Hi Ernst,

--- In [email protected], "Ernst Mayerhofer"
<[EMAIL PROTECTED]> wrote:
>
> Hi Geert,
> 
> After I read your email I think of course you're right. ;-)
> 
> My english is actually very poor and I cannot express myself very well. 

That's no problem.  English is my mother tongue neither, so I have the
same "handicap"...

> But what I don't understand - if you use it dynamically, why you have to
> write the node only once? What's about if you get at every startup a
> different major number?

Well, since I'm taking the "dynamical" approach, the node can indeed
be "recreated" every time the kernel boots.  And that's what I want.
Hence, I've seriously thought about moving the I2C device driver node
into /var.  And that's what I've done now (see my wiki page regarding
this subject:
http://www.gevawebsolutions.com/wiki/index.php?title=FB_General_Issues#Adding_a_dynamic_MAJOR_number:_the_steps_to_take...).
 It's working perfect and since we have anyhow quite a bit of ram,
that should not harm too much...
Next to this, I'm also not writing to flash any more (which I was
doing when using /etc/dev instead...), I'm working purely in RAM.
And yes, I know /var/dev is not such a trivial place for a device
driver, but since I'm using the trick with the link from /dev/i2c to
wherever the real node is located, the user is not "annoyed" with
this, he simply "doesn't know/doesn't have to know" where the real
node is.  He can still use open( "/dev/i2c"...

Best rgds,

--Geert

Reply via email to