On 2015/5/29 16:11, Chen, Jing D wrote:
> From: "Chen Jing D(Mark)" <jing.d.chen at intel.com>
>
> In fm10k, PF driver needs to communicate with switch through
> mailbox if it needs to add/delete MAC address.
> This fix will validate if switch is ready before going forward.
> Then, it is necessary to acquire LPORT_MAP info after issuing
> MAC addr request to switch.
>
> Signed-off-by: Chen Jing D(Mark) <jing.d.chen at intel.com>
> ---
> drivers/net/fm10k/fm10k_ethdev.c | 34 +++++++++++++++++++++++++++++++---
> 1 files changed, 31 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/fm10k/fm10k_ethdev.c
> b/drivers/net/fm10k/fm10k_ethdev.c
> index 19e718b..dedfbb4 100644
> --- a/drivers/net/fm10k/fm10k_ethdev.c
> +++ b/drivers/net/fm10k/fm10k_ethdev.c
> @@ -45,6 +45,10 @@
> #define FM10K_MBXLOCK_DELAY_US 20
> #define UINT64_LOWER_32BITS_MASK 0x00000000ffffffffULL
>
> +/* Max try times to aquire switch status */
> +#define MAX_QUERY_SWITCH_STATE_TIMES 10
> +/* Wait interval to get switch status */
> +#define WAIT_SWITCH_MSG_US 100000
> /* Number of chars per uint32 type */
> #define CHARS_PER_UINT32 (sizeof(uint32_t))
> #define BIT_MASK_PER_UINT32 ((1 << CHARS_PER_UINT32) - 1)
> @@ -1802,6 +1806,32 @@ eth_fm10k_dev_init(struct rte_eth_dev *dev)
> fm10k_dev_enable_intr_vf(dev);
> }
>
> + /* Enable uio intr after callback registered */
> + rte_intr_enable(&(dev->pci_dev->intr_handle));
> +
> + hw->mac.ops.update_int_moderator(hw);
> +
> + /* Make sure Switch Manager is ready before going forward. */
> + if (hw->mac.type == fm10k_mac_pf) {
> + int switch_ready = 0;
> + int i;
> +
> + for (i = 0; i < MAX_QUERY_SWITCH_STATE_TIMES; i++) {
> + fm10k_mbx_lock(hw);
> + hw->mac.ops.get_host_state(hw, &switch_ready);
> + fm10k_mbx_unlock(hw);
> + if (switch_ready)
> + break;
> + /* Delay some time to acquire async LPORT_MAP info. */
> + rte_delay_us(WAIT_SWITCH_MSG_US);
> + }
> +
> + if (switch_ready == 0) {
> + PMD_INIT_LOG(ERR, "switch is not ready");
> + return -1;
Here better to return -EIO or other error code? "-1" seems not keep the
same style with other return routine in this function.
Thanks,
Michael
> + }
> + }
> +
> /*
> * Below function will trigger operations on mailbox, acquire lock to
> * avoid race condition from interrupt handler. Operations on mailbox
> @@ -1811,7 +1841,7 @@ eth_fm10k_dev_init(struct rte_eth_dev *dev)
> */
> fm10k_mbx_lock(hw);
> /* Enable port first */
> - hw->mac.ops.update_lport_state(hw, 0, 0, 1);
> + hw->mac.ops.update_lport_state(hw, hw->mac.dglort_map, 1, 1);
>
> /* Update default vlan */
> hw->mac.ops.update_vlan(hw, hw->mac.default_vid, 0, true);
> @@ -1831,8 +1861,6 @@ eth_fm10k_dev_init(struct rte_eth_dev *dev)
>
> fm10k_mbx_unlock(hw);
>
> - /* enable uio intr after callback registered */
> - rte_intr_enable(&(dev->pci_dev->intr_handle));
>
> return 0;
> }
