[sane-devel] saned issues for remote scanner access

[Henning Meier-Geinitz]

Hi,

On Sun, Feb 22, 2004 at 09:30:47PM +0100, Julien BLACHE wrote:
> "Jim George" <jim.george at blueyonder.co.uk> wrote:
> 
> > [saned] process_request: waiting for request
> > [saned] process_request: got request 8
> > [saned] decode_handle: w->status: 0, h: 0, num_handles: 0
> > [saned] decode_handle: cancel: error while decoding handle argument (h=0, Success)
> 
> Ok, now that is interesting. How can num_handles fall to 0, although
> there was a successful sane_open() before that ?

Error because assumption is wrong (I think) :-)

Let's have a look at the original log again:

[saned] process_request: got request 2
[saned] process_request: (open) strlen(resource) == 0
[gt68xx] sane_get_devices: start: local_only = true
[gt68xx] sane_get_devices: exit
[saned] process_request: access to resource gt68xx' granted
[saned] process_request: waiting for request

There is something wrong here. After "access to resource `gt68xx'
granted" there should be a log entry from the gt68xx backend.
Something like:

[gt68xx] sane_open: start (devicename = libusb:001:004')
[gt68xx] sane_open: found mustek-bearpaw-1200-cu' in devlist
[gt68xx] gt68xx_device_open: enter: dev=0x80812e8
[...]

So sane_open() wasn't ever called in the gt68xx backend. I even think
that sane_open() returned with an error (from the dll backend, because
no backend reported any scanner).

So this code returns SANE_STATUS_INVALID:

       reply.status = sane_open (name, &be_handle);

And therefore get_handle is never called.        

So it looks like an error of the frontend/client to call
SANE_NET_CANCEL with an invalid handle.

saned shouldn't crash anyway.

I think the scanner was just not detected on the server by saned for
some reason (permission problem?).

Bye,
  Henning