[lkml]   [2004]   [Dec]   [27]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    Subject[RFC] I2C: Remove the i2c_client id field
    Hi Greg, hi all,

    While porting various hardware monitoring drivers to Linux 2.6 and
    otherwise working on i2c drivers in 2.6, I found that the i2c_client
    structure has an "id" field (of type int) which is mostly unused. I am
    not exactly sure why it was introduced in the first place, and since the
    i2c subsystem code was significantly reworked since, it might not
    actually matter.

    Most hardware monitoring drivers allocate a unique (per driver) id
    through an incremented static global variable, and never use it. Some
    (lm85 and most notably adm1026) use the value in debug messages. I saw
    various video drivers appending the id value to the client name between
    square brackets, while others would set the id field to -1 and then
    leave it alone. The i2c core itself doesn't use this field.

    Using this field to identify a client doesn't make much sense to me, for
    the following reasons:

    1* A client is already uniquely identified by the combination of the
    number of the bus it sits on and the address it is located at on this

    2* With the implementation described above, the id will possibly change
    depending on which i2c bus drivers are loaded and the order they were
    loaded in. As a consequence, you can't rely on its value from
    user-space, and its usability in kernel-space isn't obvious either.

    3* As a matter of fact, no driver in the kernel tree uses this field
    except for debugging (and even then with no obvious benefit), with only
    a few exceptions where I could easily change the code so it wouldn't
    need this field anymore.

    Thus, I propose that we simply get rid of this field, so as to save some
    memory space and kill some useless code. If anyone really ever needs to
    carry some sort of id attached to an i2c_client structure, this is
    private data and can be added to whatever structure the data field is
    pointing to for this particular driver.

    Unless someone objects with valid reasons, I am going to send patches to
    kill the i2c_client id field. I have everything ready, but don't know
    exactly how I should send them. The difficulty comes from the relatively
    large number of affected drivers (50) and the fact that they spread over
    very different subsystems (the big ones are i2c and media/video, plus
    half a dozen drivers in acorn/char, macintoch and sound).

    Greg, can you tell me if you would take such a patch, and how I would
    have to split it for you to accept it?

    Jean Delvare
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    More majordomo info at
    Please read the FAQ at

     \ /
      Last update: 2005-03-22 14:09    [W:0.022 / U:21.012 seconds]
    ©2003-2017 Jasper Spaans. hosted at Digital OceanAdvertise on this site