Re: [PATCH] USB: serial: ftdi_sio: fix extreme low-latency setting

[Johan Hovold <johan@kernel.org>]

On Wed, Jan 25, 2017 at 04:42:01PM +0100, Greg Kroah-Hartman wrote:
> On Wed, Jan 25, 2017 at 04:14:36PM +0100, Johan Hovold wrote:
> > On Wed, Jan 25, 2017 at 04:06:47PM +0100, Greg Kroah-Hartman wrote:
> > > On Wed, Jan 25, 2017 at 03:35:20PM +0100, Johan Hovold wrote:
> > > > Since commit 557aaa7ffab6 ("ft232: support the ASYNC_LOW_LATENCY
> > > > flag") the FTDI driver has been using a receive latency-timer value of
> > > > 1 ms instead of the device default of 16 ms.
> > > > 
> > > > The latency timer is used to periodically empty a non-full receive
> > > > buffer, but a status header is always sent when the timer expires
> > > > including when the buffer is empty. This means that a two-byte bulk
> > > > message is received every millisecond also for an otherwise idle port as
> > > > long as it is open.
> > > > 
> > > > Let's restore the pre-2009 behaviour which reduces rate of status
> > > > messages to 1/16th (e.g. interrupt frequency drops from 1 kHz to 62.5
> > > > Hz) by not setting ASYNC_LOW_LATENCY by default.
> > > > 
> > > > Anyone willing to pay the price for the minimum-latency behaviour should
> > > > set the flag explicitly instead using the TIOCSSERIAL ioctl or a tool
> > > > such as setserial (e.g. setserial /dev/ttyUSB0 low_latency).
> > > > 
> > > > Note that since commit 0cbd81a9f6ba ("USB: ftdi_sio: remove
> > > > tty->low_latency") the ASYNC_LOW_LATENCY flag has no other effects but
> > > > to set a minimal latency timer.
> > > > 
> > > > Reported-by: Antoine Aubert <a.aubert@overkiz.com>
> > > > Fixes: 557aaa7ffab6 ("ft232: support the ASYNC_LOW_LATENCY flag")
> > > > Cc: stable <stable@vger.kernel.org>	# v2.6.31
> > > > Signed-off-by: Johan Hovold <johan@kernel.org>
> > > > ---
> > > > 
> > > > Greg,
> > > > 
> > > > I've been aware of this overhead for a while, but never realised it was
> > > > actually a regression introduced in 2009.
> > > > 
> > > > Fixing something like this after such a long time obviously means
> > > > risking a regression for anyone who is now relying on the new default
> > > > behaviour instead. I still think it's reasonable in this case to restore
> > > > the earlier behaviour given the penalty everyone else is paying for a
> > > > minimal-latency behaviour that they likely do not need or want.
> > > > 
> > > > Whether this should go to stable is a different question. Perhaps the
> > > > stable tag is not warranted, and this should just be the default
> > > > behaviour going forward? What do you think?
> > > 
> > > I think the stable tag is warrented here.  Do you want me to take this
> > > patch now into my usb-linus tree, or will you include it in a pull
> > > request?
> > 
> > I'll include it in a pull request. I was gonna apply it for -next, but
> > if you prefer I can send it along with some new device ids I have queued
> > up for 4.10-rc?
> 
> -next is probably good, it's not like there is a rush for it :)

True. Don't really wanna think about how much power has been wasted on
these insane interrupt rates considering the number of these devices
that are out there though...

Now applied for -next, with a stable dependency on patch already in my
tree which did not need to be backported while ASYNC_LOW_LATENCY was
set.

Thanks,
Johan