Re: [PATCH net-next v2] net: Print hashed skb addresses for all net and qdisc events

[Cong Wang <xiyou.wangcong@gmail.com>]

On Fri, Jun 24, 2022 at 03:34:25PM -0600, Subash Abhinov Kasiviswanathan wrote:
> The following commits added support for printing the real address-
> 65875073eddd ("net: use %px to print skb address in trace_netif_receive_skb")
> 70713dddf3d2 ("net_sched: introduce tracepoint trace_qdisc_enqueue()")
> 851f36e40962 ("net_sched: use %px to print skb address in trace_qdisc_dequeue()")
> 
> However, tracing the packet traversal shows a mix of hashes and real
> addresses. Pasting a sample trace for reference-
> 
> ping-14249   [002] .....  3424.046612: netif_rx_entry: dev=lo napi_id=0x3 queue_mapping=0
> skbaddr=00000000dcbed83e vlan_tagged=0 vlan_proto=0x0000 vlan_tci=0x0000 protocol=0x0800
> ip_summed=0 hash=0x00000000 l4_hash=0 len=84 data_len=0 truesize=768 mac_header_valid=1
> mac_header=-14 nr_frags=0 gso_size=0 gso_type=0x0
> ping-14249   [002] .....  3424.046615: netif_rx: dev=lo skbaddr=ffffff888e5d1000 len=84
> 
> Switch the trace print formats to %p for all the events to have a
> consistent format of printing the hashed addresses in all cases.
> 

This is obscured...

What exactly is the inconsistency here? Both are apparently hex, from
user's point of view. The only difference is one is an apparently
invalid kernel address, the other is not. This difference only matters
when you try to dereference it, but I don't think you should do it here,
this is not a raw tracepoint at all. You can always use raw tracepoints
to dereference it without even bothering whatever we print.

Thanks.