> Date: Sun, 30 Mar 2008 13:54:09 +0200
> From: [EMAIL PROTECTED]
> To: [EMAIL PROTECTED]
> CC: [email protected]
> Subject: Re: [i2c] i2c Questions
>
> Hi Christian,
>
> On Sat, 29 Mar 2008 22:58:43 +0000, badguy development wrote:
> > > On Thu, 27 Mar 2008 21:14:53 +0000, badguy development wrote:
> > > > I have to now develop a driver for bus address 0x01
> > > > and in the code I have located that to be an invalid address.
> > > >
> > > > the code tests for address < 0x03 && address > 0x77
> > > > then promptly fails with invalid address.
> > > >
> > > > I am hoping to find an understanding for this.
> > > > this is a programmable address, and I will have it changed if I have
> > > > to..
> > >
> > > Please read the I2C specification. Address 0x01 is for CBUS
> > > compatibility and cannot be used by I2C devices.
> >
> > Thank you for the CBUS tip, that is what lead me to it in the spec.
> > I guess I will have to continue digging to learn why 0x02 is also
> > disallowed.
>
> 0x02 is a bit of mystery to me as well, the specification doesn't
> prohibit its use completely, but you can only use it for devices that
> are compatible with "other protocols" but they don't say which ones.
> 0x03 is mysterious as well, it's reserved for "future purposes" since
> I2C was defined (two dozen years ago.) 0x04-0x07 is for "HS-mode master
> code", whatever it is. 0x08 is used frequently as the SMBus slave
> address, 0x09-0x0b for battery stuff, and 0x0c for SMBus alert
> response. So my advice would be to stay away of the whole 0x00 row. If
> you can program the address, 0x10 would be a sane choice I think, if no
> other device on your system needs it.
>
Ok Thanks for the knowledge here.
> > > > Also, I have 3 i2c busses on this iMX31 SOM
> > > >
> > > > how can I tell my device to only query / operate on a specific bus.
> > > > do I need to?
> > >
> > > The device can only operate on the one bus to which it is connected, so
> > > I can't see any problem here.
> > >
> >
> > I guess I did not use correct terms there.
> > What I meant to ask was: how can I tell my device driver to only query a
> > given bus.
> > but I also guess that is not necessary, and the first bus to respond on
> > that address will be carrying my device.
>
> Depends on whether you write a legacy i2c driver (with attach_adapter
> and detach_client methods) or a new-style one (with probe and remove
> methods.)
>
> If you write a legacy i2c driver (discouraged), you'll call i2c_probe()
> in your attach_adapter method, and it will call you back only if a
> device is found at the specified address(es). If your device can only
> live at one address and only one bus has a device responding at this
> address, then the driver will only attach to this device on that bus.
> If you are unlucky and you have other devices using the same address on
> the other buses, you have to add some detection code in your i2c_probe
> callback, otherwise your driver will incorrectly attach to the other
> devices. This can be very tricky if the devices don't have
> identification registers.
>
> If you write a new-style i2c driver (encouraged), you first number your
> i2c buses, then you explicitly instantiate the i2c devices on the bus
> you know they are connected to. There is no probing or guessing
> involved, you just tell "I have device X on I2C bus N at address A" and
> you're done. Much cleaner, and faster too.
>
Well I now plan to write a new style as soon as my test spike code written
using legacy code is working :)
heh, I am using Kernel 2.6.19.2 is the new style available here?
and if so can you recommend some code I can use for an example?
> > > This is the right list. It would be nice if you could post under your
> > > real name.
> >
> > Thanks, I currently do not have an email that uses my name, its top
> > secret.
> > but I can place my name at the end of my messages.
> >
> > is this perhaps why I have not recieved confirmation for joining the list?
> > am I disallowed without an email revealing a more realistic name?
>
> No, mailman isn't smart enough to check this, and even if it was, there
> would be no point in doing this... Everyone can read the list archive,
> so putting restrictions on who can subscribe to the list would be silly.
>
> I have no idea why your subscription did not work. Please try again.
>
> > if that is the case I guess I will have to finally create an email account
> > using a name :(
>
> As far as I am concerned, it's just a matter of politeness. When I
> spend time helping people, I think it's fair that I know their name.
> Not to mention that you won't be able to contribute anything to the
> kernel without a real name.
>
> --
> Jean Delvare
Thanks for the help.
-Christian Hunt
_________________________________________________________________
Turn every day into $1000. Learn more at SignInAndWIN.ca
http://g.msn.ca/ca55/213_______________________________________________
i2c mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/i2c
