[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <639b56c577a0433cb49c85ce0abd5dca@realtek.com>
Date: Mon, 4 Aug 2025 01:31:31 +0000
From: Ping-Ke Shih <pkshih@...ltek.com>
To: Sean Anderson <sean.anderson@...ux.dev>,
"linux-wireless@...r.kernel.org"
<linux-wireless@...r.kernel.org>
CC: "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"Bitterblue
Smith" <rtl8821cerfe2@...il.com>
Subject: RE: [PATCH v2] wifi: rtw89: Print just once for unknown C2H classes
Sean Anderson <sean.anderson@...ux.dev> wrote:
> On 7/31/25 20:30, Ping-Ke Shih wrote:
> > Sean Anderson <sean.anderson@...ux.dev> wrote:
> >> On 7/29/25 20:36, Ping-Ke Shih wrote:
> >> > Sean Anderson <sean.anderson@...ux.dev> wrote:
> >> >> There are more unsupported functions than just LOWRT_RTY. Improve on
> >> >> commit 3b66519b023b ("wifi: rtw89: phy: add dummy c2h handler to avoid
> >> >> warning message") by printing a message just once when we first
> >> >> encounter an unsupported class.
> >> >
> >> > Once I encounter an unsupported class/func, I'll check firmware team if the
> >> > C2H events can be ignored. If so, I add a dummy function to avoid the message.
> >> > If not, I should add code to handle the event.
> >> >
> >> > Do you want to see the message even though it only appears once?
> >>
> >> I mean, maybe it should just be a debug? Are these messages useful for anyone
> >> other than the developers?
> >
> > Yes, this could just be a debug. However, developers normally don't turn on
> > debug mask, so using rtw89_info is to clearly remind developers to pay
> > attention on this lack of C2H handler. And, I suppose developers must handle
> > this when they see flooding messages.
>
> Well, regular users get this too. It is really unnecessary to print
> thousands of messages when they are completely benign.
>
> >>
> >> Maybe we should just print only the very first unsupported message at info level
> >> and print the rest at debug.
> >
> > I'm afraid developers will ignore or miss the messages. To reduce messages
> > is fine to me , but more important is to look up vendor driver to see if
> > the C2H handler is necessary.
>
> OK, so we should print exactly once for each class/func.
>
I sent a patchset [1] to print exactly once, but I don't add bitmap for every
printed class/func, because it seems to be unnecessary to add much code to
handle the case which should be handled during development.
Please help to try if this patchset can resolve thousands of message in your
side.
[1] https://lore.kernel.org/linux-wireless/20250804012234.8913-1-pkshih@realtek.com/T/#t
Powered by blists - more mailing lists