Re: MM 0.6.0 conflicts with USB GPS device?

[poma <pomidorabelisima gmail com>]

On 07.06.2013 23:28, Dan Williams wrote:
> On Fri, 2013-06-07 at 18:26 +0200, Aleksander Morgado wrote:
> > On 06/07/2013 06:16 PM, poma wrote:
> > > > > Maybe in the very distant future, MM would not probe the serial
> > > > > > > device at all - instead I imagine a "serial detector" software
> > > > > > > component, that does all kind of probing and only sends signals on
> > > > > > > the DBus about the result ("modem found at /dev/ttyUSB0", "GPS
> > > > > > > found at /dev/ttyUSB1", "voltmeter found...").
> > > > >
> > > > > In MM git master, USB<->RS232 adaptors are never probed by default when
> > > > > detected, so there shouldn't be any direct conflict. Instead, they will
> > > > > only be probed to look for modems if Scan() is called explicitly.
> > > Are you saying that there is no more need for acrobatics[1] regarding
> > > analog modems and UPSes connected to the serial ports or via adapters.
> >
> > That should be the case, yes.
> >
> > Platform RS232 ports need to be explicitly tagged with
> > ID_MM_PLATFORM_DRIVER_PROBE to get automatically probed; and USB devices
> > listed in the greylist of USB<->Serial devices won't get automatically
> > probed by MM, only when Scan() is called.
>
> That said, if some UI decided to add a "Look for all modems" button,
> then you'd still have problems if a non-modem was connected to one of
> the greylisted serial adapters.  It doesn't mean MM won't touch them, it
> just means that MM won't touch them *by itself*.  But if there's some UI
> or some script that tells MM to do so, then MM will because it was told
> to.

OK, that can be overcome via 'ID_MM_DEVICE_IGNORE' within a custom rule.
As for this device,
http://www.koutech.com/proddetail.asp?linenumber=56[1]
'ID_MM_DEVICE_IGNORE' *must* be applied, if we want to prevent scanning
connected ups and nut-driver/server breakage.
And all this as a result of
/lib/udev/rules.d/80-mm-candidate.rules
SUBSYSTEM=="tty", ENV{ID_MM_CANDIDATE}="1"
Therefore 'ID_MM_PLATFORM_DRIVER_PROBE' is redundant, in the first
place, in this case.

mmcli --version
mmcli 0.7.991


poma


[1]
DEVPATH=/class/pci_bus/0000:01/device/0000:01:06.0
DRIVER=serial
ID_MODEL_FROM_DATABASE=1P2S
ID_PCI_CLASS_FROM_DATABASE=Communication controller
ID_PCI_SUBCLASS_FROM_DATABASE=Communication controller
ID_VENDOR_FROM_DATABASE=MosChip Semiconductor Technology Ltd.
MODALIAS=pci:v00009710d00009835sv00001000sd00000012bc07sc80i02
PCI_CLASS=78002
PCI_ID=9710:9835
PCI_SLOT_NAME=0000:01:06.0
PCI_SUBSYS_ID=1000:0012
SUBSYSTEM=pci
USEC_INITIALIZED=22068