* [PATCH v18 net-next 0/2] octeontx2: mqprio bandwidth offload for NIX TX schedulers
@ 2026-09-29 2:29 Ratheesh Kannoth
2026-09-29 2:29 ` [PATCH v18 net-next 1/2] octeontx2: use atomic bitops for PF/VF and rep flags Ratheesh Kannoth
` (3 more replies)
0 siblings, 4 replies; 7+ messages in thread
From: Ratheesh Kannoth @ 2026-09-29 2:29 UTC (permalink / raw)
To: bpf, linux-kernel, netdev
Cc: andrew+netdev, ast, daniel, davem, edumazet, hawk,
john.fastabend, kuba, pabeni, sdf, sgoutham, Ratheesh Kannoth
This series adds hardware offload for channel-mode mqprio with
TC_MQPRIO_SHAPER_BW_RATE on Marvell octeontx2/cn10k PF and VF RVU
netdevices. Each non-QoS transmit queue is shaped by programming MDQ CIR/PIR
on the NIX TX scheduler. When bandwidth offload is active, the driver
allocates one SMQ per queue, parents every MDQ under TL4[0], and maps each
traffic class min/max rate to the queue(s) in that class.
The NIX TX scheduler hierarchy cannot be reprogrammed live today, so
mqprio add, replace, delete, and failed-replace rollback rebuild it by
bouncing the netdev through ndo_stop()/ndo_open(). That intentionally
drops in-flight traffic on each change. otx2_mqprio_restart_netdev() clears
__LINK_STATE_START before ndo_stop() and does not call
dev_deactivate()/dev_activate(); carrier and TX queues are restored after
ndo_open() via the normal link-event path when the link is up. Cache the
active rates and restore MDQ shapers from otx2_mqprio_up() during ndo_open();
fail closed if restoration fails, leaving ndo_open() unsuccessful and the
interface administratively down.
Track mqprio configuration in mq_offload_snap snapshots (TC layout and
rates). On tc qdisc replace, stage the new configuration while keeping
the previous snapshot for rollback: failed setup restores the old
snapshot via netdev restart when the interface is running, successful
graft is recorded through TC_ROOT_GRAFT, and teardown of the replaced
qdisc instance commits the staged snapshot without tearing down the live
offload.
Patch 1 converts PF/VF and representor flag access to atomic bitops.
Patch 2 depends on it for safe OTX2_FLAG_INTF_DOWN and OTX2_FLAG_PORT_UP
updates on asynchronous mbox paths and during the mqprio netdev bounce.
The driver rejects offload unless the interface is running and the device
advertises CIR+PIR support. PF and VF RVU netdevices share the same TC
offload path via ndo_setup_tc / otx2_open(); SDP representors are not
supported. Per-TC rates are rejected when a traffic class spans more than
one queue. Concurrent PFC, XDP, SDP rep, or HTB use is blocked, and ethtool
channel count changes are blocked while mqprio bandwidth offload is active.
Ratheesh Kannoth (2):
octeontx2: use atomic bitops for PF/VF and rep flags
octeontx2: add mqprio bandwidth offload for NIX TX schedulers
.../ethernet/marvell/octeontx2/nic/cn10k_ipsec.c | 8 +-
.../ethernet/marvell/octeontx2/nic/otx2_common.c | 154 +++-
.../ethernet/marvell/octeontx2/nic/otx2_common.h | 107 ++-
.../ethernet/marvell/octeontx2/nic/otx2_dcbnl.c | 6 +
.../ethernet/marvell/octeontx2/nic/otx2_devlink.c | 2 +-
.../ethernet/marvell/octeontx2/nic/otx2_ethtool.c | 29 +-
.../ethernet/marvell/octeontx2/nic/otx2_flows.c | 34 +-
.../net/ethernet/marvell/octeontx2/nic/otx2_pf.c | 95 ++-
.../net/ethernet/marvell/octeontx2/nic/otx2_tc.c | 838 ++++++++++++++++++++-
.../net/ethernet/marvell/octeontx2/nic/otx2_txrx.c | 16 +-
.../net/ethernet/marvell/octeontx2/nic/otx2_vf.c | 10 +-
.../net/ethernet/marvell/octeontx2/nic/otx2_xsk.c | 4 +-
drivers/net/ethernet/marvell/octeontx2/nic/qos.c | 11 +
.../net/ethernet/marvell/octeontx2/nic/qos_sq.c | 4 +-
drivers/net/ethernet/marvell/octeontx2/nic/rep.c | 32 +-
drivers/net/ethernet/marvell/octeontx2/nic/rep.h | 3 +-
16 files changed, 1203 insertions(+), 150 deletions(-)
---
v17 -> v18: Addressed sashiko comments on v17.
- Commit staged mqprio replace snapshots when TC_ROOT_GRAFT is skipped
because hw-tc-offload is off in netdev->features (!tc_can_offload()),
instead of rolling back a replace that actually succeeded.
- Keep NETIF_F_HW_TC in netdev->hw_features only (PF and VF); gate mqprio
setup/teardown on otx2_tc_can_offload() so hw-tc-offload stays opt-in via
ethtool -K and flower/matchall are not pushed to ndo_setup_tc by default.
- Split TC shutdown around netdev unregister: cancel mqprio deferred work and
free mqprio snapshots before unregister_netdev(), but destroy the TC flower
flow list only after unregister so clsact teardown can still run
otx2_tc_del_flow() and free MCAM/mcast/policer state.
- Reject ethtool -L TX queue reduction while mqprio offload snapshots remain,
so a later replace rollback cannot restore stale per-queue rates past the
current queue count.
- Clarify mqprio shaper and netdev-restart comments: ndo_open() may restore
cached MDQ shapers via otx2_mqprio_up() before setup clears them, and a
failed mqprio restart relies on OTX2_FLAG_INTF_DOWN so otx2_stop() returns
early through netif_close(), not on skipping ndo_stop().
https://lore.kernel.org/netdev/20260923032217.1732753-1-rkannoth@marvell.com/
v16 -> v17: Addressed sashiko comments on v16.
- Replace the per-bit otx2_sync_flags_from_rep() loop with a masked
READ_ONCE/WRITE_ONCE publish of OTX2_REP_SYNC_FLAGS_MASK so lockless NAPI
readers never observe torn PF/representor flag combinations.
- Evaluate mqprio.rate_limit and old_mq_snap inside rtnl_lock in
otx2_mqprio_netdev_tc_work() so a concurrent qdisc delete cannot leave
stale netdev TC mappings after offload teardown.
- Advertise NETIF_F_HW_TC in netdev->features (PF and VF) when TC flower
offload is supported, so tc_can_offload() succeeds without ethtool -K
hw-tc-offload on; move otx2_init_tc() before register_netdev() and fix
probe/remove teardown ordering.
- Reject mqprio add when a software mqprio root is already installed
(otx2_mqprio_keep_netdev_tc()) and defer netdev TC restore from a new
fail_validate path on failed replace validation before any hardware
change.
- Fix otx2_mqprio_max_rate_bytes_ps() to cap against the NIX TLX maximum
rate instead of the burst-bucket size; use the 65536 byte HTB default
burst when programming MDQ shapers; guard otx2_get_smq_idx() when
txschq_cnt[NIX_TXSCH_LVL_SMQ] is zero after otx2_txschq_stop().
- Rename patch 2 to octeontx2: (driver-wide PF/VF offload, not PF-only).
https://lore.kernel.org/netdev/20260918015906.1255204-1-rkannoth@marvell.com/
v15 -> v16: Addressed sashiko comments on v15 and aligned documentation with code.
- Sync representor flags through OTX2_FLAG_MAX in otx2_sync_flags_from_rep()
instead of hard-coding OTX2_REP_VF_INITIALIZED as the loop bound.
- Drop the rvu_nix.c is_valid_txschq() ratelimited error print from the mqprio
patch; remove the misplaced atomic-bitops and AF-debug paragraphs from the
mqprio commit message (they belong to patch 1 or are out of scope).
- Extend mq_offload_snap to record prio_tc_map[] and mqprio rate flags; restore
the full netdev TC layout (num_tc, queue ranges, and priority map) via
otx2_mqprio_apply_snap_netdev() on rollback paths.
- Defer netdev TC restore on failed replace (otx2_mqprio_netdev_tc_work) so
rollback survives mqprio_destroy() clearing dev->num_tc after setup errors
once the core unwinds the failed qdisc instance.
- Stop calling dev_deactivate()/dev_activate() from otx2_mqprio_restart_netdev();
bounce the interface with ndo_stop()/ndo_open() only and restore carrier
through the normal link-event path after ndo_open(), avoiding qdisc
reentrancy during tc replace graft.
- Preserve netdev TC mappings when tearing down an offloaded instance that is
replaced by a software mqprio graft (otx2_mqprio_keep_netdev_tc()) instead
of always calling netdev_set_num_tc(0) and breaking the live replacement.
- Return an error from otx2_mqprio_down() when clearing hardware shapers fails
and keep offload software state, instead of v15's behaviour of clearing
rate_limit while stale MDQ limits may remain programmed.
- Rebuild the TX scheduler via netdev restart in otx2_mqprio_restore_old() on
a running interface after failed-replace rollback so partially applied MDQ
shapers are not left running with mismatched software state.
- Clear txschq_cnt[] in otx2_txschq_stop() after freeing scheduler nodes so
post-stop shaper mailbox operations do not consult stale counts.
- Document fail-closed ndo_open() when otx2_mqprio_up() cannot restore shapers,
and that PF/VF RVU netdevices share the ndo_setup_tc / otx2_open() offload
path (SDP representors remain unsupported); downgrade the mqprio restart
notice to netdev_dbg().
https://lore.kernel.org/netdev/20260911105521.689565-1-rkannoth@marvell.com/
v14 -> v15: Addressed sashiko comments.
- Split atomic PF/VF and representor flag access into a preparatory patch
so mqprio netdev-restart and mbox paths can update OTX2_FLAG_INTF_DOWN
and OTX2_FLAG_PORT_UP without data races on the shared flags word.
- Clear mqprio software state when hardware shaper teardown fails, warn,
and still bounce the netdev on delete so offload does not remain stuck
active after a mailbox error.
https://lore.kernel.org/netdev/20260904031553.3196916-1-rkannoth@marvell.com/
v13 -> v14: Addressed sashiko comments.
- Use atomic set_bit()/clear_bit() for OTX2_FLAG_INTF_DOWN and
OTX2_FLAG_PORT_UP updates on netdev-restart and mbox paths.
- Block concurrent mqprio bandwidth offload and HTB shaping.
- Fail ndo_open() if otx2_mqprio_up() cannot restore MDQ shapers.
- Rebuild the TX scheduler via netdev restart in otx2_mqprio_restore_old()
when rolling back a failed replace on a running interface.
- Return an error from otx2_mqprio_down() if clearing hardware shapers
fails instead of clearing software state anyway.
https://lore.kernel.org/netdev/20260904031553.3196916-1-rkannoth@marvell.com/
v12 -> v13: Addressed sashiko comments.
https://sashiko.dev/#/patchset/20260903023324.3078284-1-rkannoth%40marvell.com
v11 -> v12: Addressed sashiko comments.
https://sashiko.dev/#/patchset/20260902015500.2985371-1-rkannoth%40marvell.com
v10 -> v11: Addressed sashiko comments.
https://sashiko.dev/#/patchset/20260831131014.2639581-1-rkannoth%40marvell.com
v9 -> v10: Addressed sashiko/jacub comments.
https://sashiko.dev/#/message/20260817032747.1765883-1-rkannoth%40marvell.com
v8 -> v9: Addressed Sashiko comments
https://lore.kernel.org/netdev/aoJ6FhtWue0FHDQV@rkannoth-OptiPlex-7090/
v7 -> v8: Addressed Sashiko comments
https://sashiko.dev/#/patchset/20260811085050.3212280-1-rkannoth%40marvell.com
v6 -> v7: Addressed Sashiko comments
https://sashiko.dev/#/message/20260810034738.1786029-1-rkannoth%40marvell.com
v5 -> v6: Addressed Sashiko comments
https://lore.kernel.org/netdev/20260806095434.1144397-1-rkannoth@marvell.com/
v4 -> v5: Addressed sashiko comments
https://sashiko.dev/#/patchset/20260803042724.3380209-1-rkannoth%40marvell.com
v3 -> v4: Addressed sashiko comments
https://lore.kernel.org/netdev/20260729105139.2302908-1-rkannoth@marvell.com/
v2 -> v3: Addressed sashiko comments
https://lore.kernel.org/netdev/amnYX866mYx02cBe@rkannoth-OptiPlex-7090/T/#m67310cbec48b21c7720858ab3a1ea083a0f8dc10
v1 -> v2: Addressed sashiko comments
https://lore.kernel.org/netdev/20260724075010.2665758-1-rkannoth@marvell.com/
--
2.43.0
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH v18 net-next 1/2] octeontx2: use atomic bitops for PF/VF and rep flags
2026-09-29 2:29 [PATCH v18 net-next 0/2] octeontx2: mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
@ 2026-09-29 2:29 ` Ratheesh Kannoth
2026-10-02 8:31 ` netdev-bot+sashiko
2026-09-29 2:29 ` [PATCH v18 net-next 2/2] octeontx2: add mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
` (2 subsequent siblings)
3 siblings, 1 reply; 7+ messages in thread
From: Ratheesh Kannoth @ 2026-09-29 2:29 UTC (permalink / raw)
To: bpf, linux-kernel, netdev
Cc: andrew+netdev, ast, daniel, davem, edumazet, hawk,
john.fastabend, kuba, pabeni, sdf, sgoutham, Ratheesh Kannoth
Switch PF/VF and representor flags from non-atomic u64 masks to unsigned
long bitmaps with set_bit/clear_bit/test_bit. Add helpers and
otx2_sync_flags_from_rep() to publish representor-owned flags onto the PF
mailbox context with a masked WRITE_ONCE, giving lockless NAPI readers a
consistent word without clearing PF-owned bits.
Relocate representor VF initialization to OTX2_FLAG_REP_VF_INITIALIZED
(bit 21).
Signed-off-by: Ratheesh Kannoth <rkannoth@marvell.com>
---
.../marvell/octeontx2/nic/cn10k_ipsec.c | 8 +-
.../marvell/octeontx2/nic/otx2_common.c | 8 +-
.../marvell/octeontx2/nic/otx2_common.h | 84 ++++++++++++++-----
.../marvell/octeontx2/nic/otx2_devlink.c | 2 +-
.../marvell/octeontx2/nic/otx2_ethtool.c | 21 ++---
.../marvell/octeontx2/nic/otx2_flows.c | 34 ++++----
.../ethernet/marvell/octeontx2/nic/otx2_pf.c | 78 ++++++++---------
.../ethernet/marvell/octeontx2/nic/otx2_tc.c | 30 +++----
.../marvell/octeontx2/nic/otx2_txrx.c | 16 ++--
.../ethernet/marvell/octeontx2/nic/otx2_vf.c | 10 +--
.../ethernet/marvell/octeontx2/nic/otx2_xsk.c | 4 +-
.../ethernet/marvell/octeontx2/nic/qos_sq.c | 4 +-
.../net/ethernet/marvell/octeontx2/nic/rep.c | 32 +++----
.../net/ethernet/marvell/octeontx2/nic/rep.h | 3 +-
14 files changed, 185 insertions(+), 149 deletions(-)
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c
index 77543d472345..50ec4542c418 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c
@@ -334,7 +334,7 @@ static int cn10k_outb_cpt_init(struct net_device *netdev)
CN10K_CPT_LF_NQX(0));
/* Set ipsec offload enabled for this device */
- pf->flags |= OTX2_FLAG_IPSEC_OFFLOAD_ENABLED;
+ otx2_set_flag(pf, OTX2_FLAG_IPSEC_OFFLOAD_ENABLED);
cn10k_cpt_device_set_available(pf);
return 0;
@@ -356,7 +356,7 @@ static int cn10k_outb_cpt_clean(struct otx2_nic *pf)
}
/* Set ipsec offload disabled for this device */
- pf->flags &= ~OTX2_FLAG_IPSEC_OFFLOAD_ENABLED;
+ otx2_clear_flag(pf, OTX2_FLAG_IPSEC_OFFLOAD_ENABLED);
/* Disable CPTLF Instruction Queue (IQ) */
cn10k_outb_cptlf_iq_disable(pf);
@@ -820,7 +820,7 @@ void cn10k_ipsec_clean(struct otx2_nic *pf)
if (!is_dev_support_ipsec_offload(pf->pdev))
return;
- if (!(pf->flags & OTX2_FLAG_IPSEC_OFFLOAD_ENABLED))
+ if (!otx2_test_flag(pf, OTX2_FLAG_IPSEC_OFFLOAD_ENABLED))
return;
if (pf->ipsec.sa_workq) {
@@ -945,7 +945,7 @@ bool cn10k_ipsec_transmit(struct otx2_nic *pf, struct netdev_queue *txq,
u16 dlen;
/* Check for IPSEC offload enabled */
- if (!(pf->flags & OTX2_FLAG_IPSEC_OFFLOAD_ENABLED))
+ if (!otx2_test_flag(pf, OTX2_FLAG_IPSEC_OFFLOAD_ENABLED))
goto drop;
sp = skb_sec_path(skb);
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
index 175992188c18..b421cb75e44b 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
@@ -220,10 +220,10 @@ int otx2_set_mac_address(struct net_device *netdev, void *p)
eth_hw_addr_set(netdev, addr->sa_data);
/* update dmac field in vlan offload rule */
if (netif_running(netdev) &&
- pfvf->flags & OTX2_FLAG_RX_VLAN_SUPPORT)
+ otx2_test_flag(pfvf, OTX2_FLAG_RX_VLAN_SUPPORT))
otx2_install_rxvlan_offload_flow(pfvf);
/* update dmac address in ntuple and DMAC filter list */
- if (pfvf->flags & OTX2_FLAG_DMACFLTR_SUPPORT)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_DMACFLTR_SUPPORT))
otx2_dmacflt_update_pfmac_flow(pfvf);
} else {
return -EPERM;
@@ -275,8 +275,8 @@ int otx2_config_pause_frm(struct otx2_nic *pfvf)
goto unlock;
}
- req->rx_pause = !!(pfvf->flags & OTX2_FLAG_RX_PAUSE_ENABLED);
- req->tx_pause = !!(pfvf->flags & OTX2_FLAG_TX_PAUSE_ENABLED);
+ req->rx_pause = otx2_test_flag(pfvf, OTX2_FLAG_RX_PAUSE_ENABLED);
+ req->tx_pause = otx2_test_flag(pfvf, OTX2_FLAG_TX_PAUSE_ENABLED);
req->set = 1;
err = otx2_sync_mbox_msg(&pfvf->mbox);
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
index 5850bc1870a1..b146ca4be5db 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
@@ -482,6 +482,41 @@ struct pf_irq_data {
int mdevs;
};
+enum otx2_flag_bits {
+ OTX2_FLAG_RX_TSTAMP_ENABLED,
+ OTX2_FLAG_TX_TSTAMP_ENABLED,
+ OTX2_FLAG_INTF_DOWN,
+ OTX2_FLAG_MCAM_ENTRIES_ALLOC,
+ OTX2_FLAG_NTUPLE_SUPPORT,
+ OTX2_FLAG_UCAST_FLTR_SUPPORT,
+ OTX2_FLAG_RX_VLAN_SUPPORT,
+ OTX2_FLAG_VF_VLAN_SUPPORT,
+ OTX2_FLAG_PF_SHUTDOWN,
+ OTX2_FLAG_RX_PAUSE_ENABLED,
+ OTX2_FLAG_TX_PAUSE_ENABLED,
+ OTX2_FLAG_TC_FLOWER_SUPPORT,
+ OTX2_FLAG_TC_MATCHALL_EGRESS_ENABLED,
+ OTX2_FLAG_TC_MATCHALL_INGRESS_ENABLED,
+ OTX2_FLAG_DMACFLTR_SUPPORT,
+ OTX2_FLAG_PTP_ONESTEP_SYNC,
+ OTX2_FLAG_ADPTV_INT_COAL_ENABLED,
+ OTX2_FLAG_TC_MARK_ENABLED,
+ OTX2_FLAG_REP_MODE_ENABLED,
+ OTX2_FLAG_PORT_UP,
+ OTX2_FLAG_IPSEC_OFFLOAD_ENABLED,
+ OTX2_FLAG_REP_VF_INITIALIZED,
+ OTX2_FLAG_MAX,
+};
+
+/* Representor-owned flags copied onto the PF mailbox context in
+ * rvu_rep_setup_tc_cb(). All other bits are owned by the PF/VF netdev.
+ */
+#define OTX2_REP_SYNC_FLAGS_MASK \
+ (BIT(OTX2_FLAG_MCAM_ENTRIES_ALLOC) | \
+ BIT(OTX2_FLAG_NTUPLE_SUPPORT) | \
+ BIT(OTX2_FLAG_TC_FLOWER_SUPPORT) | \
+ BIT(OTX2_FLAG_REP_VF_INITIALIZED))
+
struct otx2_nic {
void __iomem *reg_base;
struct net_device *netdev;
@@ -490,28 +525,7 @@ struct otx2_nic {
u16 tx_max_pktlen;
u16 rbsize; /* Receive buffer size */
-#define OTX2_FLAG_RX_TSTAMP_ENABLED BIT_ULL(0)
-#define OTX2_FLAG_TX_TSTAMP_ENABLED BIT_ULL(1)
-#define OTX2_FLAG_INTF_DOWN BIT_ULL(2)
-#define OTX2_FLAG_MCAM_ENTRIES_ALLOC BIT_ULL(3)
-#define OTX2_FLAG_NTUPLE_SUPPORT BIT_ULL(4)
-#define OTX2_FLAG_UCAST_FLTR_SUPPORT BIT_ULL(5)
-#define OTX2_FLAG_RX_VLAN_SUPPORT BIT_ULL(6)
-#define OTX2_FLAG_VF_VLAN_SUPPORT BIT_ULL(7)
-#define OTX2_FLAG_PF_SHUTDOWN BIT_ULL(8)
-#define OTX2_FLAG_RX_PAUSE_ENABLED BIT_ULL(9)
-#define OTX2_FLAG_TX_PAUSE_ENABLED BIT_ULL(10)
-#define OTX2_FLAG_TC_FLOWER_SUPPORT BIT_ULL(11)
-#define OTX2_FLAG_TC_MATCHALL_EGRESS_ENABLED BIT_ULL(12)
-#define OTX2_FLAG_TC_MATCHALL_INGRESS_ENABLED BIT_ULL(13)
-#define OTX2_FLAG_DMACFLTR_SUPPORT BIT_ULL(14)
-#define OTX2_FLAG_PTP_ONESTEP_SYNC BIT_ULL(15)
-#define OTX2_FLAG_ADPTV_INT_COAL_ENABLED BIT_ULL(16)
-#define OTX2_FLAG_TC_MARK_ENABLED BIT_ULL(17)
-#define OTX2_FLAG_REP_MODE_ENABLED BIT_ULL(18)
-#define OTX2_FLAG_PORT_UP BIT_ULL(19)
-#define OTX2_FLAG_IPSEC_OFFLOAD_ENABLED BIT_ULL(20)
- u64 flags;
+ unsigned long flags;
u64 *cq_op_addr;
struct bpf_prog *xdp_prog;
@@ -593,6 +607,32 @@ struct otx2_nic {
unsigned long *af_xdp_zc_qidx;
};
+static inline void otx2_set_flag(struct otx2_nic *nic, unsigned int flag)
+{
+ set_bit(flag, &nic->flags);
+}
+
+static inline void otx2_clear_flag(struct otx2_nic *nic, unsigned int flag)
+{
+ clear_bit(flag, &nic->flags);
+}
+
+static inline bool otx2_test_flag(struct otx2_nic *nic, unsigned int flag)
+{
+ return test_bit(flag, &nic->flags);
+}
+
+static inline void otx2_sync_flags_from_rep(struct otx2_nic *dst,
+ unsigned long *src_flags)
+{
+ unsigned long src = READ_ONCE(*src_flags);
+ unsigned long new_flags;
+
+ new_flags = (READ_ONCE(dst->flags) & ~OTX2_REP_SYNC_FLAGS_MASK) |
+ (src & OTX2_REP_SYNC_FLAGS_MASK);
+ WRITE_ONCE(dst->flags, new_flags);
+}
+
static inline bool is_otx2_lbkvf(struct pci_dev *pdev)
{
return (pdev->device == PCI_DEVID_OCTEONTX2_RVU_AFVF) ||
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.c
index 4a5ce0e67dda..863a5ced9a26 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.c
@@ -104,7 +104,7 @@ static int otx2_dl_ucast_flt_cnt_validate(struct devlink *devlink, u32 id,
struct otx2_nic *pfvf = otx2_dl->pfvf;
/* Check for UNICAST filter support*/
- if (!(pfvf->flags & OTX2_FLAG_UCAST_FLTR_SUPPORT)) {
+ if (!otx2_test_flag(pfvf, OTX2_FLAG_UCAST_FLTR_SUPPORT)) {
NL_SET_ERR_MSG_MOD(extack,
"Unicast filter not enabled");
return -EINVAL;
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
index a05dee0085a3..4fe473d9ea0d 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
@@ -354,14 +354,14 @@ static int otx2_set_pauseparam(struct net_device *netdev,
return -EOPNOTSUPP;
if (pause->rx_pause)
- pfvf->flags |= OTX2_FLAG_RX_PAUSE_ENABLED;
+ otx2_set_flag(pfvf, OTX2_FLAG_RX_PAUSE_ENABLED);
else
- pfvf->flags &= ~OTX2_FLAG_RX_PAUSE_ENABLED;
+ otx2_clear_flag(pfvf, OTX2_FLAG_RX_PAUSE_ENABLED);
if (pause->tx_pause)
- pfvf->flags |= OTX2_FLAG_TX_PAUSE_ENABLED;
+ otx2_set_flag(pfvf, OTX2_FLAG_TX_PAUSE_ENABLED);
else
- pfvf->flags &= ~OTX2_FLAG_TX_PAUSE_ENABLED;
+ otx2_clear_flag(pfvf, OTX2_FLAG_TX_PAUSE_ENABLED);
return otx2_config_pause_frm(pfvf);
}
@@ -470,8 +470,7 @@ static int otx2_get_coalesce(struct net_device *netdev,
cmd->rx_max_coalesced_frames = hw->cq_ecount_wait;
cmd->tx_coalesce_usecs = hw->cq_time_wait;
cmd->tx_max_coalesced_frames = hw->cq_ecount_wait;
- if ((pfvf->flags & OTX2_FLAG_ADPTV_INT_COAL_ENABLED) ==
- OTX2_FLAG_ADPTV_INT_COAL_ENABLED) {
+ if (otx2_test_flag(pfvf, OTX2_FLAG_ADPTV_INT_COAL_ENABLED)) {
cmd->use_adaptive_rx_coalesce = 1;
cmd->use_adaptive_tx_coalesce = 1;
} else {
@@ -502,15 +501,14 @@ static int otx2_set_coalesce(struct net_device *netdev,
}
/* Check and update coalesce status */
- if ((pfvf->flags & OTX2_FLAG_ADPTV_INT_COAL_ENABLED) ==
- OTX2_FLAG_ADPTV_INT_COAL_ENABLED) {
+ if (otx2_test_flag(pfvf, OTX2_FLAG_ADPTV_INT_COAL_ENABLED)) {
priv_coalesce_status = 1;
if (!ec->use_adaptive_rx_coalesce)
- pfvf->flags &= ~OTX2_FLAG_ADPTV_INT_COAL_ENABLED;
+ otx2_clear_flag(pfvf, OTX2_FLAG_ADPTV_INT_COAL_ENABLED);
} else {
priv_coalesce_status = 0;
if (ec->use_adaptive_rx_coalesce)
- pfvf->flags |= OTX2_FLAG_ADPTV_INT_COAL_ENABLED;
+ otx2_set_flag(pfvf, OTX2_FLAG_ADPTV_INT_COAL_ENABLED);
}
/* 'cq_time_wait' is 8bit and is in multiple of 100ns,
@@ -556,8 +554,7 @@ static int otx2_set_coalesce(struct net_device *netdev,
* 'on' to 'off'.
*/
if (priv_coalesce_status &&
- ((pfvf->flags & OTX2_FLAG_ADPTV_INT_COAL_ENABLED) !=
- OTX2_FLAG_ADPTV_INT_COAL_ENABLED)) {
+ (!otx2_test_flag(pfvf, OTX2_FLAG_ADPTV_INT_COAL_ENABLED))) {
hw->cq_time_wait = CQ_TIMER_THRESH_DEFAULT;
hw->cq_ecount_wait = CQ_CQE_THRESH_DEFAULT;
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c
index 99d78fc5a2c4..b8ff49f0f6e3 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_flows.c
@@ -270,9 +270,9 @@ int otx2_alloc_mcam_entries(struct otx2_nic *pfvf, u16 count)
flow_cfg->max_flows = allocated;
if (allocated) {
- pfvf->flags |= OTX2_FLAG_MCAM_ENTRIES_ALLOC;
- pfvf->flags |= OTX2_FLAG_NTUPLE_SUPPORT;
- pfvf->flags |= OTX2_FLAG_TC_FLOWER_SUPPORT;
+ otx2_set_flag(pfvf, OTX2_FLAG_MCAM_ENTRIES_ALLOC);
+ otx2_set_flag(pfvf, OTX2_FLAG_NTUPLE_SUPPORT);
+ otx2_set_flag(pfvf, OTX2_FLAG_TC_FLOWER_SUPPORT);
}
if (allocated != count)
@@ -376,7 +376,7 @@ int otx2_mcam_entry_init(struct otx2_nic *pfvf)
flow_cfg->unicast_offset = vf_vlan_max_flows;
flow_cfg->rx_vlan_offset = flow_cfg->unicast_offset +
flow_cfg->ucast_flt_cnt;
- pfvf->flags |= OTX2_FLAG_UCAST_FLTR_SUPPORT;
+ otx2_set_flag(pfvf, OTX2_FLAG_UCAST_FLTR_SUPPORT);
/* Check if NPC_DMAC field is supported
* by the mkex profile before setting VLAN support flag.
@@ -401,11 +401,11 @@ int otx2_mcam_entry_init(struct otx2_nic *pfvf)
}
if (frsp->enable) {
- pfvf->flags |= OTX2_FLAG_RX_VLAN_SUPPORT;
- pfvf->flags |= OTX2_FLAG_VF_VLAN_SUPPORT;
+ otx2_set_flag(pfvf, OTX2_FLAG_RX_VLAN_SUPPORT);
+ otx2_set_flag(pfvf, OTX2_FLAG_VF_VLAN_SUPPORT);
}
- pfvf->flags |= OTX2_FLAG_MCAM_ENTRIES_ALLOC;
+ otx2_set_flag(pfvf, OTX2_FLAG_MCAM_ENTRIES_ALLOC);
mutex_unlock(&pfvf->mbox.lock);
/* Allocate entries for Ntuple filters */
@@ -415,7 +415,7 @@ int otx2_mcam_entry_init(struct otx2_nic *pfvf)
return 0;
}
- pfvf->flags |= OTX2_FLAG_TC_FLOWER_SUPPORT;
+ otx2_set_flag(pfvf, OTX2_FLAG_TC_FLOWER_SUPPORT);
refcount_set(&flow_cfg->mark_flows, 1);
return 0;
@@ -479,7 +479,7 @@ int otx2_mcam_flow_init(struct otx2_nic *pf)
return err;
/* Check if MCAM entries are allocate or not */
- if (!(pf->flags & OTX2_FLAG_UCAST_FLTR_SUPPORT))
+ if (!otx2_test_flag(pf, OTX2_FLAG_UCAST_FLTR_SUPPORT))
return 0;
pf->mac_table = devm_kzalloc(pf->dev, sizeof(struct otx2_mac_table)
@@ -501,7 +501,7 @@ int otx2_mcam_flow_init(struct otx2_nic *pf)
if (!pf->flow_cfg->bmap_to_dmacindex)
return -ENOMEM;
- pf->flags |= OTX2_FLAG_DMACFLTR_SUPPORT;
+ otx2_set_flag(pf, OTX2_FLAG_DMACFLTR_SUPPORT);
return 0;
}
@@ -521,7 +521,7 @@ static int otx2_do_add_macfilter(struct otx2_nic *pf, const u8 *mac)
struct npc_install_flow_req *req;
int err, i;
- if (!(pf->flags & OTX2_FLAG_UCAST_FLTR_SUPPORT))
+ if (!otx2_test_flag(pf, OTX2_FLAG_UCAST_FLTR_SUPPORT))
return -ENOMEM;
/* dont have free mcam entries or uc list is greater than alloted */
@@ -1167,7 +1167,7 @@ static int otx2_is_flow_rule_dmacfilter(struct otx2_nic *pfvf,
u64 ring_cookie = fsp->ring_cookie;
u32 flow_type;
- if (!(pfvf->flags & OTX2_FLAG_DMACFLTR_SUPPORT))
+ if (!otx2_test_flag(pfvf, OTX2_FLAG_DMACFLTR_SUPPORT))
return false;
flow_type = fsp->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT | FLOW_RSS);
@@ -1364,7 +1364,7 @@ int otx2_add_flow(struct otx2_nic *pfvf, struct ethtool_rxnfc *nfc)
}
ring = ethtool_get_flow_spec_ring(fsp->ring_cookie);
- if (!(pfvf->flags & OTX2_FLAG_NTUPLE_SUPPORT))
+ if (!otx2_test_flag(pfvf, OTX2_FLAG_NTUPLE_SUPPORT))
return -ENOMEM;
/* Number of queues on a VF can be greater or less than
@@ -1596,7 +1596,7 @@ int otx2_destroy_ntuple_flows(struct otx2_nic *pfvf)
struct otx2_flow *iter, *tmp;
int err;
- if (!(pfvf->flags & OTX2_FLAG_NTUPLE_SUPPORT))
+ if (!otx2_test_flag(pfvf, OTX2_FLAG_NTUPLE_SUPPORT))
return 0;
if (!flow_cfg->max_flows)
@@ -1629,7 +1629,7 @@ int otx2_destroy_mcam_flows(struct otx2_nic *pfvf)
struct otx2_flow *iter, *tmp;
int err;
- if (!(pfvf->flags & OTX2_FLAG_MCAM_ENTRIES_ALLOC))
+ if (!otx2_test_flag(pfvf, OTX2_FLAG_MCAM_ENTRIES_ALLOC))
return 0;
/* remove all flows */
@@ -1658,7 +1658,7 @@ int otx2_destroy_mcam_flows(struct otx2_nic *pfvf)
return err;
}
- pfvf->flags &= ~OTX2_FLAG_MCAM_ENTRIES_ALLOC;
+ otx2_clear_flag(pfvf, OTX2_FLAG_MCAM_ENTRIES_ALLOC);
flow_cfg->max_flows = 0;
mutex_unlock(&pfvf->mbox.lock);
@@ -1721,7 +1721,7 @@ int otx2_enable_rxvlan(struct otx2_nic *pf, bool enable)
int err;
/* Dont have enough mcam entries */
- if (!(pf->flags & OTX2_FLAG_RX_VLAN_SUPPORT))
+ if (!otx2_test_flag(pf, OTX2_FLAG_RX_VLAN_SUPPORT))
return -ENOMEM;
if (enable) {
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
index c0e2100de1d9..32582b6347ea 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
@@ -879,7 +879,7 @@ static void otx2_handle_link_event(struct otx2_nic *pf)
struct cgx_link_user_info *linfo = &pf->linfo;
struct net_device *netdev = pf->netdev;
- if (pf->flags & OTX2_FLAG_PORT_UP)
+ if (otx2_test_flag(pf, OTX2_FLAG_PORT_UP))
return;
pr_info("%s NIC Link is %s %d Mbps %s duplex\n", netdev->name,
@@ -907,11 +907,11 @@ static int otx2_mbox_up_handler_rep_event_up_notify(struct otx2_nic *pf,
if (info->event == RVU_EVENT_PORT_STATE) {
if (info->evt_data.port_state) {
- pf->flags |= OTX2_FLAG_PORT_UP;
+ otx2_set_flag(pf, OTX2_FLAG_PORT_UP);
netif_carrier_on(netdev);
netif_tx_start_all_queues(netdev);
} else {
- pf->flags &= ~OTX2_FLAG_PORT_UP;
+ otx2_clear_flag(pf, OTX2_FLAG_PORT_UP);
netif_tx_stop_all_queues(netdev);
netif_carrier_off(netdev);
}
@@ -953,7 +953,7 @@ int otx2_mbox_up_handler_cgx_link_event(struct otx2_nic *pf,
}
/* interface has not been fully configured yet */
- if (pf->flags & OTX2_FLAG_INTF_DOWN)
+ if (otx2_test_flag(pf, OTX2_FLAG_INTF_DOWN))
return 0;
otx2_handle_link_event(pf);
@@ -1828,7 +1828,7 @@ void otx2_free_hw_resources(struct otx2_nic *pf)
free_req = otx2_mbox_alloc_msg_nix_lf_free(mbox);
if (free_req) {
free_req->flags = NIX_LF_DISABLE_FLOWS | NIX_LF_DONT_FREE_DFT_IDXS;
- if (!(pf->flags & OTX2_FLAG_PF_SHUTDOWN))
+ if (!otx2_test_flag(pf, OTX2_FLAG_PF_SHUTDOWN))
free_req->flags |= NIX_LF_DONT_FREE_TX_VTAG;
if (otx2_sync_mbox_msg(mbox))
dev_err(pf->dev, "%s failed to free nixlf\n", __func__);
@@ -2135,21 +2135,21 @@ int otx2_open(struct net_device *netdev)
}
otx2_write64(pf, NIX_LF_RAS_ENA_W1S, NIX_LF_RAS_MASK);
- if (pf->flags & OTX2_FLAG_RX_VLAN_SUPPORT)
+ if (otx2_test_flag(pf, OTX2_FLAG_RX_VLAN_SUPPORT))
otx2_enable_rxvlan(pf, true);
/* When reinitializing enable time stamping if it is enabled before */
- if (pf->flags & OTX2_FLAG_TX_TSTAMP_ENABLED) {
- pf->flags &= ~OTX2_FLAG_TX_TSTAMP_ENABLED;
+ if (otx2_test_flag(pf, OTX2_FLAG_TX_TSTAMP_ENABLED)) {
+ otx2_clear_flag(pf, OTX2_FLAG_TX_TSTAMP_ENABLED);
otx2_config_hw_tx_tstamp(pf, true);
}
- if (pf->flags & OTX2_FLAG_RX_TSTAMP_ENABLED) {
- pf->flags &= ~OTX2_FLAG_RX_TSTAMP_ENABLED;
+ if (otx2_test_flag(pf, OTX2_FLAG_RX_TSTAMP_ENABLED)) {
+ otx2_clear_flag(pf, OTX2_FLAG_RX_TSTAMP_ENABLED);
otx2_config_hw_rx_tstamp(pf, true);
}
- pf->flags &= ~OTX2_FLAG_INTF_DOWN;
- pf->flags &= ~OTX2_FLAG_PORT_UP;
+ otx2_clear_flag(pf, OTX2_FLAG_INTF_DOWN);
+ otx2_clear_flag(pf, OTX2_FLAG_PORT_UP);
/* 'intf_down' may be checked on any cpu */
smp_wmb();
@@ -2161,7 +2161,7 @@ int otx2_open(struct net_device *netdev)
otx2_handle_link_event(pf);
/* Install DMAC Filters */
- if (pf->flags & OTX2_FLAG_DMACFLTR_SUPPORT)
+ if (otx2_test_flag(pf, OTX2_FLAG_DMACFLTR_SUPPORT))
otx2_dmacflt_reinstall_flows(pf);
otx2_tc_apply_ingress_police_rules(pf);
@@ -2186,7 +2186,7 @@ int otx2_open(struct net_device *netdev)
err_tx_stop_queues:
netif_tx_stop_all_queues(netdev);
netif_carrier_off(netdev);
- pf->flags |= OTX2_FLAG_INTF_DOWN;
+ otx2_set_flag(pf, OTX2_FLAG_INTF_DOWN);
/* free NIXLF POISON irq */
vec = pci_irq_vector(pf->pdev,
pf->hw.nix_msixoff + NIX_LF_POISON_VEC);
@@ -2220,13 +2220,13 @@ int otx2_stop(struct net_device *netdev)
int qidx, vec, wrk;
/* If the DOWN flag is set resources are already freed */
- if (pf->flags & OTX2_FLAG_INTF_DOWN)
+ if (otx2_test_flag(pf, OTX2_FLAG_INTF_DOWN))
return 0;
netif_carrier_off(netdev);
netif_tx_stop_all_queues(netdev);
- pf->flags |= OTX2_FLAG_INTF_DOWN;
+ otx2_set_flag(pf, OTX2_FLAG_INTF_DOWN);
/* 'intf_down' may be checked on any cpu */
smp_wmb();
@@ -2457,7 +2457,7 @@ static int otx2_config_hw_rx_tstamp(struct otx2_nic *pfvf, bool enable)
struct msg_req *req;
int err;
- if (pfvf->flags & OTX2_FLAG_RX_TSTAMP_ENABLED && enable)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_RX_TSTAMP_ENABLED) && enable)
return 0;
mutex_lock(&pfvf->mbox.lock);
@@ -2478,9 +2478,9 @@ static int otx2_config_hw_rx_tstamp(struct otx2_nic *pfvf, bool enable)
mutex_unlock(&pfvf->mbox.lock);
if (enable)
- pfvf->flags |= OTX2_FLAG_RX_TSTAMP_ENABLED;
+ otx2_set_flag(pfvf, OTX2_FLAG_RX_TSTAMP_ENABLED);
else
- pfvf->flags &= ~OTX2_FLAG_RX_TSTAMP_ENABLED;
+ otx2_clear_flag(pfvf, OTX2_FLAG_RX_TSTAMP_ENABLED);
return 0;
}
@@ -2489,7 +2489,7 @@ static int otx2_config_hw_tx_tstamp(struct otx2_nic *pfvf, bool enable)
struct msg_req *req;
int err;
- if (pfvf->flags & OTX2_FLAG_TX_TSTAMP_ENABLED && enable)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_TX_TSTAMP_ENABLED) && enable)
return 0;
mutex_lock(&pfvf->mbox.lock);
@@ -2510,9 +2510,9 @@ static int otx2_config_hw_tx_tstamp(struct otx2_nic *pfvf, bool enable)
mutex_unlock(&pfvf->mbox.lock);
if (enable)
- pfvf->flags |= OTX2_FLAG_TX_TSTAMP_ENABLED;
+ otx2_set_flag(pfvf, OTX2_FLAG_TX_TSTAMP_ENABLED);
else
- pfvf->flags &= ~OTX2_FLAG_TX_TSTAMP_ENABLED;
+ otx2_clear_flag(pfvf, OTX2_FLAG_TX_TSTAMP_ENABLED);
return 0;
}
@@ -2537,8 +2537,8 @@ int otx2_config_hwtstamp_set(struct net_device *netdev,
switch (config->tx_type) {
case HWTSTAMP_TX_OFF:
- if (pfvf->flags & OTX2_FLAG_PTP_ONESTEP_SYNC)
- pfvf->flags &= ~OTX2_FLAG_PTP_ONESTEP_SYNC;
+ if (otx2_test_flag(pfvf, OTX2_FLAG_PTP_ONESTEP_SYNC))
+ otx2_clear_flag(pfvf, OTX2_FLAG_PTP_ONESTEP_SYNC);
cancel_delayed_work(&pfvf->ptp->synctstamp_work);
otx2_config_hw_tx_tstamp(pfvf, false);
@@ -2549,7 +2549,7 @@ int otx2_config_hwtstamp_set(struct net_device *netdev,
"One-step time stamping is not supported");
return -ERANGE;
}
- pfvf->flags |= OTX2_FLAG_PTP_ONESTEP_SYNC;
+ otx2_set_flag(pfvf, OTX2_FLAG_PTP_ONESTEP_SYNC);
schedule_delayed_work(&pfvf->ptp->synctstamp_work,
msecs_to_jiffies(500));
fallthrough;
@@ -2835,7 +2835,7 @@ static int otx2_set_vf_vlan(struct net_device *netdev, int vf, u16 vlan, u8 qos,
if (proto != htons(ETH_P_8021Q))
return -EPROTONOSUPPORT;
- if (!(pf->flags & OTX2_FLAG_VF_VLAN_SUPPORT))
+ if (!otx2_test_flag(pf, OTX2_FLAG_VF_VLAN_SUPPORT))
return -EOPNOTSUPP;
return otx2_do_set_vf_vlan(pf, vf, vlan, qos, proto);
@@ -3086,7 +3086,7 @@ int otx2_realloc_msix_vectors(struct otx2_nic *pf)
* interrupt range (QINT, CINT, GINT, ERR and POISON vectors).
*/
num_vec = hw->nix_msixoff;
- if (pf->flags & OTX2_FLAG_REP_MODE_ENABLED)
+ if (otx2_test_flag(pf, OTX2_FLAG_REP_MODE_ENABLED))
num_vec += NIX_LF_CINT_VEC_START + hw->max_queues;
else
num_vec += NIX_LF_POISON_VEC + 1;
@@ -3273,7 +3273,7 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id)
pf->pdev = pdev;
pf->dev = dev;
pf->total_vfs = pci_sriov_get_totalvfs(pdev);
- pf->flags |= OTX2_FLAG_INTF_DOWN;
+ otx2_set_flag(pf, OTX2_FLAG_INTF_DOWN);
hw = &pf->hw;
hw->pdev = pdev;
@@ -3328,23 +3328,23 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id)
if (err)
goto err_del_mcam_entries;
- if (pf->flags & OTX2_FLAG_NTUPLE_SUPPORT)
+ if (otx2_test_flag(pf, OTX2_FLAG_NTUPLE_SUPPORT))
netdev->hw_features |= NETIF_F_NTUPLE;
- if (pf->flags & OTX2_FLAG_UCAST_FLTR_SUPPORT)
+ if (otx2_test_flag(pf, OTX2_FLAG_UCAST_FLTR_SUPPORT))
netdev->priv_flags |= IFF_UNICAST_FLT;
/* Support TSO on tag interface */
netdev->vlan_features |= netdev->features;
netdev->hw_features |= NETIF_F_HW_VLAN_CTAG_TX |
NETIF_F_HW_VLAN_STAG_TX;
- if (pf->flags & OTX2_FLAG_RX_VLAN_SUPPORT)
+ if (otx2_test_flag(pf, OTX2_FLAG_RX_VLAN_SUPPORT))
netdev->hw_features |= NETIF_F_HW_VLAN_CTAG_RX |
NETIF_F_HW_VLAN_STAG_RX;
netdev->features |= netdev->hw_features;
/* HW supports tc offload but mutually exclusive with n-tuple filters */
- if (pf->flags & OTX2_FLAG_TC_FLOWER_SUPPORT)
+ if (otx2_test_flag(pf, OTX2_FLAG_TC_FLOWER_SUPPORT))
netdev->hw_features |= NETIF_F_HW_TC;
netdev->hw_features |= NETIF_F_LOOPBACK | NETIF_F_RXALL;
@@ -3595,18 +3595,18 @@ static void otx2_remove(struct pci_dev *pdev)
pf = netdev_priv(netdev);
- pf->flags |= OTX2_FLAG_PF_SHUTDOWN;
+ otx2_set_flag(pf, OTX2_FLAG_PF_SHUTDOWN);
- if (pf->flags & OTX2_FLAG_TX_TSTAMP_ENABLED)
+ if (otx2_test_flag(pf, OTX2_FLAG_TX_TSTAMP_ENABLED))
otx2_config_hw_tx_tstamp(pf, false);
- if (pf->flags & OTX2_FLAG_RX_TSTAMP_ENABLED)
+ if (otx2_test_flag(pf, OTX2_FLAG_RX_TSTAMP_ENABLED))
otx2_config_hw_rx_tstamp(pf, false);
/* Disable 802.3x pause frames */
- if (pf->flags & OTX2_FLAG_RX_PAUSE_ENABLED ||
- (pf->flags & OTX2_FLAG_TX_PAUSE_ENABLED)) {
- pf->flags &= ~OTX2_FLAG_RX_PAUSE_ENABLED;
- pf->flags &= ~OTX2_FLAG_TX_PAUSE_ENABLED;
+ if (otx2_test_flag(pf, OTX2_FLAG_RX_PAUSE_ENABLED) ||
+ otx2_test_flag(pf, OTX2_FLAG_TX_PAUSE_ENABLED)) {
+ otx2_clear_flag(pf, OTX2_FLAG_RX_PAUSE_ENABLED);
+ otx2_clear_flag(pf, OTX2_FLAG_TX_PAUSE_ENABLED);
otx2_config_pause_frm(pf);
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
index 039fd47ebf52..ddb46b580c3b 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
@@ -159,7 +159,7 @@ static int otx2_tc_validate_flow(struct otx2_nic *nic,
struct flow_action *actions,
struct netlink_ext_ack *extack)
{
- if (nic->flags & OTX2_FLAG_INTF_DOWN) {
+ if (otx2_test_flag(nic, OTX2_FLAG_INTF_DOWN)) {
NL_SET_ERR_MSG_MOD(extack, "Interface not initialized");
return -EINVAL;
}
@@ -223,7 +223,7 @@ static int otx2_tc_egress_matchall_install(struct otx2_nic *nic,
if (err)
return err;
- if (nic->flags & OTX2_FLAG_TC_MATCHALL_EGRESS_ENABLED) {
+ if (otx2_test_flag(nic, OTX2_FLAG_TC_MATCHALL_EGRESS_ENABLED)) {
NL_SET_ERR_MSG_MOD(extack,
"Only one Egress MATCHALL ratelimiter can be offloaded");
return -ENOMEM;
@@ -244,7 +244,7 @@ static int otx2_tc_egress_matchall_install(struct otx2_nic *nic,
otx2_convert_rate(entry->police.rate_bytes_ps));
if (err)
return err;
- nic->flags |= OTX2_FLAG_TC_MATCHALL_EGRESS_ENABLED;
+ otx2_set_flag(nic, OTX2_FLAG_TC_MATCHALL_EGRESS_ENABLED);
break;
default:
NL_SET_ERR_MSG_MOD(extack,
@@ -261,13 +261,13 @@ static int otx2_tc_egress_matchall_delete(struct otx2_nic *nic,
struct netlink_ext_ack *extack = cls->common.extack;
int err;
- if (nic->flags & OTX2_FLAG_INTF_DOWN) {
+ if (otx2_test_flag(nic, OTX2_FLAG_INTF_DOWN)) {
NL_SET_ERR_MSG_MOD(extack, "Interface not initialized");
return -EINVAL;
}
err = otx2_set_matchall_egress_rate(nic, 0, 0);
- nic->flags &= ~OTX2_FLAG_TC_MATCHALL_EGRESS_ENABLED;
+ otx2_clear_flag(nic, OTX2_FLAG_TC_MATCHALL_EGRESS_ENABLED);
return err;
}
@@ -505,7 +505,7 @@ static int otx2_tc_parse_actions(struct otx2_nic *nic,
mark = act->mark;
req->match_id = mark & OTX2_RX_MATCH_ID_MASK;
req->op = NIX_RX_ACTION_DEFAULT;
- nic->flags |= OTX2_FLAG_TC_MARK_ENABLED;
+ otx2_set_flag(nic, OTX2_FLAG_TC_MARK_ENABLED);
refcount_inc(&nic->flow_cfg->mark_flows);
break;
@@ -942,7 +942,7 @@ static void otx2_destroy_tc_flow_list(struct otx2_nic *pfvf)
struct otx2_flow_config *flow_cfg = pfvf->flow_cfg;
struct otx2_tc_flow *iter, *tmp;
- if (!(pfvf->flags & OTX2_FLAG_MCAM_ENTRIES_ALLOC))
+ if (!otx2_test_flag(pfvf, OTX2_FLAG_MCAM_ENTRIES_ALLOC))
return;
list_for_each_entry_safe(iter, tmp, &flow_cfg->flow_list_tc, list) {
@@ -1195,12 +1195,12 @@ static int otx2_tc_del_flow(struct otx2_nic *nic,
/* Disable TC MARK flag if they are no rules with skbedit mark action */
if (flow_node->req.match_id)
if (!refcount_dec_and_test(&flow_cfg->mark_flows))
- nic->flags &= ~OTX2_FLAG_TC_MARK_ENABLED;
+ otx2_clear_flag(nic, OTX2_FLAG_TC_MARK_ENABLED);
if (flow_node->is_act_police) {
__clear_bit(flow_node->rq, &nic->rq_bmap);
- if (nic->flags & OTX2_FLAG_INTF_DOWN)
+ if (otx2_test_flag(nic, OTX2_FLAG_INTF_DOWN))
goto free_mcam_flow;
mutex_lock(&nic->mbox.lock);
@@ -1246,10 +1246,10 @@ static int otx2_tc_add_flow(struct otx2_nic *nic,
struct npc_install_flow_req *req, dummy;
int rc, err, entry;
- if (!(nic->flags & OTX2_FLAG_TC_FLOWER_SUPPORT))
+ if (!otx2_test_flag(nic, OTX2_FLAG_TC_FLOWER_SUPPORT))
return -ENOMEM;
- if (nic->flags & OTX2_FLAG_INTF_DOWN) {
+ if (otx2_test_flag(nic, OTX2_FLAG_INTF_DOWN)) {
NL_SET_ERR_MSG_MOD(extack, "Interface not initialized");
return -EINVAL;
}
@@ -1444,7 +1444,7 @@ static int otx2_tc_ingress_matchall_install(struct otx2_nic *nic,
if (err)
return err;
- if (nic->flags & OTX2_FLAG_TC_MATCHALL_INGRESS_ENABLED) {
+ if (otx2_test_flag(nic, OTX2_FLAG_TC_MATCHALL_INGRESS_ENABLED)) {
NL_SET_ERR_MSG_MOD(extack,
"Only one ingress MATCHALL ratelimitter can be offloaded");
return -ENOMEM;
@@ -1469,7 +1469,7 @@ static int otx2_tc_ingress_matchall_install(struct otx2_nic *nic,
err = cn10k_set_matchall_ipolicer_rate(nic, entry->police.burst, rate);
if (err)
return err;
- nic->flags |= OTX2_FLAG_TC_MATCHALL_INGRESS_ENABLED;
+ otx2_set_flag(nic, OTX2_FLAG_TC_MATCHALL_INGRESS_ENABLED);
break;
default:
NL_SET_ERR_MSG_MOD(extack,
@@ -1486,13 +1486,13 @@ static int otx2_tc_ingress_matchall_delete(struct otx2_nic *nic,
struct netlink_ext_ack *extack = cls->common.extack;
int err;
- if (nic->flags & OTX2_FLAG_INTF_DOWN) {
+ if (otx2_test_flag(nic, OTX2_FLAG_INTF_DOWN)) {
NL_SET_ERR_MSG_MOD(extack, "Interface not initialized");
return -EINVAL;
}
err = cn10k_free_matchall_ipolicer(nic);
- nic->flags &= ~OTX2_FLAG_TC_MATCHALL_INGRESS_ENABLED;
+ otx2_clear_flag(nic, OTX2_FLAG_TC_MATCHALL_INGRESS_ENABLED);
return err;
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c
index 8d2d607bc92f..f65ba44db60b 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c
@@ -171,7 +171,7 @@ static void otx2_set_rxtstamp(struct otx2_nic *pfvf,
u64 timestamp, tsns;
int err;
- if (!(pfvf->flags & OTX2_FLAG_RX_TSTAMP_ENABLED))
+ if (!otx2_test_flag(pfvf, OTX2_FLAG_RX_TSTAMP_ENABLED))
return;
timestamp = pfvf->ptp->convert_rx_ptp_tstmp(*(u64 *)data);
@@ -374,13 +374,13 @@ static void otx2_rcv_pkt_handler(struct otx2_nic *pfvf,
}
otx2_set_rxhash(pfvf, cqe, skb);
- if (!(pfvf->flags & OTX2_FLAG_REP_MODE_ENABLED)) {
+ if (!otx2_test_flag(pfvf, OTX2_FLAG_REP_MODE_ENABLED)) {
skb_record_rx_queue(skb, cq->cq_idx);
if (pfvf->netdev->features & NETIF_F_RXCSUM)
skb->ip_summed = CHECKSUM_UNNECESSARY;
}
- if (pfvf->flags & OTX2_FLAG_TC_MARK_ENABLED)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_TC_MARK_ENABLED))
skb->mark = parse->match_id;
skb_mark_for_recycle(skb);
@@ -513,7 +513,7 @@ static int otx2_tx_napi_handler(struct otx2_nic *pfvf,
((u64)cq->cq_idx << 32) | processed_cqe);
#if IS_ENABLED(CONFIG_RVU_ESWITCH)
- if (pfvf->flags & OTX2_FLAG_REP_MODE_ENABLED)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_REP_MODE_ENABLED))
ndev = pfvf->reps[qidx]->netdev;
else
#endif
@@ -526,7 +526,7 @@ static int otx2_tx_napi_handler(struct otx2_nic *pfvf,
if (qidx >= pfvf->hw.tx_queues)
qidx -= pfvf->hw.xdp_queues;
- if (pfvf->flags & OTX2_FLAG_REP_MODE_ENABLED)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_REP_MODE_ENABLED))
qidx = 0;
txq = netdev_get_tx_queue(ndev, qidx);
netdev_tx_completed_queue(txq, tx_pkts, tx_bytes);
@@ -599,11 +599,11 @@ int otx2_napi_handler(struct napi_struct *napi, int budget)
if (workdone < budget && napi_complete_done(napi, workdone)) {
/* If interface is going down, don't re-enable IRQ */
- if (pfvf->flags & OTX2_FLAG_INTF_DOWN)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_INTF_DOWN))
return workdone;
/* Adjust irq coalese using net_dim */
- if (pfvf->flags & OTX2_FLAG_ADPTV_INT_COAL_ENABLED)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_ADPTV_INT_COAL_ENABLED))
otx2_adjust_adaptive_coalese(pfvf, cq_poll);
if (likely(cq))
@@ -1137,7 +1137,7 @@ static void otx2_set_txtstamp(struct otx2_nic *pfvf, struct sk_buff *skb,
if (unlikely(!skb_shinfo(skb)->gso_size &&
(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP))) {
- if (unlikely(pfvf->flags & OTX2_FLAG_PTP_ONESTEP_SYNC &&
+ if (unlikely(otx2_test_flag(pfvf, OTX2_FLAG_PTP_ONESTEP_SYNC) &&
otx2_ptp_is_sync(skb, &ptp_offset, &udp_csum_crt))) {
origin_tstamp = (struct ptpv2_tstamp *)
((u8 *)skb->data + ptp_offset +
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c
index f7765e19d78a..5f7915231ca3 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c
@@ -610,7 +610,7 @@ static int otx2vf_probe(struct pci_dev *pdev, const struct pci_device_id *id)
vf->dev = dev;
vf->iommu_domain = iommu_get_domain_for_dev(dev);
- vf->flags |= OTX2_FLAG_INTF_DOWN;
+ otx2_set_flag(vf, OTX2_FLAG_INTF_DOWN);
hw = &vf->hw;
hw->pdev = vf->pdev;
hw->rx_queues = qcount;
@@ -824,10 +824,10 @@ static void otx2vf_remove(struct pci_dev *pdev)
vf = netdev_priv(netdev);
/* Disable 802.3x pause frames */
- if (vf->flags & OTX2_FLAG_RX_PAUSE_ENABLED ||
- (vf->flags & OTX2_FLAG_TX_PAUSE_ENABLED)) {
- vf->flags &= ~OTX2_FLAG_RX_PAUSE_ENABLED;
- vf->flags &= ~OTX2_FLAG_TX_PAUSE_ENABLED;
+ if (otx2_test_flag(vf, OTX2_FLAG_RX_PAUSE_ENABLED) ||
+ otx2_test_flag(vf, OTX2_FLAG_TX_PAUSE_ENABLED)) {
+ otx2_clear_flag(vf, OTX2_FLAG_RX_PAUSE_ENABLED);
+ otx2_clear_flag(vf, OTX2_FLAG_TX_PAUSE_ENABLED);
otx2_config_pause_frm(vf);
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.c
index 0e8a6a6486c4..7808588a0234 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.c
@@ -96,7 +96,7 @@ static void otx2_clean_up_rq(struct otx2_nic *pfvf, int qidx)
u64 iova;
/* If the DOWN flag is set SQs are already freed */
- if (pfvf->flags & OTX2_FLAG_INTF_DOWN)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_INTF_DOWN))
return;
cq = &qset->cq[qidx];
@@ -172,7 +172,7 @@ int otx2_xsk_wakeup(struct net_device *dev, u32 queue_id, u32 flags)
struct otx2_cq_poll *cq_poll = NULL;
struct otx2_qset *qset = &pf->qset;
- if (pf->flags & OTX2_FLAG_INTF_DOWN)
+ if (otx2_test_flag(pf, OTX2_FLAG_INTF_DOWN))
return -ENETDOWN;
if (queue_id >= pf->hw.rx_queues || queue_id >= pf->hw.tx_queues)
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/qos_sq.c b/drivers/net/ethernet/marvell/octeontx2/nic/qos_sq.c
index 2872adabc830..5f09e2960144 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/qos_sq.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/qos_sq.c
@@ -238,7 +238,7 @@ int otx2_qos_enable_sq(struct otx2_nic *pfvf, int qidx)
struct otx2_hw *hw = &pfvf->hw;
int pool_id, sq_idx, err;
- if (pfvf->flags & OTX2_FLAG_INTF_DOWN)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_INTF_DOWN))
return -EPERM;
sq_idx = hw->non_qos_queues + qidx;
@@ -288,7 +288,7 @@ void otx2_qos_disable_sq(struct otx2_nic *pfvf, int qidx)
sq_idx = hw->non_qos_queues + qidx;
/* If the DOWN flag is set SQs are already freed */
- if (pfvf->flags & OTX2_FLAG_INTF_DOWN)
+ if (otx2_test_flag(pfvf, OTX2_FLAG_INTF_DOWN))
return;
sq = &pfvf->qset.sq[sq_idx];
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/rep.c b/drivers/net/ethernet/marvell/octeontx2/nic/rep.c
index 0f5d5642d3f7..7df82c22cc12 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/rep.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/rep.c
@@ -93,9 +93,9 @@ static int rvu_rep_mcam_flow_init(struct rep_dev *rep)
rep->flow_cfg->max_flows = allocated;
if (allocated) {
- rep->flags |= OTX2_FLAG_MCAM_ENTRIES_ALLOC;
- rep->flags |= OTX2_FLAG_NTUPLE_SUPPORT;
- rep->flags |= OTX2_FLAG_TC_FLOWER_SUPPORT;
+ set_bit(OTX2_FLAG_MCAM_ENTRIES_ALLOC, &rep->flags);
+ set_bit(OTX2_FLAG_NTUPLE_SUPPORT, &rep->flags);
+ set_bit(OTX2_FLAG_TC_FLOWER_SUPPORT, &rep->flags);
}
INIT_LIST_HEAD(&rep->flow_cfg->flow_list);
@@ -109,14 +109,14 @@ static int rvu_rep_setup_tc_cb(enum tc_setup_type type,
struct rep_dev *rep = cb_priv;
struct otx2_nic *priv = rep->mdev;
- if (!(rep->flags & RVU_REP_VF_INITIALIZED))
+ if (!test_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags))
return -EINVAL;
- if (!(rep->flags & OTX2_FLAG_TC_FLOWER_SUPPORT))
+ if (!test_bit(OTX2_FLAG_TC_FLOWER_SUPPORT, &rep->flags))
rvu_rep_mcam_flow_init(rep);
priv->netdev = rep->netdev;
- priv->flags = rep->flags;
+ otx2_sync_flags_from_rep(priv, &rep->flags);
priv->pcifunc = rep->pcifunc;
priv->flow_cfg = rep->flow_cfg;
@@ -303,9 +303,9 @@ static void rvu_rep_state_evt_handler(struct otx2_nic *priv,
rep_id = rvu_rep_get_repid(priv, info->pcifunc);
rep = priv->reps[rep_id];
if (info->evt_data.vf_state)
- rep->flags |= RVU_REP_VF_INITIALIZED;
+ set_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags);
else
- rep->flags &= ~RVU_REP_VF_INITIALIZED;
+ clear_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags);
}
int rvu_event_up_notify(struct otx2_nic *pf, struct rep_event *info)
@@ -382,7 +382,7 @@ static void rvu_rep_get_stats64(struct net_device *dev,
{
struct rep_dev *rep = netdev_priv(dev);
- if (!(rep->flags & RVU_REP_VF_INITIALIZED))
+ if (!test_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags))
return;
stats->rx_packets = rep->stats.rx_frames;
@@ -453,7 +453,7 @@ static int rvu_rep_open(struct net_device *dev)
struct otx2_nic *priv = rep->mdev;
struct rep_event evt = {0};
- if (!(rep->flags & RVU_REP_VF_INITIALIZED))
+ if (!test_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags))
return 0;
netif_carrier_on(dev);
@@ -472,7 +472,7 @@ static int rvu_rep_stop(struct net_device *dev)
struct otx2_nic *priv = rep->mdev;
struct rep_event evt = {0};
- if (!(rep->flags & RVU_REP_VF_INITIALIZED))
+ if (!test_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags))
return 0;
netif_carrier_off(dev);
@@ -547,7 +547,7 @@ static int rvu_rep_napi_init(struct otx2_nic *priv,
otx2_write64(priv, NIX_LF_CINTX_INT(qidx), BIT_ULL(0));
otx2_write64(priv, NIX_LF_CINTX_ENA_W1S(qidx), BIT_ULL(0));
}
- priv->flags &= ~OTX2_FLAG_INTF_DOWN;
+ otx2_clear_flag(priv, OTX2_FLAG_INTF_DOWN);
return 0;
err_free_cints:
@@ -632,7 +632,7 @@ void rvu_rep_destroy(struct otx2_nic *priv)
int rep_id;
rvu_eswitch_config(priv, false);
- priv->flags |= OTX2_FLAG_INTF_DOWN;
+ otx2_set_flag(priv, OTX2_FLAG_INTF_DOWN);
rvu_rep_free_cq_rsrc(priv);
for (rep_id = 0; rep_id < priv->rep_cnt; rep_id++) {
rep = priv->reps[rep_id];
@@ -801,8 +801,8 @@ static int rvu_rep_probe(struct pci_dev *pdev, const struct pci_device_id *id)
pci_set_drvdata(pdev, priv);
priv->pdev = pdev;
priv->dev = dev;
- priv->flags |= OTX2_FLAG_INTF_DOWN;
- priv->flags |= OTX2_FLAG_REP_MODE_ENABLED;
+ otx2_set_flag(priv, OTX2_FLAG_INTF_DOWN);
+ otx2_set_flag(priv, OTX2_FLAG_REP_MODE_ENABLED);
hw = &priv->hw;
hw->pdev = pdev;
@@ -845,7 +845,7 @@ static void rvu_rep_remove(struct pci_dev *pdev)
struct otx2_nic *priv = pci_get_drvdata(pdev);
otx2_unregister_dl(priv);
- if (!(priv->flags & OTX2_FLAG_INTF_DOWN))
+ if (!otx2_test_flag(priv, OTX2_FLAG_INTF_DOWN))
rvu_rep_destroy(priv);
otx2_detach_resources(&priv->mbox);
if (priv->hw.lmt_info)
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/rep.h b/drivers/net/ethernet/marvell/octeontx2/nic/rep.h
index 5bc9e2c7d800..45707c434d89 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/rep.h
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/rep.h
@@ -37,8 +37,7 @@ struct rep_dev {
struct delayed_work stats_wrk;
struct devlink_port dl_port;
struct otx2_flow_config *flow_cfg;
-#define RVU_REP_VF_INITIALIZED BIT_ULL(0)
- u64 flags;
+ unsigned long flags;
u16 rep_id;
u16 pcifunc;
u8 mac[ETH_ALEN];
--
2.43.0
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH v18 net-next 2/2] octeontx2: add mqprio bandwidth offload for NIX TX schedulers
2026-09-29 2:29 [PATCH v18 net-next 0/2] octeontx2: mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
2026-09-29 2:29 ` [PATCH v18 net-next 1/2] octeontx2: use atomic bitops for PF/VF and rep flags Ratheesh Kannoth
@ 2026-09-29 2:29 ` Ratheesh Kannoth
2026-10-02 8:31 ` netdev-bot+sashiko
2026-10-02 1:22 ` [PATCH v18 net-next 0/2] octeontx2: " Jakub Kicinski
2026-10-02 9:37 ` David Laight
3 siblings, 1 reply; 7+ messages in thread
From: Ratheesh Kannoth @ 2026-09-29 2:29 UTC (permalink / raw)
To: bpf, linux-kernel, netdev
Cc: andrew+netdev, ast, daniel, davem, edumazet, hawk,
john.fastabend, kuba, pabeni, sdf, sgoutham, Ratheesh Kannoth
Add TC_SETUP_QDISC_MQPRIO offload with TC_MQPRIO_SHAPER_BW_RATE on PF and
VF netdevs. Program per-queue MDQ CIR/PIR via the NIX TX scheduler mailbox;
when active, allocate one SMQ per non-QoS queue under TL4[0].
The TX scheduler cannot be reprogrammed live, so add/replace/delete and
failed setup bounce the netdev through ndo_stop()/ndo_open(). Cache rates
in software and restore them from otx2_mqprio_up() on open, failing closed
on error.
Stage tc replace in mq_offload_snap snapshots committed on TC_ROOT_GRAFT or
replaced-qdisc teardown.
Require a running interface with CIR+PIR support. Reject offload with PFC,
XDP, HTB, SDP representors, per-TC rates on multi-queue classes, and
ethtool channel changes while active.
Signed-off-by: Ratheesh Kannoth <rkannoth@marvell.com>
---
.../marvell/octeontx2/nic/otx2_common.c | 143 ++-
.../marvell/octeontx2/nic/otx2_common.h | 33 +
.../marvell/octeontx2/nic/otx2_dcbnl.c | 6 +
.../marvell/octeontx2/nic/otx2_ethtool.c | 15 +
.../ethernet/marvell/octeontx2/nic/otx2_pf.c | 34 +-
.../ethernet/marvell/octeontx2/nic/otx2_tc.c | 855 ++++++++++++++++++
.../ethernet/marvell/octeontx2/nic/otx2_vf.c | 28 +-
.../net/ethernet/marvell/octeontx2/nic/qos.c | 11 +
8 files changed, 1104 insertions(+), 21 deletions(-)
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
index b421cb75e44b..511efb0dcc75 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
@@ -615,6 +615,139 @@ void otx2_get_mac_from_af(struct net_device *netdev)
}
EXPORT_SYMBOL(otx2_get_mac_from_af);
+static int
+otx2_nix_tmq_reg_write(struct otx2_nic *pfvf, int cnt,
+ u64 reg_addr[MAX_REGS_PER_MBOX_MSG],
+ u64 reg_val[MAX_REGS_PER_MBOX_MSG])
+{
+ struct mbox *mbox = &pfvf->mbox;
+ struct nix_txschq_config *req;
+ int i, err;
+
+ mutex_lock(&mbox->lock);
+ req = otx2_mbox_alloc_msg_nix_txschq_cfg(mbox);
+ if (!req) {
+ mutex_unlock(&mbox->lock);
+ return -ENOMEM;
+ }
+
+ req->lvl = NIX_TXSCH_LVL_MDQ;
+ req->num_regs = cnt;
+
+ for (i = 0; i < cnt; i++) {
+ req->reg[i] = reg_addr[i];
+ req->regval[i] = reg_val[i];
+ }
+
+ err = otx2_sync_mbox_msg(mbox);
+ mutex_unlock(&mbox->lock);
+
+ return err;
+}
+
+int otx2_nix_tm_clear_queue_shaper(struct otx2_nic *pfvf)
+{
+ u64 reg_addr[MAX_REGS_PER_MBOX_MSG];
+ u64 reg_val[MAX_REGS_PER_MBOX_MSG];
+ int err, smq, i, cnt = 0;
+
+ for (i = 0; i < pfvf->hw.txschq_cnt[NIX_TXSCH_LVL_SMQ]; i++) {
+ smq = pfvf->hw.txschq_list[NIX_TXSCH_LVL_SMQ][i];
+
+ reg_addr[cnt] = NIX_AF_MDQX_PIR(smq);
+ reg_val[cnt] = 0;
+ cnt++;
+
+ reg_addr[cnt] = NIX_AF_MDQX_CIR(smq);
+ reg_val[cnt] = 0;
+ cnt++;
+
+ if (cnt < MAX_REGS_PER_MBOX_MSG - 1)
+ continue;
+
+ err = otx2_nix_tmq_reg_write(pfvf, cnt,
+ reg_addr, reg_val);
+ if (err)
+ goto fail;
+ cnt = 0;
+ }
+
+ if (cnt) {
+ err = otx2_nix_tmq_reg_write(pfvf, cnt,
+ reg_addr, reg_val);
+ if (err)
+ goto fail;
+ }
+
+ return 0;
+fail:
+ return err;
+}
+
+int otx2_nix_tm_set_queue_shaper(struct otx2_nic *pfvf,
+ int txq, u64 minrate, u64 maxrate)
+{
+ struct mbox *mbox = &pfvf->mbox;
+ struct nix_txschq_config *req;
+ int err, smq, n = 0;
+ u64 reg_addr[2];
+ u64 reg_val[2];
+ u64 rate;
+
+ if (!maxrate && !minrate) {
+ smq = otx2_get_smq_idx(pfvf, txq);
+ reg_addr[0] = NIX_AF_MDQX_PIR(smq);
+ reg_val[0] = 0;
+ reg_addr[1] = NIX_AF_MDQX_CIR(smq);
+ reg_val[1] = 0;
+ return otx2_nix_tmq_reg_write(pfvf, 2, reg_addr, reg_val);
+ }
+
+ smq = otx2_get_smq_idx(pfvf, txq);
+
+ mutex_lock(&mbox->lock);
+ req = otx2_mbox_alloc_msg_nix_txschq_cfg(mbox);
+ if (!req) {
+ mutex_unlock(&mbox->lock);
+ return -ENOMEM;
+ }
+
+ req->lvl = NIX_TXSCH_LVL_MDQ;
+
+ /* MQPRIO exposes only min/max rate, not burst. Use the same 65536
+ * byte default as the HTB shaper path.
+ *
+ * mqprio setup restarts the netdev (otx2_mqprio_restart_netdev);
+ * ndo_open() may restore cached MDQ shapers via otx2_mqprio_up(), and
+ * setup clears every MDQ shaper before applying the new mapping.
+ * Program both PIR and CIR on every update so omitted rates are
+ * applied explicitly rather than relying on stale hardware state.
+ */
+ req->reg[n] = NIX_AF_MDQX_PIR(smq);
+ if (maxrate) {
+ rate = otx2_convert_rate(maxrate);
+ req->regval[n] = otx2_get_txschq_rate_regval(pfvf, rate, 65536);
+ } else {
+ req->regval[n] = 0;
+ }
+ n++;
+
+ /* CIR+PIR support is required and checked at mqprio setup. */
+ req->reg[n] = NIX_AF_MDQX_CIR(smq);
+ if (minrate) {
+ rate = otx2_convert_rate(minrate);
+ req->regval[n] = otx2_get_txschq_rate_regval(pfvf, rate, 65536);
+ } else {
+ req->regval[n] = 0;
+ }
+ n++;
+ req->num_regs = n;
+
+ err = otx2_sync_mbox_msg(mbox);
+ mutex_unlock(&mbox->lock);
+ return err;
+}
+
int otx2_txschq_config(struct otx2_nic *pfvf, int lvl, int prio, bool txschq_for_pfc)
{
u16 (*schq_list)[MAX_TXSCHQ_PER_FUNC];
@@ -651,7 +784,11 @@ int otx2_txschq_config(struct otx2_nic *pfvf, int lvl, int prio, bool txschq_for
(u64)hw->smq_link_type);
req->num_regs++;
/* MDQ config */
- parent = schq_list[NIX_TXSCH_LVL_TL4][prio];
+ if (pfvf->mqprio.rate_limit)
+ parent = schq_list[NIX_TXSCH_LVL_TL4][0];
+ else
+ parent = schq_list[NIX_TXSCH_LVL_TL4][prio];
+
req->reg[1] = NIX_AF_MDQX_PARENT(schq);
req->regval[1] = parent << 16;
req->num_regs++;
@@ -779,6 +916,9 @@ int otx2_txsch_alloc(struct otx2_nic *pfvf)
req->schq[NIX_TXSCH_LVL_TL4] = chan_cnt;
}
+ if (pfvf->mqprio.rate_limit)
+ req->schq[NIX_TXSCH_LVL_SMQ] = pfvf->hw.non_qos_queues;
+
rc = otx2_sync_mbox_msg(&pfvf->mbox);
if (rc)
return rc;
@@ -844,6 +984,7 @@ void otx2_txschq_stop(struct otx2_nic *pfvf)
/* Clear the txschq list */
for (lvl = 0; lvl < NIX_TXSCH_LVL_CNT; lvl++) {
+ pfvf->hw.txschq_cnt[lvl] = 0;
for (schq = 0; schq < MAX_TXSCHQ_PER_FUNC; schq++)
pfvf->hw.txschq_list[lvl][schq] = 0;
}
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
index b146ca4be5db..c0ed8b827f27 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
@@ -17,6 +17,7 @@
#include <linux/soc/marvell/silicons.h>
#include <linux/soc/marvell/octeontx2/asm.h>
#include <net/macsec.h>
+#include <uapi/linux/pkt_sched.h>
#include <net/pkt_cls.h>
#include <net/devlink.h>
#include <linux/time64.h>
@@ -517,6 +518,26 @@ enum otx2_flag_bits {
BIT(OTX2_FLAG_TC_FLOWER_SUPPORT) | \
BIT(OTX2_FLAG_REP_VF_INITIALIZED))
+struct mq_offload_snap {
+ u64 min_rate[TC_QOPT_MAX_QUEUE];
+ u64 max_rate[TC_QOPT_MAX_QUEUE];
+ u32 flags;
+ __u8 num_tc;
+ __u16 count[TC_QOPT_MAX_QUEUE];
+ __u16 offset[TC_QOPT_MAX_QUEUE];
+ __u8 prio_tc_map[TC_QOPT_BITMASK + 1];
+};
+
+struct otx2_mqprio {
+ u32 flags;
+ u64 *min_rate;
+ u64 *max_rate;
+ bool rate_limit;
+ bool replace_setup_done;
+ bool replace_graft_done;
+ struct work_struct netdev_tc_work;
+};
+
struct otx2_nic {
void __iomem *reg_base;
struct net_device *netdev;
@@ -528,6 +549,10 @@ struct otx2_nic {
unsigned long flags;
u64 *cq_op_addr;
+ struct otx2_mqprio mqprio;
+ struct mq_offload_snap *cur_mq_snap;
+ struct mq_offload_snap *old_mq_snap;
+
struct bpf_prog *xdp_prog;
struct otx2_qset qset;
struct otx2_hw hw;
@@ -1036,6 +1061,8 @@ static inline u16 otx2_get_smq_idx(struct otx2_nic *pfvf, u16 qidx)
if (qidx >= pfvf->hw.non_qos_queues) {
smq = pfvf->qos.qid_to_sqmap[qidx - pfvf->hw.non_qos_queues];
} else {
+ if (!pfvf->hw.txschq_cnt[NIX_TXSCH_LVL_SMQ])
+ return 0;
idx = qidx % pfvf->hw.txschq_cnt[NIX_TXSCH_LVL_SMQ];
smq = pfvf->hw.txschq_list[NIX_TXSCH_LVL_SMQ][idx];
}
@@ -1214,6 +1241,7 @@ int otx2_mcam_entry_init(struct otx2_nic *pfvf);
/* tc support */
int otx2_init_tc(struct otx2_nic *nic);
+void otx2_shutdown_tc_mqprio(struct otx2_nic *nic);
void otx2_shutdown_tc(struct otx2_nic *nic);
int otx2_setup_tc(struct net_device *netdev, enum tc_setup_type type,
void *type_data);
@@ -1285,6 +1313,11 @@ dma_addr_t otx2_dma_map_skb_frag(struct otx2_nic *pfvf,
struct sk_buff *skb, int seg, int *len);
void otx2_dma_unmap_skb_frags(struct otx2_nic *pfvf, struct sg_list *sg);
int otx2_read_free_sqe(struct otx2_nic *pfvf, u16 qidx);
+int otx2_nix_tm_set_queue_shaper(struct otx2_nic *pfvf, int txq,
+ u64 minrate, u64 maxrate);
+int otx2_nix_tm_clear_queue_shaper(struct otx2_nic *pfvf);
+int otx2_mqprio_down(struct otx2_nic *pfvf);
+int otx2_mqprio_up(struct otx2_nic *pfvf);
void otx2_queue_vf_work(struct mbox *mw, struct workqueue_struct *mbox_wq,
int first, int mdevs, u64 intr);
int otx2_del_mcam_flow_entry(struct otx2_nic *nic, u16 entry,
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_dcbnl.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_dcbnl.c
index 91d346d114af..b7bd08129fb6 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_dcbnl.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_dcbnl.c
@@ -413,6 +413,12 @@ static int otx2_dcbnl_ieee_setpfc(struct net_device *dev, struct ieee_pfc *pfc)
u8 old_pfc_en;
int err;
+ if (pfvf->mqprio.rate_limit && pfc->pfc_en) {
+ netdev_err(dev,
+ "PFC: cannot enable while mqprio bandwidth offload is active\n");
+ return -EOPNOTSUPP;
+ }
+
old_pfc_en = pfvf->pfc_en;
pfvf->pfc_en = pfc->pfc_en;
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
index 4fe473d9ea0d..836582631127 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
@@ -287,6 +287,21 @@ static int otx2_set_channels(struct net_device *dev,
return -EINVAL;
}
+ if (pfvf->mqprio.rate_limit &&
+ (channel->tx_count != pfvf->hw.tx_queues ||
+ channel->rx_count != pfvf->hw.rx_queues)) {
+ netdev_info(dev,
+ "Not permitted to change channel count while MQ prio is active\n");
+ return -EINVAL;
+ }
+
+ if ((pfvf->old_mq_snap || pfvf->cur_mq_snap) &&
+ channel->tx_count < pfvf->hw.tx_queues) {
+ netdev_err(dev,
+ "Cannot reduce TX queues after mqprio bandwidth offload was configured\n");
+ return -EINVAL;
+ }
+
if (if_up)
dev->netdev_ops->ndo_stop(dev);
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
index 32582b6347ea..eba1a26593c5 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
@@ -2007,6 +2007,15 @@ int otx2_open(struct net_device *netdev)
if (err)
goto err_free_mem;
+ /* Fail closed: abort open if cached mqprio shapers cannot be restored. */
+ err = otx2_mqprio_up(pf);
+ if (err) {
+ netdev_err(pf->netdev,
+ "mqprio: failed to restore shapers during open: %d\n",
+ err);
+ goto err_free_hw;
+ }
+
/* Register NAPI handler */
for (qidx = 0; qidx < pf->hw.cint_cnt; qidx++) {
cq_poll = &qset->napi[qidx];
@@ -2205,6 +2214,7 @@ int otx2_open(struct net_device *netdev)
free_irq(vec, pf);
err_disable_napi:
otx2_disable_napi(pf);
+err_free_hw:
otx2_free_hw_resources(pf);
err_free_mem:
otx2_free_queue_mem(qset);
@@ -2280,6 +2290,7 @@ int otx2_stop(struct net_device *netdev)
for (qidx = 0; qidx < netdev->num_tx_queues; qidx++)
netdev_tx_reset_queue(netdev_get_tx_queue(netdev, qidx));
+ synchronize_net();
otx2_free_queue_mem(qset);
/* Do not clear RQ/SQ ringsize settings */
memset_startat(qset, 0, sqe_cnt);
@@ -2923,6 +2934,12 @@ static int otx2_xdp_setup(struct otx2_nic *pf, struct bpf_prog *prog)
bool if_up = netif_running(pf->netdev);
struct bpf_prog *old_prog;
+ if (prog && pf->mqprio.rate_limit) {
+ netdev_err(dev,
+ "XDP: cannot attach while mqprio bandwidth offload is active\n");
+ return -EOPNOTSUPP;
+ }
+
if (prog && dev->mtu > MAX_XDP_MTU) {
netdev_warn(dev, "Jumbo frames not yet supported with XDP\n");
return -EOPNOTSUPP;
@@ -3368,10 +3385,14 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id)
if (err)
goto err_mcs_free;
+ err = otx2_init_tc(pf);
+ if (err)
+ goto err_ipsec_clean;
+
err = register_netdev(netdev);
if (err) {
dev_err(dev, "Failed to register netdevice\n");
- goto err_ipsec_clean;
+ goto err_shutdown_tc;
}
err = otx2_wq_init(pf);
@@ -3380,10 +3401,6 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id)
otx2_set_ethtool_ops(netdev);
- err = otx2_init_tc(pf);
- if (err)
- goto err_mcam_flow_del;
-
err = otx2_register_dl(pf);
if (err)
goto err_mcam_flow_del;
@@ -3420,11 +3437,13 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id)
otx2_sriov_vfcfg_cleanup(pf);
err_pf_sriov_init:
otx2_unregister_dl(pf);
- otx2_shutdown_tc(pf);
err_mcam_flow_del:
otx2_mcam_flow_del(pf);
err_unreg_netdev:
+ otx2_shutdown_tc_mqprio(pf);
unregister_netdev(netdev);
+err_shutdown_tc:
+ otx2_shutdown_tc(pf);
err_ipsec_clean:
cn10k_ipsec_clean(pf);
err_mcs_free:
@@ -3622,7 +3641,9 @@ static void otx2_remove(struct pci_dev *pdev)
otx2_cgx_config_linkevents(pf, false);
otx2_unregister_dl(pf);
+ otx2_shutdown_tc_mqprio(pf);
unregister_netdev(netdev);
+ otx2_shutdown_tc(pf);
cn10k_ipsec_clean(pf);
cn10k_mcs_free(pf);
otx2_sriov_disable(pf->pdev);
@@ -3632,7 +3653,6 @@ static void otx2_remove(struct pci_dev *pdev)
otx2_ptp_destroy(pf);
otx2_mcam_flow_del(pf);
- otx2_shutdown_tc(pf);
otx2_shutdown_qos(pf);
otx2_ndc_sync(pf);
otx2_detach_resources(&pf->mbox);
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
index ddb46b580c3b..4bb5b49d4dc7 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
@@ -6,6 +6,8 @@
*/
#include <linux/netdevice.h>
+#include <linux/rtnetlink.h>
+#include <linux/string.h>
#include <linux/etherdevice.h>
#include <linux/inetdevice.h>
#include <linux/rhashtable.h>
@@ -16,6 +18,7 @@
#include <net/tc_act/tc_mirred.h>
#include <net/tc_act/tc_vlan.h>
#include <net/ipv6.h>
+#include <net/pkt_sched.h>
#include "cn10k.h"
#include "otx2_common.h"
@@ -31,6 +34,19 @@
#define MCAST_INVALID_GRP (-1U)
#define RATE_MANTISSA_BITS 8
+/* Min per-queue egress shaping rate the NIX TLX encoder supports (2 Mbps). */
+#define OTX2_MQPRIO_MIN_RATE_BYTES_PS 250000ULL
+
+static u64 otx2_mqprio_max_rate_bytes_ps(void)
+{
+ u64 max_mbps;
+
+ /* NIX TLX maximum rate (Mbps), not the burst bucket cap. */
+ max_mbps = 2ULL * ((256ULL + MAX_RATE_MANTISSA) << MAX_RATE_EXPONENT) /
+ 256ULL;
+
+ return div_u64(max_mbps * 1000000ULL, 8ULL);
+}
static void otx2_get_egress_burst_cfg(struct otx2_nic *nic, u32 burst,
u32 *burst_exp, u32 *burst_mantissa)
@@ -61,6 +77,9 @@ static void otx2_get_egress_burst_cfg(struct otx2_nic *nic, u32 burst,
*burst_mantissa = tmp / (1ULL << (*burst_exp - 7));
}
} else {
+ /* burst 0: largest encodable burst (CN10K_MAX_BURST_SIZE on
+ * CN10K), not a minimal burst.
+ */
*burst_exp = MAX_BURST_EXPONENT;
*burst_mantissa = max_mantissa;
}
@@ -1600,14 +1619,841 @@ static int otx2_setup_tc_block(struct net_device *netdev,
nic, nic, ingress);
}
+/* Free the per-queue min/max rate caches. */
+static void otx2_mqprio_free_cache(struct otx2_nic *pfvf)
+{
+ devm_kfree(pfvf->dev, pfvf->mqprio.min_rate);
+ devm_kfree(pfvf->dev, pfvf->mqprio.max_rate);
+ pfvf->mqprio.min_rate = NULL;
+ pfvf->mqprio.max_rate = NULL;
+ pfvf->mqprio.flags = 0;
+}
+
+static int otx2_mqprio_alloc_cache(struct otx2_nic *pfvf, bool replacing)
+{
+ u16 num_txq = pfvf->hw.non_qos_queues;
+ u64 *min_rate, *max_rate;
+
+ if (replacing && pfvf->mqprio.min_rate && pfvf->mqprio.max_rate) {
+ memset(pfvf->mqprio.min_rate, 0,
+ num_txq * sizeof(*pfvf->mqprio.min_rate));
+ memset(pfvf->mqprio.max_rate, 0,
+ num_txq * sizeof(*pfvf->mqprio.max_rate));
+ pfvf->mqprio.flags = 0;
+ return 0;
+ }
+
+ min_rate = devm_kcalloc(pfvf->dev, num_txq, sizeof(*min_rate), GFP_KERNEL);
+ max_rate = devm_kcalloc(pfvf->dev, num_txq, sizeof(*max_rate), GFP_KERNEL);
+ if (!min_rate || !max_rate) {
+ devm_kfree(pfvf->dev, min_rate);
+ devm_kfree(pfvf->dev, max_rate);
+ return -ENOMEM;
+ }
+
+ otx2_mqprio_free_cache(pfvf);
+ pfvf->mqprio.min_rate = min_rate;
+ pfvf->mqprio.max_rate = max_rate;
+
+ return 0;
+}
+
+static void otx2_mqprio_snap_free(struct otx2_nic *pfvf,
+ struct mq_offload_snap **snap)
+{
+ if (!*snap)
+ return;
+
+ devm_kfree(pfvf->dev, *snap);
+ *snap = NULL;
+}
+
+static int otx2_mqprio_snap_copy(struct otx2_nic *pfvf,
+ struct mq_offload_snap **dst,
+ const struct tc_mqprio_qopt_offload *mqprio)
+{
+ const struct tc_mqprio_qopt *qopt = &mqprio->qopt;
+ struct mq_offload_snap *snap;
+ int tc;
+
+ if (!*dst) {
+ snap = devm_kzalloc(pfvf->dev, sizeof(*snap), GFP_KERNEL);
+ if (!snap)
+ return -ENOMEM;
+ *dst = snap;
+ } else {
+ snap = *dst;
+ }
+
+ snap->num_tc = qopt->num_tc;
+ snap->flags = mqprio->flags;
+ for (tc = 0; tc < TC_QOPT_MAX_QUEUE; tc++) {
+ snap->count[tc] = qopt->count[tc];
+ snap->offset[tc] = qopt->offset[tc];
+ snap->min_rate[tc] = 0;
+ snap->max_rate[tc] = 0;
+ }
+
+ for (tc = 0; tc < qopt->num_tc; tc++) {
+ if (mqprio->flags & TC_MQPRIO_F_MIN_RATE)
+ snap->min_rate[tc] = mqprio->min_rate[tc];
+ if (mqprio->flags & TC_MQPRIO_F_MAX_RATE)
+ snap->max_rate[tc] = mqprio->max_rate[tc];
+ }
+ memcpy(snap->prio_tc_map, qopt->prio_tc_map, sizeof(snap->prio_tc_map));
+
+ return 0;
+}
+
+static int otx2_mqprio_stage_cur(struct otx2_nic *pfvf,
+ const struct tc_mqprio_qopt_offload *mqprio)
+{
+ return otx2_mqprio_snap_copy(pfvf, &pfvf->cur_mq_snap, mqprio);
+}
+
+static void otx2_mqprio_snap_commit(struct otx2_nic *pfvf)
+{
+ otx2_mqprio_snap_free(pfvf, &pfvf->old_mq_snap);
+ pfvf->old_mq_snap = pfvf->cur_mq_snap;
+ pfvf->cur_mq_snap = NULL;
+}
+
+static void otx2_mqprio_clear_replace_state(struct otx2_nic *pfvf)
+{
+ pfvf->mqprio.replace_setup_done = false;
+ pfvf->mqprio.replace_graft_done = false;
+}
+
+static bool otx2_mqprio_mdq_allocated(struct otx2_nic *pfvf)
+{
+ return pfvf->hw.txschq_cnt[NIX_TXSCH_LVL_MDQ] != 0;
+}
+
+static int otx2_mqprio_restart_netdev(struct net_device *netdev, bool rate_limit);
+
+static void otx2_mqprio_apply_snap_netdev(struct net_device *netdev,
+ const struct mq_offload_snap *snap)
+{
+ int tc;
+
+ if (!snap)
+ return;
+
+ netdev_set_num_tc(netdev, snap->num_tc);
+ for (tc = 0; tc < snap->num_tc; tc++)
+ netdev_set_tc_queue(netdev, tc, snap->count[tc],
+ snap->offset[tc]);
+ for (tc = 0; tc < TC_QOPT_BITMASK + 1; tc++)
+ netdev_set_prio_tc_map(netdev, tc, snap->prio_tc_map[tc]);
+}
+
+static void otx2_mqprio_netdev_tc_work(struct work_struct *work)
+{
+ struct otx2_mqprio *mqprio = container_of(work, struct otx2_mqprio,
+ netdev_tc_work);
+ struct otx2_nic *pfvf = container_of(mqprio, struct otx2_nic, mqprio);
+
+ rtnl_lock();
+ if (pfvf->mqprio.rate_limit && pfvf->old_mq_snap)
+ otx2_mqprio_apply_snap_netdev(pfvf->netdev, pfvf->old_mq_snap);
+ rtnl_unlock();
+}
+
+static void otx2_mqprio_defer_netdev_tc_restore(struct otx2_nic *pfvf)
+{
+ schedule_work(&pfvf->mqprio.netdev_tc_work);
+}
+
+static int otx2_mqprio_restore_old(struct otx2_nic *pfvf)
+{
+ struct mq_offload_snap *snap = pfvf->old_mq_snap;
+ struct net_device *netdev = pfvf->netdev;
+ u16 num_txq = pfvf->hw.non_qos_queues;
+ int tc, txq, err;
+
+ if (!snap)
+ return 0;
+
+ err = otx2_mqprio_alloc_cache(pfvf, false);
+ if (err)
+ return err;
+
+ memset(pfvf->mqprio.min_rate, 0, num_txq * sizeof(*pfvf->mqprio.min_rate));
+ memset(pfvf->mqprio.max_rate, 0, num_txq * sizeof(*pfvf->mqprio.max_rate));
+ pfvf->mqprio.flags = snap->flags;
+
+ for (tc = 0; tc < snap->num_tc; tc++) {
+ u64 min_rate = snap->min_rate[tc];
+ u64 max_rate = snap->max_rate[tc];
+
+ for (txq = snap->offset[tc];
+ txq < snap->offset[tc] + snap->count[tc]; txq++) {
+ pfvf->mqprio.min_rate[txq] = min_rate;
+ pfvf->mqprio.max_rate[txq] = max_rate;
+ }
+ }
+
+ otx2_mqprio_apply_snap_netdev(netdev, snap);
+
+ if (otx2_mqprio_mdq_allocated(pfvf)) {
+ err = otx2_nix_tm_clear_queue_shaper(pfvf);
+ if (err)
+ return err;
+ }
+
+ /* Rebuild the TX scheduler via netdev restart when running; otx2_mqprio_up()
+ * alone is insufficient after a failed replace that already bounced the
+ * interface. If open failed, TX schedulers were freed; defer shaper restore
+ * to the next successful ndo_open() via otx2_mqprio_up().
+ */
+ pfvf->mqprio.rate_limit = true;
+
+ if (netif_running(netdev)) {
+ err = otx2_mqprio_restart_netdev(netdev, true);
+ if (err)
+ return err;
+ } else if (pfvf->hw.txschq_cnt[NIX_TXSCH_LVL_SMQ]) {
+ err = otx2_mqprio_up(pfvf);
+ if (err)
+ return err;
+ }
+
+ otx2_mqprio_snap_free(pfvf, &pfvf->cur_mq_snap);
+
+ return 0;
+}
+
+static void otx2_mqprio_snap_destroy(struct otx2_nic *pfvf)
+{
+ otx2_mqprio_snap_free(pfvf, &pfvf->cur_mq_snap);
+ otx2_mqprio_snap_free(pfvf, &pfvf->old_mq_snap);
+}
+
+/* Offloaded mqprio replaced by software mqprio installs netdev TC layout in
+ * mqprio_init() before the old offload instance is destroyed during graft.
+ */
+static bool otx2_mqprio_keep_netdev_tc(struct otx2_nic *pfvf)
+{
+ struct Qdisc *qdisc = rtnl_dereference(pfvf->netdev->qdisc);
+
+ return qdisc && qdisc->ops && !strcmp(qdisc->ops->id, "mqprio");
+}
+
+static void otx2_mqprio_clear_sw(struct otx2_nic *pfvf)
+{
+ struct net_device *netdev = pfvf->netdev;
+
+ pfvf->mqprio.rate_limit = false;
+ otx2_mqprio_clear_replace_state(pfvf);
+ if (!otx2_mqprio_keep_netdev_tc(pfvf))
+ netdev_set_num_tc(netdev, 0);
+ otx2_mqprio_free_cache(pfvf);
+}
+
+/* Tear down mqprio bandwidth offload: clear per-queue shapers,
+ * mqprio_rate_limit, netdev TC mappings, and the cached rates. Called on
+ * explicit mqprio teardown (tc qdisc del) and error cleanup, not on
+ * routine netdev stop/open cycles where the offload stays active.
+ */
+int otx2_mqprio_down(struct otx2_nic *pfvf)
+{
+ int err = 0;
+
+ if (!pfvf->mqprio.rate_limit)
+ return 0;
+
+ if (netif_running(pfvf->netdev) &&
+ otx2_mqprio_mdq_allocated(pfvf))
+ err = otx2_nix_tm_clear_queue_shaper(pfvf);
+
+ if (err) {
+ netdev_warn(pfvf->netdev,
+ "mqprio: failed to clear hardware shapers: %d; keeping offload state\n",
+ err);
+ return err;
+ }
+
+ otx2_mqprio_clear_sw(pfvf);
+
+ return 0;
+}
+
+/* Restore cached mqprio MDQ shapers after ndo_open() reprograms the TX
+ * scheduler. Called from otx2_open() when bandwidth offload stays active
+ * across admin down/up or an mqprio netdev bounce.
+ *
+ * Returns an error if any shaper mailbox operation fails. otx2_open()
+ * fail-closes on that error: it aborts open and leaves the interface down
+ * rather than running with partial or missing bandwidth limits.
+ */
+int otx2_mqprio_up(struct otx2_nic *pfvf)
+{
+ struct net_device *netdev = pfvf->netdev;
+ int txq, err;
+
+ if (!pfvf->mqprio.rate_limit)
+ return 0;
+
+ if (!pfvf->mqprio.min_rate || !pfvf->mqprio.max_rate)
+ return 0;
+
+ for (txq = 0; txq < pfvf->hw.non_qos_queues; txq++) {
+ u64 min_rate = 0, max_rate = 0;
+
+ if (pfvf->mqprio.flags & TC_MQPRIO_F_MIN_RATE)
+ min_rate = pfvf->mqprio.min_rate[txq];
+ if (pfvf->mqprio.flags & TC_MQPRIO_F_MAX_RATE)
+ max_rate = pfvf->mqprio.max_rate[txq];
+
+ if (!min_rate && !max_rate)
+ continue;
+
+ err = otx2_nix_tm_set_queue_shaper(pfvf, txq, min_rate,
+ max_rate);
+ if (err) {
+ netdev_err(netdev,
+ "mqprio: failed to restore shaper for txq %d: %d\n",
+ txq, err);
+ if (otx2_mqprio_mdq_allocated(pfvf) &&
+ otx2_nix_tm_clear_queue_shaper(pfvf))
+ netdev_warn(netdev,
+ "mqprio: failed to clear shapers after partial restore\n");
+ return err;
+ }
+ }
+
+ return 0;
+}
+
+/* Restart the netdev to reprogram the TX scheduler hierarchy for mqprio
+ * bandwidth offload. Both mqprio add and delete (when offload was active)
+ * take this path via ndo_stop()/ndo_open() so VF-specific open logic (e.g.
+ * LBK carrier on) runs correctly.
+ *
+ * Intentional behaviour: this full stop/open cycle drops in-flight traffic
+ * (carrier off, IRQ/NAPI teardown, queue drain). The NIX TX scheduler must
+ * be reallocated (e.g. one SMQ per non-QoS queue) and cannot be reprogrammed
+ * live today, so a netdev bounce is required on every mqprio add, replace,
+ * delete, and rollback. Users see a brief connectivity blip; this is not a
+ * bug to "fix" without implementing the live-reprogramming path noted below.
+ * If open fails, OTX2_FLAG_INTF_DOWN is set before netif_close() so
+ * otx2_stop() returns early and does not touch resources already torn
+ * down by the open error path.
+ *
+ * Do not call dev_deactivate()/dev_activate() here. On replace,
+ * qdisc_graft() already deactivates qdiscs around offload teardown;
+ * dev_activate() from ndo_setup_tc() would republish qdiscs before graft
+ * completes and race __qdisc_run() on the old root qdisc. After
+ * ndo_open(), carrier and TX queues are restored via otx2_handle_link_event()
+ * when link is up, same as otx2_change_mtu(), not via dev_activate().
+ *
+ * Clear __LINK_STATE_START before ndo_stop() so netif_running() is false
+ * for the duration of the bounce.
+ */
+static int otx2_mqprio_restart_netdev(struct net_device *netdev, bool rate_limit)
+{
+ struct otx2_nic *pfvf = netdev_priv(netdev);
+ const struct net_device_ops *ops = netdev->netdev_ops;
+ bool running = netif_running(netdev);
+ int err;
+
+ /* TODO: Explore live TX scheduler reprogramming to avoid a full
+ * ndo_stop()/ndo_open() bounce on every mqprio change.
+ */
+ netdev_dbg(netdev,
+ "mqprio: restarting interface to reprogram TX scheduler; in-flight traffic will be dropped\n");
+
+ if (running) {
+ clear_bit(__LINK_STATE_START, &netdev->state);
+ smp_mb__after_atomic(); /* Commit netif_running(). */
+ }
+
+ err = ops->ndo_stop(netdev);
+ if (err) {
+ if (running)
+ set_bit(__LINK_STATE_START, &netdev->state);
+ return err;
+ }
+
+ /* Set before ndo_open() so otx2_txsch_alloc() widens SMQ allocation.
+ * On teardown, drop mqprio software state so ndo_open() does not
+ * re-apply bandwidth limits via otx2_mqprio_up() after the kernel
+ * removed the qdisc.
+ */
+ if (rate_limit)
+ pfvf->mqprio.rate_limit = true;
+ else
+ otx2_mqprio_clear_sw(pfvf);
+
+ err = ops->ndo_open(netdev);
+ if (!err && running) {
+ set_bit(__LINK_STATE_START, &netdev->state);
+ } else if (err) {
+ netdev_err(netdev,
+ "Failed to restart device after mqprio change: %d\n",
+ err);
+ /* ndo_open() already freed the TX schedulers on failure while
+ * netif_running() may still be true; drop mqprio software state
+ * only instead of sending shaper clears to freed queues.
+ */
+ otx2_mqprio_clear_sw(pfvf);
+ /* ndo_open() rolls back on failure; mark the interface down so
+ * otx2_stop() returns early when netif_close() runs. Caller
+ * holds RTNL; dev_close() would deadlock.
+ */
+ otx2_set_flag(pfvf, OTX2_FLAG_INTF_DOWN);
+ /* visible to otx2_stop() on other cpus */
+ smp_wmb();
+ netif_close(netdev);
+ }
+
+ return err;
+}
+
+static int otx2_mqprio_validate_tc_rate(struct net_device *netdev,
+ struct netlink_ext_ack *extack,
+ u64 rate, u32 qcount, int tc,
+ const char *name)
+{
+ if (!rate)
+ return 0;
+
+ if (qcount <= 1)
+ return 0;
+
+ /* TODO: per-TC TL4 shapers or equal per-queue MDQ split for multi-queue TC rates. */
+ netdev_err(netdev,
+ "mqprio: %s rate for tc %d not supported with %u queues\n",
+ name, tc, qcount);
+ NL_SET_ERR_MSG_FMT_MOD(extack,
+ "mqprio: %s rate for tc %d not supported with %u queues",
+ name, tc, qcount);
+ return -EOPNOTSUPP;
+}
+
+static int otx2_mqprio_validate_txqs(struct net_device *netdev,
+ struct netlink_ext_ack *extack,
+ struct tc_mqprio_qopt *qopt)
+{
+ struct otx2_nic *pfvf = netdev_priv(netdev);
+ u16 num_txq = pfvf->hw.non_qos_queues;
+ int tc, txq;
+
+ if (qopt->num_tc > num_txq) {
+ netdev_err(netdev, "Number of TCs (%u) exceeds hw queues %u\n",
+ qopt->num_tc, num_txq);
+ NL_SET_ERR_MSG_FMT_MOD(extack,
+ "Number of TCs (%u) exceeds hw queues %u",
+ qopt->num_tc, num_txq);
+ return -EINVAL;
+ }
+
+ if (num_txq > MAX_TXSCHQ_PER_FUNC) {
+ netdev_err(netdev,
+ "Number of queues (%u) exceeds max scheduler queues %u\n",
+ num_txq, MAX_TXSCHQ_PER_FUNC);
+ NL_SET_ERR_MSG_FMT_MOD(extack,
+ "Number of queues (%u) exceeds max scheduler queues %u",
+ num_txq, MAX_TXSCHQ_PER_FUNC);
+ return -EINVAL;
+ }
+
+ for (tc = 0; tc < qopt->num_tc; tc++) {
+ u32 qcount = qopt->count[tc];
+
+ for (txq = qopt->offset[tc];
+ txq < qopt->offset[tc] + qcount; txq++) {
+ if (txq >= num_txq) {
+ netdev_err(netdev,
+ "mqprio: txq %d exceeds offload queue count %u\n",
+ txq, num_txq);
+ NL_SET_ERR_MSG_FMT_MOD(extack,
+ "mqprio: txq %d exceeds offload queue count %u",
+ txq, num_txq);
+ return -EINVAL;
+ }
+ }
+ }
+
+ return 0;
+}
+
+static bool otx2_mqprio_rate_valid(struct otx2_nic *pfvf, u64 rate_bytes_ps)
+{
+ u64 mbps;
+
+ if (!rate_bytes_ps)
+ return true;
+
+ if (rate_bytes_ps < OTX2_MQPRIO_MIN_RATE_BYTES_PS)
+ return false;
+
+ if (rate_bytes_ps > otx2_mqprio_max_rate_bytes_ps())
+ return false;
+
+ if (rate_bytes_ps > div_u64(U64_MAX, 8))
+ return false;
+
+ mbps = otx2_convert_rate(rate_bytes_ps);
+ return ilog2(mbps / 2) <= MAX_RATE_EXPONENT;
+}
+
+static bool otx2_tc_can_offload(struct net_device *netdev)
+{
+ return !!(netdev->hw_features & NETIF_F_HW_TC);
+}
+
+static int otx2_teardown_tc_mqprio(struct otx2_nic *pfvf,
+ struct tc_mqprio_qopt_offload *mqprio)
+{
+ struct tc_mqprio_qopt *qopt = &mqprio->qopt;
+ bool had_mqprio = pfvf->mqprio.rate_limit;
+ struct net_device *netdev = pfvf->netdev;
+ bool if_up = netif_running(netdev);
+ int err;
+
+ if (!otx2_tc_can_offload(netdev))
+ return -EOPNOTSUPP;
+
+ qopt->hw = 0;
+
+ /* tc qdisc replace runs setup on the new mqprio before destroying the
+ * old one. replace_setup_done and TC_ROOT_GRAFT distinguish stale
+ * old-instance teardown from graft failure after setup.
+ * qdisc_offload_graft_helper() skips TC_ROOT_GRAFT unless hw-tc-offload
+ * is enabled in netdev->features; commit without graft in that case.
+ */
+ if (pfvf->mqprio.replace_setup_done && pfvf->cur_mq_snap) {
+ err = 0;
+ if (pfvf->mqprio.replace_graft_done || !tc_can_offload(netdev))
+ otx2_mqprio_snap_commit(pfvf);
+ else
+ err = otx2_mqprio_restore_old(pfvf);
+ otx2_mqprio_clear_replace_state(pfvf);
+ return err;
+ }
+
+ /* Skip the netdev restart when mqprio offload was not active. */
+ if (!had_mqprio)
+ return 0;
+
+ if (if_up) {
+ err = otx2_mqprio_down(pfvf);
+ if (err)
+ return err;
+
+ return otx2_mqprio_restart_netdev(netdev, false);
+ }
+
+ /* ndo_stop() already freed the TX scheduler TL nodes; drop software
+ * state only.
+ */
+ otx2_mqprio_clear_sw(pfvf);
+ return 0;
+}
+
+static int otx2_setup_tc_mqprio(struct net_device *netdev,
+ struct tc_mqprio_qopt_offload *mqprio)
+{
+ struct netlink_ext_ack *extack = mqprio->extack;
+ struct otx2_nic *pfvf = netdev_priv(netdev);
+ struct tc_mqprio_qopt *qopt = &mqprio->qopt;
+ bool replacing = pfvf->mqprio.rate_limit;
+ bool if_up = netif_running(netdev);
+ int tc, txq, err, i;
+
+ if (!otx2_tc_can_offload(netdev))
+ return -EOPNOTSUPP;
+
+ if (!qopt->hw)
+ return otx2_teardown_tc_mqprio(pfvf, mqprio);
+
+ if (!if_up) {
+ netdev_err(netdev, "mqprio: setup requires interface UP\n");
+ NL_SET_ERR_MSG_MOD(extack, "mqprio: setup requires interface UP");
+ err = -EOPNOTSUPP;
+ goto fail_validate;
+ }
+
+ if (!replacing && otx2_mqprio_keep_netdev_tc(pfvf)) {
+ netdev_err(netdev,
+ "mqprio: delete existing mqprio before re-enabling hw/sw offload\n");
+ NL_SET_ERR_MSG_MOD(extack,
+ "mqprio: delete existing mqprio before re-enabling hw/sw offload\n");
+ return -EOPNOTSUPP;
+ }
+
+ if (mqprio->shaper != TC_MQPRIO_SHAPER_BW_RATE) {
+ netdev_err(netdev, "Unsupported mqprio shaper %#x\n", mqprio->shaper);
+ NL_SET_ERR_MSG_FMT_MOD(extack, "Unsupported mqprio shaper %#x",
+ mqprio->shaper);
+ err = -EOPNOTSUPP;
+ goto fail_validate;
+ }
+
+ if (!test_bit(QOS_CIR_PIR_SUPPORT, &pfvf->hw.cap_flag)) {
+ netdev_err(netdev,
+ "mqprio: bandwidth offload requires CIR+PIR support\n");
+ NL_SET_ERR_MSG_MOD(extack,
+ "mqprio: bandwidth offload requires CIR+PIR support");
+ err = -EOPNOTSUPP;
+ goto fail_validate;
+ }
+
+ if (is_otx2_sdp_rep(pfvf->pdev)) {
+ netdev_err(netdev, "mqprio: bandwidth offload not supported on SDP rep\n");
+ NL_SET_ERR_MSG_MOD(extack,
+ "mqprio: bandwidth offload not supported on SDP rep");
+ err = -EOPNOTSUPP;
+ goto fail_validate;
+ }
+
+ if (pfvf->pfc_en) {
+ netdev_err(netdev,
+ "mqprio: cannot enable offload while PFC is enabled\n");
+ NL_SET_ERR_MSG_MOD(extack,
+ "mqprio: cannot enable offload while PFC is enabled");
+ err = -EOPNOTSUPP;
+ goto fail_validate;
+ }
+
+ if (pfvf->xdp_prog) {
+ netdev_err(netdev,
+ "mqprio: cannot enable offload while XDP is active\n");
+ NL_SET_ERR_MSG_MOD(extack,
+ "mqprio: cannot enable offload while XDP is active");
+ err = -EOPNOTSUPP;
+ goto fail_validate;
+ }
+
+ if (!list_empty(&pfvf->qos.qos_tree)) {
+ netdev_err(netdev,
+ "mqprio: cannot enable offload while HTB is active\n");
+ NL_SET_ERR_MSG_MOD(extack,
+ "mqprio: cannot enable offload while HTB is active");
+ err = -EOPNOTSUPP;
+ goto fail_validate;
+ }
+
+ for (tc = 0; tc < qopt->num_tc; tc++) {
+ u64 min_rate = 0, max_rate = 0;
+ u32 qcount = qopt->count[tc];
+
+ if (mqprio->flags & TC_MQPRIO_F_MIN_RATE)
+ min_rate = mqprio->min_rate[tc];
+ if (mqprio->flags & TC_MQPRIO_F_MAX_RATE)
+ max_rate = mqprio->max_rate[tc];
+
+ if (min_rate && max_rate && min_rate > max_rate) {
+ netdev_err(netdev,
+ "min_rate %llu exceeds max_rate %llu for tc %d\n",
+ min_rate, max_rate, tc);
+ NL_SET_ERR_MSG_FMT_MOD(extack,
+ "min_rate %llu exceeds max_rate %llu for tc %d",
+ min_rate, max_rate, tc);
+ err = -EINVAL;
+ goto fail_validate;
+ }
+
+ if (mqprio->flags & TC_MQPRIO_F_MIN_RATE) {
+ err = otx2_mqprio_validate_tc_rate(netdev, extack, min_rate,
+ qcount, tc, "min");
+ if (err)
+ goto fail_validate;
+ }
+
+ if (mqprio->flags & TC_MQPRIO_F_MAX_RATE) {
+ err = otx2_mqprio_validate_tc_rate(netdev, extack, max_rate,
+ qcount, tc, "max");
+ if (err)
+ goto fail_validate;
+ }
+
+ if (mqprio->flags & TC_MQPRIO_F_MIN_RATE &&
+ !otx2_mqprio_rate_valid(pfvf, min_rate)) {
+ netdev_err(netdev,
+ "mqprio: min_rate %llu for tc %d is outside hardware limits\n",
+ min_rate, tc);
+ NL_SET_ERR_MSG_FMT_MOD(extack,
+ "mqprio: min_rate %llu for tc %d is outside hardware limits",
+ min_rate, tc);
+ err = -EINVAL;
+ goto fail_validate;
+ }
+
+ if (mqprio->flags & TC_MQPRIO_F_MAX_RATE &&
+ !otx2_mqprio_rate_valid(pfvf, max_rate)) {
+ netdev_err(netdev,
+ "mqprio: max_rate %llu for tc %d is outside hardware limits\n",
+ max_rate, tc);
+ NL_SET_ERR_MSG_FMT_MOD(extack,
+ "mqprio: max_rate %llu for tc %d is outside hardware limits",
+ max_rate, tc);
+ err = -EINVAL;
+ goto fail_validate;
+ }
+ }
+
+ err = otx2_mqprio_validate_txqs(netdev, extack, qopt);
+ if (err)
+ goto fail_validate;
+
+ err = otx2_mqprio_stage_cur(pfvf, mqprio);
+ if (err)
+ goto fail_validate;
+
+ err = otx2_mqprio_restart_netdev(pfvf->netdev, true);
+ if (err)
+ goto cleanup;
+
+ err = otx2_mqprio_alloc_cache(pfvf, replacing);
+ if (err)
+ goto cleanup;
+
+ /* otx2_mqprio_up() may have restored the previous configuration during
+ * the restart above. Clear every MDQ shaper before applying the new
+ * mapping so queues dropped from the TC layout do not keep stale
+ * limits in hardware.
+ */
+ if (otx2_mqprio_mdq_allocated(pfvf)) {
+ err = otx2_nix_tm_clear_queue_shaper(pfvf);
+ if (err)
+ goto cleanup;
+ }
+
+ pfvf->mqprio.flags = mqprio->flags;
+
+ for (tc = 0; tc < qopt->num_tc; tc++) {
+ u64 min_rate = 0, max_rate = 0;
+ u32 qcount = qopt->count[tc];
+
+ /* Rates omitted from tc mqprio are passed as zero and both MDQ
+ * shaper registers are programmed; see
+ * otx2_nix_tm_set_queue_shaper().
+ * TODO: multi-queue TC rates need per-queue split; see
+ * otx2_mqprio_validate_tc_rate().
+ */
+ if (mqprio->flags & TC_MQPRIO_F_MIN_RATE)
+ min_rate = mqprio->min_rate[tc];
+ if (mqprio->flags & TC_MQPRIO_F_MAX_RATE)
+ max_rate = mqprio->max_rate[tc];
+
+ for (txq = qopt->offset[tc];
+ txq < qopt->offset[tc] + qcount; txq++) {
+ netdev_dbg(netdev,
+ "mqprio: tc %d txq %d min_rate %llu max_rate %llu\n",
+ tc, txq, min_rate, max_rate);
+
+ pfvf->mqprio.min_rate[txq] = min_rate;
+ pfvf->mqprio.max_rate[txq] = max_rate;
+
+ err = otx2_nix_tm_set_queue_shaper(pfvf, txq,
+ min_rate, max_rate);
+ if (err)
+ goto cleanup;
+ }
+ }
+
+ netdev_set_num_tc(netdev, pfvf->cur_mq_snap->num_tc);
+ for (i = 0; i < pfvf->cur_mq_snap->num_tc; i++)
+ netdev_set_tc_queue(netdev, i, pfvf->cur_mq_snap->count[i],
+ qopt->offset[i]);
+
+ qopt->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
+
+ if (replacing) {
+ pfvf->mqprio.replace_setup_done = true;
+ pfvf->mqprio.replace_graft_done = false;
+ } else {
+ otx2_mqprio_snap_commit(pfvf);
+ }
+
+ return 0;
+
+fail_validate:
+ /* Failed replace destroys the new qdisc with hw_offload unset, so
+ * mqprio_destroy() clears netdev TC after we return. Re-apply the
+ * prior layout when validation fails before any hardware change.
+ */
+ if (replacing)
+ otx2_mqprio_defer_netdev_tc_restore(pfvf);
+ return err;
+
+cleanup:
+ qopt->hw = 0;
+ if (replacing) {
+ int restore_err = otx2_mqprio_restore_old(pfvf);
+
+ otx2_mqprio_clear_replace_state(pfvf);
+ if (restore_err) {
+ netdev_err(netdev,
+ "mqprio: replace failed and prior configuration rollback failed: %d\n",
+ restore_err);
+ if (extack)
+ NL_SET_ERR_MSG_FMT_MOD(extack,
+ "mqprio: replace failed and prior configuration rollback failed: %d",
+ restore_err);
+ } else {
+ netdev_err(netdev,
+ "mqprio: replace failed; hardware limits restored, netdev TC layout restore deferred\n");
+ if (extack)
+ NL_SET_ERR_MSG_MOD(extack,
+ "mqprio: replace failed; hardware limits restored, netdev TC layout restore deferred");
+ /* Failed replace destroys the new qdisc with hw_offload
+ * unset, so mqprio_destroy() clears netdev TC after we
+ * return. Re-apply the restored layout once that unwind
+ * finishes.
+ */
+ otx2_mqprio_defer_netdev_tc_restore(pfvf);
+ }
+ return err ? err : -EIO;
+ }
+ otx2_mqprio_snap_free(pfvf, &pfvf->cur_mq_snap);
+ otx2_teardown_tc_mqprio(pfvf, mqprio);
+ return err;
+}
+
+static int otx2_setup_tc_root(struct otx2_nic *pfvf,
+ struct tc_root_qopt_offload *root)
+{
+ switch (root->command) {
+ case TC_ROOT_GRAFT:
+ if (pfvf->mqprio.replace_setup_done)
+ pfvf->mqprio.replace_graft_done = true;
+ return 0;
+ default:
+ return -EOPNOTSUPP;
+ }
+}
+
+static int otx2_setup_tc_query_caps(void *type_data)
+{
+ struct tc_query_caps_base *base = type_data;
+ struct tc_mqprio_caps *caps;
+
+ if (base->type != TC_SETUP_QDISC_MQPRIO)
+ return -EOPNOTSUPP;
+
+ caps = base->caps;
+ caps->validate_queue_counts = true;
+
+ return 0;
+}
+
int otx2_setup_tc(struct net_device *netdev, enum tc_setup_type type,
void *type_data)
{
switch (type) {
+ case TC_QUERY_CAPS:
+ return otx2_setup_tc_query_caps(type_data);
case TC_SETUP_BLOCK:
return otx2_setup_tc_block(netdev, type_data);
case TC_SETUP_QDISC_HTB:
return otx2_setup_tc_htb(netdev, type_data);
+ case TC_SETUP_QDISC_MQPRIO:
+ return otx2_setup_tc_mqprio(netdev, type_data);
+ case TC_SETUP_ROOT_QDISC:
+ return otx2_setup_tc_root(netdev_priv(netdev), type_data);
default:
return -EOPNOTSUPP;
}
@@ -1625,10 +2471,19 @@ int otx2_init_tc(struct otx2_nic *nic)
return -EINVAL;
}
+ INIT_WORK(&nic->mqprio.netdev_tc_work, otx2_mqprio_netdev_tc_work);
+
return 0;
}
EXPORT_SYMBOL(otx2_init_tc);
+void otx2_shutdown_tc_mqprio(struct otx2_nic *nic)
+{
+ cancel_work_sync(&nic->mqprio.netdev_tc_work);
+ otx2_mqprio_snap_destroy(nic);
+}
+EXPORT_SYMBOL(otx2_shutdown_tc_mqprio);
+
void otx2_shutdown_tc(struct otx2_nic *nic)
{
otx2_destroy_tc_flow_list(nic);
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c
index 5f7915231ca3..6ca08b12520a 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_vf.c
@@ -740,25 +740,25 @@ static int otx2vf_probe(struct pci_dev *pdev, const struct pci_device_id *id)
if (err)
goto err_ipsec_clean;
+ err = otx2vf_mcam_flow_init(vf);
+ if (err)
+ goto err_wq_destroy;
+
+ err = otx2_init_tc(vf);
+ if (err)
+ goto err_wq_destroy;
+
err = register_netdev(netdev);
if (err) {
dev_err(dev, "Failed to register netdevice\n");
- goto err_wq_destroy;
+ goto err_shutdown_tc;
}
otx2vf_set_ethtool_ops(netdev);
- err = otx2vf_mcam_flow_init(vf);
- if (err)
- goto err_unreg_netdev;
-
- err = otx2_init_tc(vf);
- if (err)
- goto err_unreg_netdev;
-
err = otx2_register_dl(vf);
if (err)
- goto err_shutdown_tc;
+ goto err_unreg_netdev;
vf->af_xdp_zc_qidx = bitmap_zalloc(qcount, GFP_KERNEL);
if (!vf->af_xdp_zc_qidx) {
@@ -784,10 +784,11 @@ static int otx2vf_probe(struct pci_dev *pdev, const struct pci_device_id *id)
#endif
err_unreg_devlink:
otx2_unregister_dl(vf);
-err_shutdown_tc:
- otx2_shutdown_tc(vf);
err_unreg_netdev:
+ otx2_shutdown_tc_mqprio(vf);
unregister_netdev(netdev);
+err_shutdown_tc:
+ otx2_shutdown_tc(vf);
err_wq_destroy:
cancel_work_sync(&vf->reset_task);
cancel_work_sync(&vf->rx_mode_work);
@@ -840,7 +841,9 @@ static void otx2vf_remove(struct pci_dev *pdev)
#endif
otx2_unregister_dl(vf);
+ otx2_shutdown_tc_mqprio(vf);
unregister_netdev(netdev);
+ otx2_shutdown_tc(vf);
if (vf->otx2_wq) {
cancel_work_sync(&vf->reset_task);
cancel_work_sync(&vf->rx_mode_work);
@@ -849,7 +852,6 @@ static void otx2vf_remove(struct pci_dev *pdev)
cn10k_ipsec_clean(vf);
otx2_ptp_destroy(vf);
otx2_mcam_flow_del(vf);
- otx2_shutdown_tc(vf);
otx2_shutdown_qos(vf);
otx2_detach_resources(&vf->mbox);
otx2vf_disable_mbox_intr(vf);
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/qos.c b/drivers/net/ethernet/marvell/octeontx2/nic/qos.c
index f160b1618efa..9ef55a6db50b 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/qos.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/qos.c
@@ -118,6 +118,9 @@ static void otx2_config_sched_shaping(struct otx2_nic *pfvf,
/* configure PIR */
maxrate = (node->rate > node->ceil) ? node->rate : node->ceil;
+ /* 65536 is the kernel-side default burst when HTB does not supply an
+ * explicit value, not the NIX hardware maximum (CN10K_MAX_BURST_SIZE).
+ */
cfg->regval[*num_regs] =
otx2_get_txschq_rate_regval(pfvf, maxrate, 65536);
(*num_regs)++;
@@ -1088,6 +1091,14 @@ static int otx2_qos_root_add(struct otx2_nic *pfvf, u16 htb_maj_id, u16 htb_defc
"TC_HTB_CREATE: handle=0x%x defcls=0x%x\n",
htb_maj_id, htb_defcls);
+ if (pfvf->mqprio.rate_limit) {
+ netdev_err(pfvf->netdev,
+ "HTB: cannot enable while mqprio bandwidth offload is active\n");
+ NL_SET_ERR_MSG_MOD(extack,
+ "HTB: cannot enable while mqprio bandwidth offload is active");
+ return -EOPNOTSUPP;
+ }
+
root = otx2_qos_alloc_root(pfvf);
if (IS_ERR(root)) {
err = PTR_ERR(root);
--
2.43.0
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v18 net-next 0/2] octeontx2: mqprio bandwidth offload for NIX TX schedulers
2026-09-29 2:29 [PATCH v18 net-next 0/2] octeontx2: mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
2026-09-29 2:29 ` [PATCH v18 net-next 1/2] octeontx2: use atomic bitops for PF/VF and rep flags Ratheesh Kannoth
2026-09-29 2:29 ` [PATCH v18 net-next 2/2] octeontx2: add mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
@ 2026-10-02 1:22 ` Jakub Kicinski
2026-10-02 9:37 ` David Laight
3 siblings, 0 replies; 7+ messages in thread
From: Jakub Kicinski @ 2026-10-02 1:22 UTC (permalink / raw)
To: Ratheesh Kannoth
Cc: bpf, linux-kernel, netdev, andrew+netdev, ast, daniel, davem,
edumazet, hawk, john.fastabend, pabeni, sdf, sgoutham
On Tue, 29 Sep 2026 07:59:13 +0530 Ratheesh Kannoth wrote:
> This series adds hardware offload for channel-mode mqprio with
> TC_MQPRIO_SHAPER_BW_RATE on Marvell octeontx2/cn10k PF and VF RVU
> netdevices. Each non-QoS transmit queue is shaped by programming MDQ CIR/PIR
> on the NIX TX scheduler. When bandwidth offload is active, the driver
> allocates one SMQ per queue, parents every MDQ under TL4[0], and maps each
> traffic class min/max rate to the queue(s) in that class.
>
> The NIX TX scheduler hierarchy cannot be reprogrammed live today, so
> mqprio add, replace, delete, and failed-replace rollback rebuild it by
> bouncing the netdev through ndo_stop()/ndo_open(). That intentionally
> drops in-flight traffic on each change. otx2_mqprio_restart_netdev() clears
> __LINK_STATE_START before ndo_stop() and does not call
> dev_deactivate()/dev_activate(); carrier and TX queues are restored after
> ndo_open() via the normal link-event path when the link is up. Cache the
> active rates and restore MDQ shapers from otx2_mqprio_up() during ndo_open();
> fail closed if restoration fails, leaving ndo_open() unsuccessful and the
> interface administratively down.
>
> Track mqprio configuration in mq_offload_snap snapshots (TC layout and
> rates). On tc qdisc replace, stage the new configuration while keeping
> the previous snapshot for rollback: failed setup restores the old
> snapshot via netdev restart when the interface is running, successful
> graft is recorded through TC_ROOT_GRAFT, and teardown of the replaced
> qdisc instance commits the staged snapshot without tearing down the live
> offload.
>
> Patch 1 converts PF/VF and representor flag access to atomic bitops.
> Patch 2 depends on it for safe OTX2_FLAG_INTF_DOWN and OTX2_FLAG_PORT_UP
> updates on asynchronous mbox paths and during the mqprio netdev bounce.
>
> The driver rejects offload unless the interface is running and the device
> advertises CIR+PIR support. PF and VF RVU netdevices share the same TC
> offload path via ndo_setup_tc / otx2_open(); SDP representors are not
> supported. Per-TC rates are rejected when a traffic class spans more than
> one queue. Concurrent PFC, XDP, SDP rep, or HTB use is blocked, and ethtool
> channel count changes are blocked while mqprio bandwidth offload is active.
coccicheck
drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c:2182:7-74: WARNING avoid newline at end of message in NL_SET_ERR_MSG_MOD
Obviously please still wait for the AI feedback before sending v19
--
pw-bot: cr
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v18 net-next 1/2] octeontx2: use atomic bitops for PF/VF and rep flags
2026-09-29 2:29 ` [PATCH v18 net-next 1/2] octeontx2: use atomic bitops for PF/VF and rep flags Ratheesh Kannoth
@ 2026-10-02 8:31 ` netdev-bot+sashiko
0 siblings, 0 replies; 7+ messages in thread
From: netdev-bot+sashiko @ 2026-10-02 8:31 UTC (permalink / raw)
To: rkannoth
Cc: bpf, linux-kernel, netdev, andrew+netdev, ast, daniel, davem,
edumazet, hawk, john.fastabend, kuba, pabeni, sdf, sgoutham
Thank you for your contribution! Sashiko AI review found 2 potential
issue(s) to consider.
Critical: 0 · High: 1 (3) · Medium: 1 (5) · Low: 0
If you believe none of the issues are legitimate, please reply to each
issue of Medium severity or higher and explain why the AI is wrong.
If any of them are legitimate and you plan to respin, please let patchwork
know by including "pw-bot: cr" as a separate line at the end of your reply
(one such reply per series is enough).
- [High] otx2_sync_flags_from_rep() (otx2_common.h:625-634) updates the
shared rep-PF flags word with a plain, non-atomic read-modify-write:…
- [Medium] The patch quietly fixes a serious representor bug but gives it
neither a Fixes: tag nor a description.
Pre-existing issues:
- [High] rvu_rep_destroy() (rep.c:637-643) calls free_netdev(rep->netdev)
and then kfree(rep->flow_cfg).
- [High] rep->stats_wrk is a delayed_work inside rep_dev, which is
net_device private data.
- [High] rvu_rep_mcam_flow_init() (rep.c:54-91) builds and sends NPC MCAM
alloc mailbox messages (otx2_mbox_alloc_msg_npc_mcam_alloc_entry(),…
- [Medium] otx2_tc_del_flow() (otx2_tc.c:1197-1198) clears
OTX2_FLAG_TC_MARK_ENABLED on every delete of a mark flow, even when…
- [Medium] rvu_rep_setup_tc_cb() (rep.c:115-121) ignores the return value
of rvu_rep_mcam_flow_init() and then publishes rep->flow_cfg to…
- [Medium] rvu_rep_mcam_flow_init() (rep.c:43-52) always overwrites
rep->flow_cfg with a new kzalloc and never frees the previous one.
- [Medium] rvu_rep_mcam_flow_init() allocates rep->flow_cfg with kzalloc
and never calls refcount_set(&flow_cfg->mark_flows, 1).
- [Medium] rvu_rep_state_evt_handler() (rep.c:297-309) uses the result of
rvu_rep_get_repid() as an index into priv->reps[] without checking it,…
--
octeontx2: use atomic bitops for PF/VF and rep flags
This commit changes the otx2_nic and rep_dev flags words from u64 masks
to unsigned long bitmaps, accessed with set_bit/clear_bit/test_bit. It adds
otx2_sync_flags_from_rep(), which merges the representor-owned flags into
the representor PF's flags with a masked WRITE_ONCE. It also moves the
representor VF-initialized flag to bit 21.
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
> index 5850bc1870a1d..b146ca4be5db1 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
[ ... ]
> @@ -593,6 +607,32 @@ struct otx2_nic {
> unsigned long *af_xdp_zc_qidx;
> };
>
> +static inline void otx2_set_flag(struct otx2_nic *nic, unsigned int flag)
> +{
> + set_bit(flag, &nic->flags);
> +}
[ ... ]
> +static inline void otx2_sync_flags_from_rep(struct otx2_nic *dst,
> + unsigned long *src_flags)
> +{
> + unsigned long src = READ_ONCE(*src_flags);
> + unsigned long new_flags;
> +
> + new_flags = (READ_ONCE(dst->flags) & ~OTX2_REP_SYNC_FLAGS_MASK) |
> + (src & OTX2_REP_SYNC_FLAGS_MASK);
> + WRITE_ONCE(dst->flags, new_flags);
> +}
[Severity: High]
Can this read-modify-write lose a concurrent atomic update to the same word?
Every other writer of priv->flags now uses set_bit()/clear_bit(). This one
reads the whole word and writes it back non-atomically.
rvu_rep_setup_tc_cb() calls this under rtnl from the flower block callback.
rvu_rep_destroy() can run from otx2_devlink_eswitch_mode_set() with only the
devlink lock held, while the rep netdevs are still registered:
CPU1 (tc callback, rtnl held)
otx2_sync_flags_from_rep()
READ_ONCE(dst->flags) <- INTF_DOWN clear
CPU2 (devlink eswitch mode legacy)
rvu_rep_destroy()
otx2_set_flag(priv, OTX2_FLAG_INTF_DOWN);
CPU1
WRITE_ONCE(dst->flags, new_flags) <- INTF_DOWN cleared again
After that, otx2_napi_handler() can see INTF_DOWN clear and write
NIX_LF_CINTX_ENA_W1S during rvu_rep_free_cq_rsrc(). Later,
rvu_rep_remove() sees !INTF_DOWN and calls rvu_rep_destroy() a second
time, on the freed priv->reps and the NULLed qset->napi.
The commit message says the helper publishes the rep flags "without
clearing PF-owned bits". To guarantee that, would it need a cmpxchg loop,
or assign_bit() for each bit in OTX2_REP_SYNC_FLAGS_MASK?
[ ... ]
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
> index 039fd47ebf522..ddb46b580c3b6 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
[ ... ]
> @@ -1195,12 +1195,12 @@ static int otx2_tc_del_flow(struct otx2_nic *nic,
> /* Disable TC MARK flag if they are no rules with skbedit mark action */
> if (flow_node->req.match_id)
> if (!refcount_dec_and_test(&flow_cfg->mark_flows))
> - nic->flags &= ~OTX2_FLAG_TC_MARK_ENABLED;
> + otx2_clear_flag(nic, OTX2_FLAG_TC_MARK_ENABLED);
[Severity: Medium]
This is a pre-existing issue, not introduced by this patch, but is this
check inverted?
otx2_mcam_entry_init() starts mark_flows at 1, and otx2_tc_parse_actions()
increments it for every FLOW_ACTION_MARK rule. So refcount_dec_and_test()
never reaches zero here, and the flag is cleared on every mark flow delete.
With two mark rules installed (count 3), deleting one leaves the count at
2 but still clears OTX2_FLAG_TC_MARK_ENABLED. otx2_rcv_pkt_handler() then
stops setting skb->mark for the rule that remains:
if (otx2_test_flag(pfvf, OTX2_FLAG_TC_MARK_ENABLED))
skb->mark = parse->match_id;
[ ... ]
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/rep.c b/drivers/net/ethernet/marvell/octeontx2/nic/rep.c
> index 0f5d5642d3f7a..7df82c22cc12b 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/rep.c
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/rep.c
> @@ -93,9 +93,9 @@ static int rvu_rep_mcam_flow_init(struct rep_dev *rep)
> rep->flow_cfg->max_flows = allocated;
>
> if (allocated) {
> - rep->flags |= OTX2_FLAG_MCAM_ENTRIES_ALLOC;
> - rep->flags |= OTX2_FLAG_NTUPLE_SUPPORT;
> - rep->flags |= OTX2_FLAG_TC_FLOWER_SUPPORT;
> + set_bit(OTX2_FLAG_MCAM_ENTRIES_ALLOC, &rep->flags);
> + set_bit(OTX2_FLAG_NTUPLE_SUPPORT, &rep->flags);
> + set_bit(OTX2_FLAG_TC_FLOWER_SUPPORT, &rep->flags);
> }
>
> INIT_LIST_HEAD(&rep->flow_cfg->flow_list);
[Severity: High]
This isn't a bug introduced by this patch, but where is priv->mbox.lock
taken in rvu_rep_mcam_flow_init()? The function sends the NPC MCAM alloc
messages without the lock, then unlocks at the exit label:
exit:
...
mutex_unlock(&priv->mbox.lock);
Its only caller, rvu_rep_setup_tc_cb(), does not hold the lock either.
The rvu_rep_get_stats() delayed work uses the same mailbox under mbox.lock
and runs without rtnl. Suppose it holds the lock when the first flower rule
is added on a representor. Can both contexts then build messages in the
shared mailbox at the same time, with this mutex_unlock() releasing the
worker's critical section?
Even without that overlap, lockdep would report a bad unlock balance on
the first flower rule added on each representor.
[Severity: Medium]
This is also a pre-existing issue, not introduced by this patch.
rep->flow_cfg is zero-allocated here and mark_flows is never set, unlike
in the PF path in otx2_mcam_entry_init():
refcount_set(&flow_cfg->mark_flows, 1);
When a representor installs a flower rule with FLOW_ACTION_MARK,
otx2_tc_parse_actions() does:
refcount_inc(&nic->flow_cfg->mark_flows);
on a counter that is zero. Does that trigger the "refcount_t: addition on
0; use-after-free" warning and saturate the counter?
With this patch, OTX2_FLAG_TC_MARK_ENABLED is no longer wiped by each
representor tc callback. That makes this refcount the only thing that
clears the flag on the representor PF.
> @@ -109,14 +109,14 @@ static int rvu_rep_setup_tc_cb(enum tc_setup_type type,
> struct rep_dev *rep = cb_priv;
> struct otx2_nic *priv = rep->mdev;
>
> - if (!(rep->flags & RVU_REP_VF_INITIALIZED))
> + if (!test_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags))
> return -EINVAL;
>
> - if (!(rep->flags & OTX2_FLAG_TC_FLOWER_SUPPORT))
> + if (!test_bit(OTX2_FLAG_TC_FLOWER_SUPPORT, &rep->flags))
> rvu_rep_mcam_flow_init(rep);
>
[Severity: Medium]
This is a pre-existing issue, not introduced by this patch. The return
value of rvu_rep_mcam_flow_init() is ignored, and rep->flow_cfg is
published to priv->flow_cfg regardless.
If the kzalloc of flow_cfg fails, rep->flow_cfg is NULL. If the kcalloc of
flow_ent fails, the list heads stay zeroed, because INIT_LIST_HEAD() is
only reached at the end of the function.
FLOW_CLS_REPLACE returns early in otx2_tc_add_flow(), since
TC_FLOWER_SUPPORT is clear. But fl_hw_replace_filter() then calls
fl_hw_destroy_filter(). An ordinary tc filter del and stats queries also
reach this callback. In those cases:
otx2_tc_del_flow() / otx2_tc_get_flow_stats()
otx2_tc_get_entry_by_cookie()
list_for_each_entry(tmp, &flow_cfg->flow_list_tc, list)
Can this dereference the NULL or zeroed flow_cfg?
[Severity: Medium]
This is a pre-existing issue as well, not introduced by this patch.
rvu_rep_mcam_flow_init() always does:
rep->flow_cfg = kzalloc_objs(struct otx2_flow_config, 1);
and never frees the previous flow_cfg. If the MCAM allocation returns zero
entries, or an allocation fails, TC_FLOWER_SUPPORT stays clear. Does every
later tc callback on that representor then leak a flow_cfg and its
flow_ent array?
rvu_rep_destroy() also only does kfree(rep->flow_cfg). That seems to leak
flow_cfg->flow_ent on every switchdev/legacy cycle.
> priv->netdev = rep->netdev;
> - priv->flags = rep->flags;
> + otx2_sync_flags_from_rep(priv, &rep->flags);
[Severity: Medium]
This change fixes a real bug, but the commit message describes it as a
cleanup. Before this patch, the line:
priv->flags = rep->flags;
overwrote the whole flags word of the shared representor PF. rep->flags
only held RVU_REP_VF_INITIALIZED (BIT_ULL(0)) and the MCAM, NTUPLE and
TC_FLOWER bits. So the first tc callback cleared the
OTX2_FLAG_REP_MODE_ENABLED bit that rvu_rep_probe() set. It also set bit 0,
which in the otx2_nic namespace is OTX2_FLAG_RX_TSTAMP_ENABLED.
With REP_MODE_ENABLED gone, otx2_tx_napi_handler() uses pfvf->netdev, which
is now the last representor's single-queue netdev. It also no longer
forces qidx to 0, so netdev_get_tx_queue(ndev, qidx) can index past that
netdev's _tx array. The RX path also starts recording the rx queue and
setting CHECKSUM_UNNECESSARY based on the wrong netdev's features.
The stray RX_TSTAMP bit also lets otx2_set_rxtstamp() dereference
pfvf->ptp, which is never set up on the rep PF.
The commit message only says:
Relocate representor VF initialization to OTX2_FLAG_REP_VF_INITIALIZED
(bit 21).
Could the commit message describe the bug being fixed and add:
Fixes: 6c40ca957fe5 ("octeontx2-pf: Adds TC offload support")
Could the fix also be split out from the mechanical bitops conversion, so
it can be picked up for stable?
The change also affects how long OTX2_FLAG_TC_MARK_ENABLED stays set on
the rep PF. It used to be wiped on every tc callback, and now it persists.
> priv->pcifunc = rep->pcifunc;
> priv->flow_cfg = rep->flow_cfg;
>
> @@ -303,9 +303,9 @@ static void rvu_rep_state_evt_handler(struct otx2_nic *priv,
> rep_id = rvu_rep_get_repid(priv, info->pcifunc);
> rep = priv->reps[rep_id];
> if (info->evt_data.vf_state)
> - rep->flags |= RVU_REP_VF_INITIALIZED;
> + set_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags);
> else
> - rep->flags &= ~RVU_REP_VF_INITIALIZED;
> + clear_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags);
> }
[Severity: Medium]
This isn't a bug introduced by this patch, but rvu_rep_get_repid() returns
-EINVAL for an unknown pcifunc. rep_id is then used to index priv->reps[]
without a check, before set_bit()/clear_bit() on rep->flags.
Separately, rvu_rep_destroy() does kfree(priv->reps) without clearing the
pointer. The AF could send a PFVF_STATE up-message just before
rvu_eswitch_config(priv, false) is processed. Could that message be
handled after the reps are freed, and write into freed memory here?
[ ... ]
> @@ -382,7 +382,7 @@ static void rvu_rep_get_stats64(struct net_device *dev,
> {
> struct rep_dev *rep = netdev_priv(dev);
>
> - if (!(rep->flags & RVU_REP_VF_INITIALIZED))
> + if (!test_bit(OTX2_FLAG_REP_VF_INITIALIZED, &rep->flags))
> return;
>
> stats->rx_packets = rep->stats.rx_frames;
[Severity: High]
This is a pre-existing issue, not introduced by this patch.
rvu_rep_get_stats64() ends with:
schedule_delayed_work(&rep->stats_wrk, msecs_to_jiffies(100));
rep->stats_wrk lives in the net_device private data. Neither
rvu_rep_destroy() nor the exit unwind in rvu_rep_create() calls
cancel_delayed_work_sync() before free_netdev().
The teardown path itself arms the work:
unregister_netdev()
unregister_netdevice_many_notify()
rtmsg_ifinfo_build_skb(RTM_DELLINK)
rtnl_fill_ifinfo()
rtnl_fill_stats()
dev_get_stats()
rvu_rep_get_stats64()
schedule_delayed_work(&rep->stats_wrk, ...)
Does the timer then fire about 100ms after free_netdev(), on freed memory,
with rvu_rep_get_stats() dereferencing rep->mdev and rep->stats?
[ ... ]
> @@ -632,7 +632,7 @@ void rvu_rep_destroy(struct otx2_nic *priv)
> int rep_id;
>
> rvu_eswitch_config(priv, false);
> - priv->flags |= OTX2_FLAG_INTF_DOWN;
> + otx2_set_flag(priv, OTX2_FLAG_INTF_DOWN);
> rvu_rep_free_cq_rsrc(priv);
> for (rep_id = 0; rep_id < priv->rep_cnt; rep_id++) {
> rep = priv->reps[rep_id];
[Severity: High]
This isn't a bug introduced by this patch, but further down this loop
does:
free_netdev(rep->netdev);
kfree(rep->flow_cfg);
rep is the netdev_priv() area of the netdev that rvu_rep_create() allocates
with alloc_etherdev(sizeof(*rep)). After unregister_netdev(),
free_netdev() normally drops the last device reference and frees that
memory.
Is rep->flow_cfg read here after rep has been freed? Freeing flow_cfg
(and flow_cfg->flow_ent) before free_netdev() would avoid that.
[ ... ]
--
Sashiko AI review · https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260929022915.2704627-1-rkannoth%40marvell.com
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v18 net-next 2/2] octeontx2: add mqprio bandwidth offload for NIX TX schedulers
2026-09-29 2:29 ` [PATCH v18 net-next 2/2] octeontx2: add mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
@ 2026-10-02 8:31 ` netdev-bot+sashiko
0 siblings, 0 replies; 7+ messages in thread
From: netdev-bot+sashiko @ 2026-10-02 8:31 UTC (permalink / raw)
To: rkannoth
Cc: bpf, linux-kernel, netdev, andrew+netdev, ast, daniel, davem,
edumazet, hawk, john.fastabend, kuba, pabeni, sdf, sgoutham
Thank you for your contribution! Sashiko AI review found 9 potential
issue(s) to consider.
Critical: 0 · High: 1 · Medium: 6 (1) · Low: 2
If you believe none of the issues are legitimate, please reply to each
issue of Medium severity or higher and explain why the AI is wrong.
If any of them are legitimate and you plan to respin, please let patchwork
know by including "pw-bot: cr" as a separate line at the end of your reply
(one such reply per series is enough).
- [High] otx2_remove() and otx2vf_remove(), and the err_unreg_netdev
probe error labels, call otx2_shutdown_tc_mqprio() before…
- [Medium] Deleting the offload leaves old_mq_snap allocated, and that
stale snapshot permanently blocks TX channel reduction.
- [Medium] Teardown can fail, but the tc core cannot veto it, so the
driver ends up holding offload state for a qdisc that no longer exists.
- [Medium] With default features (NETIF_F_HW_TC is in hw_features but not
in features on both PF and VF, because it is added to hw_features…
- [Medium] When rollback of a failed replace itself fails, the old qdisc
can stay attached with partial or missing hardware limits.
- [Medium] With mqprio.rate_limit set, otx2_txsch_alloc() requests
hw.non_qos_queues SMQs but still only one TL4.
- [Medium] otx2_nix_tm_clear_queue_shaper() is not atomic.
- [Low] In otx2_setup_tc_mqprio(), the extack message for the `!replacing
&& otx2_mqprio_keep_netdev_tc(pfvf)` rejection ends with a newline:…
- [Low] A comment in the ndo_open() failure branch of
otx2_mqprio_restart_netdev() says "ndo_open() already freed the TX…
Pre-existing issues:
- [Medium] otx2_config_nix_queues() allocates pfvf->refill_wrk with
devm_kcalloc() on every open.
--
octeontx2: add mqprio bandwidth offload for NIX TX schedulers
This commit adds TC_SETUP_QDISC_MQPRIO offload with the bw_rate shaper for
octeontx2 PF and VF netdevs, programming per-queue MDQ CIR/PIR through the
AF mailbox. The TX scheduler cannot be reprogrammed live, so every add,
replace and delete bounces the netdev through ndo_stop()/ndo_open(), and
otx2_mqprio_up() restores the cached rates on open. A replace is staged in
mq_offload_snap snapshots, which are committed on TC_ROOT_GRAFT or when the
replaced qdisc is torn down.
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
> index b421cb75e44b5..511efb0dcc752 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c
> @@ -615,6 +615,139 @@ void otx2_get_mac_from_af(struct net_device *netdev)
[ ... ]
> +int otx2_nix_tm_clear_queue_shaper(struct otx2_nic *pfvf)
> +{
> + u64 reg_addr[MAX_REGS_PER_MBOX_MSG];
> + u64 reg_val[MAX_REGS_PER_MBOX_MSG];
> + int err, smq, i, cnt = 0;
> +
> + for (i = 0; i < pfvf->hw.txschq_cnt[NIX_TXSCH_LVL_SMQ]; i++) {
> + smq = pfvf->hw.txschq_list[NIX_TXSCH_LVL_SMQ][i];
> +
> + reg_addr[cnt] = NIX_AF_MDQX_PIR(smq);
> + reg_val[cnt] = 0;
> + cnt++;
> +
> + reg_addr[cnt] = NIX_AF_MDQX_CIR(smq);
> + reg_val[cnt] = 0;
> + cnt++;
> +
> + if (cnt < MAX_REGS_PER_MBOX_MSG - 1)
> + continue;
> +
> + err = otx2_nix_tmq_reg_write(pfvf, cnt,
> + reg_addr, reg_val);
> + if (err)
> + goto fail;
> + cnt = 0;
> + }
[Severity: Medium]
Is it a problem that this clear is not atomic across batches? Each mailbox
message covers 10 MDQs. As soon as a batch is acknowledged, PIR/CIR for
those queues are zero.
With more than 10 non-QoS queues, a later batch can fail in
otx2_mbox_alloc_msg_nix_txschq_cfg() or otx2_sync_mbox_msg(). By then the
earlier queues are already unshaped, and nothing reprograms them.
In that case otx2_mqprio_down() keeps rate_limit and the cached rates:
if (err) {
netdev_warn(pfvf->netdev,
"mqprio: failed to clear hardware shapers: %d; keeping offload state\n",
err);
return err;
}
otx2_mqprio_restore_old() has already written the old rates into the
software cache. It then returns here, before it reaches
otx2_mqprio_restart_netdev():
if (otx2_mqprio_mdq_allocated(pfvf)) {
err = otx2_nix_tm_clear_queue_shaper(pfvf);
if (err)
return err;
}
Could a subset of queues end up enforcing neither the old nor the new
limits while the driver still reports the offload as active?
[ ... ]
> @@ -651,7 +784,11 @@ int otx2_txschq_config(struct otx2_nic *pfvf, int lvl, int prio, bool txschq_for
> (u64)hw->smq_link_type);
> req->num_regs++;
> /* MDQ config */
> - parent = schq_list[NIX_TXSCH_LVL_TL4][prio];
> + if (pfvf->mqprio.rate_limit)
> + parent = schq_list[NIX_TXSCH_LVL_TL4][0];
> + else
> + parent = schq_list[NIX_TXSCH_LVL_TL4][prio];
> +
> req->reg[1] = NIX_AF_MDQX_PARENT(schq);
> req->regval[1] = parent << 16;
> req->num_regs++;
[Severity: Medium]
With rate_limit set, every MDQ gets TL4[0] as its parent. otx2_txsch_alloc()
now requests hw.non_qos_queues SMQs but still gets a single TL4.
NIX_AF_MDQX_SCHEDULE is written as dwrr_val only, so every MDQ has PRIO=0.
Nothing on this path programs NIX_AF_TL4X_TOPOLOGY for TL4[0]. In the NIC
driver, only the TL1 setup and the HTB code in qos.c write TOPOLOGY. On the
AF side, nix_reset_tx_schedule() clears only the PARENT and SCHEDULE
registers:
if (tlx_parent)
rvu_write64(rvu, blkaddr, tlx_parent, 0x0);
if (tlx_schedule)
rvu_write64(rvu, blkaddr, tlx_schedule, 0x0);
The TOPOLOGY clear in otx2_qos_free_hw_schq() is best-effort, and another
user of the same scheduler queue could leave any value behind.
If TL4[0] is handed out with a non-zero RR_PRIO or PRIO_ANCHOR, do the N
priority-0 MDQs still form one DWRR group? Or do they look like N
static-priority children claiming the same slot? Should TL4[0]'s topology
be programmed explicitly when rate_limit is set?
[ ... ]
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
> index 4fe473d9ea0dd..8365826311279 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c
> @@ -287,6 +287,21 @@ static int otx2_set_channels(struct net_device *dev,
> return -EINVAL;
> }
>
> + if (pfvf->mqprio.rate_limit &&
> + (channel->tx_count != pfvf->hw.tx_queues ||
> + channel->rx_count != pfvf->hw.rx_queues)) {
> + netdev_info(dev,
> + "Not permitted to change channel count while MQ prio is active\n");
> + return -EINVAL;
> + }
> +
> + if ((pfvf->old_mq_snap || pfvf->cur_mq_snap) &&
> + channel->tx_count < pfvf->hw.tx_queues) {
> + netdev_err(dev,
> + "Cannot reduce TX queues after mqprio bandwidth offload was configured\n");
> + return -EINVAL;
> + }
> +
[Severity: Medium]
Does this check block TX channel reduction permanently after a single
add/delete cycle?
A successful first add ends in otx2_mqprio_snap_commit(), which leaves
pfvf->old_mq_snap set. A later tc qdisc del then goes through:
otx2_teardown_tc_mqprio()
otx2_mqprio_down()
otx2_mqprio_clear_sw()
otx2_mqprio_restart_netdev(netdev, false)
otx2_mqprio_clear_sw() only calls otx2_mqprio_free_cache(). old_mq_snap and
cur_mq_snap are only freed when a later add commits or when
otx2_shutdown_tc_mqprio() runs on remove.
This condition doesn't look at rate_limit. So ethtool -L dev tx N with a
smaller N keeps returning -EINVAL even though no offload is active or
pending. The commit message only claims:
Reject offload with PFC,
XDP, HTB, SDP representors, per-TC rates on multi-queue classes, and
ethtool channel changes while active.
The commit message also doesn't mention that the new
!replacing && otx2_mqprio_keep_netdev_tc(pfvf) check in
otx2_setup_tc_mqprio() now rejects replacing a software mqprio with an
offloaded one. Could that be documented as well?
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
> index 32582b6347ea4..eba1a26593c57 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
> @@ -2007,6 +2007,15 @@ int otx2_open(struct net_device *netdev)
> if (err)
> goto err_free_mem;
>
> + /* Fail closed: abort open if cached mqprio shapers cannot be restored. */
> + err = otx2_mqprio_up(pf);
> + if (err) {
> + netdev_err(pf->netdev,
> + "mqprio: failed to restore shapers during open: %d\n",
> + err);
> + goto err_free_hw;
> + }
> +
[ ... ]
> @@ -2205,6 +2214,7 @@ int otx2_open(struct net_device *netdev)
> free_irq(vec, pf);
> err_disable_napi:
> otx2_disable_napi(pf);
> +err_free_hw:
> otx2_free_hw_resources(pf);
> err_free_mem:
> otx2_free_queue_mem(qset);
[Severity: Medium]
This is a pre-existing issue and was not introduced by this patch.
otx2_config_nix_queues() allocates pfvf->refill_wrk on every open:
pfvf->refill_wrk = devm_kcalloc(pfvf->dev, pfvf->qset.cq_cnt,
sizeof(struct refill_work), GFP_KERNEL);
None of the otx2_open() error labels free it, and otx2_free_hw_resources()
doesn't release it either. Only otx2_stop() calls
devm_kfree(pf->dev, pf->refill_wrk). After a failed open,
OTX2_FLAG_INTF_DOWN makes a later otx2_stop() return early.
The new err_free_hw branch adds another way into this unwind, and tc can
now trigger stop/open cycles. Does each failed open leave one refill_wrk
array on the devres list until unbind?
[ ... ]
> @@ -3622,7 +3641,9 @@ static void otx2_remove(struct pci_dev *pdev)
> otx2_cgx_config_linkevents(pf, false);
>
> otx2_unregister_dl(pf);
> + otx2_shutdown_tc_mqprio(pf);
> unregister_netdev(netdev);
> + otx2_shutdown_tc(pf);
> cn10k_ipsec_clean(pf);
> cn10k_mcs_free(pf);
> otx2_sriov_disable(pf->pdev);
[Severity: High]
Can this ordering lead to a use-after-free? otx2_shutdown_tc_mqprio() runs
without RTNL while the netdev is still registered, so tc requests can still
reach otx2_setup_tc():
void otx2_shutdown_tc_mqprio(struct otx2_nic *nic)
{
cancel_work_sync(&nic->mqprio.netdev_tc_work);
otx2_mqprio_snap_destroy(nic);
}
Suppose a concurrent tc qdisc replace fails while an offload is active,
either at fail_validate or at cleanup with replacing true. It then calls:
otx2_mqprio_defer_netdev_tc_restore()
schedule_work()
This can happen after cancel_work_sync() has returned. Nothing cancels the
work again before free_netdev(). otx2_mqprio_netdev_tc_work() then does a
container_of() into the freed netdev_priv and reads
pfvf->mqprio.rate_limit, pfvf->old_mq_snap and pfvf->netdev.
There is a second problem. otx2_mqprio_snap_destroy() devm_kfree()s
old_mq_snap and cur_mq_snap with no lock held. Every other user runs under
RTNL: otx2_mqprio_restore_old(), otx2_mqprio_snap_copy(),
otx2_mqprio_snap_commit(), the worker and otx2_set_channels(). For example,
otx2_mqprio_restore_old() could load the pointer just before it is freed:
struct mq_offload_snap *snap = pfvf->old_mq_snap;
otx2vf_remove() uses the same ordering, and so do the err_unreg_netdev
labels in otx2_probe() and otx2vf_probe(). Should the cancel and the
snapshot free happen after unregister_netdev(), once no more ndo_setup_tc()
calls can arrive?
[ ... ]
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
> index ddb46b580c3b6..4bb5b49d4dc70 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
[ ... ]
> @@ -1600,14 +1619,841 @@ static int otx2_setup_tc_block(struct net_device *netdev,
[ ... ]
> + err = ops->ndo_open(netdev);
> + if (!err && running) {
> + set_bit(__LINK_STATE_START, &netdev->state);
> + } else if (err) {
> + netdev_err(netdev,
> + "Failed to restart device after mqprio change: %d\n",
> + err);
> + /* ndo_open() already freed the TX schedulers on failure while
> + * netif_running() may still be true; drop mqprio software state
> + * only instead of sending shaper clears to freed queues.
> + */
> + otx2_mqprio_clear_sw(pfvf);
[Severity: Low]
Is this comment accurate? Every caller invokes the function with the device
running. __LINK_STATE_START is cleared before ndo_stop() and set again only
when ndo_open() succeeds. So netif_running() is always false in this
branch, as the function's header comment says:
Clear __LINK_STATE_START before ndo_stop() so netif_running() is false
for the duration of the bounce.
[ ... ]
> +static int otx2_teardown_tc_mqprio(struct otx2_nic *pfvf,
> + struct tc_mqprio_qopt_offload *mqprio)
> +{
[ ... ]
> + if (pfvf->mqprio.replace_setup_done && pfvf->cur_mq_snap) {
> + err = 0;
> + if (pfvf->mqprio.replace_graft_done || !tc_can_offload(netdev))
> + otx2_mqprio_snap_commit(pfvf);
> + else
> + err = otx2_mqprio_restore_old(pfvf);
> + otx2_mqprio_clear_replace_state(pfvf);
> + return err;
> + }
[Severity: Medium]
Does the !tc_can_offload(netdev) case commit a replace that actually failed?
On both PF and VF, NETIF_F_HW_TC is added to hw_features after
features |= hw_features. By default it is therefore not in features, and
qdisc_offload_graft_helper() never sends TC_ROOT_GRAFT:
if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
return;
Consider a qdisc_create() that fails after mqprio_init() succeeded, for
example:
tc qdisc replace dev X root handle 1: estimator 1sec 8sec mqprio ... hw 1 shaper bw_rate ...
mqprio_init() has already set TCQ_F_MQROOT, so this check fails:
if (sch->flags & TCQ_F_MQROOT) {
NL_SET_ERR_MSG(extack, "Cannot attach rate estimator to a multi-queue root qdisc");
goto err_out4;
}
By then otx2_setup_tc_mqprio() has already restarted the netdev,
programmed the new MDQ shapers and netdev TC layout, and set
replace_setup_done. err_out4 destroys the new qdisc and
mqprio_disable_offload() ends up here. otx2_mqprio_snap_commit() runs
instead of otx2_mqprio_restore_old().
tc reports failure and the old qdisc stays root. The hardware, the rate
cache, old_mq_snap and the netdev TC layout keep the rejected
configuration, and later ndo_open() calls re-apply it through
otx2_mqprio_up().
In this configuration, is there a way to tell a successful graft apart from
a failed create?
> +
> + /* Skip the netdev restart when mqprio offload was not active. */
> + if (!had_mqprio)
> + return 0;
> +
> + if (if_up) {
> + err = otx2_mqprio_down(pfvf);
> + if (err)
> + return err;
> +
> + return otx2_mqprio_restart_netdev(netdev, false);
> + }
[Severity: Medium]
What happens to the driver state when otx2_mqprio_down() fails here? The tc
core can't veto the teardown. mqprio_disable_offload() ignores the return
value and the qdisc is freed anyway:
case TC_MQPRIO_MODE_CHANNEL:
dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_MQPRIO,
&mqprio);
break;
When otx2_nix_tm_clear_queue_shaper() fails, otx2_mqprio_down() returns
early on purpose ("keeping offload state"). This code then returns before
otx2_mqprio_restart_netdev(netdev, false). So rate_limit, the rate caches,
the widened SMQ allocation and the programmed MDQ shapers all outlive a
qdisc that no longer exists.
The non-replace cleanup: label in otx2_setup_tc_mqprio() gets the same
result. It ignores otx2_teardown_tc_mqprio()'s return value, and the failed
mqprio_init() never records hw_offload, so no second teardown ever comes.
After that:
- XDP, HTB, PFC and ethtool -L stay blocked.
- Each ndo_open() re-applies the stale shapers via otx2_mqprio_up().
- The next tc qdisc add ... mqprio hw 1 sees
replacing = pfvf->mqprio.rate_limit as true. It sets replace_setup_done
and doesn't commit.
- A later tc qdisc del takes the replace_setup_done && cur_mq_snap branch
above, commits, and returns 0 without clearing rate_limit or the hardware
shapers.
Should teardown drop the software state and bounce the netdev even when
the shaper clear fails?
> +static int otx2_setup_tc_mqprio(struct net_device *netdev,
> + struct tc_mqprio_qopt_offload *mqprio)
> +{
[ ... ]
> + if (!replacing && otx2_mqprio_keep_netdev_tc(pfvf)) {
> + netdev_err(netdev,
> + "mqprio: delete existing mqprio before re-enabling hw/sw offload\n");
> + NL_SET_ERR_MSG_MOD(extack,
> + "mqprio: delete existing mqprio before re-enabling hw/sw offload\n");
> + return -EOPNOTSUPP;
> + }
[Severity: Low]
This isn't a bug, but this extack string ends with a newline, unlike the
other extack messages in this patch. The string is sent to user space in
NLMSGERR_ATTR_MSG.
[ ... ]
> +cleanup:
> + qopt->hw = 0;
> + if (replacing) {
> + int restore_err = otx2_mqprio_restore_old(pfvf);
> +
> + otx2_mqprio_clear_replace_state(pfvf);
> + if (restore_err) {
> + netdev_err(netdev,
> + "mqprio: replace failed and prior configuration rollback failed: %d\n",
> + restore_err);
> + if (extack)
> + NL_SET_ERR_MSG_FMT_MOD(extack,
> + "mqprio: replace failed and prior configuration rollback failed: %d",
> + restore_err);
> + } else {
[Severity: Medium]
When the rollback itself fails, what state is the old qdisc left in? This
branch only reports the error.
Case one: otx2_nix_tm_set_queue_shaper() fails partway through the new
mapping. otx2_mqprio_restore_old() then returns early, either with -ENOMEM
from otx2_mqprio_alloc_cache(pfvf, false) or with an error from
otx2_nix_tm_clear_queue_shaper(). The interface keeps running with the new
rates only partly programmed. On top of that, mqprio_destroy() on the new
instance (hw_offload is 0) clears netdev num_tc.
Case two: the rollback reaches otx2_mqprio_restart_netdev(netdev, true)
and ndo_open() fails. The helper then calls otx2_mqprio_clear_sw(), which
sets rate_limit to false and frees the caches, and then netif_close(). The
old offloaded mqprio stays root. The next ip link set up skips
otx2_mqprio_up() because rate_limit is false, so none of the old qdisc's
rates are enforced.
In either case, could the hardware limits silently disagree with the
offloaded qdisc that is still attached?
[ ... ]
--
Sashiko AI review · https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260929022915.2704627-1-rkannoth%40marvell.com
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v18 net-next 0/2] octeontx2: mqprio bandwidth offload for NIX TX schedulers
2026-09-29 2:29 [PATCH v18 net-next 0/2] octeontx2: mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
` (2 preceding siblings ...)
2026-10-02 1:22 ` [PATCH v18 net-next 0/2] octeontx2: " Jakub Kicinski
@ 2026-10-02 9:37 ` David Laight
3 siblings, 0 replies; 7+ messages in thread
From: David Laight @ 2026-10-02 9:37 UTC (permalink / raw)
To: Ratheesh Kannoth
Cc: bpf, linux-kernel, netdev, andrew+netdev, ast, daniel, davem,
edumazet, hawk, john.fastabend, kuba, pabeni, sdf, sgoutham
On Tue, 29 Sep 2026 07:59:13 +0530
Ratheesh Kannoth <rkannoth@marvell.com> wrote:
> This series adds hardware offload for channel-mode mqprio with
> TC_MQPRIO_SHAPER_BW_RATE on Marvell octeontx2/cn10k PF and VF RVU
> netdevices. Each non-QoS transmit queue is shaped by programming MDQ CIR/PIR
> on the NIX TX scheduler. When bandwidth offload is active, the driver
> allocates one SMQ per queue, parents every MDQ under TL4[0], and maps each
> traffic class min/max rate to the queue(s) in that class.
>
> The NIX TX scheduler hierarchy cannot be reprogrammed live today, so
> mqprio add, replace, delete, and failed-replace rollback rebuild it by
> bouncing the netdev through ndo_stop()/ndo_open(). That intentionally
> drops in-flight traffic on each change. otx2_mqprio_restart_netdev() clears
> __LINK_STATE_START before ndo_stop() and does not call
> dev_deactivate()/dev_activate(); carrier and TX queues are restored after
> ndo_open() via the normal link-event path when the link is up. Cache the
> active rates and restore MDQ shapers from otx2_mqprio_up() during ndo_open();
> fail closed if restoration fails, leaving ndo_open() unsuccessful and the
> interface administratively down.
>
> Track mqprio configuration in mq_offload_snap snapshots (TC layout and
> rates). On tc qdisc replace, stage the new configuration while keeping
> the previous snapshot for rollback: failed setup restores the old
> snapshot via netdev restart when the interface is running, successful
> graft is recorded through TC_ROOT_GRAFT, and teardown of the replaced
> qdisc instance commits the staged snapshot without tearing down the live
> offload.
>
> Patch 1 converts PF/VF and representor flag access to atomic bitops.
> Patch 2 depends on it for safe OTX2_FLAG_INTF_DOWN and OTX2_FLAG_PORT_UP
> updates on asynchronous mbox paths and during the mqprio netdev bounce.
Can't you just move those two flags to a separate structure member?
In at least one place the code separately clears one and sets the other.
That makes me think it should a a three-valued state not two bits.
That would save all the expensive locked operations.
David
>
> The driver rejects offload unless the interface is running and the device
> advertises CIR+PIR support. PF and VF RVU netdevices share the same TC
> offload path via ndo_setup_tc / otx2_open(); SDP representors are not
> supported. Per-TC rates are rejected when a traffic class spans more than
> one queue. Concurrent PFC, XDP, SDP rep, or HTB use is blocked, and ethtool
> channel count changes are blocked while mqprio bandwidth offload is active.
>
> Ratheesh Kannoth (2):
> octeontx2: use atomic bitops for PF/VF and rep flags
> octeontx2: add mqprio bandwidth offload for NIX TX schedulers
>
> .../ethernet/marvell/octeontx2/nic/cn10k_ipsec.c | 8 +-
> .../ethernet/marvell/octeontx2/nic/otx2_common.c | 154 +++-
> .../ethernet/marvell/octeontx2/nic/otx2_common.h | 107 ++-
> .../ethernet/marvell/octeontx2/nic/otx2_dcbnl.c | 6 +
> .../ethernet/marvell/octeontx2/nic/otx2_devlink.c | 2 +-
> .../ethernet/marvell/octeontx2/nic/otx2_ethtool.c | 29 +-
> .../ethernet/marvell/octeontx2/nic/otx2_flows.c | 34 +-
> .../net/ethernet/marvell/octeontx2/nic/otx2_pf.c | 95 ++-
> .../net/ethernet/marvell/octeontx2/nic/otx2_tc.c | 838 ++++++++++++++++++++-
> .../net/ethernet/marvell/octeontx2/nic/otx2_txrx.c | 16 +-
> .../net/ethernet/marvell/octeontx2/nic/otx2_vf.c | 10 +-
> .../net/ethernet/marvell/octeontx2/nic/otx2_xsk.c | 4 +-
> drivers/net/ethernet/marvell/octeontx2/nic/qos.c | 11 +
> .../net/ethernet/marvell/octeontx2/nic/qos_sq.c | 4 +-
> drivers/net/ethernet/marvell/octeontx2/nic/rep.c | 32 +-
> drivers/net/ethernet/marvell/octeontx2/nic/rep.h | 3 +-
> 16 files changed, 1203 insertions(+), 150 deletions(-)
>
> ---
>
> v17 -> v18: Addressed sashiko comments on v17.
> - Commit staged mqprio replace snapshots when TC_ROOT_GRAFT is skipped
> because hw-tc-offload is off in netdev->features (!tc_can_offload()),
> instead of rolling back a replace that actually succeeded.
> - Keep NETIF_F_HW_TC in netdev->hw_features only (PF and VF); gate mqprio
> setup/teardown on otx2_tc_can_offload() so hw-tc-offload stays opt-in via
> ethtool -K and flower/matchall are not pushed to ndo_setup_tc by default.
> - Split TC shutdown around netdev unregister: cancel mqprio deferred work and
> free mqprio snapshots before unregister_netdev(), but destroy the TC flower
> flow list only after unregister so clsact teardown can still run
> otx2_tc_del_flow() and free MCAM/mcast/policer state.
> - Reject ethtool -L TX queue reduction while mqprio offload snapshots remain,
> so a later replace rollback cannot restore stale per-queue rates past the
> current queue count.
> - Clarify mqprio shaper and netdev-restart comments: ndo_open() may restore
> cached MDQ shapers via otx2_mqprio_up() before setup clears them, and a
> failed mqprio restart relies on OTX2_FLAG_INTF_DOWN so otx2_stop() returns
> early through netif_close(), not on skipping ndo_stop().
> https://lore.kernel.org/netdev/20260923032217.1732753-1-rkannoth@marvell.com/
>
> v16 -> v17: Addressed sashiko comments on v16.
> - Replace the per-bit otx2_sync_flags_from_rep() loop with a masked
> READ_ONCE/WRITE_ONCE publish of OTX2_REP_SYNC_FLAGS_MASK so lockless NAPI
> readers never observe torn PF/representor flag combinations.
> - Evaluate mqprio.rate_limit and old_mq_snap inside rtnl_lock in
> otx2_mqprio_netdev_tc_work() so a concurrent qdisc delete cannot leave
> stale netdev TC mappings after offload teardown.
> - Advertise NETIF_F_HW_TC in netdev->features (PF and VF) when TC flower
> offload is supported, so tc_can_offload() succeeds without ethtool -K
> hw-tc-offload on; move otx2_init_tc() before register_netdev() and fix
> probe/remove teardown ordering.
> - Reject mqprio add when a software mqprio root is already installed
> (otx2_mqprio_keep_netdev_tc()) and defer netdev TC restore from a new
> fail_validate path on failed replace validation before any hardware
> change.
> - Fix otx2_mqprio_max_rate_bytes_ps() to cap against the NIX TLX maximum
> rate instead of the burst-bucket size; use the 65536 byte HTB default
> burst when programming MDQ shapers; guard otx2_get_smq_idx() when
> txschq_cnt[NIX_TXSCH_LVL_SMQ] is zero after otx2_txschq_stop().
> - Rename patch 2 to octeontx2: (driver-wide PF/VF offload, not PF-only).
> https://lore.kernel.org/netdev/20260918015906.1255204-1-rkannoth@marvell.com/
>
> v15 -> v16: Addressed sashiko comments on v15 and aligned documentation with code.
> - Sync representor flags through OTX2_FLAG_MAX in otx2_sync_flags_from_rep()
> instead of hard-coding OTX2_REP_VF_INITIALIZED as the loop bound.
> - Drop the rvu_nix.c is_valid_txschq() ratelimited error print from the mqprio
> patch; remove the misplaced atomic-bitops and AF-debug paragraphs from the
> mqprio commit message (they belong to patch 1 or are out of scope).
> - Extend mq_offload_snap to record prio_tc_map[] and mqprio rate flags; restore
> the full netdev TC layout (num_tc, queue ranges, and priority map) via
> otx2_mqprio_apply_snap_netdev() on rollback paths.
> - Defer netdev TC restore on failed replace (otx2_mqprio_netdev_tc_work) so
> rollback survives mqprio_destroy() clearing dev->num_tc after setup errors
> once the core unwinds the failed qdisc instance.
> - Stop calling dev_deactivate()/dev_activate() from otx2_mqprio_restart_netdev();
> bounce the interface with ndo_stop()/ndo_open() only and restore carrier
> through the normal link-event path after ndo_open(), avoiding qdisc
> reentrancy during tc replace graft.
> - Preserve netdev TC mappings when tearing down an offloaded instance that is
> replaced by a software mqprio graft (otx2_mqprio_keep_netdev_tc()) instead
> of always calling netdev_set_num_tc(0) and breaking the live replacement.
> - Return an error from otx2_mqprio_down() when clearing hardware shapers fails
> and keep offload software state, instead of v15's behaviour of clearing
> rate_limit while stale MDQ limits may remain programmed.
> - Rebuild the TX scheduler via netdev restart in otx2_mqprio_restore_old() on
> a running interface after failed-replace rollback so partially applied MDQ
> shapers are not left running with mismatched software state.
> - Clear txschq_cnt[] in otx2_txschq_stop() after freeing scheduler nodes so
> post-stop shaper mailbox operations do not consult stale counts.
> - Document fail-closed ndo_open() when otx2_mqprio_up() cannot restore shapers,
> and that PF/VF RVU netdevices share the ndo_setup_tc / otx2_open() offload
> path (SDP representors remain unsupported); downgrade the mqprio restart
> notice to netdev_dbg().
> https://lore.kernel.org/netdev/20260911105521.689565-1-rkannoth@marvell.com/
>
> v14 -> v15: Addressed sashiko comments.
> - Split atomic PF/VF and representor flag access into a preparatory patch
> so mqprio netdev-restart and mbox paths can update OTX2_FLAG_INTF_DOWN
> and OTX2_FLAG_PORT_UP without data races on the shared flags word.
> - Clear mqprio software state when hardware shaper teardown fails, warn,
> and still bounce the netdev on delete so offload does not remain stuck
> active after a mailbox error.
> https://lore.kernel.org/netdev/20260904031553.3196916-1-rkannoth@marvell.com/
>
> v13 -> v14: Addressed sashiko comments.
> - Use atomic set_bit()/clear_bit() for OTX2_FLAG_INTF_DOWN and
> OTX2_FLAG_PORT_UP updates on netdev-restart and mbox paths.
> - Block concurrent mqprio bandwidth offload and HTB shaping.
> - Fail ndo_open() if otx2_mqprio_up() cannot restore MDQ shapers.
> - Rebuild the TX scheduler via netdev restart in otx2_mqprio_restore_old()
> when rolling back a failed replace on a running interface.
> - Return an error from otx2_mqprio_down() if clearing hardware shapers
> fails instead of clearing software state anyway.
> https://lore.kernel.org/netdev/20260904031553.3196916-1-rkannoth@marvell.com/
>
> v12 -> v13: Addressed sashiko comments.
> https://sashiko.dev/#/patchset/20260903023324.3078284-1-rkannoth%40marvell.com
>
> v11 -> v12: Addressed sashiko comments.
> https://sashiko.dev/#/patchset/20260902015500.2985371-1-rkannoth%40marvell.com
> v10 -> v11: Addressed sashiko comments.
> https://sashiko.dev/#/patchset/20260831131014.2639581-1-rkannoth%40marvell.com
>
> v9 -> v10: Addressed sashiko/jacub comments.
> https://sashiko.dev/#/message/20260817032747.1765883-1-rkannoth%40marvell.com
>
> v8 -> v9: Addressed Sashiko comments
> https://lore.kernel.org/netdev/aoJ6FhtWue0FHDQV@rkannoth-OptiPlex-7090/
> v7 -> v8: Addressed Sashiko comments
> https://sashiko.dev/#/patchset/20260811085050.3212280-1-rkannoth%40marvell.com
> v6 -> v7: Addressed Sashiko comments
> https://sashiko.dev/#/message/20260810034738.1786029-1-rkannoth%40marvell.com
> v5 -> v6: Addressed Sashiko comments
> https://lore.kernel.org/netdev/20260806095434.1144397-1-rkannoth@marvell.com/
> v4 -> v5: Addressed sashiko comments
> https://sashiko.dev/#/patchset/20260803042724.3380209-1-rkannoth%40marvell.com
> v3 -> v4: Addressed sashiko comments
> https://lore.kernel.org/netdev/20260729105139.2302908-1-rkannoth@marvell.com/
> v2 -> v3: Addressed sashiko comments
> https://lore.kernel.org/netdev/amnYX866mYx02cBe@rkannoth-OptiPlex-7090/T/#m67310cbec48b21c7720858ab3a1ea083a0f8dc10
> v1 -> v2: Addressed sashiko comments
> https://lore.kernel.org/netdev/20260724075010.2665758-1-rkannoth@marvell.com/
>
> --
> 2.43.0
>
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2026-10-02 9:37 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-09-29 2:29 [PATCH v18 net-next 0/2] octeontx2: mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
2026-09-29 2:29 ` [PATCH v18 net-next 1/2] octeontx2: use atomic bitops for PF/VF and rep flags Ratheesh Kannoth
2026-10-02 8:31 ` netdev-bot+sashiko
2026-09-29 2:29 ` [PATCH v18 net-next 2/2] octeontx2: add mqprio bandwidth offload for NIX TX schedulers Ratheesh Kannoth
2026-10-02 8:31 ` netdev-bot+sashiko
2026-10-02 1:22 ` [PATCH v18 net-next 0/2] octeontx2: " Jakub Kicinski
2026-10-02 9:37 ` David Laight
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®