| Message ID | 1456476401-23398-1-git-send-email-ivecera@redhat.com |
|---|---|
| State | Superseded, archived |
| Delegated to: | David Miller |
| Headers | show |
On Fri, Feb 26, 2016 at 2:16 PM, Ivan Vecera <ivecera@redhat.com> wrote: > The FILTMGMT privilege is necessary on BE3 chip to manipulare filters > like MC, UC list management, VLAN filter, promisc mode... This privilege > is dropped for VFs when transparrent tagging is enabled on them. This > prevents to make interface up for such VF because be_enable_if_filters() > called from be_open() fails thus be_open() also fails. > > Cc: Sathya Perla <sathya.perla@broadcom.com> > Cc: Sriharsha Basavapatna <sriharsha.basavapatna@broadcom.com> > Signed-off-by: Ivan Vecera <ivecera@redhat.com> > --- > drivers/net/ethernet/emulex/benet/be_main.c | 11 +++++++---- > 1 file changed, 7 insertions(+), 4 deletions(-) > > diff --git a/drivers/net/ethernet/emulex/benet/be_main.c b/drivers/net/ethernet/emulex/benet/be_main.c > index f99de36..2e176f6 100644 > --- a/drivers/net/ethernet/emulex/benet/be_main.c > +++ b/drivers/net/ethernet/emulex/benet/be_main.c > @@ -3430,7 +3430,8 @@ static int be_close(struct net_device *netdev) > if (!(adapter->flags & BE_FLAGS_SETUP_DONE)) > return 0; > > - be_disable_if_filters(adapter); > + if (!(BEx_chip(adapter) && be_pvid_tagging_enabled(adapter))) > + be_disable_if_filters(adapter); > > if (adapter->flags & BE_FLAGS_NAPI_ENABLED) { > for_all_evt_queues(adapter, eqo, i) { > @@ -3571,9 +3572,11 @@ static int be_open(struct net_device *netdev) > if (status) > goto err; > > - status = be_enable_if_filters(adapter); > - if (status) > - goto err; > + if (!(BEx_chip(adapter) && be_pvid_tagging_enabled(adapter))) { > + status = be_enable_if_filters(adapter); > + if (status) > + goto err; Ivan, in the be_enable_if_filters() routine the VF tries to enable the basic filtering flags (not promisc mode etc) via the RX_FILTER cmd, for which the VFs shouldn't need any FILTMGMT privileges. I also don't expect be_cmd_pmac_add() to fail as the PF would have already provisioned the mac-addr for the VF. We should be able to reproduce this first thing Monday morning and see what's wrong. Also, as the privilege rules are same across all chips, any fix must be applicable for all chips that be2net supports. thanks, -Sathya
On 26.2.2016 12:36, Sathya Perla wrote: > On Fri, Feb 26, 2016 at 2:16 PM, Ivan Vecera <ivecera@redhat.com> wrote: >> The FILTMGMT privilege is necessary on BE3 chip to manipulare filters >> like MC, UC list management, VLAN filter, promisc mode... This privilege >> is dropped for VFs when transparrent tagging is enabled on them. This >> prevents to make interface up for such VF because be_enable_if_filters() >> called from be_open() fails thus be_open() also fails. >> >> Cc: Sathya Perla <sathya.perla@broadcom.com> >> Cc: Sriharsha Basavapatna <sriharsha.basavapatna@broadcom.com> >> Signed-off-by: Ivan Vecera <ivecera@redhat.com> >> --- >> drivers/net/ethernet/emulex/benet/be_main.c | 11 +++++++---- >> 1 file changed, 7 insertions(+), 4 deletions(-) >> >> diff --git a/drivers/net/ethernet/emulex/benet/be_main.c b/drivers/net/ethernet/emulex/benet/be_main.c >> index f99de36..2e176f6 100644 >> --- a/drivers/net/ethernet/emulex/benet/be_main.c >> +++ b/drivers/net/ethernet/emulex/benet/be_main.c >> @@ -3430,7 +3430,8 @@ static int be_close(struct net_device *netdev) >> if (!(adapter->flags & BE_FLAGS_SETUP_DONE)) >> return 0; >> >> - be_disable_if_filters(adapter); >> + if (!(BEx_chip(adapter) && be_pvid_tagging_enabled(adapter))) >> + be_disable_if_filters(adapter); >> >> if (adapter->flags & BE_FLAGS_NAPI_ENABLED) { >> for_all_evt_queues(adapter, eqo, i) { >> @@ -3571,9 +3572,11 @@ static int be_open(struct net_device *netdev) >> if (status) >> goto err; >> >> - status = be_enable_if_filters(adapter); >> - if (status) >> - goto err; >> + if (!(BEx_chip(adapter) && be_pvid_tagging_enabled(adapter))) { >> + status = be_enable_if_filters(adapter); >> + if (status) >> + goto err; > > Ivan, in the be_enable_if_filters() routine the VF tries to enable the > basic filtering flags (not promisc mode etc) via the RX_FILTER cmd, > for which the VFs shouldn't need any FILTMGMT privileges. I also don't > expect be_cmd_pmac_add() to fail as the PF would have already > provisioned the mac-addr for the VF. > We should be able to reproduce this first thing Monday morning and see > what's wrong. Also, as the privilege rules are same across all chips, > any fix must be applicable for all chips that be2net supports. Hi Sathya, if the transparent VLAN tagging is enabled then be_enable_if_filters() will fail and be_open() will also fail. This happens only on BE3 not on Lancer as this does not support SR-IOV. I don't see any problem on Skyhawk. Log from the guest after 'ip link set ... up': [root@bootp-73-131-183 ~]# dmesg | tail -n 20 pci 0000:00:07.0: no hotplug settings from platform be2net 0000:00:07.0: be2net version is 10.6.0.3r be2net 0000:00:07.0: enabling device (0040 -> 0042) be2net 0000:00:07.0: setting latency timer to 64 be2net 0000:00:07.0: FW config: function_mode=0x14003, function_caps=0x4 be2net 0000:00:07.0: VF is not privileged to issue opcode 125-1 be2net 0000:00:07.0: Max: txqs 1, rxqs 1, rss 0, eqs 1, vfs 0 be2net 0000:00:07.0: Max: uc-macs 2, mc-macs 64, vlans 64 alloc irq_desc for 29 on node -1 alloc kstat_irqs on node -1 be2net 0000:00:07.0: irq 29 for MSI/MSI-X be2net 0000:00:07.0: enabled 1 MSI-x vector(s) for NIC be2net 0000:00:07.0: created 1 TX queue(s) be2net 0000:00:07.0: created 1 RX queue(s) be2net 0000:00:07.0: LPVID: 3 be2net 0000:00:07.0: FW version is 10.4.255.25 be2net 0000:00:07.0: HW Flow control - TX:1 RX:1 be2net 0000:00:07.0: Emulex OneConnect(be3): VF port 0 be2net 0000:00:07.0: VF is not privileged to issue opcode 34-1 be2net 0000:00:07.0: VF is not privileged to issue opcode 60-1 The be_open() calls be_enable_if_filters() -> be_cmd_rx_filter(). Command with opcode 34 alias NTWK_RX_FILTER fails thus be_open() also fails and be_close() is called as cleaning-up action. The be_close() calls then be_disable_if_filters() that calls be_cmd_pmac_del() (opcode 60). This also fails. As I has written above under Skyhawk this does not happen... so I has created BE3 specific patch. Ivan
On Fri, Feb 26, 2016 at 5:54 PM, Ivan Vecera <ivecera@redhat.com> wrote: > On 26.2.2016 12:36, Sathya Perla wrote: >> >> On Fri, Feb 26, 2016 at 2:16 PM, Ivan Vecera <ivecera@redhat.com> wrote: >>> >>> The FILTMGMT privilege is necessary on BE3 chip to manipulare filters >>> like MC, UC list management, VLAN filter, promisc mode... This privilege >>> is dropped for VFs when transparrent tagging is enabled on them. This >>> prevents to make interface up for such VF because be_enable_if_filters() >>> called from be_open() fails thus be_open() also fails. >>> >>> Cc: Sathya Perla <sathya.perla@broadcom.com> >>> Cc: Sriharsha Basavapatna <sriharsha.basavapatna@broadcom.com> >>> Signed-off-by: Ivan Vecera <ivecera@redhat.com> ... >> >> >> Ivan, in the be_enable_if_filters() routine the VF tries to enable the >> basic filtering flags (not promisc mode etc) via the RX_FILTER cmd, >> for which the VFs shouldn't need any FILTMGMT privileges. I also don't >> expect be_cmd_pmac_add() to fail as the PF would have already >> provisioned the mac-addr for the VF. >> We should be able to reproduce this first thing Monday morning and see >> what's wrong. Also, as the privilege rules are same across all chips, >> any fix must be applicable for all chips that be2net supports. > > Hi Sathya, > if the transparent VLAN tagging is enabled then be_enable_if_filters() will > fail and be_open() will also fail. This happens only on BE3 not on Lancer as > this does not support SR-IOV. I don't see any problem on Skyhawk. > > Log from the guest after 'ip link set ... up': > > [root@bootp-73-131-183 ~]# dmesg | tail -n 20 > pci 0000:00:07.0: no hotplug settings from platform > be2net 0000:00:07.0: be2net version is 10.6.0.3r > be2net 0000:00:07.0: enabling device (0040 -> 0042) > be2net 0000:00:07.0: setting latency timer to 64 > be2net 0000:00:07.0: FW config: function_mode=0x14003, function_caps=0x4 > be2net 0000:00:07.0: VF is not privileged to issue opcode 125-1 > be2net 0000:00:07.0: Max: txqs 1, rxqs 1, rss 0, eqs 1, vfs 0 > be2net 0000:00:07.0: Max: uc-macs 2, mc-macs 64, vlans 64 > alloc irq_desc for 29 on node -1 > alloc kstat_irqs on node -1 > be2net 0000:00:07.0: irq 29 for MSI/MSI-X > be2net 0000:00:07.0: enabled 1 MSI-x vector(s) for NIC > be2net 0000:00:07.0: created 1 TX queue(s) > be2net 0000:00:07.0: created 1 RX queue(s) > be2net 0000:00:07.0: LPVID: 3 > be2net 0000:00:07.0: FW version is 10.4.255.25 > be2net 0000:00:07.0: HW Flow control - TX:1 RX:1 > be2net 0000:00:07.0: Emulex OneConnect(be3): VF port 0 > be2net 0000:00:07.0: VF is not privileged to issue opcode 34-1 > be2net 0000:00:07.0: VF is not privileged to issue opcode 60-1 > > The be_open() calls be_enable_if_filters() -> be_cmd_rx_filter(). Command > with opcode 34 alias NTWK_RX_FILTER fails thus be_open() also fails and > be_close() is called as cleaning-up action. The be_close() calls then > be_disable_if_filters() that calls be_cmd_pmac_del() (opcode 60). This also > fails. > > As I has written above under Skyhawk this does not happen... so I has > created BE3 specific patch. > Ivan, thanks for the dmesg log. I think the root-cause of this issue is the MULTICAST bit in the BE_IF_EN_FLAGS. The FW is not allowing the VF driver's be_cmd_rx_filter(BE_IF_EN_FLAGS) call. In be_enable_if_filters() we are sensitive to this error, but in be_set_rx_mode() we ignore any FW errors. I'll provide an update Monday morning....thanks!
On 26.2.2016 13:44, Sathya Perla wrote: > On Fri, Feb 26, 2016 at 5:54 PM, Ivan Vecera <ivecera@redhat.com> wrote: >> On 26.2.2016 12:36, Sathya Perla wrote: >>> >>> On Fri, Feb 26, 2016 at 2:16 PM, Ivan Vecera <ivecera@redhat.com> wrote: >>>> >>>> The FILTMGMT privilege is necessary on BE3 chip to manipulare filters >>>> like MC, UC list management, VLAN filter, promisc mode... This privilege >>>> is dropped for VFs when transparrent tagging is enabled on them. This >>>> prevents to make interface up for such VF because be_enable_if_filters() >>>> called from be_open() fails thus be_open() also fails. >>>> >>>> Cc: Sathya Perla <sathya.perla@broadcom.com> >>>> Cc: Sriharsha Basavapatna <sriharsha.basavapatna@broadcom.com> >>>> Signed-off-by: Ivan Vecera <ivecera@redhat.com> > ... >>> >>> >>> Ivan, in the be_enable_if_filters() routine the VF tries to enable the >>> basic filtering flags (not promisc mode etc) via the RX_FILTER cmd, >>> for which the VFs shouldn't need any FILTMGMT privileges. I also don't >>> expect be_cmd_pmac_add() to fail as the PF would have already >>> provisioned the mac-addr for the VF. >>> We should be able to reproduce this first thing Monday morning and see >>> what's wrong. Also, as the privilege rules are same across all chips, >>> any fix must be applicable for all chips that be2net supports. >> >> Hi Sathya, >> if the transparent VLAN tagging is enabled then be_enable_if_filters() will >> fail and be_open() will also fail. This happens only on BE3 not on Lancer as >> this does not support SR-IOV. I don't see any problem on Skyhawk. >> >> Log from the guest after 'ip link set ... up': >> >> [root@bootp-73-131-183 ~]# dmesg | tail -n 20 >> pci 0000:00:07.0: no hotplug settings from platform >> be2net 0000:00:07.0: be2net version is 10.6.0.3r >> be2net 0000:00:07.0: enabling device (0040 -> 0042) >> be2net 0000:00:07.0: setting latency timer to 64 >> be2net 0000:00:07.0: FW config: function_mode=0x14003, function_caps=0x4 >> be2net 0000:00:07.0: VF is not privileged to issue opcode 125-1 >> be2net 0000:00:07.0: Max: txqs 1, rxqs 1, rss 0, eqs 1, vfs 0 >> be2net 0000:00:07.0: Max: uc-macs 2, mc-macs 64, vlans 64 >> alloc irq_desc for 29 on node -1 >> alloc kstat_irqs on node -1 >> be2net 0000:00:07.0: irq 29 for MSI/MSI-X >> be2net 0000:00:07.0: enabled 1 MSI-x vector(s) for NIC >> be2net 0000:00:07.0: created 1 TX queue(s) >> be2net 0000:00:07.0: created 1 RX queue(s) >> be2net 0000:00:07.0: LPVID: 3 >> be2net 0000:00:07.0: FW version is 10.4.255.25 >> be2net 0000:00:07.0: HW Flow control - TX:1 RX:1 >> be2net 0000:00:07.0: Emulex OneConnect(be3): VF port 0 >> be2net 0000:00:07.0: VF is not privileged to issue opcode 34-1 >> be2net 0000:00:07.0: VF is not privileged to issue opcode 60-1 >> >> The be_open() calls be_enable_if_filters() -> be_cmd_rx_filter(). Command >> with opcode 34 alias NTWK_RX_FILTER fails thus be_open() also fails and >> be_close() is called as cleaning-up action. The be_close() calls then >> be_disable_if_filters() that calls be_cmd_pmac_del() (opcode 60). This also >> fails. >> >> As I has written above under Skyhawk this does not happen... so I has >> created BE3 specific patch. >> > > Ivan, thanks for the dmesg log. I think the root-cause of this issue > is the MULTICAST bit in the BE_IF_EN_FLAGS. The FW is not allowing the > VF driver's be_cmd_rx_filter(BE_IF_EN_FLAGS) call. In > be_enable_if_filters() we are sensitive to this error, but in > be_set_rx_mode() we ignore any FW errors. > I'll provide an update Monday morning....thanks! > Should the MULTICAST bit be masked in any be_cmd_rx_filter() call on BE3's VFs if the trans. tagging is enabled? Ivan
On Fri, Feb 26, 2016 at 6:43 PM, Ivan Vecera <ivecera@redhat.com> wrote: > Should the MULTICAST bit be masked in any be_cmd_rx_filter() call on BE3's > VFs if the trans. tagging is enabled? Not on any be_cmd_rx_filter() call, but on the first call in be_open()->be_if_enable_filters() where the basic filtering flags are being enabled and the driver is explicitly checking for the return status. We have a fix for this already in our internal builds that hasn't been upstreamed yet. Will send out a patch for this asap. Thanks for the help on this Ivan!
diff --git a/drivers/net/ethernet/emulex/benet/be_main.c b/drivers/net/ethernet/emulex/benet/be_main.c index f99de36..2e176f6 100644 --- a/drivers/net/ethernet/emulex/benet/be_main.c +++ b/drivers/net/ethernet/emulex/benet/be_main.c @@ -3430,7 +3430,8 @@ static int be_close(struct net_device *netdev) if (!(adapter->flags & BE_FLAGS_SETUP_DONE)) return 0; - be_disable_if_filters(adapter); + if (!(BEx_chip(adapter) && be_pvid_tagging_enabled(adapter))) + be_disable_if_filters(adapter); if (adapter->flags & BE_FLAGS_NAPI_ENABLED) { for_all_evt_queues(adapter, eqo, i) { @@ -3571,9 +3572,11 @@ static int be_open(struct net_device *netdev) if (status) goto err; - status = be_enable_if_filters(adapter); - if (status) - goto err; + if (!(BEx_chip(adapter) && be_pvid_tagging_enabled(adapter))) { + status = be_enable_if_filters(adapter); + if (status) + goto err; + } status = be_irq_register(adapter); if (status)
The FILTMGMT privilege is necessary on BE3 chip to manipulare filters like MC, UC list management, VLAN filter, promisc mode... This privilege is dropped for VFs when transparrent tagging is enabled on them. This prevents to make interface up for such VF because be_enable_if_filters() called from be_open() fails thus be_open() also fails. Cc: Sathya Perla <sathya.perla@broadcom.com> Cc: Sriharsha Basavapatna <sriharsha.basavapatna@broadcom.com> Signed-off-by: Ivan Vecera <ivecera@redhat.com> --- drivers/net/ethernet/emulex/benet/be_main.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)