On 12/9/2024 9:08 PM, Alex Shumsky wrote:
On Tue, Nov 26, 2024 at 2:02 PM Arend van Spriel
<arend.vanspriel@xxxxxxxxxxxx> wrote:
On 11/22/2024 7:04 PM, Alex Shumsky wrote:
Currently when debug info enabled, dmesg spammed every few minutes with
misleading messages like:
brcmf_netdev_start_xmit phy0-sta0: insufficient headroom (0)
Do not log this when headroom is actually sufficient.
Thanks for your patch. The message may be misleading, but it is actually
information that we need to cow the packet. The zero value indicates
that this is needed because skb_header_cloned(skb) is true. So it is
still useful in my opinion. If you want to make the message less
misleading for that case I would be happy to ack the patch.
Regards,
Arend
Thanks for the review and sorry for the delayed response.
Do "%s: clone skb header\n" rephrase make sense to you?
I would say:
brcmf_dbg(INFO, "%s: %s headroom\n", brcmf_ifname(ifp),
head_delta ? "insufficient" : "unmodifiable");
I have no deep knowledge of this code, and if you think the original message
is actually useful, I'm ok to leave a log message as it is.
Initially I had guessed it was an unintentional log message because it has
misleading text and logs spammed every few minutes - too rarely to consider
It as a real performance issue.
If you enable debug prints you should expect performance impact. If you
want to capture debug message with negligible performance loss you
should use ftrace. Debug prints in brcmfmac are setup as trace events.
Regards,
Arend
