Nut upsdev - Sep 2007 - PowerWalker and megatec_usb

Hi!

I've (finally) decided to try and make my PowerWalker work with
megatec_usb. So far it seems the agiler subdriver is the way to go,
but it doesn't work. The initial detection process sometimes succeeds
in 1 out of 5 "Q1" commands, but that's about it (which is better
than
what the krauler subdriver can do... absolutely nothing).

So, what needs to be done to obtain more information? I don't seem to
get anything readable out of the "get_descriptor" tool. What command
line parameters should I use?

Be aware that I'm still almost completely ignorant about this USB stuff...

-- 
Carlos Rodrigues

On Thu, 13 Sep 2007 01:24:04 +0400, Carlos Rodrigues  
<carlos.efr at mail.telepac.pt> wrote:
> Hi!
>
> I've (finally) decided to try and make my PowerWalker work with
> megatec_usb. So far it seems the agiler subdriver is the way to go,
> but it doesn't work. The initial detection process sometimes succeeds
> in 1 out of 5 "Q1" commands, but that's about it (which is
better than
> what the krauler subdriver can do... absolutely nothing).
>
> So, what needs to be done to obtain more information? I don't seem to
> get anything readable out of the "get_descriptor" tool. What
command
> line parameters should I use?
>
> Be aware that I'm still almost completely ignorant about this USB  
> stuff...
>
I'll try to help, but I don't know much about agiler subdriver.
AFAIK it assumes that the device has a bogus HID descriptor (which
is to allow the device to describe it's own abilities. Still this info
could be obtained from specs/Windows drivers/etc.) So the driver
sends and receives chunks of data of a hard-coded size to a hard-coded
report descriptor.
So the "get_descriptor" tool can only help in retrieving the HID
descriptor and nothing more.

I've just commited a change intended to increase verbosity of the agiler
subdriver (it can dump all the traffic now). Please, run it with
'-DDDDD'.

P.S. If you have a Windows driver for your UPS it can help a lot.
How it's called?

-- 
   Alexander

> I've (finally) decided to try and make my PowerWalker work with
> megatec_usb. So far it seems the agiler subdriver is the way to go,
> but it doesn't work. The initial detection process sometimes succeeds
> in 1 out of 5 "Q1" commands, but that's about it (which is
better than
> what the krauler subdriver can do... absolutely nothing).
Having looked at this subdriver, I think this is duplicating what libusb
is already doing for us. Breaking up data in 8 byte chunks is probably not
needed and neither is setting the exact number of bytes you expect. So the
get_data_agiler() function could probably be reduced to something like

static int get_data_agiler(char *buffer, int buffer_size)
{
	return usb->get_interrupt(udev, (unsigned char *)buffer, buffer_size, 250);
}

Note the reduction in the timeout we set here. Although the USB 2.0
specification allows for up to 5 seconds for interrupt reports, this is
not what we want in NUT. Rather than keep waiting for information (because
the USB bus is apparently busy at the time), drivers should abort the
attempt to read data and retry the next time. This probably requires some
leniency in the upsdrv_updateinfo() function, so that it doesn't
immediately declare a UPS stale if it fails reading data a single time.
See for an example safenet.c, which allows for up to two consecutive
failures before declaring the data stale.
> So, what needs to be done to obtain more information? I don't seem to
> get anything readable out of the "get_descriptor" tool. What
command
> line parameters should I use?
Could you post the output of 'lsusb -v' for your UPS here?
> Be aware that I'm still almost completely ignorant about this USB
stuff...
I also was until a couple of weeks ago. You'll find that most of the low
level stuff will be dealt with by the libusb library, so it's probably
best to start reading there (see http://libusb.sourceforge.net/doc).

Best regards, Arjen
-- 
Eindhoven - The Netherlands
Key fingerprint - 66 4E 03 2C 9D B5 CB 9B  7A FE 7E C1 EE 88 BC 57