On 29 September, 2015 - Thiago Macieira wrote:
> On Tuesday 29 September 2015 21:12:55 Anton Lundin wrote:
> > This adds the QIODevice::Unbuffered flag to our rfcomm connections to
> > bypass the buffering layer in QIODevice.
> >
> > This fixes so firmware upgrades work against the OSTC Sport.
>
> Both patches look good.
>
> The unbuffered mode should not have made a difference. If it does, it's
> probably
> a bug somewhere (the firmware receiving the update or QtBluetooth).
>
While playing around with this i tried all kinds of different tricks.
One that mostly worked was in qt_serial_write having a block that
looked something like:
if (device->socket->bytesToWrite() > 0) {
QEventLoop loop;
QTimer timer;
...
timer.start(device->timeout * 4); // 4 is the magic number
...
}
Another one was spraying qDebug() << QByteArray((char *) data ....) in
lots of places.
Another one was :
- rc = device->socket->write((char *) data + nbytes, size -
nbytes);
+ rc = device->socket->write((char *) data + nbytes, size -
nbytes > 16 ? 16 : size - nbytes);
All tree made things better to various degrees, but unbuffered made it
100%
All tests where made on 15.04 and different 4.2/4.3 rc kernels. I even
rebooted into the ubuntu 3.19.0-28-generic kernel to verify that there
were no wierdness in my kernel either.
Bluez 4.101-0ubuntu25 and Qt 5.5.0 binaries from qt.io.
//Anton
--
Anton Lundin +46702-161604
_______________________________________________
subsurface mailing list
[email protected]
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface
