| Message ID | 20240805124651.125761-1-wojciech.drewek@intel.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [iwl-next,v2] ice: Implement ethtool reset support | expand |
| Context | Check | Description |
|---|---|---|
| netdev/tree_selection | success | Not a local patch |
On 8/5/24 14:46, Wojciech Drewek wrote: > Enable ethtool reset support. Ethtool reset flags are mapped to the > E810 reset type: > PF reset: > $ ethtool --reset <ethX> irq dma filter offload > CORE reset: > $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ > offload-shared ram-shared > GLOBAL reset: > $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ > offload-shared mac-shared phy-shared ram-shared > > Calling the same set of flags as in PF reset case on port representor > triggers VF reset. > > Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com> > Reviewed-by: Marcin Szycik <marcin.szycik@linux.intel.com> > Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com> > +/** > + * ice_repr_ethtool_reset - triggers a VF reset > + * @dev: network interface device structure > + * @flags: set of reset flags > + * > + * Return: 0 on success, > + * -EOPNOTSUPP when using unsupported set of flags > + * -EBUSY when VF is not ready for reset. nit: technically it could return also -EIO, -EINVAL, or -EFAULT if VF reset fails I don't know if this list needs to be exhaustive though > + */ > +static int ice_repr_ethtool_reset(struct net_device *dev, u32 *flags) > +{ > + struct ice_repr *repr = ice_netdev_to_repr(dev); > + struct ice_vf *vf; > + > + if (repr->type != ICE_REPR_TYPE_VF || > + *flags != ICE_ETHTOOL_VFR) > + return -EOPNOTSUPP; > + > + vf = repr->vf; > + > + if (ice_check_vf_ready_for_cfg(vf)) > + return -EBUSY; > + > + *flags = 0; I'm fine with zeroing the flags here even if the reset fails afterwards > + > + return ice_reset_vf(vf, ICE_VF_RESET_VFLR | ICE_VF_RESET_LOCK); > +} Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
On Mon, 5 Aug 2024 14:46:51 +0200 Wojciech Drewek wrote: > Enable ethtool reset support. Ethtool reset flags are mapped to the > E810 reset type: > PF reset: > $ ethtool --reset <ethX> irq dma filter offload > CORE reset: > $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ > offload-shared ram-shared > GLOBAL reset: > $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ > offload-shared mac-shared phy-shared ram-shared > > Calling the same set of flags as in PF reset case on port representor > triggers VF reset. Acked-by: Jakub Kicinski <kuba@kernel.org> Thanks!
On Mon, Aug 05, 2024 at 02:46:51PM +0200, Wojciech Drewek wrote: > Enable ethtool reset support. Ethtool reset flags are mapped to the > E810 reset type: > PF reset: > $ ethtool --reset <ethX> irq dma filter offload > CORE reset: > $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ > offload-shared ram-shared > GLOBAL reset: > $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ > offload-shared mac-shared phy-shared ram-shared > > Calling the same set of flags as in PF reset case on port representor > triggers VF reset. > > Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com> > Reviewed-by: Marcin Szycik <marcin.szycik@linux.intel.com> > Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com> > --- > .../device_drivers/ethernet/intel/ice.rst | 28 +++++++ > drivers/net/ethernet/intel/ice/ice_ethtool.c | 77 +++++++++++++++++++ > 2 files changed, 105 insertions(+) > > diff --git a/Documentation/networking/device_drivers/ethernet/intel/ice.rst b/Documentation/networking/device_drivers/ethernet/intel/ice.rst > index 934752f675ba..c043164bfacc 100644 > --- a/Documentation/networking/device_drivers/ethernet/intel/ice.rst > +++ b/Documentation/networking/device_drivers/ethernet/intel/ice.rst > @@ -102,6 +102,34 @@ rx_bytes as "X", then ethtool (hardware statistics) will display rx_bytes as > "X+40" (4 bytes CRC x 10 packets). > > > +ethtool reset > +------------- > +The driver supports 3 types of resets: > +- PF reset - resets only components associated with the given PF, does not > + impact other PFs > +- CORE reset - whole adapter is affected, reset all PFs > +- GLOBAL reset - same as CORE but mac and phy components are also reinitialized Hi Wojciech, I'm not sure, but I think that Sphinx wants blank likes between these list items. Flagged by make: htmldocs SPHINXDIRS=networking ...
On 06.08.2024 16:55, Simon Horman wrote: > On Mon, Aug 05, 2024 at 02:46:51PM +0200, Wojciech Drewek wrote: >> Enable ethtool reset support. Ethtool reset flags are mapped to the >> E810 reset type: >> PF reset: >> $ ethtool --reset <ethX> irq dma filter offload >> CORE reset: >> $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ >> offload-shared ram-shared >> GLOBAL reset: >> $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ >> offload-shared mac-shared phy-shared ram-shared >> >> Calling the same set of flags as in PF reset case on port representor >> triggers VF reset. >> >> Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com> >> Reviewed-by: Marcin Szycik <marcin.szycik@linux.intel.com> >> Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com> >> --- >> .../device_drivers/ethernet/intel/ice.rst | 28 +++++++ >> drivers/net/ethernet/intel/ice/ice_ethtool.c | 77 +++++++++++++++++++ >> 2 files changed, 105 insertions(+) >> >> diff --git a/Documentation/networking/device_drivers/ethernet/intel/ice.rst b/Documentation/networking/device_drivers/ethernet/intel/ice.rst >> index 934752f675ba..c043164bfacc 100644 >> --- a/Documentation/networking/device_drivers/ethernet/intel/ice.rst >> +++ b/Documentation/networking/device_drivers/ethernet/intel/ice.rst >> @@ -102,6 +102,34 @@ rx_bytes as "X", then ethtool (hardware statistics) will display rx_bytes as >> "X+40" (4 bytes CRC x 10 packets). >> >> >> +ethtool reset >> +------------- >> +The driver supports 3 types of resets: >> +- PF reset - resets only components associated with the given PF, does not >> + impact other PFs >> +- CORE reset - whole adapter is affected, reset all PFs >> +- GLOBAL reset - same as CORE but mac and phy components are also reinitialized > > Hi Wojciech, > > I'm not sure, but I think that Sphinx wants blank likes between these list > items. Thanks, I fixed that in v3 > > Flagged by make: htmldocs SPHINXDIRS=networking > > ... >
diff --git a/Documentation/networking/device_drivers/ethernet/intel/ice.rst b/Documentation/networking/device_drivers/ethernet/intel/ice.rst index 934752f675ba..c043164bfacc 100644 --- a/Documentation/networking/device_drivers/ethernet/intel/ice.rst +++ b/Documentation/networking/device_drivers/ethernet/intel/ice.rst @@ -102,6 +102,34 @@ rx_bytes as "X", then ethtool (hardware statistics) will display rx_bytes as "X+40" (4 bytes CRC x 10 packets). +ethtool reset +------------- +The driver supports 3 types of resets: +- PF reset - resets only components associated with the given PF, does not + impact other PFs +- CORE reset - whole adapter is affected, reset all PFs +- GLOBAL reset - same as CORE but mac and phy components are also reinitialized + +These are mapped to ethtool reset flags as follow: +- PF reset: + + # ethtool --reset <ethX> irq dma filter offload + +- CORE reset: + + # ethtool --reset <ethX> irq-shared dma-shared filter-shared offload-shared \ + ram-shared + +- GLOBAL reset: + + # ethtool --reset <ethX> irq-shared dma-shared filter-shared offload-shared \ + mac-shared phy-shared ram-shared + +In switchdev mode you can reset a VF using port representor: + + # ethtool --reset <repr> irq dma filter offload + + Viewing Link Messages --------------------- Link messages will not be displayed to the console if the distribution is diff --git a/drivers/net/ethernet/intel/ice/ice_ethtool.c b/drivers/net/ethernet/intel/ice/ice_ethtool.c index 39d2652c3ee1..14f4bd9397c5 100644 --- a/drivers/net/ethernet/intel/ice/ice_ethtool.c +++ b/drivers/net/ethernet/intel/ice/ice_ethtool.c @@ -4794,6 +4794,81 @@ static void ice_get_ts_stats(struct net_device *netdev, ts_stats->lost = ptp->tx_hwtstamp_timeouts; } +#define ICE_ETHTOOL_PFR (ETH_RESET_IRQ | ETH_RESET_DMA | \ + ETH_RESET_FILTER | ETH_RESET_OFFLOAD) + +#define ICE_ETHTOOL_CORER ((ICE_ETHTOOL_PFR | ETH_RESET_RAM) << \ + ETH_RESET_SHARED_SHIFT) + +#define ICE_ETHTOOL_GLOBR (ICE_ETHTOOL_CORER | \ + (ETH_RESET_MAC << ETH_RESET_SHARED_SHIFT) | \ + (ETH_RESET_PHY << ETH_RESET_SHARED_SHIFT)) + +#define ICE_ETHTOOL_VFR ICE_ETHTOOL_PFR + +/** + * ice_ethtool_reset - triggers a given type of reset + * @dev: network interface device structure + * @flags: set of reset flags + * + * Return: 0 on success, -EOPNOTSUPP when using unsupported set of flags. + */ +static int ice_ethtool_reset(struct net_device *dev, u32 *flags) +{ + struct ice_netdev_priv *np = netdev_priv(dev); + struct ice_pf *pf = np->vsi->back; + enum ice_reset_req reset; + + switch (*flags) { + case ICE_ETHTOOL_CORER: + reset = ICE_RESET_CORER; + break; + case ICE_ETHTOOL_GLOBR: + reset = ICE_RESET_GLOBR; + break; + case ICE_ETHTOOL_PFR: + reset = ICE_RESET_PFR; + break; + default: + netdev_info(dev, "Unsupported set of ethtool flags"); + return -EOPNOTSUPP; + } + + ice_schedule_reset(pf, reset); + + *flags = 0; + + return 0; +} + +/** + * ice_repr_ethtool_reset - triggers a VF reset + * @dev: network interface device structure + * @flags: set of reset flags + * + * Return: 0 on success, + * -EOPNOTSUPP when using unsupported set of flags + * -EBUSY when VF is not ready for reset. + */ +static int ice_repr_ethtool_reset(struct net_device *dev, u32 *flags) +{ + struct ice_repr *repr = ice_netdev_to_repr(dev); + struct ice_vf *vf; + + if (repr->type != ICE_REPR_TYPE_VF || + *flags != ICE_ETHTOOL_VFR) + return -EOPNOTSUPP; + + vf = repr->vf; + + if (ice_check_vf_ready_for_cfg(vf)) + return -EBUSY; + + *flags = 0; + + return ice_reset_vf(vf, ICE_VF_RESET_VFLR | ICE_VF_RESET_LOCK); +} + static const struct ethtool_ops ice_ethtool_ops = { .cap_rss_ctx_supported = true, .supported_coalesce_params = ETHTOOL_COALESCE_USECS | @@ -4829,6 +4904,7 @@ static const struct ethtool_ops ice_ethtool_ops = { .nway_reset = ice_nway_reset, .get_pauseparam = ice_get_pauseparam, .set_pauseparam = ice_set_pauseparam, + .reset = ice_ethtool_reset, .get_rxfh_key_size = ice_get_rxfh_key_size, .get_rxfh_indir_size = ice_get_rxfh_indir_size, .get_rxfh = ice_get_rxfh, @@ -4885,6 +4961,7 @@ static const struct ethtool_ops ice_ethtool_repr_ops = { .get_strings = ice_repr_get_strings, .get_ethtool_stats = ice_repr_get_ethtool_stats, .get_sset_count = ice_repr_get_sset_count, + .reset = ice_repr_ethtool_reset, }; /**
Enable ethtool reset support. Ethtool reset flags are mapped to the E810 reset type: PF reset: $ ethtool --reset <ethX> irq dma filter offload CORE reset: $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ offload-shared ram-shared GLOBAL reset: $ ethtool --reset <ethX> irq-shared dma-shared filter-shared \ offload-shared mac-shared phy-shared ram-shared Calling the same set of flags as in PF reset case on port representor triggers VF reset. Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com> Reviewed-by: Marcin Szycik <marcin.szycik@linux.intel.com> Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com> --- .../device_drivers/ethernet/intel/ice.rst | 28 +++++++ drivers/net/ethernet/intel/ice/ice_ethtool.c | 77 +++++++++++++++++++ 2 files changed, 105 insertions(+)