Re: [PATCH net-next 12/13] rtnetlink: make rtnl_fill_link_ifmap() RCU ready

[Jiri Pirko <jiri@resnulli.us>]

Wed, Feb 21, 2024 at 06:15:11PM CET, edumazet@google.com wrote:
>On Wed, Feb 21, 2024 at 5:03 PM Jiri Pirko <jiri@resnulli.us> wrote:
>>
>> Wed, Feb 21, 2024 at 11:59:14AM CET, edumazet@google.com wrote:
>> >Use READ_ONCE() to read the following device fields:
>> >
>> >       dev->mem_start
>> >       dev->mem_end
>> >       dev->base_addr
>> >       dev->irq
>> >       dev->dma
>> >       dev->if_port
>> >
>> >Provide IFLA_MAP attribute only if at least one of these fields
>> >is not zero. This saves some space in the output skb for most devices.
>> >
>> >Signed-off-by: Eric Dumazet <edumazet@google.com>
>> >---
>> > net/core/rtnetlink.c | 26 ++++++++++++++------------
>> > 1 file changed, 14 insertions(+), 12 deletions(-)
>> >
>> >diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
>> >index 1b26dfa5668d22fb2e30ceefbf143e98df13ae29..b91ec216c593aaebf97ea69aa0d2d265ab61c098 100644
>> >--- a/net/core/rtnetlink.c
>> >+++ b/net/core/rtnetlink.c
>> >@@ -1455,19 +1455,21 @@ static noinline_for_stack int rtnl_fill_vf(struct sk_buff *skb,
>> >       return 0;
>> > }
>> >
>> >-static int rtnl_fill_link_ifmap(struct sk_buff *skb, struct net_device *dev)
>> >+static int rtnl_fill_link_ifmap(struct sk_buff *skb,
>> >+                              const struct net_device *dev)
>> > {
>> >       struct rtnl_link_ifmap map;
>> >
>> >       memset(&map, 0, sizeof(map));
>> >-      map.mem_start   = dev->mem_start;
>> >-      map.mem_end     = dev->mem_end;
>> >-      map.base_addr   = dev->base_addr;
>> >-      map.irq         = dev->irq;
>> >-      map.dma         = dev->dma;
>> >-      map.port        = dev->if_port;
>> >-
>> >-      if (nla_put_64bit(skb, IFLA_MAP, sizeof(map), &map, IFLA_PAD))
>> >+      map.mem_start = READ_ONCE(dev->mem_start);
>> >+      map.mem_end   = READ_ONCE(dev->mem_end);
>> >+      map.base_addr = READ_ONCE(dev->base_addr);
>> >+      map.irq       = READ_ONCE(dev->irq);
>> >+      map.dma       = READ_ONCE(dev->dma);
>> >+      map.port      = READ_ONCE(dev->if_port);
>> >+      /* Only report non zero information. */
>> >+      if (memchr_inv(&map, 0, sizeof(map)) &&
>>
>> This check(optimization) is unrelated to the rest of the patch, correct?
>> If yes, could it be a separate patch?
>
>Sure thing. BTW, do you know which tool is using this ?
>
>I could not find IFLA_MAP being used in iproute2 or ethtool.

No clue. Never spotted it.