096,7 @@ i40e_dev_tx_queue_stop(struct rte_eth_dev *dev, uint16_t tx_queue_id)
i40e_tx_queue_release_mbufs(txq);
i40e_reset_tx_queue(txq);
+ dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
}
return 0;
diff --git a/drivers/net/ixgbe/ixgbe_rxtx.c b/drivers/net/ixgbe/ixgbe_rxtx.c
index a598a72..5011a6a 100644
--- a/drivers/net/ixgbe/ixgbe_rxtx.c
+++ b/drivers/net/ixgbe/ixgbe_rxtx.c
@@ -4498,6 +4498,7 @@ ixgbe_dev_rx_queue_start(struct rte_eth_dev *dev, uint16_t rx_queue_id)
rte_wmb();
IXGBE_WRITE_REG(hw, IXGBE_RDH(rxq->reg_idx), 0);
IXGBE_WRITE_REG(hw, IXGBE_RDT(rxq->reg_idx), rxq->nb_rx_desc - 1);
+ dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
} else
return -1;
@@ -4541,6 +4542,7 @@ ixgbe_dev_rx_queue_stop(struct rte_eth_dev *dev, uint16_t rx_queue_id)
ixgbe_rx_queue_release_mbufs(rxq);
ixgbe_reset_rx_queue(adapter, rxq);
+ dev->data->rx_queue_state[rx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
} else
return -1;
@@ -4583,6 +4585,7 @@ ixgbe_dev_tx_queue_start(struct rte_eth_dev *dev, uint16_t tx_queue_id)
rte_wmb();
IXGBE_WRITE_REG(hw, IXGBE_TDH(txq->reg_idx), 0);
IXGBE_WRITE_REG(hw, IXGBE_TDT(txq->reg_idx), 0);
+ dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STARTED;
} else
return -1;
@@ -4643,6 +4646,7 @@ ixgbe_dev_tx_queue_stop(struct rte_eth_dev *dev, uint16_t tx_queue_id)
txq->ops->release_mbufs(txq);
txq->ops->reset(txq);
}
+ dev->data->tx_queue_state[tx_queue_id] = RTE_ETH_QUEUE_STATE_STOPPED;
} else
return -1;
diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
index b309309..5af35c9 100644
--- a/lib/librte_ether/rte_ethdev.c
+++ b/lib/librte_ether/rte_ethdev.c
@@ -767,6 +767,13 @@ rte_eth_dev_rx_queue_start(uint8_t port_id, uint16_t rx_queue_id)
FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_start, -ENOTSUP);
+ if (dev->data->rx_queue_state[rx_queue_id] != RTE_ETH_QUEUE_STATE_STOPPED) {
+ PMD_DEBUG_TRACE("Queue %" PRIu16" of device with port_id=%" PRIu8
+ " already started\n",
+ rx_queue_id, port_id);
+ return 0;
+ }
+
return dev->dev_ops->rx_queue_start(dev, rx_queue_id);
}
@@ -790,6 +797,13 @@ rte_eth_dev_rx_queue_stop(uint8_t port_id, uint16_t rx_queue_id)
FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_stop, -ENOTSUP);
+ if (dev->data->rx_queue_state[rx_queue_id] == RTE_ETH_QUEUE_STATE_STOPPED) {
+ PMD_DEBUG_TRACE("Queue %" PRIu16" of device with port_id=%" PRIu8
+ " already stopped\n",
+ rx_queue_id, port_id);
+ return 0;
+ }
+
return dev->dev_ops->rx_queue_stop(dev, rx_queue_id);
}
@@ -813,6 +827,13 @@ rte_eth_dev_tx_queue_start(uint8_t port_id, uint16_t tx_queue_id)
FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_queue_start, -ENOTSUP);
+ if (dev->data->tx_queue_state[tx_queue_id] != RTE_ETH_QUEUE_STATE_STOPPED) {
+ PMD_DEBUG_TRACE("Queue %" PRIu16" of device with port_id=%" PRIu8
+ " already started\n",
+ tx_queue_id, port_id);
+ return 0;
+ }
+
return dev->dev_ops->tx_queue_start(dev, tx_queue_id);
}
@@ -836,6 +857,13 @@ rte_eth_dev_tx_queue_stop(uint8_t port_id, uint16_t tx_queue_id)
FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_queue_stop, -ENOTSUP);
+ if (dev->data->tx_queue_state[tx_queue_id] == RTE_ETH_QUEUE_STATE_STOPPED) {
+ PMD_DEBUG_TRACE("Queue %" PRIu16" of device with port_id=%" PRIu8
+ " already stopped\n",
+ tx_queue_id, port_id);
+ return 0;
+ }
+
return dev->dev_ops->tx_queue_stop(dev, tx_queue_id);
}
diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index fa06554..120808b 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -962,6 +962,12 @@ struct rte_eth_xstats {
uint64_t value;
};
+/**
+ * RX/TX queue states
+ */
+#define RTE_ETH_QUEUE_STATE_STOPPED 0
+#define RTE_ETH_QUEUE_STATE_STARTED 1
+
struct rte_eth_dev;
struct rte_eth_dev_callback;
@@ -1635,6 +1641,10 @@ struct rte_eth_dev_data {
all_multicast : 1, /**< RX all multicast mode ON(1) / OFF(0). */
dev_started : 1, /**< Device state: STARTED(1) / STOPPED(0). */
lro : 1; /**< RX LRO is ON(1) / OFF(0) */
+ uint8_t rx_queue_state[RTE_MAX_QUEUES_PER_PORT];
+ /** Queues state: STARTED(1) / STOPPED(0) */
+ uint8_t tx_queue_state[RTE_MAX_QUEUES_PER_PORT];
+ /** Queues state: STARTED(1) / STOPPED(0) */
};
/**
--
2.4.2
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
2015-09-16 21:51 [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data Pablo de Lara
@ 2015-09-21 22:40 ` Stephen Hemminger
2015-09-23 6:56 ` Qiu, Michael
2015-09-23 11:49 ` Ananyev, Konstantin
1 sibling, 1 reply; 7+ messages in thread
From: Stephen Hemminger @ 2015-09-21 22:40 UTC (permalink / raw)
To: Pablo de Lara; +Cc: dev
On Wed, 16 Sep 2015 22:51:24 +0100
Pablo de Lara <pablo.de.lara.guarch@intel.com> wrote:
> This is important to avoid trying to start/stop twice a queue,
> which will result in undefined behaviour
> (which may cause RX/TX disruption).
>
> Mind that only the PMDs which have queue_start/stop functions
> have been changed to update this field, as the functions will
> check the queue state before switching it.
>
> Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
I agree that the DPDK API should check for buggy manipulation
in the control path. But this should be done in generic code.
Anything where you have to change any driver is making more work
than necessary.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
2015-09-21 22:40 ` Stephen Hemminger
@ 2015-09-23 6:56 ` Qiu, Michael
2015-09-23 7:39 ` De Lara Guarch, Pablo
2015-09-23 8:22 ` Ananyev, Konstantin
0 siblings, 2 replies; 7+ messages in thread
From: Qiu, Michael @ 2015-09-23 6:56 UTC (permalink / raw)
To: Stephen Hemminger, De Lara Guarch, Pablo; +Cc: dev
On 2015/9/22 6:40, Stephen Hemminger wrote:
> On Wed, 16 Sep 2015 22:51:24 +0100
> Pablo de Lara <pablo.de.lara.guarch@intel.com> wrote:
>
>> This is important to avoid trying to start/stop twice a queue,
>> which will result in undefined behaviour
>> (which may cause RX/TX disruption).
>>
>> Mind that only the PMDs which have queue_start/stop functions
>> have been changed to update this field, as the functions will
>> check the queue state before switching it.
>>
>> Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
> I agree that the DPDK API should check for buggy manipulation
> in the control path. But this should be done in generic code.
> Anything where you have to change any driver is making more work
> than necessary.
I agree with you, but I have a question, why we need expose the queue
start and stop function to app?
In my opinion, user app will hardly to start a device but stop the
device queue. what's the purpose of it?
Thanks,
Michael
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
2015-09-23 6:56 ` Qiu, Michael
@ 2015-09-23 7:39 ` De Lara Guarch, Pablo
2015-09-23 8:22 ` Ananyev, Konstantin
1 sibling, 0 replies; 7+ messages in thread
From: De Lara Guarch, Pablo @ 2015-09-23 7:39 UTC (permalink / raw)
To: Qiu, Michael, Stephen Hemminger; +Cc: dev
> -----Original Message-----
> From: Qiu, Michael
> Sent: Wednesday, September 23, 2015 7:57 AM
> To: Stephen Hemminger; De Lara Guarch, Pablo
> Cc: dev@dpdk.org; thomas.monjalon@6wind.com
> Subject: Re: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state
> arrays in rte_eth_dev_data
>
> On 2015/9/22 6:40, Stephen Hemminger wrote:
> > On Wed, 16 Sep 2015 22:51:24 +0100
> > Pablo de Lara <pablo.de.lara.guarch@intel.com> wrote:
> >
> >> This is important to avoid trying to start/stop twice a queue,
> >> which will result in undefined behaviour
> >> (which may cause RX/TX disruption).
> >>
> >> Mind that only the PMDs which have queue_start/stop functions
> >> have been changed to update this field, as the functions will
> >> check the queue state before switching it.
> >>
> >> Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
> > I agree that the DPDK API should check for buggy manipulation
> > in the control path. But this should be done in generic code.
> > Anything where you have to change any driver is making more work
> > than necessary.
>
I see little way to set the queue state without touching the PMDs.
Basically, because queues can be started either calling directly
the function from the PMDs or calling the generic function from ethdev,
therefore some code must be placed in the PMDs.
I guess there are two possible ways to overcome this:
- Use in all PMDs rte_eth_rx_queue_start/stop, instead of directly
the internal functions: first, it is necessary the port id, which I am not sure if it is always available
(plus, when it is, it is available in the specific queue of that PMD, so it does not have to be there).
Anyway, this option requires changing the PMDs.
- Apparently, the only function that calls these queue_start/stop functions is dev_start/stop.
In some PMDs, it is possible to defer the start of some queues, so it would not initialize
these queues, but in other PMDs this is not implemented.
We will also be depending at how it is going on within the PMDs,
so the code will not be generic either.
Any ideas to make this truly generic?
> I agree with you, but I have a question, why we need expose the queue
> start and stop function to app?
>
> In my opinion, user app will hardly to start a device but stop the
> device queue. what's the purpose of it?
This API was added in 1.7 if I am not wrong. The purpose was that user could decide
not to start some of the queues in a device at start up, and then a new API to start
these queues was necessary, but there were no checks of their current state,
leading to undefined behaviour.
Thanks both Stephen and Michael,
Pablo
>
> Thanks,
> Michael
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
2015-09-23 6:56 ` Qiu, Michael
2015-09-23 7:39 ` De Lara Guarch, Pablo
@ 2015-09-23 8:22 ` Ananyev, Konstantin
1 sibling, 0 replies; 7+ messages in thread
From: Ananyev, Konstantin @ 2015-09-23 8:22 UTC (permalink / raw)
To: Qiu, Michael, Stephen Hemminger, De Lara Guarch, Pablo; +Cc: dev
> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Qiu, Michael
> Sent: Wednesday, September 23, 2015 7:57 AM
> To: Stephen Hemminger; De Lara Guarch, Pablo
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
>
> On 2015/9/22 6:40, Stephen Hemminger wrote:
> > On Wed, 16 Sep 2015 22:51:24 +0100
> > Pablo de Lara <pablo.de.lara.guarch@intel.com> wrote:
> >
> >> This is important to avoid trying to start/stop twice a queue,
> >> which will result in undefined behaviour
> >> (which may cause RX/TX disruption).
> >>
> >> Mind that only the PMDs which have queue_start/stop functions
> >> have been changed to update this field, as the functions will
> >> check the queue state before switching it.
> >>
> >> Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
> > I agree that the DPDK API should check for buggy manipulation
> > in the control path. But this should be done in generic code.
> > Anything where you have to change any driver is making more work
> > than necessary.
>
> I agree with you, but I have a question, why we need expose the queue
> start and stop function to app?
>
> In my opinion, user app will hardly to start a device but stop the
> device queue. what's the purpose of it?
>
There are use cases when application need to start/stop individual queues.
DPDK vhost app would be one of the examples.
Konstantin
> Thanks,
> Michael
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
2015-09-16 21:51 [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data Pablo de Lara
2015-09-21 22:40 ` Stephen Hemminger
@ 2015-09-23 11:49 ` Ananyev, Konstantin
2015-11-04 16:53 ` Thomas Monjalon
1 sibling, 1 reply; 7+ messages in thread
From: Ananyev, Konstantin @ 2015-09-23 11:49 UTC (permalink / raw)
To: De Lara Guarch, Pablo, dev
> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Pablo de Lara
> Sent: Wednesday, September 16, 2015 10:51 PM
> To: dev@dpdk.org
> Subject: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
>
> Following the same approach taken with dev_started field
> in rte_eth_dev_data structure, this patch adds two new fields
> in it, rx_queue_state and tx_queue_state arrays, which track
> which queues have been started and which not.
>
> This is important to avoid trying to start/stop twice a queue,
> which will result in undefined behaviour
> (which may cause RX/TX disruption).
>
> Mind that only the PMDs which have queue_start/stop functions
> have been changed to update this field, as the functions will
> check the queue state before switching it.
>
> Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
> ---
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data
2015-09-23 11:49 ` Ananyev, Konstantin
@ 2015-11-04 16:53 ` Thomas Monjalon
0 siblings, 0 replies; 7+ messages in thread
From: Thomas Monjalon @ 2015-11-04 16:53 UTC (permalink / raw)
To: De Lara Guarch, Pablo; +Cc: dev
> > Following the same approach taken with dev_started field
> > in rte_eth_dev_data structure, this patch adds two new fields
> > in it, rx_queue_state and tx_queue_state arrays, which track
> > which queues have been started and which not.
> >
> > This is important to avoid trying to start/stop twice a queue,
> > which will result in undefined behaviour
> > (which may cause RX/TX disruption).
> >
> > Mind that only the PMDs which have queue_start/stop functions
> > have been changed to update this field, as the functions will
> > check the queue state before switching it.
> >
> > Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
>
> Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Applied, thanks
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2015-11-04 16:54 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-09-16 21:51 [dpdk-dev] [PATCH v2] ethdev: add new RX/TX queue state arrays in rte_eth_dev_data Pablo de Lara
2015-09-21 22:40 ` Stephen Hemminger
2015-09-23 6:56 ` Qiu, Michael
2015-09-23 7:39 ` De Lara Guarch, Pablo
2015-09-23 8:22 ` Ananyev, Konstantin
2015-09-23 11:49 ` Ananyev, Konstantin
2015-11-04 16:53 ` Thomas Monjalon
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).