| Message ID | 20231120055138.3602102-1-sumang@marvell.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net] octeontx2-pf: Fix ntuple rule creation to direct packet to VF with higher Rx queue than its PF | expand |
On 20.11.2023 06:51, Suman Ghosh wrote: > It is possible to add a ntuple rule which would like to direct packet to > a VF whose number of queues are greater/less than its PF's queue numbers. > For example a PF can have 2 Rx queues but a VF created on that PF can have > 8 Rx queues. As of today, ntuple rule will reject rule because it is > checking the requested queue number against PF's number of Rx queues. > As a part of this fix if the action of a ntuple rule is to move a packet > to a VF's queue then the check is removed. Also, a debug information is > printed to aware user that it is user's responsibility to cross check if > the requested queue number on that VF is a valid one. > > Fixes: f0a1913f8a6f ("octeontx2-pf: Add support for ethtool ntuple filters") > Signed-off-by: Suman Ghosh <sumang@marvell.com> > --- > .../marvell/octeontx2/nic/otx2_flows.c | 21 +++++++++++++++++++ > 1 file changed, 21 insertions(+) > > diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c > index 4762dbea64a1..4200f2d387f6 100644 > --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c > +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c > @@ -1088,6 +1088,7 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct ethtool_rxnfc *nfc) > struct ethhdr *eth_hdr; > bool new = false; > int err = 0; > + u64 vf_num; > u32 ring; > > if (!flow_cfg->max_flows) { > @@ -1100,9 +1101,26 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct ethtool_rxnfc *nfc) > if (!(pfvf->flags & OTX2_FLAG_NTUPLE_SUPPORT)) > return -ENOMEM; > > + /* Number of queues on a VF can be greater or less than > + * the PF's queue. Hence no need to check for the > + * queue count. Hence no need to check queue count if PF > + * is installing for its VF. Below is the expected vf_num value > + * based on the ethtool commands. > + * > + * e.g. > + * 1. ethtool -U <netdev> ... action -1 ==> vf_num:255 > + * 2. ethtool -U <netdev> ... action <queue_num> ==> vf_num:0 > + * 3. ethtool -U <netdev> ... vf <vf_idx> queue <queue_num> ==> > + * vf_num:vf_idx+1 > + */ > + vf_num = ethtool_get_flow_spec_ring_vf(fsp->ring_cookie); > + if (!is_otx2_vf(pfvf->pcifunc) && vf_num) > + goto bypass_queue_check; Let's just add this condition to the next if, no need for goto. > + > if (ring >= pfvf->hw.rx_queues && fsp->ring_cookie != RX_CLS_FLOW_DISC) > return -EINVAL; > > +bypass_queue_check: > if (fsp->location >= otx2_get_maxflows(flow_cfg)) > return -EINVAL; > > @@ -1182,6 +1200,9 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct ethtool_rxnfc *nfc) > flow_cfg->nr_flows++; > } > > + if (flow->is_vf) > + netdev_info(pfvf->netdev, > + "Make sure that VF's queue number is within its queue limit\n"); > return 0; > } >
>> Signed-off-by: Suman Ghosh <sumang@marvell.com> >> --- >> .../marvell/octeontx2/nic/otx2_flows.c | 21 >+++++++++++++++++++ >> 1 file changed, 21 insertions(+) >> >> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c >> b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c >> index 4762dbea64a1..4200f2d387f6 100644 >> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c >> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c >> @@ -1088,6 +1088,7 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct >ethtool_rxnfc *nfc) >> struct ethhdr *eth_hdr; >> bool new = false; >> int err = 0; >> + u64 vf_num; >> u32 ring; >> >> if (!flow_cfg->max_flows) { >> @@ -1100,9 +1101,26 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct >ethtool_rxnfc *nfc) >> if (!(pfvf->flags & OTX2_FLAG_NTUPLE_SUPPORT)) >> return -ENOMEM; >> >> + /* Number of queues on a VF can be greater or less than >> + * the PF's queue. Hence no need to check for the >> + * queue count. Hence no need to check queue count if PF >> + * is installing for its VF. Below is the expected vf_num value >> + * based on the ethtool commands. >> + * >> + * e.g. >> + * 1. ethtool -U <netdev> ... action -1 ==> vf_num:255 >> + * 2. ethtool -U <netdev> ... action <queue_num> ==> vf_num:0 >> + * 3. ethtool -U <netdev> ... vf <vf_idx> queue <queue_num> ==> >> + * vf_num:vf_idx+1 >> + */ >> + vf_num = ethtool_get_flow_spec_ring_vf(fsp->ring_cookie); >> + if (!is_otx2_vf(pfvf->pcifunc) && vf_num) >> + goto bypass_queue_check; > >Let's just add this condition to the next if, no need for goto. [Suman] I kept it a separate check to make the code more readable. Otherwise the next if condition will be complicated. > >> + >> if (ring >= pfvf->hw.rx_queues && fsp->ring_cookie != >RX_CLS_FLOW_DISC) >> return -EINVAL; >> >> +bypass_queue_check: >> if (fsp->location >= otx2_get_maxflows(flow_cfg)) >> return -EINVAL; >> >> @@ -1182,6 +1200,9 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct >ethtool_rxnfc *nfc) >> flow_cfg->nr_flows++; >> } >> >> + if (flow->is_vf) >> + netdev_info(pfvf->netdev, >> + "Make sure that VF's queue number is within its queue >> +limit\n"); >> return 0; >> } >>
On Tue, Nov 21, 2023 at 09:54:00AM +0000, Suman Ghosh wrote: > >> Signed-off-by: Suman Ghosh <sumang@marvell.com> > >> --- > >> .../marvell/octeontx2/nic/otx2_flows.c | 21 > >+++++++++++++++++++ > >> 1 file changed, 21 insertions(+) > >> > >> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c > >> b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c > >> index 4762dbea64a1..4200f2d387f6 100644 > >> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c > >> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c > >> @@ -1088,6 +1088,7 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct > >ethtool_rxnfc *nfc) > >> struct ethhdr *eth_hdr; > >> bool new = false; > >> int err = 0; > >> + u64 vf_num; > >> u32 ring; > >> > >> if (!flow_cfg->max_flows) { > >> @@ -1100,9 +1101,26 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct > >ethtool_rxnfc *nfc) > >> if (!(pfvf->flags & OTX2_FLAG_NTUPLE_SUPPORT)) > >> return -ENOMEM; > >> > >> + /* Number of queues on a VF can be greater or less than > >> + * the PF's queue. Hence no need to check for the > >> + * queue count. Hence no need to check queue count if PF > >> + * is installing for its VF. Below is the expected vf_num value > >> + * based on the ethtool commands. > >> + * > >> + * e.g. > >> + * 1. ethtool -U <netdev> ... action -1 ==> vf_num:255 > >> + * 2. ethtool -U <netdev> ... action <queue_num> ==> vf_num:0 > >> + * 3. ethtool -U <netdev> ... vf <vf_idx> queue <queue_num> ==> > >> + * vf_num:vf_idx+1 > >> + */ > >> + vf_num = ethtool_get_flow_spec_ring_vf(fsp->ring_cookie); > >> + if (!is_otx2_vf(pfvf->pcifunc) && vf_num) > >> + goto bypass_queue_check; > > > >Let's just add this condition to the next if, no need for goto. > [Suman] I kept it a separate check to make the code more readable. Otherwise the next if condition will be complicated. Readability is subjective, but, FWIIW, I'd also prefer to avoid a goto here. > >> + > >> if (ring >= pfvf->hw.rx_queues && fsp->ring_cookie != > >RX_CLS_FLOW_DISC) > >> return -EINVAL; > >> > >> +bypass_queue_check: > >> if (fsp->location >= otx2_get_maxflows(flow_cfg)) > >> return -EINVAL; > >> > >> @@ -1182,6 +1200,9 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct > >ethtool_rxnfc *nfc) > >> flow_cfg->nr_flows++; > >> } > >> > >> + if (flow->is_vf) > >> + netdev_info(pfvf->netdev, > >> + "Make sure that VF's queue number is within its queue > >> +limit\n"); > >> return 0; > >> } > >>
>> >ethtool_rxnfc *nfc) >> >> struct ethhdr *eth_hdr; >> >> bool new = false; >> >> int err = 0; >> >> + u64 vf_num; >> >> u32 ring; >> >> >> >> if (!flow_cfg->max_flows) { >> >> @@ -1100,9 +1101,26 @@ int otx2_add_flow(struct otx2_nic *pfvf, >> >> struct >> >ethtool_rxnfc *nfc) >> >> if (!(pfvf->flags & OTX2_FLAG_NTUPLE_SUPPORT)) >> >> return -ENOMEM; >> >> >> >> + /* Number of queues on a VF can be greater or less than >> >> + * the PF's queue. Hence no need to check for the >> >> + * queue count. Hence no need to check queue count if PF >> >> + * is installing for its VF. Below is the expected vf_num >value >> >> + * based on the ethtool commands. >> >> + * >> >> + * e.g. >> >> + * 1. ethtool -U <netdev> ... action -1 ==> vf_num:255 >> >> + * 2. ethtool -U <netdev> ... action <queue_num> ==> vf_num:0 >> >> + * 3. ethtool -U <netdev> ... vf <vf_idx> queue <queue_num> >==> >> >> + * vf_num:vf_idx+1 >> >> + */ >> >> + vf_num = ethtool_get_flow_spec_ring_vf(fsp->ring_cookie); >> >> + if (!is_otx2_vf(pfvf->pcifunc) && vf_num) >> >> + goto bypass_queue_check; >> > >> >Let's just add this condition to the next if, no need for goto. >> [Suman] I kept it a separate check to make the code more readable. >Otherwise the next if condition will be complicated. > >Readability is subjective, but, FWIIW, I'd also prefer to avoid a goto >here. [Suman] Okay. Since both of you are suggesting the same change, I will update the same in v2. > >> >> + >> >> if (ring >= pfvf->hw.rx_queues && fsp->ring_cookie != >> >RX_CLS_FLOW_DISC) >> >> return -EINVAL; >> >> >> >> +bypass_queue_check: >> >> if (fsp->location >= otx2_get_maxflows(flow_cfg)) >> >> return -EINVAL; >> >> >> >> @@ -1182,6 +1200,9 @@ int otx2_add_flow(struct otx2_nic *pfvf, >> >> struct >> >ethtool_rxnfc *nfc) >> >> flow_cfg->nr_flows++; >> >> } >> >> >> >> + if (flow->is_vf) >> >> + netdev_info(pfvf->netdev, >> >> + "Make sure that VF's queue number is within its >queue >> >> +limit\n"); >> >> return 0; >> >> } >> >>
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c index 4762dbea64a1..4200f2d387f6 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c @@ -1088,6 +1088,7 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct ethtool_rxnfc *nfc) struct ethhdr *eth_hdr; bool new = false; int err = 0; + u64 vf_num; u32 ring; if (!flow_cfg->max_flows) { @@ -1100,9 +1101,26 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct ethtool_rxnfc *nfc) if (!(pfvf->flags & OTX2_FLAG_NTUPLE_SUPPORT)) return -ENOMEM; + /* Number of queues on a VF can be greater or less than + * the PF's queue. Hence no need to check for the + * queue count. Hence no need to check queue count if PF + * is installing for its VF. Below is the expected vf_num value + * based on the ethtool commands. + * + * e.g. + * 1. ethtool -U <netdev> ... action -1 ==> vf_num:255 + * 2. ethtool -U <netdev> ... action <queue_num> ==> vf_num:0 + * 3. ethtool -U <netdev> ... vf <vf_idx> queue <queue_num> ==> + * vf_num:vf_idx+1 + */ + vf_num = ethtool_get_flow_spec_ring_vf(fsp->ring_cookie); + if (!is_otx2_vf(pfvf->pcifunc) && vf_num) + goto bypass_queue_check; + if (ring >= pfvf->hw.rx_queues && fsp->ring_cookie != RX_CLS_FLOW_DISC) return -EINVAL; +bypass_queue_check: if (fsp->location >= otx2_get_maxflows(flow_cfg)) return -EINVAL; @@ -1182,6 +1200,9 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct ethtool_rxnfc *nfc) flow_cfg->nr_flows++; } + if (flow->is_vf) + netdev_info(pfvf->netdev, + "Make sure that VF's queue number is within its queue limit\n"); return 0; }
It is possible to add a ntuple rule which would like to direct packet to a VF whose number of queues are greater/less than its PF's queue numbers. For example a PF can have 2 Rx queues but a VF created on that PF can have 8 Rx queues. As of today, ntuple rule will reject rule because it is checking the requested queue number against PF's number of Rx queues. As a part of this fix if the action of a ntuple rule is to move a packet to a VF's queue then the check is removed. Also, a debug information is printed to aware user that it is user's responsibility to cross check if the requested queue number on that VF is a valid one. Fixes: f0a1913f8a6f ("octeontx2-pf: Add support for ethtool ntuple filters") Signed-off-by: Suman Ghosh <sumang@marvell.com> --- .../marvell/octeontx2/nic/otx2_flows.c | 21 +++++++++++++++++++ 1 file changed, 21 insertions(+)