netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Wojciech Drewek <wojciech.drewek@intel.com>
To: Karthik Sundaravel <ksundara@redhat.com>,
	Suman Ghosh <sumang@marvell.com>,
	Jacob Keller <jacob.e.keller@intel.com>
Cc: "pmenzel@molgen.mpg.de" <pmenzel@molgen.mpg.de>,
	"aharivel@redhat.com" <aharivel@redhat.com>,
	"michal.swiatkowski@linux.intel.com"
	<michal.swiatkowski@linux.intel.com>,
	"jiri@resnulli.us" <jiri@resnulli.us>,
	"cfontain@redhat.com" <cfontain@redhat.com>,
	"intel-wired-lan@lists.osuosl.org"
	<intel-wired-lan@lists.osuosl.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"edumazet@google.com" <edumazet@google.com>,
	"anthony.l.nguyen@intel.com" <anthony.l.nguyen@intel.com>,
	"vchundur@redhat.com" <vchundur@redhat.com>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"kuba@kernel.org" <kuba@kernel.org>,
	"rjarry@redhat.com" <rjarry@redhat.com>,
	"pabeni@redhat.com" <pabeni@redhat.com>,
	"davem@davemloft.net" <davem@davemloft.net>
Subject: Re: [Intel-wired-lan] [EXTERNAL] [PATCH v5] ice: Add get/set hw address for VFs using devlink commands
Date: Mon, 18 Mar 2024 16:04:56 +0100	[thread overview]
Message-ID: <0a152a46-3800-443a-b370-50cf071f7c13@intel.com> (raw)
In-Reply-To: <CAPh+B4JH-Wb1qetsqfnCeLcxTpOM72TSTwreqP9H6sxnP9Kn9A@mail.gmail.com>



On 18.03.2024 12:55, Karthik Sundaravel wrote:
> On Fri, Mar 8, 2024 at 3:28 PM Suman Ghosh <sumang@marvell.com> wrote:
>>
>>> ----------------------------------------------------------------------
>>> Changing the MAC address of the VFs are not available via devlink. Add
>>> the function handlers to set and get the HW address for the VFs.
>>>
>>> Signed-off-by: Karthik Sundaravel <ksundara@redhat.com>
>>> ---
>>> drivers/net/ethernet/intel/ice/ice_devlink.c | 78 +++++++++++++++++++-
>>> drivers/net/ethernet/intel/ice/ice_sriov.c   | 62 ++++++++++++++++
>>> drivers/net/ethernet/intel/ice/ice_sriov.h   |  8 ++
>>> 3 files changed, 147 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/net/ethernet/intel/ice/ice_devlink.c
>>> b/drivers/net/ethernet/intel/ice/ice_devlink.c
>>> index 80dc5445b50d..39d4d79ac731 100644
>>> --- a/drivers/net/ethernet/intel/ice/ice_devlink.c
>>> +++ b/drivers/net/ethernet/intel/ice/ice_devlink.c
>>> @@ -1576,6 +1576,81 @@ void ice_devlink_destroy_pf_port(struct ice_pf
>>> *pf)
>>>       devlink_port_unregister(&pf->devlink_port);
>>> }
>>>
>>> +/**
>>> + * ice_devlink_port_get_vf_fn_mac - .port_fn_hw_addr_get devlink
>>> +handler
>>> + * @port: devlink port structure
>>> + * @hw_addr: MAC address of the port
>>> + * @hw_addr_len: length of MAC address
>>> + * @extack: extended netdev ack structure
>>> + *
>>> + * Callback for the devlink .port_fn_hw_addr_get operation
>>> + * Return: zero on success or an error code on failure.
>>> + */
>>> +
>>> +static int ice_devlink_port_get_vf_fn_mac(struct devlink_port *port,
>>> +                                        u8 *hw_addr, int *hw_addr_len,
>>> +                                        struct netlink_ext_ack *extack)
>>> +{
>>> +      struct devlink_port_attrs *attrs = &port->attrs;
>> [Suman] I agree with Wojciech about using container_of:
> 
> [Karthik] when I use container_of(), on some occasions I get core dump
> in get and set functions.
> These issues were not seen in the earlier versions.
> Can you please suggest any pointers on what could have gone wrong ?
> 
> struct ice_vf *vf = container_of(port, struct ice_vf, devlink_port);
> 
> [  597.658325] ------------[ cut here ]------------
> [  597.658329] refcount_t: underflow; use-after-free.
> [  597.658430] CPU: 18 PID: 1926 Comm: devlink Not tainted 6.8.0-rc5-dirty #1
> [  ...]
> [  597.658506]  ? refcount_warn_saturate+0xbe/0x110
> [  597.658509]  ice_devlink_port_get_vf_fn_mac+0x39/0x70 [ice]
> [  597.658607]  ? __pfx_ice_devlink_port_get_vf_fn_mac+0x10/0x10 [ice]
> [  597.658676]  devlink_nl_port_fill+0x314/0xa30
> [  ...]
> [  597.658835] ---[ end trace 0000000000000000 ]---
> 
> 
> [  859.989482] ------------[ cut here ]------------
> [  859.989485] refcount_t: saturated; leaking memory.
> [  859.989500] WARNING: CPU: 0 PID: 1965 at lib/refcount.c:19
> refcount_warn_saturate+0x9b/0x110
> [  ...]
> [  859.989671]  ? refcount_warn_saturate+0x9b/0x110
> [  859.989674]  ice_get_vf_by_id+0x87/0xa0 [ice]
> [  859.989777]  ice_set_vf_fn_mac+0x33/0x150 [ice]
> [  859.989858]  ice_devlink_port_set_vf_fn_mac+0x61/0x90 [ice]
> [  859.989940]  devlink_nl_port_set_doit+0x1d3/0x610
> [  ...]
> [  952.413933] ---[ end trace 0000000000000000 ]---

Ok, I think we forgot about kref here.
Once you have a VF pointer you have to inc the ref count using
kref_get_unless_zero and you have to check return value because the VF
might be already freed. When you don't need the VF's pointer anymore call ice_put_vf.
Would be cool to have Jake's opinion on that though since he implemented it.

> 
>>
>>> +      struct devlink_port_pci_vf_attrs *pci_vf;
>>> +      struct devlink *devlink = port->devlink;
>>> +      struct ice_pf *pf;
>>> +      struct ice_vf *vf;
>>> +      int vf_id;
>>> +
>>> +      pf = devlink_priv(devlink);
>>> +      pci_vf = &attrs->pci_vf;
>>> +      vf_id = pci_vf->vf;
>>> +
>>> +      vf = ice_get_vf_by_id(pf, vf_id);
>>> +      if (!vf) {
>>> +              NL_SET_ERR_MSG_MOD(extack, "Unable to get the vf");
>>> +              return -EINVAL;
>>> +      }
>>> +      ether_addr_copy(hw_addr, vf->dev_lan_addr);
>>> +      *hw_addr_len = ETH_ALEN;
>>> +
>>> +      ice_put_vf(vf);
>>> +      return 0;
>>> +}
>>> +
>>> +/**
>>> + * ice_devlink_port_set_vf_fn_mac - .port_fn_hw_addr_set devlink
>>> +handler
>>> + * @port: devlink port structure
>>> + * @hw_addr: MAC address of the port
>>> + * @hw_addr_len: length of MAC address
>>> + * @extack: extended netdev ack structure
>>> + *
>>> + * Callback for the devlink .port_fn_hw_addr_set operation
>>> + * Return: zero on success or an error code on failure.
>>> + */
>>> +static int ice_devlink_port_set_vf_fn_mac(struct devlink_port *port,
>>> +                                        const u8 *hw_addr,
>>> +                                        int hw_addr_len,
>>> +                                        struct netlink_ext_ack *extack)
>>> +
>>> +{
>>> +      struct devlink_port_attrs *attrs = &port->attrs;
>>> +      struct devlink_port_pci_vf_attrs *pci_vf;
>>> +      struct devlink *devlink = port->devlink;
>>> +      struct ice_pf *pf;
>>> +      u8 mac[ETH_ALEN];
>>> +      int vf_id;
>>> +
>>> +      pf = devlink_priv(devlink);
>>> +      pci_vf = &attrs->pci_vf;
>>> +      vf_id = pci_vf->vf;
>>> +
>>> +      ether_addr_copy(mac, hw_addr);
>>> +
>>> +      return ice_set_vf_fn_mac(pf, vf_id, mac); }
>>> +
>>> +static const struct devlink_port_ops ice_devlink_vf_port_ops = {
>>> +      .port_fn_hw_addr_get = ice_devlink_port_get_vf_fn_mac,
>>> +      .port_fn_hw_addr_set = ice_devlink_port_set_vf_fn_mac, };
>>> +
>>> /**
>>>  * ice_devlink_create_vf_port - Create a devlink port for this VF
>>>  * @vf: the VF to create a port for
>>> @@ -1611,7 +1686,8 @@ int ice_devlink_create_vf_port(struct ice_vf *vf)
>>>       devlink_port_attrs_set(devlink_port, &attrs);
>>>       devlink = priv_to_devlink(pf);
>>>
>>> -      err = devlink_port_register(devlink, devlink_port, vsi->idx);
>>> +      err = devlink_port_register_with_ops(devlink, devlink_port,
>>> +                                           vsi->idx, &ice_devlink_vf_port_ops);
>>>       if (err) {
>>>               dev_err(dev, "Failed to create devlink port for VF %d, error
>>> %d\n",
>>>                       vf->vf_id, err);
>>> diff --git a/drivers/net/ethernet/intel/ice/ice_sriov.c
>>> b/drivers/net/ethernet/intel/ice/ice_sriov.c
>>> index 31314e7540f8..73cf1d9e9daa 100644
>>> --- a/drivers/net/ethernet/intel/ice/ice_sriov.c
>>> +++ b/drivers/net/ethernet/intel/ice/ice_sriov.c
>>> @@ -1216,6 +1216,68 @@ ice_get_vf_cfg(struct net_device *netdev, int
>>> vf_id, struct ifla_vf_info *ivi)
>>>       return ret;
>>> }
>>>
>>> +/**
>>> + * ice_set_vf_fn_mac
>>> + * @pf: PF to be configure
>>> + * @vf_id: VF identifier
>>> + * @mac: MAC address
>>> + *
>>> + * program VF MAC address
>>> + */
>>> +int ice_set_vf_fn_mac(struct ice_pf *pf, int vf_id, u8 *mac) {
>>> +      struct device *dev;
>>> +      struct ice_vf *vf;
>>> +      int ret;
>>> +
>>> +      dev = ice_pf_to_dev(pf);
>>> +      if (is_multicast_ether_addr(mac)) {
>>> +              dev_err(dev, "%pM not a valid unicast address\n", mac);
>>> +              return -EINVAL;
>>> +      }
>> [Suman] I would suggest put all the validation checks at the beginning of the function.
>>> +
>>> +      vf = ice_get_vf_by_id(pf, vf_id);
>> [Suman] Any reason we are passing vf_id instead of the vf itself? If you decide to pass vf itself please move the ether_addr_equal() check at the beginning also.
>>
>>> +      if (!vf)
>>> +              return -EINVAL;
>>> +
>>> +      /* nothing left to do, unicast MAC already set */
>>> +      if (ether_addr_equal(vf->dev_lan_addr, mac) &&
>>> +          ether_addr_equal(vf->hw_lan_addr, mac)) {
>>> +              ret = 0;
>>> +              goto out_put_vf;
>>> +      }
>>> +
>>> +      ret = ice_check_vf_ready_for_cfg(vf);
>>> +      if (ret)
>>> +              goto out_put_vf;
>>> +
>>> +      mutex_lock(&vf->cfg_lock);
>>> +
>>> +      /* VF is notified of its new MAC via the PF's response to the
>>> +       * VIRTCHNL_OP_GET_VF_RESOURCES message after the VF has been reset
>>> +       */
>>> +      ether_addr_copy(vf->dev_lan_addr, mac);
>>> +      ether_addr_copy(vf->hw_lan_addr, mac);
>>> +      if (is_zero_ether_addr(mac)) {
>>> +              /* VF will send VIRTCHNL_OP_ADD_ETH_ADDR message with its MAC
>>> */
>>> +              vf->pf_set_mac = false;
>>> +              dev_info(dev, "Removing MAC on VF %d. VF driver will be
>>> reinitialized\n",
>>> +                       vf->vf_id);
>>> +      } else {
>>> +              /* PF will add MAC rule for the VF */
>>> +              vf->pf_set_mac = true;
>>> +              dev_info(dev, "Setting MAC %pM on VF %d. VF driver will be
>>> reinitialized\n",
>>> +                       mac, vf_id);
>>> +      }
>>> +
>>> +      ice_reset_vf(vf, ICE_VF_RESET_NOTIFY);
>>> +      mutex_unlock(&vf->cfg_lock);
>>> +
>>> +out_put_vf:
>>> +      ice_put_vf(vf);
>>> +      return ret;
>>> +}
>>> +
>>> /**
>>>  * ice_set_vf_mac
>>>  * @netdev: network interface device structure diff --git
>>> a/drivers/net/ethernet/intel/ice/ice_sriov.h
>>> b/drivers/net/ethernet/intel/ice/ice_sriov.h
>>> index 346cb2666f3a..a03be184a806 100644
>>> --- a/drivers/net/ethernet/intel/ice/ice_sriov.h
>>> +++ b/drivers/net/ethernet/intel/ice/ice_sriov.h
>>> @@ -28,6 +28,7 @@
>>> #ifdef CONFIG_PCI_IOV
>>> void ice_process_vflr_event(struct ice_pf *pf);  int
>>> ice_sriov_configure(struct pci_dev *pdev, int num_vfs);
>>> +int ice_set_vf_fn_mac(struct ice_pf *pf, int vf_id, u8 *mac);
>>> int ice_set_vf_mac(struct net_device *netdev, int vf_id, u8 *mac);  int
>>> ice_get_vf_cfg(struct net_device *netdev, int vf_id, struct ifla_vf_info
>>> *ivi); @@ -76,6 +77,13 @@ ice_sriov_configure(struct pci_dev
>>> __always_unused *pdev,
>>>       return -EOPNOTSUPP;
>>> }
>>>
>>> +static inline int
>>> +ice_set_vf_fn_mac(struct ice_pf __always_unused *pf,
>>> +                int __always_unused vf_id, u8 __always_unused *mac) {
>>> +      return -EOPNOTSUPP;
>>> +}
>>> +
>>> static inline int
>>> ice_set_vf_mac(struct net_device __always_unused *netdev,
>>>              int __always_unused vf_id, u8 __always_unused *mac)
>>> --
>>> 2.39.3 (Apple Git-145)
>>>
>>
> 

  reply	other threads:[~2024-03-18 15:07 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-03-05 15:26 [PATCH v5] ice: Add get/set hw address for VFs using devlink commands Karthik Sundaravel
2024-03-05 15:35 ` Paul Menzel
2024-03-06  8:34 ` [Intel-wired-lan] " Wojciech Drewek
2024-03-08  9:58 ` [EXTERNAL] " Suman Ghosh
2024-03-18 11:55   ` Karthik Sundaravel
2024-03-18 15:04     ` Wojciech Drewek [this message]
2024-03-19 22:52       ` [Intel-wired-lan] " Jacob Keller

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=0a152a46-3800-443a-b370-50cf071f7c13@intel.com \
    --to=wojciech.drewek@intel.com \
    --cc=aharivel@redhat.com \
    --cc=anthony.l.nguyen@intel.com \
    --cc=cfontain@redhat.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=intel-wired-lan@lists.osuosl.org \
    --cc=jacob.e.keller@intel.com \
    --cc=jiri@resnulli.us \
    --cc=ksundara@redhat.com \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=michal.swiatkowski@linux.intel.com \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=pmenzel@molgen.mpg.de \
    --cc=rjarry@redhat.com \
    --cc=sumang@marvell.com \
    --cc=vchundur@redhat.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).