Configuring one queue per port (#define RTE_MAX_QUEUES_PER_PORT 1) fails
compilation with e.g.:
../drivers/net/bnxt/bnxt_rxq.c: In function 'bnxt_rx_queue_stop':
../drivers/net/bnxt/bnxt_rxq.c:587:34: error: array subscript 1 is above array bounds of 'uint8_t[1]' {aka 'unsigned char[1]'} [-Werror=array-bounds=]
587 | dev->data->rx_queue_state[q_id] = RTE_ETH_QUEUE_STATE_STOPPED;
| ~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
In file included from ../drivers/net/bnxt/bnxt.h:16,
from ../drivers/net/bnxt/bnxt_rxq.c:10:
../lib/ethdev/ethdev_driver.h:168:17: note: while referencing 'rx_queue_state'
168 | uint8_t rx_queue_state[RTE_MAX_QUEUES_PER_PORT];
| ^~~~~~~~~~~~~~
Use the CI to test my ideas to fix this.
Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
---
v7:
* Introduce __rte_assume(e) in rte_common.h, and use this instead.
v6:
* Use __attribute__((assume(i < RTE_MAX_QUEUES_PER_PORT))) where the
indexing warning occurs, instead of pragma GCC optimize("no-peel-loops")
for the entire function.
v5:
* Wrap GCC optimizer pragmas in if defined(RTE_TOOLCHAIN_GCC).
v4:
* Workaound GCC optimizer incorrectly throwing a warning in these network
drivers:
* bnxt
* e1000
* failsafe
* hns3
v3:
* Fix net/ixgbe driver.
v2:
* Fix net/vmxnet3 driver.
---
config/rte_config.h | 4 ++--
drivers/net/bnxt/bnxt_ethdev.c | 2 ++
drivers/net/bnxt/bnxt_rxq.c | 1 +
drivers/net/e1000/igb_rxtx.c | 2 ++
drivers/net/failsafe/failsafe_ops.c | 10 +++++++--
drivers/net/hns3/hns3_rxtx.c | 2 ++
drivers/net/ixgbe/ixgbe_ethdev.c | 3 ++-
drivers/net/mana/tx.c | 1 +
drivers/net/vmxnet3/vmxnet3_ethdev.c | 32 +++++++++++++++++-----------
lib/eal/include/rte_common.h | 11 ++++++++++
10 files changed, 51 insertions(+), 17 deletions(-)
@@ -65,8 +65,8 @@
#define RTE_MBUF_DEFAULT_MEMPOOL_OPS "ring_mp_mc"
/* ether defines */
-#define RTE_MAX_QUEUES_PER_PORT 1024
-#define RTE_ETHDEV_QUEUE_STAT_CNTRS 16 /* max 256 */
+#define RTE_MAX_QUEUES_PER_PORT 1 /* default 1024 */
+#define RTE_ETHDEV_QUEUE_STAT_CNTRS 1 /* max 256, default 16 */
#define RTE_ETHDEV_RXTX_CALLBACKS 1
#define RTE_MAX_MULTI_HOST_CTRLS 4
@@ -910,6 +910,7 @@ static int bnxt_start_nic(struct bnxt *bp)
struct bnxt_rx_queue *rxq = bp->rx_queues[j];
if (!rxq->rx_deferred_start) {
+ __rte_assume(j < RTE_MAX_QUEUES_PER_PORT);
bp->eth_dev->data->rx_queue_state[j] =
RTE_ETH_QUEUE_STATE_STARTED;
rxq->rx_started = true;
@@ -930,6 +931,7 @@ static int bnxt_start_nic(struct bnxt *bp)
struct bnxt_tx_queue *txq = bp->tx_queues[j];
if (!txq->tx_deferred_start) {
+ __rte_assume(j < RTE_MAX_QUEUES_PER_PORT);
bp->eth_dev->data->tx_queue_state[j] =
RTE_ETH_QUEUE_STATE_STARTED;
txq->tx_started = true;
@@ -584,6 +584,7 @@ int bnxt_rx_queue_stop(struct rte_eth_dev *dev, uint16_t rx_queue_id)
return -EINVAL;
}
+ __rte_assume(q_id < RTE_MAX_QUEUES_PER_PORT);
dev->data->rx_queue_state[q_id] = RTE_ETH_QUEUE_STATE_STOPPED;
rxq->rx_started = false;
PMD_DRV_LOG_LINE(DEBUG, "Rx queue stopped");
@@ -1868,6 +1868,7 @@ igb_dev_clear_queues(struct rte_eth_dev *dev)
struct igb_rx_queue *rxq;
for (i = 0; i < dev->data->nb_tx_queues; i++) {
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
txq = dev->data->tx_queues[i];
if (txq != NULL) {
igb_tx_queue_release_mbufs(txq);
@@ -1877,6 +1878,7 @@ igb_dev_clear_queues(struct rte_eth_dev *dev)
}
for (i = 0; i < dev->data->nb_rx_queues; i++) {
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
rxq = dev->data->rx_queues[i];
if (rxq != NULL) {
igb_rx_queue_release_mbufs(rxq);
@@ -111,12 +111,14 @@ fs_set_queues_state_start(struct rte_eth_dev *dev)
uint16_t i;
for (i = 0; i < dev->data->nb_rx_queues; i++) {
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
rxq = dev->data->rx_queues[i];
if (rxq != NULL && !rxq->info.conf.rx_deferred_start)
dev->data->rx_queue_state[i] =
RTE_ETH_QUEUE_STATE_STARTED;
}
for (i = 0; i < dev->data->nb_tx_queues; i++) {
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
txq = dev->data->tx_queues[i];
if (txq != NULL && !txq->info.conf.tx_deferred_start)
dev->data->tx_queue_state[i] =
@@ -176,14 +178,18 @@ fs_set_queues_state_stop(struct rte_eth_dev *dev)
{
uint16_t i;
- for (i = 0; i < dev->data->nb_rx_queues; i++)
+ for (i = 0; i < dev->data->nb_rx_queues; i++) {
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
if (dev->data->rx_queues[i] != NULL)
dev->data->rx_queue_state[i] =
RTE_ETH_QUEUE_STATE_STOPPED;
- for (i = 0; i < dev->data->nb_tx_queues; i++)
+ }
+ for (i = 0; i < dev->data->nb_tx_queues; i++) {
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
if (dev->data->tx_queues[i] != NULL)
dev->data->tx_queue_state[i] =
RTE_ETH_QUEUE_STATE_STOPPED;
+ }
}
static int
@@ -1309,6 +1309,7 @@ hns3_start_tqps(struct hns3_hw *hw)
hns3_enable_all_queues(hw, true);
for (i = 0; i < hw->data->nb_tx_queues; i++) {
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
txq = hw->data->tx_queues[i];
if (txq->enabled)
hw->data->tx_queue_state[i] =
@@ -1316,6 +1317,7 @@ hns3_start_tqps(struct hns3_hw *hw)
}
for (i = 0; i < hw->data->nb_rx_queues; i++) {
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
rxq = hw->data->rx_queues[i];
if (rxq->enabled)
hw->data->rx_queue_state[i] =
@@ -3385,7 +3385,8 @@ ixgbe_dev_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
stats->opackets = hw_stats->gptc;
stats->obytes = hw_stats->gotc;
- for (i = 0; i < IXGBE_QUEUE_STAT_COUNTERS; i++) {
+ for (i = 0; i < RTE_MIN_T(IXGBE_QUEUE_STAT_COUNTERS,
+ RTE_ETHDEV_QUEUE_STAT_CNTRS, typeof(i)); i++) {
stats->q_ipackets[i] = hw_stats->qprc[i];
stats->q_opackets[i] = hw_stats->qptc[i];
stats->q_ibytes[i] = hw_stats->qbrc[i];
@@ -154,6 +154,7 @@ mana_start_tx_queues(struct rte_eth_dev *dev)
txq->gdma_cq.count, txq->gdma_cq.size,
txq->gdma_cq.head);
+ __rte_assume(i < RTE_MAX_QUEUES_PER_PORT);
dev->data->tx_queue_state[i] = RTE_ETH_QUEUE_STATE_STARTED;
}
@@ -1470,42 +1470,52 @@ vmxnet3_dev_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
struct vmxnet3_hw *hw = dev->data->dev_private;
struct UPT1_TxStats txStats;
struct UPT1_RxStats rxStats;
+ uint64_t packets, bytes;
VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD, VMXNET3_CMD_GET_STATS);
for (i = 0; i < hw->num_tx_queues; i++) {
vmxnet3_tx_stats_get(hw, i, &txStats);
- stats->q_opackets[i] = txStats.ucastPktsTxOK +
+ packets = txStats.ucastPktsTxOK +
txStats.mcastPktsTxOK +
txStats.bcastPktsTxOK;
- stats->q_obytes[i] = txStats.ucastBytesTxOK +
+ bytes = txStats.ucastBytesTxOK +
txStats.mcastBytesTxOK +
txStats.bcastBytesTxOK;
- stats->opackets += stats->q_opackets[i];
- stats->obytes += stats->q_obytes[i];
+ stats->opackets += packets;
+ stats->obytes += bytes;
stats->oerrors += txStats.pktsTxError + txStats.pktsTxDiscard;
+
+ if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
+ stats->q_opackets[i] = packets;
+ stats->q_obytes[i] = bytes;
+ }
}
for (i = 0; i < hw->num_rx_queues; i++) {
vmxnet3_rx_stats_get(hw, i, &rxStats);
- stats->q_ipackets[i] = rxStats.ucastPktsRxOK +
+ packets = rxStats.ucastPktsRxOK +
rxStats.mcastPktsRxOK +
rxStats.bcastPktsRxOK;
- stats->q_ibytes[i] = rxStats.ucastBytesRxOK +
+ bytes = rxStats.ucastBytesRxOK +
rxStats.mcastBytesRxOK +
rxStats.bcastBytesRxOK;
- stats->ipackets += stats->q_ipackets[i];
- stats->ibytes += stats->q_ibytes[i];
-
- stats->q_errors[i] = rxStats.pktsRxError;
+ stats->ipackets += packets;
+ stats->ibytes += bytes;
stats->ierrors += rxStats.pktsRxError;
stats->imissed += rxStats.pktsRxOutOfBuf;
+
+ if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
+ stats->q_ipackets[i] = packets;
+ stats->q_ibytes[i] = bytes;
+ stats->q_errors[i] = rxStats.pktsRxError;
+ }
}
return 0;
@@ -1521,8 +1531,6 @@ vmxnet3_dev_stats_reset(struct rte_eth_dev *dev)
VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD, VMXNET3_CMD_GET_STATS);
- RTE_BUILD_BUG_ON(RTE_ETHDEV_QUEUE_STAT_CNTRS < VMXNET3_MAX_TX_QUEUES);
-
for (i = 0; i < hw->num_tx_queues; i++) {
vmxnet3_hw_tx_stats_get(hw, i, &txStats);
memcpy(&hw->snapshot_tx_stats[i], &txStats,
@@ -50,6 +50,17 @@ extern "C" {
#define __rte_constant(e) __extension__(__builtin_constant_p(e))
#endif
+/**
+ * Hint to the optimizer.
+ */
+#if defined(RTE_TOOLCHAIN_GCC)
+#define __rte_assume(e) __attribute__((assume(e)))
+#elif defined(RTE_TOOLCHAIN_CLANG)
+#define __rte_assume(e) __extension__(__builtin_assume(e))
+#else
+#define __rte_assume(e) do {} while (0)
+#endif
+
/*
* RTE_TOOLCHAIN_GCC is defined if the target is built with GCC,
* while a host application (like pmdinfogen) may have another compiler.