DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Zhang, Qi Z" <qi.z.zhang@intel.com>
To: "Ye, MingjinX" <mingjinx.ye@intel.com>, "dev@dpdk.org" <dev@dpdk.org>
Cc: "Yang, Qiming" <qiming.yang@intel.com>,
	"Zhou, YidingX" <yidingx.zhou@intel.com>,
	"stable@dpdk.org" <stable@dpdk.org>
Subject: RE: [PATCH v5] net/ice: fix crash on closing representor ports
Date: Mon, 6 Nov 2023 12:05:35 +0000	[thread overview]
Message-ID: <DM4PR11MB59943473433BD5689E523A2FD7AAA@DM4PR11MB5994.namprd11.prod.outlook.com> (raw)
In-Reply-To: <20231106100001.40213-1-mingjinx.ye@intel.com>



> -----Original Message-----
> From: Ye, MingjinX <mingjinx.ye@intel.com>
> Sent: Monday, November 6, 2023 6:00 PM
> To: dev@dpdk.org
> Cc: Yang, Qiming <qiming.yang@intel.com>; Zhou, YidingX
> <yidingx.zhou@intel.com>; Ye, MingjinX <mingjinx.ye@intel.com>;
> stable@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>
> Subject: [PATCH v5] net/ice: fix crash on closing representor ports
> 
> The data resource in struct rte_eth_dev is cleared and points to NULL when
> the DCF port is closed.
> 
> If the DCF representor port is closed after the DCF port is closed, a
> segmentation fault occurs because the representor port accesses the data
> resource released by the DCF port.
> 
> This patch fixes this issue by synchronizing the state of DCF ports and
> representor ports to the peer in real time when their state changes.
> 
> Fixes: 5674465a32c8 ("net/ice: add DCF VLAN handling")
> Fixes: da9cdcd1f372 ("net/ice: fix crash on representor port closing")
> Fixes: 7564d5509611 ("net/ice: add DCF hardware initialization")
> Fixes: c7e1a1a3bfeb ("net/ice: refactor DCF VLAN handling")
> Fixes: 1a86f4dbdf42 ("net/ice: support DCF device reset")

These fixlines does not make sense to me, I believe the issue comes from when we enabled port representor for DCF, 
A patch expose the issue does not imply it cause the issue.

> Cc: stable@dpdk.org
> 
> Signed-off-by: Mingjin Ye <mingjinx.ye@intel.com>
> ---
> v2: Reformat code to remove unneeded fixlines.
> ---
> v3: New solution.
> ---
> v4: Optimize v2 patch.
> ---
> v5: optimization.
> ---
>  drivers/net/ice/ice_dcf_ethdev.c         | 30 ++++++++++++--
>  drivers/net/ice/ice_dcf_ethdev.h         |  3 ++
>  drivers/net/ice/ice_dcf_vf_representor.c | 50 ++++++++++++++++++++++--
>  3 files changed, 77 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/net/ice/ice_dcf_ethdev.c b/drivers/net/ice/ice_dcf_ethdev.c
> index 065ec728c2..eea24ee3a9 100644
> --- a/drivers/net/ice/ice_dcf_ethdev.c
> +++ b/drivers/net/ice/ice_dcf_ethdev.c
> @@ -1618,6 +1618,26 @@ ice_dcf_free_repr_info(struct ice_dcf_adapter
> *dcf_adapter)
>  	}
>  }
> 
> +int
> +ice_dcf_handle_vf_repr_close(struct ice_dcf_adapter *dcf_adapter,
> +				uint16_t vf_id)
> +{
> +	struct ice_dcf_repr_info *vf_rep_info;
> +
> +	if (dcf_adapter->num_reprs >= vf_id) {
> +		PMD_DRV_LOG(ERR, "Invalid VF id: %d", vf_id);
> +		return -1;
> +	}
> +
> +	if (!dcf_adapter->repr_infos)
> +		return 0;
> +
> +	vf_rep_info = &dcf_adapter->repr_infos[vf_id];
> +	vf_rep_info->vf_rep_eth_dev = NULL;
> +
> +	return 0;
> +}
> +
>  static int
>  ice_dcf_init_repr_info(struct ice_dcf_adapter *dcf_adapter)  { @@ -1641,11
> +1661,10 @@ ice_dcf_dev_close(struct rte_eth_dev *dev)
>  	if (rte_eal_process_type() != RTE_PROC_PRIMARY)
>  		return 0;
> 
> +	ice_dcf_vf_repr_notify_all(adapter, false);
>  	(void)ice_dcf_dev_stop(dev);
> 
>  	ice_free_queues(dev);
> -
> -	ice_dcf_free_repr_info(adapter);
>  	ice_dcf_uninit_parent_adapter(dev);
>  	ice_dcf_uninit_hw(dev, &adapter->real_hw);
> 
> @@ -1835,7 +1854,7 @@ ice_dcf_dev_reset(struct rte_eth_dev *dev)
>  		ice_dcf_reset_hw(dev, hw);
>  	}
> 
> -	ret = ice_dcf_dev_uninit(dev);
> +	ret = ice_dcf_dev_close(dev);
>  	if (ret)
>  		return ret;
> 
> @@ -1938,12 +1957,17 @@ ice_dcf_dev_init(struct rte_eth_dev *eth_dev)
>  	}
> 
>  	dcf_config_promisc(adapter, false, false);
> +	ice_dcf_vf_repr_notify_all(adapter, true);
> +
>  	return 0;
>  }
> 
>  static int
>  ice_dcf_dev_uninit(struct rte_eth_dev *eth_dev)  {
> +	struct ice_dcf_adapter *adapter = eth_dev->data->dev_private;
> +
> +	ice_dcf_free_repr_info(adapter);
>  	ice_dcf_dev_close(eth_dev);
> 
>  	return 0;
> diff --git a/drivers/net/ice/ice_dcf_ethdev.h
> b/drivers/net/ice/ice_dcf_ethdev.h
> index 4baaec4b8b..6dcbaac5eb 100644
> --- a/drivers/net/ice/ice_dcf_ethdev.h
> +++ b/drivers/net/ice/ice_dcf_ethdev.h
> @@ -60,6 +60,7 @@ struct ice_dcf_vf_repr {
>  	struct rte_ether_addr mac_addr;
>  	uint16_t switch_domain_id;
>  	uint16_t vf_id;
> +	bool dcf_valid;
> 
>  	struct ice_dcf_vlan outer_vlan_info; /* DCF always handle outer VLAN
> */  }; @@ -80,6 +81,8 @@ int ice_dcf_vf_repr_init(struct rte_eth_dev
> *vf_rep_eth_dev, void *init_param);  int ice_dcf_vf_repr_uninit(struct
> rte_eth_dev *vf_rep_eth_dev);  int ice_dcf_vf_repr_init_vlan(struct
> rte_eth_dev *vf_rep_eth_dev);  void ice_dcf_vf_repr_stop_all(struct
> ice_dcf_adapter *dcf_adapter);
> +void ice_dcf_vf_repr_notify_all(struct ice_dcf_adapter *dcf_adapter,
> +bool valid); int ice_dcf_handle_vf_repr_close(struct ice_dcf_adapter
> +*dcf_adapter, uint16_t vf_id);
>  bool ice_dcf_adminq_need_retry(struct ice_adapter *ad);
> 
>  #endif /* _ICE_DCF_ETHDEV_H_ */
> diff --git a/drivers/net/ice/ice_dcf_vf_representor.c
> b/drivers/net/ice/ice_dcf_vf_representor.c
> index b9fcfc80ad..6c342798ac 100644
> --- a/drivers/net/ice/ice_dcf_vf_representor.c
> +++ b/drivers/net/ice/ice_dcf_vf_representor.c
> @@ -50,9 +50,32 @@ ice_dcf_vf_repr_dev_stop(struct rte_eth_dev *dev)
>  	return 0;
>  }
> 
> +static bool

This function return Boolean, but it never be checked.

> +ice_dcf_vf_repr_set_dcf_valid(struct rte_eth_dev *dev, bool valid) {
> +	struct ice_dcf_vf_repr *repr = dev->data->dev_private;
> +
> +	if (!repr)
> +		return false;

Is this check necessary? If repr = null, means the representor port already be closed, the function will not be called for that port.

> +
> +	repr->dcf_valid = valid;
> +
> +	return true;
> +}
> +
>  static int
>  ice_dcf_vf_repr_dev_close(struct rte_eth_dev *dev)  {
> +	struct ice_dcf_vf_repr *repr = dev->data->dev_private;
> +	struct ice_dcf_adapter *dcf_adapter;
> +
> +	if (repr->dcf_valid) {
> +		dcf_adapter = repr->dcf_eth_dev->data->dev_private;
> +		ice_dcf_handle_vf_repr_close(dcf_adapter, repr->vf_id);
> +		/* Commitment to not access DCF port resources */
> +		repr->dcf_valid = false;

This is not necessary, as you already called ice_dcf_handle_vf_repr_close to notify dcf.
And the repr will be freed in ice_dcf_vf_repr_uninit anyway.


> +	}
> +
>  	return ice_dcf_vf_repr_uninit(dev);
>  }
> 
> @@ -111,14 +134,15 @@ ice_dcf_vf_repr_link_update(__rte_unused struct
> rte_eth_dev *ethdev,  static __rte_always_inline struct ice_dcf_hw *
> ice_dcf_vf_repr_hw(struct ice_dcf_vf_repr *repr)  {
> -	struct ice_dcf_adapter *dcf_adapter =
> -			repr->dcf_eth_dev->data->dev_private;
> +	struct ice_dcf_adapter *dcf_adapter;
> 
> -	if (!dcf_adapter) {
> +	if (!repr->dcf_valid) {
>  		PMD_DRV_LOG(ERR, "DCF for VF representor has been
> released\n");
>  		return NULL;
>  	}
> 
> +	dcf_adapter = repr->dcf_eth_dev->data->dev_private;
> +
>  	return &dcf_adapter->real_hw;
>  }
> 
> @@ -414,6 +438,7 @@ ice_dcf_vf_repr_init(struct rte_eth_dev
> *vf_rep_eth_dev, void *init_param)
>  	repr->dcf_eth_dev = param->dcf_eth_dev;
>  	repr->switch_domain_id = param->switch_domain_id;
>  	repr->vf_id = param->vf_id;
> +	repr->dcf_valid = true;
>  	repr->outer_vlan_info.port_vlan_ena = false;
>  	repr->outer_vlan_info.stripping_ena = false;
>  	repr->outer_vlan_info.tpid = RTE_ETHER_TYPE_VLAN; @@ -488,3
> +513,22 @@ ice_dcf_vf_repr_stop_all(struct ice_dcf_adapter *dcf_adapter)
>  			vf_rep_eth_dev->data->dev_started = 0;
>  	}
>  }
> +
> +void
> +ice_dcf_vf_repr_notify_all(struct ice_dcf_adapter *dcf_adapter, bool
> +valid) {
> +	uint16_t vf_id;
> +	struct rte_eth_dev *vf_rep_eth_dev;
> +
> +	if (!dcf_adapter->repr_infos)
> +		return;
> +
> +	for (vf_id = 0; vf_id < dcf_adapter->real_hw.num_vfs; vf_id++) {
> +		vf_rep_eth_dev = dcf_adapter-
> >repr_infos[vf_id].vf_rep_eth_dev;
> +
> +		if (!vf_rep_eth_dev)
> +			continue;
> +
> +		ice_dcf_vf_repr_set_dcf_valid(vf_rep_eth_dev, valid);
> +	}
> +}
> --
> 2.25.1


  reply	other threads:[~2023-11-06 12:05 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-10-26  9:51 [PATCH] net/iavf: " Mingjin Ye
2023-10-27  7:17 ` David Marchand
2023-10-30  2:35 ` [PATCH v2] " Mingjin Ye
2023-10-30  8:44 ` [PATCH v2] net/ice: " Mingjin Ye
2023-11-01  1:11   ` Zhang, Qi Z
2023-11-01 10:13   ` [PATCH v3] " Mingjin Ye
2023-11-01 10:48     ` Zhang, Qi Z
2023-11-02  2:38       ` Ye, MingjinX
2023-11-02  3:59         ` Zhang, Qi Z
2023-11-02 10:11     ` [PATCH v4] " Mingjin Ye
2023-11-06  1:23       ` Zhang, Qi Z
2023-11-06 10:00       ` [PATCH v5] " Mingjin Ye
2023-11-06 12:05         ` Zhang, Qi Z [this message]
2023-11-07 10:12         ` [PATCH v6] " Mingjin Ye
2023-11-07 12:18           ` Zhang, Qi Z
2023-11-08 11:39           ` [PATCH v7] " Mingjin Ye
2023-11-09  0:26             ` Zhang, Qi Z

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=DM4PR11MB59943473433BD5689E523A2FD7AAA@DM4PR11MB5994.namprd11.prod.outlook.com \
    --to=qi.z.zhang@intel.com \
    --cc=dev@dpdk.org \
    --cc=mingjinx.ye@intel.com \
    --cc=qiming.yang@intel.com \
    --cc=stable@dpdk.org \
    --cc=yidingx.zhou@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).