Bug #1807: sensors with kernel 2.4, but not with kernel 2.6

Jean Delvare khali at linux-fr.org
Mon Nov 8 10:31:04 CET 2004


Hi again Mark,

>After working on a patch for a bit... now I hate it.  Why does i2c_detect()
>need to work that way?
>
>Back to where I started: anyone mind if I rewrite i2c_detect()?  A grep
>said that drivers/i2c/chips/*.c are the only callers.

Fine with me as long as it is done in a consistent way, and the
documentation is updated accordingly (both writing- and porting-clients).

As a side note, such a change is not suitable for 2.4/CVS, so fixing the
drivers you listed in your other post there is still on the to-do list.

As yet another side note, I have noticed that i2c_detect checks all
addresses from 0 to 0x7f for business before processing them. This
sounds bad in terms of efficiency. The test should be done before
actually attempting to detect chips at any given address.

Thanks,
Jean



More information about the lm-sensors mailing list