[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200325134147.GB32284@localhost>
Date:   Wed, 25 Mar 2020 06:41:47 -0700
From:   Richard Cochran <richardcochran@...il.com>
To:     "Y.b. Lu" <yangbo.lu@....com>
Cc:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "David S . Miller" <davem@...emloft.net>,
        Vladimir Oltean <vladimir.oltean@....com>,
        Claudiu Manoil <claudiu.manoil@....com>,
        Andrew Lunn <andrew@...n.ch>,
        Vivien Didelot <vivien.didelot@...il.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        Alexandre Belloni <alexandre.belloni@...tlin.com>,
        Microchip Linux Driver Support <UNGLinuxDriver@...rochip.com>
Subject: Re: [PATCH 6/6] ptp_ocelot: support 4 programmable pins
On Wed, Mar 25, 2020 at 03:08:46AM +0000, Y.b. Lu wrote:
> The calling should be like this,
> ptp_set_pinfunc (hold pincfg_mux)
> ---> ptp_disable_pinfunc
>    ---> .enable
>       ---> ptp_find_pin (hold pincfg_mux)
I see.  The call
    ptp_disable_pinfunc() --> .enable()
is really
    ptp_disable_pinfunc() --> .enable(on=0)
or disable.
All of the other drivers (except mv88e6xxx which has a bug) avoid the
deadlock by only calling ptp_find_pin() when invoked by .enable(on=1);
Of course, that is horrible, and I am going to find a way to fix it.
For now, maybe you can drop the "programmable pins" feature for your
driver?  After all, the pins are not programmable.
Thanks,
Richard
Powered by blists - more mailing lists
 
