| Message ID | 20230122161602.1958577-3-vadfed@meta.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | mlx5: bugfixes for ptp fifo queue | expand |
| Context | Check | Description |
|---|---|---|
| netdev/tree_selection | success | Clearly marked for net |
| netdev/apply | fail | Patch does not apply to net |
On 22/01/2023 18:16, Vadim Fedorenko wrote: > During ptp resync operation SKBs were poped from the fifo but were never > freed neither by napi_consume nor by dev_kfree_skb_any. Add call to > napi_consume_skb to properly free SKBs. > > Fixes: 19b43a432e3e ("net/mlx5e: Extend SKB room check to include PTP-SQ") Same comment as previous patch? > Signed-off-by: Vadim Fedorenko <vadfed@meta.com> > --- > drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c > index 11a99e0f00c6..d60bb997c53b 100644 > --- a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c > +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c > @@ -102,6 +102,7 @@ static bool mlx5e_ptp_skb_fifo_ts_cqe_resync(struct mlx5e_ptpsq *ptpsq, u16 skb_ > hwts.hwtstamp = mlx5e_skb_cb_get_hwts(skb)->cqe_hwtstamp; > skb_tstamp_tx(skb, &hwts); > ptpsq->cq_stats->resync_cqe++; > + napi_consume_skb(skb, 1); Was wondering whether we should pass the actual budget here instead of 1, but looking at napi_consume_skb() it doesn't really matter.. Anyway: Reviewed-by: Gal Pressman <gal@nvidia.com> > skb_cc = PTP_WQE_CTR2IDX(ptpsq->skb_fifo_cc); > } >
On 23/01/2023 12:38, Gal Pressman wrote: > On 22/01/2023 18:16, Vadim Fedorenko wrote: >> During ptp resync operation SKBs were poped from the fifo but were never >> freed neither by napi_consume nor by dev_kfree_skb_any. Add call to >> napi_consume_skb to properly free SKBs. >> >> Fixes: 19b43a432e3e ("net/mlx5e: Extend SKB room check to include PTP-SQ") > > Same comment as previous patch? Yeah, and it's correct for this patch too. Commit mentioned in Fixes introduced several bugs apart from architectural problem. The first bug of wrong checks and possible overflow/underflow of FIFO is fixed by previous patch. This patch fixes another issue of leaking SKBs from FIFO during synchronisation process. > >> Signed-off-by: Vadim Fedorenko <vadfed@meta.com> >> --- >> drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c | 1 + >> 1 file changed, 1 insertion(+) >> >> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c >> index 11a99e0f00c6..d60bb997c53b 100644 >> --- a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c >> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c >> @@ -102,6 +102,7 @@ static bool mlx5e_ptp_skb_fifo_ts_cqe_resync(struct mlx5e_ptpsq *ptpsq, u16 skb_ >> hwts.hwtstamp = mlx5e_skb_cb_get_hwts(skb)->cqe_hwtstamp; >> skb_tstamp_tx(skb, &hwts); >> ptpsq->cq_stats->resync_cqe++; >> + napi_consume_skb(skb, 1); > > Was wondering whether we should pass the actual budget here instead of > 1, but looking at napi_consume_skb() it doesn't really matter.. > > Anyway: > Reviewed-by: Gal Pressman <gal@nvidia.com> > >> skb_cc = PTP_WQE_CTR2IDX(ptpsq->skb_fifo_cc); >> } >>
On Mon, 23 Jan 2023 14:38:35 +0200 Gal Pressman wrote: > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c > > index 11a99e0f00c6..d60bb997c53b 100644 > > --- a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c > > +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c > > @@ -102,6 +102,7 @@ static bool mlx5e_ptp_skb_fifo_ts_cqe_resync(struct mlx5e_ptpsq *ptpsq, u16 skb_ > > hwts.hwtstamp = mlx5e_skb_cb_get_hwts(skb)->cqe_hwtstamp; > > skb_tstamp_tx(skb, &hwts); > > ptpsq->cq_stats->resync_cqe++; > > + napi_consume_skb(skb, 1); > > Was wondering whether we should pass the actual budget here instead of > 1, but looking at napi_consume_skb() it doesn't really matter.. We should pass the real budget in. The exact value does not matter, but it could matter whether it's zero or not. Budget is zero when NAPI gets polled from an IRQ context.
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c index 11a99e0f00c6..d60bb997c53b 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c @@ -102,6 +102,7 @@ static bool mlx5e_ptp_skb_fifo_ts_cqe_resync(struct mlx5e_ptpsq *ptpsq, u16 skb_ hwts.hwtstamp = mlx5e_skb_cb_get_hwts(skb)->cqe_hwtstamp; skb_tstamp_tx(skb, &hwts); ptpsq->cq_stats->resync_cqe++; + napi_consume_skb(skb, 1); skb_cc = PTP_WQE_CTR2IDX(ptpsq->skb_fifo_cc); }
During ptp resync operation SKBs were poped from the fifo but were never freed neither by napi_consume nor by dev_kfree_skb_any. Add call to napi_consume_skb to properly free SKBs. Fixes: 19b43a432e3e ("net/mlx5e: Extend SKB room check to include PTP-SQ") Signed-off-by: Vadim Fedorenko <vadfed@meta.com> --- drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c | 1 + 1 file changed, 1 insertion(+)