Date: Wed, 10 Jan 2018 09:49:14 +0100
From: Johan Hovold <johan@kernel.org>
To: "Ji-Ze Hong (Peter Hong)" <hpeter@gmail.com>
Cc: Johan Hovold <johan@kernel.org>, gregkh@linuxfoundation.org,
        linux-usb@vger.kernel.org, linux-kernel@vger.kernel.org,
        peter_hong@fintek.com.tw,
        "Ji-Ze Hong (Peter Hong)" <hpeter+linux_kernel@gmail.com>
Subject: Re: [PATCH V2 1/5] usb: serial: f81534: add high baud rate support
Message-ID: <20180110084914.GV11344@localhost>
References: <1515032961-29131-1-git-send-email-hpeter+linux_kernel@gmail.com>
 <20180109110841.GN11344@localhost>
 <18d4171a-1e1a-5476-dc9d-02a522cbf077@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <18d4171a-1e1a-5476-dc9d-02a522cbf077@gmail.com>
User-Agent: Mutt/1.7.2 (2016-11-26)
Sender: linux-kernel-owner@vger.kernel.org

On Wed, Jan 10, 2018 at 01:30:18PM +0800, Ji-Ze Hong (Peter Hong) wrote:
> Hi Johan,
> 
> Johan Hovold 於 2018/1/9 下午 07:08 寫道:
> > On Thu, Jan 04, 2018 at 10:29:17AM +0800, Ji-Ze Hong (Peter Hong) wrote:
> >> The F81532/534 had 4 clocksource 1.846/18.46/14.77/24MHz and baud rates
> >> can be up to 1.5Mbits with 24MHz.
> >>
> >> This device may generate data overrun when baud rate setting to 921600bps
> >> or higher with old UART trigger level setting (8x14=112) with full
> >> loading. We'll change trigger level from 8x14=112 to 8x8=64 to avoid data
> >> overrun.
> >>
> >> Also the read/write of EP0 will be affected by this patch. The worst case
> >> of responding time is 20s when all serial port are full loading and trying
> >> to access EP0, so we change EP0 timeout from 10 to 20s.
> > 
> > Surely you meant 1 and 2 seconds respectively here? And if you have
> > indeed measured response times close to 2000 ms then perhaps you want to
> > add even more margin?
> 
> Normally, the communication with F81534 ep0 will take less than 1 sec
> (even only some milliseconds), but It maybe take much long time with
> huge loading with UART functional.
> 
> We had tested it on BurnInTest, 4 ports with 921600bps + MSR status
> check to perform huge loading test. The worst case to read MSR register
> via ep0 will take 15~18 seconds. So We'll still remain the max waiting
> time for access ep0 with 2x10=20s in high baud rate mode.

Wow, that's unfortunate. But note that your patch only doubles the
timeout to 2000 ms, that is, two seconds and not twenty:

-#define F81534_USB_TIMEOUT             1000
+#define F81534_USB_TIMEOUT             2000

If you really intended to increase this to twenty seconds, then please
do so in a separate (preparatory) patch where you describe why that is
needed (e.g. what you wrote above).

Johan