openwrt/target/linux/mvebu/patches-5.15/700-mvneta-tx-queue-workaround.patch
Marek Behún 76cabb95da kernel: Backport mvneta crash fix to 5.15
Backport Russell King's series [1]
  net: mvneta: reduce size of TSO header allocation
to pending-5.15 to fix random crashes on Turris Omnia.

This also backports two patches that are dependencies to this series:
  net: mvneta: Delete unused variable
  net: mvneta: fix potential double-frees in mvneta_txq_sw_deinit()

[1] https://lore.kernel.org/netdev/ZCsbJ4nG+So%2Fn9qY@shell.armlinux.org.uk/

Signed-off-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com> (squashed)
(cherry picked from commit 7b31c2e9ed)
2023-06-08 15:33:14 +02:00

39 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
@@ -5006,6 +5006,16 @@ static int mvneta_setup_tc(struct net_de
}
}
+#ifndef CONFIG_ARM64
+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;
+}
+#endif
+
static const struct net_device_ops mvneta_netdev_ops = {
.ndo_open = mvneta_open,
.ndo_stop = mvneta_stop,
@@ -5016,6 +5026,9 @@ static const struct net_device_ops mvnet
.ndo_fix_features = mvneta_fix_features,
.ndo_get_stats64 = mvneta_get_stats64,
.ndo_eth_ioctl = mvneta_ioctl,
+#ifndef CONFIG_ARM64
+ .ndo_select_queue = mvneta_select_queue,
+#endif
.ndo_bpf = mvneta_xdp,
.ndo_xdp_xmit = mvneta_xdp_xmit,
.ndo_setup_tc = mvneta_setup_tc,