Re: Re: [Kernel-bugs] h5400 patch - merge h5400_bt.c back to h5400.c

Hello,

On Mon, 18 Feb 2008 10:49:37 -0800
Erik Hovland <erik_at_hovland.org> wrote:

> On Mon, Feb 18, 2008 at 08:26:55PM +0200, Paul Sokolovsky wrote:
> > I personally don't have clear picture how HWUART is better BTUART.
> > IIRC, HWUART allows for higher speed range, but even BTUART should
> > be able to do 900K. So, finding 10mins to read PXA manual may be
> > beneficial here (and the whole matter of figuring this out should
> > be big fun ;-) ).
>
> The bluetooth device in the h5xxx iPAQs is basically flawed. So
> using the HWUART is not a complete solution to the problems found
> there. But empirical evidence provided by past developers shows that
> the hardware flow control provided by HWUART does improve the
> stability of bluetooth usage.

Aha, so that's where BTUART and HWUART differ: HWUART provides
"autoflow" feature which is kinda allows it to pause FIFO spooling on
its own, driving RTS/reacting to CTS. PXA255 DevMan, Section 17.4.3. I
checked that Linux enables that mode on ttyS3 ;-). Still, not sure if
"full" or "half" mode per that section would be used, it is apparently
can be controlled by userspace via ioctl, but no special control of
that in the driver.

Bottom line: it's worth to detect HWUART, but fortunately it's as easy
as checking if /dev/ttyS3 exists and if not, falling back to /dev/ttyS1.

>
> So if the HWUART can be used, it should be. And since it is physically
> wired to the same pins, there is little impact to the user.
>
> Since this is still work in progress, I have no problem with accepting
> it and making BTUART the default. But at some time the issue of
> bluetooth stability will come up and it will have to addressed.
>
> E
>
> --
> Erik Hovland
[]

-- 
Best regards,
 Paul                          mailto:pmiscml_at_gmail.com