openwrt/target/linux/mvebu/patches-5.4/700-mvneta-tx-queue-workaround.patch
Jason A. Donenfeld 196f3d586f kernel-5.4: bump to 5.4.102 and refresh patches
5.4.102 backported a lot of stuff that our WireGuard backport already
did, in addition to other patches we had, so those patches were
removed from that part of the series. In the process other patches were
refreshed or reworked to account for upstream changes.

This commit involved `update_kernel.sh -v -u 5.4`.

Cc: John Audia <graysky@archlinux.us>
Cc: David Bauer <mail@david-bauer.net>
Cc: Petr Štetiar <ynezz@true.cz>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
2021-03-04 22:06:53 +01:00

35 lines
1.2 KiB
Diff

The hardware queue scheduling is apparently configured with fixed
priorities, which creates a nasty fairness issue where traffic from one
CPU can starve traffic from all other CPUs.
Work around this issue by forcing all tx packets to go through one CPU,
until this issue is fixed properly.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
--- a/drivers/net/ethernet/marvell/mvneta.c
+++ b/drivers/net/ethernet/marvell/mvneta.c
@@ -4691,6 +4691,14 @@ static int mvneta_ethtool_set_eee(struct
return phylink_ethtool_set_eee(pp->phylink, eee);
}
+static u16 mvneta_select_queue(struct net_device *dev, struct sk_buff *skb,
+ struct net_device *sb_dev)
+{
+ /* XXX: hardware queue scheduling is broken,
+ * use only one queue until it is fixed */
+ return 0;
+}
+
static const struct net_device_ops mvneta_netdev_ops = {
.ndo_open = mvneta_open,
.ndo_stop = mvneta_stop,
@@ -4701,6 +4709,7 @@ static const struct net_device_ops mvnet
.ndo_fix_features = mvneta_fix_features,
.ndo_get_stats64 = mvneta_get_stats64,
.ndo_do_ioctl = mvneta_ioctl,
+ .ndo_select_queue = mvneta_select_queue,
.ndo_bpf = mvneta_xdp,
.ndo_xdp_xmit = mvneta_xdp_xmit,
};