tjh.dev / kernel
Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

ip_tunnel: adapt iptunnel_xmit_stats() to NETDEV_PCPU_STAT_DSTATS

Blamed commits forgot that vxlan/geneve use udp_tunnel[6]_xmit_skb() which
call iptunnel_xmit_stats().

iptunnel_xmit_stats() was assuming tunnels were only using
NETDEV_PCPU_STAT_TSTATS.

@syncp offset in pcpu_sw_netstats and pcpu_dstats is different.

32bit kernels would either have corruptions or freezes if the syncp
sequence was overwritten.

This patch also moves pcpu_stat_type closer to dev->{t,d}stats to avoid
a potential cache line miss since iptunnel_xmit_stats() needs to read it.

Fixes: 6fa6de302246 ("geneve: Handle stats using NETDEV_PCPU_STAT_DSTATS.")
Fixes: be226352e8dc ("vxlan: Handle stats using NETDEV_PCPU_STAT_DSTATS.")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Guillaume Nault <gnault@redhat.com>
Link: https://patch.msgid.link/20260311123110.1471930-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

authored by

Eric Dumazet and committed by
Jakub Kicinski 8431c602 e1f0a18c

+23 -8
+1 -2
include/linux/netdevice.h
··· 2155 2155 unsigned long state; 2156 2156 unsigned int flags; 2157 2157 unsigned short hard_header_len; 2158 + enum netdev_stat_type pcpu_stat_type:8; 2158 2159 netdev_features_t features; 2159 2160 struct inet6_dev __rcu *ip6_ptr; 2160 2161 __cacheline_group_end(net_device_read_txrx); ··· 2404 2403 /* mid-layer private */ 2405 2404 void *ml_priv; 2406 2405 enum netdev_ml_priv_type ml_priv_type; 2407 - 2408 - enum netdev_stat_type pcpu_stat_type:8; 2409 2406 2410 2407 #if IS_ENABLED(CONFIG_GARP) 2411 2408 struct garp_port __rcu *garp_port;
+22 -6
include/net/ip_tunnels.h
··· 665 665 static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len) 666 666 { 667 667 if (pkt_len > 0) { 668 - struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats); 668 + if (dev->pcpu_stat_type == NETDEV_PCPU_STAT_DSTATS) { 669 + struct pcpu_dstats *dstats = get_cpu_ptr(dev->dstats); 669 670 670 - u64_stats_update_begin(&tstats->syncp); 671 - u64_stats_add(&tstats->tx_bytes, pkt_len); 672 - u64_stats_inc(&tstats->tx_packets); 673 - u64_stats_update_end(&tstats->syncp); 674 - put_cpu_ptr(tstats); 671 + u64_stats_update_begin(&dstats->syncp); 672 + u64_stats_add(&dstats->tx_bytes, pkt_len); 673 + u64_stats_inc(&dstats->tx_packets); 674 + u64_stats_update_end(&dstats->syncp); 675 + put_cpu_ptr(dstats); 676 + return; 677 + } 678 + if (dev->pcpu_stat_type == NETDEV_PCPU_STAT_TSTATS) { 679 + struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats); 680 + 681 + u64_stats_update_begin(&tstats->syncp); 682 + u64_stats_add(&tstats->tx_bytes, pkt_len); 683 + u64_stats_inc(&tstats->tx_packets); 684 + u64_stats_update_end(&tstats->syncp); 685 + put_cpu_ptr(tstats); 686 + return; 687 + } 688 + pr_err_once("iptunnel_xmit_stats pcpu_stat_type=%d\n", 689 + dev->pcpu_stat_type); 690 + WARN_ON_ONCE(1); 675 691 return; 676 692 } 677 693