| Message ID | 20240610074434.1962735-5-michal.swiatkowski@linux.intel.com (mailing list archive) |
|---|---|
| State | Awaiting Upstream |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | ice: prepare representor for SF support | expand |
On Mon, Jun 10, 2024 at 09:44:34AM +0200, Michal Swiatkowski wrote: > In case of reset of VF VSI can be reallocated. To handle this case it > should be properly updated. > > Reload representor as vsi->vsi_num can be different than the one stored > when representor was created. > > Instead of only changing antispoof do whole VSI configuration for > eswitch. > > Signed-off-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com> Thanks Michal, My comment below notwithstanding, this looks good to me. Reviewed-by: Simon Horman <horms@kernel.org> > --- > drivers/net/ethernet/intel/ice/ice_eswitch.c | 21 +++++++++++++------- > drivers/net/ethernet/intel/ice/ice_eswitch.h | 4 ++-- > drivers/net/ethernet/intel/ice/ice_vf_lib.c | 2 +- > 3 files changed, 17 insertions(+), 10 deletions(-) > > diff --git a/drivers/net/ethernet/intel/ice/ice_eswitch.c b/drivers/net/ethernet/intel/ice/ice_eswitch.c > index 3f73f46111fc..4f539b1c7781 100644 > --- a/drivers/net/ethernet/intel/ice/ice_eswitch.c > +++ b/drivers/net/ethernet/intel/ice/ice_eswitch.c > @@ -178,16 +178,16 @@ void ice_eswitch_decfg_vsi(struct ice_vsi *vsi, const u8 *mac) > * @repr_id: representor ID > * @vsi: VSI for which port representor is configured > */ > -void ice_eswitch_update_repr(unsigned long repr_id, struct ice_vsi *vsi) > +void ice_eswitch_update_repr(unsigned long *repr_id, struct ice_vsi *vsi) > { > struct ice_pf *pf = vsi->back; > struct ice_repr *repr; > - int ret; > + int err; > > if (!ice_is_switchdev_running(pf)) > return; > > - repr = xa_load(&pf->eswitch.reprs, repr_id); > + repr = xa_load(&pf->eswitch.reprs, *repr_id); > if (!repr) > return; > > @@ -197,12 +197,19 @@ void ice_eswitch_update_repr(unsigned long repr_id, struct ice_vsi *vsi) > if (repr->br_port) > repr->br_port->vsi = vsi; > > - ret = ice_vsi_update_security(vsi, ice_vsi_ctx_clear_antispoof); > - if (ret) { > - ice_fltr_add_mac_and_broadcast(vsi, repr->parent_mac, > - ICE_FWD_TO_VSI); > + err = ice_eswitch_cfg_vsi(vsi, repr->parent_mac); > + if (err) > dev_err(ice_pf_to_dev(pf), "Failed to update VSI of port representor %d", > repr->id); > + > + /* The VSI number is different, reload the PR with new id */ > + if (repr->id != vsi->vsi_num) { > + xa_erase(&pf->eswitch.reprs, repr->id); > + repr->id = vsi->vsi_num; > + if (xa_insert(&pf->eswitch.reprs, repr->id, repr, GFP_KERNEL)) > + dev_err(ice_pf_to_dev(pf), "Failed to reload port representor %d", > + repr->id); > + *repr_id = repr->id; > } > } FWIIW, I think it would be nicer if ice_eswitch_decfg_vsi returned the repr_id rather than passing repr_id by reference. This is because, in general, I think it's nicer to reduce side-effects of functions. ...
On Fri, Jun 14, 2024 at 01:47:16PM +0100, Simon Horman wrote: > On Mon, Jun 10, 2024 at 09:44:34AM +0200, Michal Swiatkowski wrote: > > In case of reset of VF VSI can be reallocated. To handle this case it > > should be properly updated. > > > > Reload representor as vsi->vsi_num can be different than the one stored > > when representor was created. > > > > Instead of only changing antispoof do whole VSI configuration for > > eswitch. > > > > Signed-off-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com> > > Thanks Michal, > > My comment below notwithstanding, this looks good to me. > > Reviewed-by: Simon Horman <horms@kernel.org> > > > --- > > drivers/net/ethernet/intel/ice/ice_eswitch.c | 21 +++++++++++++------- > > drivers/net/ethernet/intel/ice/ice_eswitch.h | 4 ++-- > > drivers/net/ethernet/intel/ice/ice_vf_lib.c | 2 +- > > 3 files changed, 17 insertions(+), 10 deletions(-) > > > > diff --git a/drivers/net/ethernet/intel/ice/ice_eswitch.c b/drivers/net/ethernet/intel/ice/ice_eswitch.c > > index 3f73f46111fc..4f539b1c7781 100644 > > --- a/drivers/net/ethernet/intel/ice/ice_eswitch.c > > +++ b/drivers/net/ethernet/intel/ice/ice_eswitch.c > > @@ -178,16 +178,16 @@ void ice_eswitch_decfg_vsi(struct ice_vsi *vsi, const u8 *mac) > > * @repr_id: representor ID > > * @vsi: VSI for which port representor is configured > > */ > > -void ice_eswitch_update_repr(unsigned long repr_id, struct ice_vsi *vsi) > > +void ice_eswitch_update_repr(unsigned long *repr_id, struct ice_vsi *vsi) > > { > > struct ice_pf *pf = vsi->back; > > struct ice_repr *repr; > > - int ret; > > + int err; > > > > if (!ice_is_switchdev_running(pf)) > > return; > > > > - repr = xa_load(&pf->eswitch.reprs, repr_id); > > + repr = xa_load(&pf->eswitch.reprs, *repr_id); > > if (!repr) > > return; > > > > @@ -197,12 +197,19 @@ void ice_eswitch_update_repr(unsigned long repr_id, struct ice_vsi *vsi) > > if (repr->br_port) > > repr->br_port->vsi = vsi; > > > > - ret = ice_vsi_update_security(vsi, ice_vsi_ctx_clear_antispoof); > > - if (ret) { > > - ice_fltr_add_mac_and_broadcast(vsi, repr->parent_mac, > > - ICE_FWD_TO_VSI); > > + err = ice_eswitch_cfg_vsi(vsi, repr->parent_mac); > > + if (err) > > dev_err(ice_pf_to_dev(pf), "Failed to update VSI of port representor %d", > > repr->id); > > + > > + /* The VSI number is different, reload the PR with new id */ > > + if (repr->id != vsi->vsi_num) { > > + xa_erase(&pf->eswitch.reprs, repr->id); > > + repr->id = vsi->vsi_num; > > + if (xa_insert(&pf->eswitch.reprs, repr->id, repr, GFP_KERNEL)) > > + dev_err(ice_pf_to_dev(pf), "Failed to reload port representor %d", > > + repr->id); > > + *repr_id = repr->id; > > } > > } > > FWIIW, I think it would be nicer if ice_eswitch_decfg_vsi returned > the repr_id rather than passing repr_id by reference. This is because, > in general, I think it's nicer to reduce side-effects of functions. > I wanted to change repr->id and the vf (or sf) repr id in the same place. But in general I agree with your comment. Thanks, Michal > ...
diff --git a/drivers/net/ethernet/intel/ice/ice_eswitch.c b/drivers/net/ethernet/intel/ice/ice_eswitch.c index 3f73f46111fc..4f539b1c7781 100644 --- a/drivers/net/ethernet/intel/ice/ice_eswitch.c +++ b/drivers/net/ethernet/intel/ice/ice_eswitch.c @@ -178,16 +178,16 @@ void ice_eswitch_decfg_vsi(struct ice_vsi *vsi, const u8 *mac) * @repr_id: representor ID * @vsi: VSI for which port representor is configured */ -void ice_eswitch_update_repr(unsigned long repr_id, struct ice_vsi *vsi) +void ice_eswitch_update_repr(unsigned long *repr_id, struct ice_vsi *vsi) { struct ice_pf *pf = vsi->back; struct ice_repr *repr; - int ret; + int err; if (!ice_is_switchdev_running(pf)) return; - repr = xa_load(&pf->eswitch.reprs, repr_id); + repr = xa_load(&pf->eswitch.reprs, *repr_id); if (!repr) return; @@ -197,12 +197,19 @@ void ice_eswitch_update_repr(unsigned long repr_id, struct ice_vsi *vsi) if (repr->br_port) repr->br_port->vsi = vsi; - ret = ice_vsi_update_security(vsi, ice_vsi_ctx_clear_antispoof); - if (ret) { - ice_fltr_add_mac_and_broadcast(vsi, repr->parent_mac, - ICE_FWD_TO_VSI); + err = ice_eswitch_cfg_vsi(vsi, repr->parent_mac); + if (err) dev_err(ice_pf_to_dev(pf), "Failed to update VSI of port representor %d", repr->id); + + /* The VSI number is different, reload the PR with new id */ + if (repr->id != vsi->vsi_num) { + xa_erase(&pf->eswitch.reprs, repr->id); + repr->id = vsi->vsi_num; + if (xa_insert(&pf->eswitch.reprs, repr->id, repr, GFP_KERNEL)) + dev_err(ice_pf_to_dev(pf), "Failed to reload port representor %d", + repr->id); + *repr_id = repr->id; } } diff --git a/drivers/net/ethernet/intel/ice/ice_eswitch.h b/drivers/net/ethernet/intel/ice/ice_eswitch.h index 9a25606e9740..09194d514f9b 100644 --- a/drivers/net/ethernet/intel/ice/ice_eswitch.h +++ b/drivers/net/ethernet/intel/ice/ice_eswitch.h @@ -18,7 +18,7 @@ ice_eswitch_mode_set(struct devlink *devlink, u16 mode, struct netlink_ext_ack *extack); bool ice_is_eswitch_mode_switchdev(struct ice_pf *pf); -void ice_eswitch_update_repr(unsigned long repr_id, struct ice_vsi *vsi); +void ice_eswitch_update_repr(unsigned long *repr_id, struct ice_vsi *vsi); void ice_eswitch_stop_all_tx_queues(struct ice_pf *pf); @@ -47,7 +47,7 @@ ice_eswitch_set_target_vsi(struct sk_buff *skb, struct ice_tx_offload_params *off) { } static inline void -ice_eswitch_update_repr(unsigned long repr_id, struct ice_vsi *vsi) { } +ice_eswitch_update_repr(unsigned long *repr_id, struct ice_vsi *vsi) { } static inline int ice_eswitch_configure(struct ice_pf *pf) { diff --git a/drivers/net/ethernet/intel/ice/ice_vf_lib.c b/drivers/net/ethernet/intel/ice/ice_vf_lib.c index 48a8d462d76a..5635e9da2212 100644 --- a/drivers/net/ethernet/intel/ice/ice_vf_lib.c +++ b/drivers/net/ethernet/intel/ice/ice_vf_lib.c @@ -948,7 +948,7 @@ int ice_reset_vf(struct ice_vf *vf, u32 flags) goto out_unlock; } - ice_eswitch_update_repr(vf->repr_id, vsi); + ice_eswitch_update_repr(&vf->repr_id, vsi); /* if the VF has been reset allow it to come up again */ ice_mbx_clear_malvf(&vf->mbx_info);
In case of reset of VF VSI can be reallocated. To handle this case it should be properly updated. Reload representor as vsi->vsi_num can be different than the one stored when representor was created. Instead of only changing antispoof do whole VSI configuration for eswitch. Signed-off-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com> --- drivers/net/ethernet/intel/ice/ice_eswitch.c | 21 +++++++++++++------- drivers/net/ethernet/intel/ice/ice_eswitch.h | 4 ++-- drivers/net/ethernet/intel/ice/ice_vf_lib.c | 2 +- 3 files changed, 17 insertions(+), 10 deletions(-)