"I would hope that, unless there are specific reasons why
sungem_eth.device and via_rhinepci.device do not open in "online" mode,
these drivers will be updated to conform to the standard practice."
There is no clear specification whether driver should be online by default
or not. As such I will not change this behaviour. The correct solution is
for the TCP/IP stack (or whatever software using the SANA2 driver) to
explicitly S2_ONLINE the driver after S2_CONFIGURE has been done. Doing so
has no ill effects as if the device is online already it will just ignore
The reason applications using SANA2 should go online explicitly can be
tracked to ppp.device which does not go online automatically. In fact, for
example with AmiTCP/IP you were supposed to add manual bin/online call to
cover this. Example: http://www.amitrix.com/faqs.html
Later AmiTCP/IP authors decided to move away from startnet script as such,
and rather automate and consolidate this functionality to avoid the
cumbersome startnet script generation that was quite inflexible and hard
to maintain in an automated way (see AmiTCP installer script for some
"netconfig" command was added. It could do for example BOOTP
autoconfiguration. This command also covers the online-problem by always
taking the device online, unless if explicitly told not to do so by a
command line switch:
NOONLINE/S - Do not put the Sana2 device online.
Normally netconfig puts the Sana2
device configured for interface-name
online. This option causes the online
command not to be sent.
Miami/MiamiDX also puts the SANA2 device online automatically.
As such I'd say it wouldn't be totally uncalled for for the TCP/IP stack
to automagically S2_ONLINE the driver, nor should it be considered a
problem in a device driver if it does not go online by default.