Roger Binns
2003-12-09 06:00:16 UTC
Summary:
The VX4400 phone itself has an issue with its
USB interface that causes timeouts. This happens
with any straight USB cable on any platform. The
only drivers that work around this issue are produced
by LG Electronics (LGE) which are only available for
Windows, and only as part of Verizon's Mobile Office
Kit.
I WILL BE COMPLETELY IGNORING ALL SUPPORT ISSUES,
HELP REQUESTS ETC FROM ANYONE USING A STRAIGHT
USB CABLE THAT IS NOT USING THE LGE DRIVER.
Detail:
The phone itself will often stall when reading data
from its USB interface. This is an issue with the
phone itself. This issue is not present in the
VX6000.
After some investigation, I have found what I believe
the method LGE uses to get around the problem in their
own drivers. Unfortunately that method can only
be done inside drivers. Even on Linux, it appears
to require kernel drivers.
The actual method is that the LGE drivers send a
read USB request to the phone and leave it outstanding
before sending a write. They then get the write
confirmation back, followed by the response to the
read request containing the data. It would appear that
pre-emptive reads before writes prevent the read
timeouts. I have not been able to cause the same
sequence of USB requests from user space code.
It appears that only kernel drivers can send the
necessary requests in the necessary order.
Roger
The VX4400 phone itself has an issue with its
USB interface that causes timeouts. This happens
with any straight USB cable on any platform. The
only drivers that work around this issue are produced
by LG Electronics (LGE) which are only available for
Windows, and only as part of Verizon's Mobile Office
Kit.
I WILL BE COMPLETELY IGNORING ALL SUPPORT ISSUES,
HELP REQUESTS ETC FROM ANYONE USING A STRAIGHT
USB CABLE THAT IS NOT USING THE LGE DRIVER.
Detail:
The phone itself will often stall when reading data
from its USB interface. This is an issue with the
phone itself. This issue is not present in the
VX6000.
After some investigation, I have found what I believe
the method LGE uses to get around the problem in their
own drivers. Unfortunately that method can only
be done inside drivers. Even on Linux, it appears
to require kernel drivers.
The actual method is that the LGE drivers send a
read USB request to the phone and leave it outstanding
before sending a write. They then get the write
confirmation back, followed by the response to the
read request containing the data. It would appear that
pre-emptive reads before writes prevent the read
timeouts. I have not been able to cause the same
sequence of USB requests from user space code.
It appears that only kernel drivers can send the
necessary requests in the necessary order.
Roger