On Wed, 2020-03-04 at 17:17 +0100, Markus Theil wrote: > Should cfg80211_rx_control_port then also adopt this behavior? > It currently does the conversion in the way I changed > cfg80211_tx_control port to. [...] > TRACE_EVENT(cfg80211_rx_control_port, > ... > __entry->proto = be16_to_cpu(skb->protocol); > __entry->unencrypted = unencrypted; > ), > TP_printk(NETDEV_PR_FMT ", len=%d, " MAC_PR_FMT ", proto: 0x%x, > unencrypted: %s", > NETDEV_PR_ARG, __entry->len, MAC_PR_ARG(from), > __entry->proto, BOOL_TO_STR(__entry->unencrypted)) Great ... I dunno. I guess I don't like touching these so much, because it may be that somebody is relying on them for some kind of tooling? But I guess that's actually unlikely and we can just try, but then I'd rather change this one to BE than the other way around. > IMHO, as long as the trace events are just printed, it is cleaner to > do the endiannes conversion already in the kernel. But they aren't just printed, trace-cmd can record them in binary form, and you can have python plugin code to consume that, for example. johannes
