Re: [PATCH net-next v7 6/7] net: stmmac: support fp parameter of tc-taprio
From: Vladimir Oltean <olteanv@gmail.com>
Date: 2024-09-04 15:55:45
Also in:
linux-arm-kernel, lkml
On Wed, Sep 04, 2024 at 05:21:21PM +0800, Furong Xu wrote:
quoted hunk
.../net/ethernet/stmicro/stmmac/stmmac_tc.c | 23 +++++++------------ 1 file changed, 8 insertions(+), 15 deletions(-)diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_tc.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_tc.c index 9ec2e6ab81aa..2bdb22e175bc 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_tc.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_tc.c@@ -931,9 +931,9 @@ static int tc_taprio_configure(struct stmmac_priv *priv, struct tc_taprio_qopt_offload *qopt) { u32 size, wid = priv->dma_cap.estwid, dep = priv->dma_cap.estdep; + struct netlink_ext_ack *extack = qopt->mqprio.extack; struct timespec64 time, current_time, qopt_time; ktime_t current_time_ns; - bool fpe = false; int i, ret = 0; u64 ctr;@@ -1018,16 +1018,12 @@ static int tc_taprio_configure(struct stmmac_priv *priv, switch (qopt->entries[i].command) { case TC_TAPRIO_CMD_SET_GATES: - if (fpe) - return -EINVAL; break; case TC_TAPRIO_CMD_SET_AND_HOLD: gates |= BIT(0); - fpe = true; break; case TC_TAPRIO_CMD_SET_AND_RELEASE: gates &= ~BIT(0); - fpe = true; break; default: return -EOPNOTSUPP;@@ -1058,11 +1054,6 @@ static int tc_taprio_configure(struct stmmac_priv *priv, tc_taprio_map_maxsdu_txq(priv, qopt); - if (fpe && !priv->dma_cap.fpesel) { - mutex_unlock(&priv->est_lock); - return -EOPNOTSUPP; - } - ret = stmmac_est_configure(priv, priv, priv->est, priv->plat->clk_ptp_rate); mutex_unlock(&priv->est_lock);@@ -1071,6 +1062,11 @@ static int tc_taprio_configure(struct stmmac_priv *priv, goto disable; } + ret = stmmac_fpe_map_preemption_class(priv, priv->dev, extack, + qopt->mqprio.preemptible_tcs); + if (ret) + goto disable; +
Doesn't this break taprio for those callers of tc_setup_taprio() which do not implement fpe_map_preemption_class(), but at least want taprio without FPE nonetheless? As in the earlier mqprio patch, they will return -EINVAL here. Through code inspection, those users are: - All users of .tc = dwxgmac_tc_ops: they have .mac = &dwxlgmac2_ops or .mac = &dwxgmac210_ops, neither of which implements fpe_map_preemption_class(). - The users of .tc = &dwmac510_tc_ops which have .mac = &dwmac4_ops. Again, this does not implement .fpe_map_preemption_class(). I can only rely on code inspection, because I don't know what is the priv->dma_cap.estsel value that the above cores were synthesized with. Thus, I don't know if we break a feature which was available before, or one that was already unavailable. My suggestion is similar: allow mqprio.preemptible_tcs to go to non-zero only if fpe_map_preemption_class() is implemented. Then, program it with any value (including 0) only if fpe_map_preemption_class() is implemented.
netdev_info(priv->dev, "configured EST\n"); return 0;