During QoS scheduling testing with multiple strict priority flows, the
netdev tx watchdog timeout routine is invoked when a low priority QoS
queue doesn't get a chance to transmit the packets because other high
priority flows are completely subscribing the transmit link. The netdev
tx watchdog timeout routine will stop MAC RX and TX functionality in
otx2_stop() routine before cleanup of HW TX queues which results in SMQ
flush errors because the packets belonging to low priority queues will
never gets flushed since MAC TX is disabled. This patch fixes the issue
by re-enabling MAC TX to ensure the packets in HW pipeline gets flushed
properly.
Fixes: a7faa68b4e7f ("octeontx2-af: Start/Stop traffic in CGX along with NPC")
Signed-off-by: Naveen Mamindlapalli <[email protected]>
Signed-off-by: Sunil Kovvuri Goutham <[email protected]>
---
drivers/net/ethernet/marvell/octeontx2/af/rvu.h | 1 +
.../net/ethernet/marvell/octeontx2/af/rvu_cgx.c | 17 +++++++++++++++++
.../net/ethernet/marvell/octeontx2/af/rvu_nix.c | 8 +++++++-
3 files changed, 25 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
index cce2806aaa50..8802961b8889 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
@@ -905,6 +905,7 @@ u32 rvu_cgx_get_fifolen(struct rvu *rvu);
void *rvu_first_cgx_pdata(struct rvu *rvu);
int cgxlmac_to_pf(struct rvu *rvu, int cgx_id, int lmac_id);
int rvu_cgx_config_tx(void *cgxd, int lmac_id, bool enable);
+int rvu_cgx_tx_enable(struct rvu *rvu, u16 pcifunc, bool enable);
int rvu_cgx_prio_flow_ctrl_cfg(struct rvu *rvu, u16 pcifunc, u8 tx_pause, u8 rx_pause,
u16 pfc_en);
int rvu_cgx_cfg_pause_frm(struct rvu *rvu, u16 pcifunc, u8 tx_pause, u8 rx_pause);
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
index 15a319684ed3..38acdc7a73bb 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
@@ -465,6 +465,23 @@ int rvu_cgx_config_rxtx(struct rvu *rvu, u16 pcifunc, bool start)
return mac_ops->mac_rx_tx_enable(cgxd, lmac_id, start);
}
+int rvu_cgx_tx_enable(struct rvu *rvu, u16 pcifunc, bool enable)
+{
+ int pf = rvu_get_pf(pcifunc);
+ struct mac_ops *mac_ops;
+ u8 cgx_id, lmac_id;
+ void *cgxd;
+
+ if (!is_cgx_config_permitted(rvu, pcifunc))
+ return LMAC_AF_ERR_PERM_DENIED;
+
+ rvu_get_cgx_lmac_id(rvu->pf2cgxlmac_map[pf], &cgx_id, &lmac_id);
+ cgxd = rvu_cgx_pdata(cgx_id, rvu);
+ mac_ops = get_mac_ops(cgxd);
+
+ return mac_ops->mac_tx_enable(cgxd, lmac_id, enable);
+}
+
int rvu_cgx_config_tx(void *cgxd, int lmac_id, bool enable)
{
struct mac_ops *mac_ops;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
index 2b6ab748ce25..58744313f0eb 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
@@ -4737,7 +4737,13 @@ int rvu_mbox_handler_nix_lf_stop_rx(struct rvu *rvu, struct msg_req *req,
pfvf = rvu_get_pfvf(rvu, pcifunc);
clear_bit(NIXLF_INITIALIZED, &pfvf->flags);
- return rvu_cgx_start_stop_io(rvu, pcifunc, false);
+ err = rvu_cgx_start_stop_io(rvu, pcifunc, false);
+ if (err)
+ return err;
+
+ rvu_cgx_tx_enable(rvu, pcifunc, true);
+
+ return 0;
}
#define RX_SA_BASE GENMASK_ULL(52, 7)
--
2.39.0.198.ga38d39a4c5
Hello:
This patch was applied to netdev/net.git (main)
by David S. Miller <[email protected]>:
On Tue, 2 Jan 2024 19:44:00 +0530 you wrote:
> During QoS scheduling testing with multiple strict priority flows, the
> netdev tx watchdog timeout routine is invoked when a low priority QoS
> queue doesn't get a chance to transmit the packets because other high
> priority flows are completely subscribing the transmit link. The netdev
> tx watchdog timeout routine will stop MAC RX and TX functionality in
> otx2_stop() routine before cleanup of HW TX queues which results in SMQ
> flush errors because the packets belonging to low priority queues will
> never gets flushed since MAC TX is disabled. This patch fixes the issue
> by re-enabling MAC TX to ensure the packets in HW pipeline gets flushed
> properly.
>
> [...]
Here is the summary with links:
- [net] octeontx2-af: Re-enable MAC TX in otx2_stop processing
https://git.kernel.org/netdev/net/c/818ed8933bd1
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html