DPDK patches and discussions
 help / color / mirror / Atom feed
From: Ye Xiaolong <xiaolong.ye@intel.com>
To: Jiang Mao <maox.jiang@intel.com>
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH 1/2] net/i40e: fix binding interrupt without msix vectors
Date: Tue, 9 Jun 2020 10:02:19 +0800	[thread overview]
Message-ID: <20200609020219.GA7842@intel.com> (raw)
In-Reply-To: <20200609013712.27026-1-maox.jiang@intel.com>

On 06/09, Jiang Mao wrote:
>The value of vsi->nb_msix shouldn`t be zero, otherwise, all of interrupts will be bind to vector 0.
>
>Signed-off-by: Jiang Mao <maox.jiang@intel.com>
>
>Fixes: 4861cde461 (i40e: new poll mode driver)

Fixes tag should be put before SOB line and please cc stable@dpdk.org.

>---
> drivers/net/i40e/i40e_ethdev.c | 41 +++++++++++++++++++++++-----------
> drivers/net/i40e/i40e_ethdev.h |  2 +-
> 2 files changed, 29 insertions(+), 14 deletions(-)
>
>diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
>index 970a31cb2..33ed556c8 100644
>--- a/drivers/net/i40e/i40e_ethdev.c
>+++ b/drivers/net/i40e/i40e_ethdev.c
>@@ -2013,7 +2013,7 @@ __vsi_queues_bind_intr(struct i40e_vsi *vsi, uint16_t msix_vect,
> 	I40E_WRITE_FLUSH(hw);
> }
> 
>-void
>+int
> i40e_vsi_queues_bind_intr(struct i40e_vsi *vsi, uint16_t itr_idx)
> {
> 	struct rte_eth_dev *dev = vsi->adapter->eth_dev;
>@@ -2033,10 +2033,14 @@ i40e_vsi_queues_bind_intr(struct i40e_vsi *vsi, uint16_t itr_idx)
> 
> 	/* VF bind interrupt */
> 	if (vsi->type == I40E_VSI_SRIOV) {
>+		if (vsi->nb_msix == 0) {
>+			PMD_DRV_LOG(ERR, "No msix resource");
>+			return -EINVAL;
>+		}
> 		__vsi_queues_bind_intr(vsi, msix_vect,
>-				       vsi->base_queue, vsi->nb_qps,
>-				       itr_idx);
>-		return;
>+							   vsi->base_queue, vsi->nb_qps,
>+							   itr_idx);

The indentation seems weird, please keep it as before, and same for other
occurrences in this patch.

>+		return 0;
> 	}
> 
> 	/* PF & VMDq bind interrupt */
>@@ -2053,16 +2057,19 @@ i40e_vsi_queues_bind_intr(struct i40e_vsi *vsi, uint16_t itr_idx)
> 	}
> 
> 	for (i = 0; i < vsi->nb_used_qps; i++) {
>-		if (nb_msix <= 1) {
>+		if (vsi->nb_msix == 0) {
>+			PMD_DRV_LOG(ERR, "No msix resource");
>+			return -EINVAL;
>+		} else if (nb_msix <= 1) {
> 			if (!rte_intr_allow_others(intr_handle))
> 				/* allow to share MISC_VEC_ID */
> 				msix_vect = I40E_MISC_VEC_ID;
> 
> 			/* no enough msix_vect, map all to one */
> 			__vsi_queues_bind_intr(vsi, msix_vect,
>-					       vsi->base_queue + i,
>-					       vsi->nb_used_qps - i,
>-					       itr_idx);
>+								   vsi->base_queue + i,
>+								   vsi->nb_used_qps - i,
>+								   itr_idx);
> 			for (; !!record && i < vsi->nb_used_qps; i++)
> 				intr_handle->intr_vec[queue_idx + i] =
> 					msix_vect;
>@@ -2078,6 +2085,8 @@ i40e_vsi_queues_bind_intr(struct i40e_vsi *vsi, uint16_t itr_idx)
> 		msix_vect++;
> 		nb_msix--;
> 	}
>+
>+	return 0;
> }
> 
> static void
>@@ -2318,21 +2327,27 @@ i40e_dev_start(struct rte_eth_dev *dev)
> 	/* Map queues with MSIX interrupt */
> 	main_vsi->nb_used_qps = dev->data->nb_rx_queues -
> 		pf->nb_cfg_vmdq_vsi * RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM;
>-	i40e_vsi_queues_bind_intr(main_vsi, I40E_ITR_INDEX_DEFAULT);
>+	ret = i40e_vsi_queues_bind_intr(main_vsi, I40E_ITR_INDEX_DEFAULT);
>+	if (ret < 0)
>+		return ret;
> 	i40e_vsi_enable_queues_intr(main_vsi);
> 
> 	/* Map VMDQ VSI queues with MSIX interrupt */
> 	for (i = 0; i < pf->nb_cfg_vmdq_vsi; i++) {
> 		pf->vmdq[i].vsi->nb_used_qps = RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM;
>-		i40e_vsi_queues_bind_intr(pf->vmdq[i].vsi,
>-					  I40E_ITR_INDEX_DEFAULT);
>+		ret = i40e_vsi_queues_bind_intr(pf->vmdq[i].vsi,
>+										I40E_ITR_INDEX_DEFAULT);
>+		if (ret < 0)
>+			return ret;
> 		i40e_vsi_enable_queues_intr(pf->vmdq[i].vsi);
> 	}
> 
> 	/* enable FDIR MSIX interrupt */
> 	if (pf->fdir.fdir_vsi) {
>-		i40e_vsi_queues_bind_intr(pf->fdir.fdir_vsi,
>-					  I40E_ITR_INDEX_NONE);
>+		ret = i40e_vsi_queues_bind_intr(pf->fdir.fdir_vsi,
>+										I40E_ITR_INDEX_NONE);
>+		if (ret < 0)
>+			return ret;
> 		i40e_vsi_enable_queues_intr(pf->fdir.fdir_vsi);
> 	}
> 
>diff --git a/drivers/net/i40e/i40e_ethdev.h b/drivers/net/i40e/i40e_ethdev.h
>index e5d0ce53f..33fbe776b 100644
>--- a/drivers/net/i40e/i40e_ethdev.h
>+++ b/drivers/net/i40e/i40e_ethdev.h
>@@ -1248,7 +1248,7 @@ void i40e_update_vsi_stats(struct i40e_vsi *vsi);
> void i40e_pf_disable_irq0(struct i40e_hw *hw);
> void i40e_pf_enable_irq0(struct i40e_hw *hw);
> int i40e_dev_link_update(struct rte_eth_dev *dev, int wait_to_complete);
>-void i40e_vsi_queues_bind_intr(struct i40e_vsi *vsi, uint16_t itr_idx);
>+int i40e_vsi_queues_bind_intr(struct i40e_vsi *vsi, uint16_t itr_idx);
> void i40e_vsi_queues_unbind_intr(struct i40e_vsi *vsi);
> int i40e_vsi_vlan_pvid_set(struct i40e_vsi *vsi,
> 			   struct i40e_vsi_vlan_pvid_info *info);
>-- 
>2.17.1
>

  parent reply	other threads:[~2020-06-09  2:10 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-06-09  1:37 Jiang Mao
2020-06-09  1:37 ` [dpdk-dev] [PATCH 2/2] net/i40e: fix fdir allocating msix resource error Jiang Mao
2020-06-09  2:02 ` Ye Xiaolong [this message]
2020-06-09  2:25 [dpdk-dev] [PATCH 1/2] net/i40e: fix binding interrupt without msix vectors Jiang Mao
2020-06-17  5:42 ` Huang, ZhiminX
2020-07-08 19:50 ` Jiang Mao
2020-07-21  3:57   ` Jeff Guo
2020-07-21  6:50     ` Jiang, MaoX

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200609020219.GA7842@intel.com \
    --to=xiaolong.ye@intel.com \
    --cc=dev@dpdk.org \
    --cc=maox.jiang@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).