Wow, this is getting to be quite a spring-cleaning now. I have incorporated all your suggestions into this patch, except for these two: * Jean: use of val & 255 in printk statement in pcf_isa_setbyte() - I think this is actually required because parameter val is an integer quantity and could have been sign-extended? I would prefer to change the function parameter to u8, but this violates the types of struct i2c_algo_pcf_data. * Greg: I haven't changed the printk and pr_debug calls yet, because there are some printk calls before a struct device is initialised (this is done when the device is registered to i2c_algo_pcf at the end of driver init). What is the policy relating to driver output before there is an associated struct device initialised? I took a look at the code with sparse, there were no reported problems. However, am I right in thinking that sparse is x86-centric? I was running the compilation on an Alpha system. With more than a little irony considering the political harangue against gcc in the sparse documentation, it didn't compile on my RH8 x86 system: gcc gave an internal compiler error... The attached patch is for code cleanup and is a direct replacement for patch 2. Regards, Stig Telfer -------------- next part -------------- A non-text attachment was scrubbed... Name: elektor-alpha-3.patch Type: application/octet-stream Size: 5054 bytes Desc: not available Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20051006/196ffd32/attachment.obj
