Hi Guenter, On Sat, 23 Jun 2012 09:02:36 -0700, Guenter Roeck wrote: > On Mon, Jun 18, 2012 at 12:07:25PM +0200, Jean Delvare wrote: > > Hi Guenter, > > > > On Wed, 13 Jun 2012 13:56:26 -0700, Guenter Roeck wrote: > > > The following compile warning may be seen if the driver is compiled with > > > -Wuninitialized: > > > > > > drivers/hwmon/w83781d.c: warning: 'sc_addr[1]' may be used uninitialized in this > > > function [-Wuninitialized] > > > > > > While this is a false positive, it is annoying in nightly builds, and may help > > > to conceal real problems. The current code is quite tricky, and and it is easy > > > to rearrage the code to make the problem disappear. So fix it. > > > > I don't see this warning here (gcc 4.6.2.) While I see why it can be > > reported, I don't really understand why your change makes it disappear, > > as you still depend on the value of a local variable, which could have > > changed between the moment you set sc_addr[1] and the moment you use > > it. I'm curious why the compiler is able to keep track of the value in > > one case and not in the other case. > > > > Out of curiosity, have you tried defining kind as const? If it makes > > the compiler equally happy, I think it would be a better way to solve > > this warning. > > did you have time to look into this some more ? Statistically, I see the warning > in about one out of ten randconfig builds. No I did not, and wont. My time is better spent on other tasks. -- Jean Delvare _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
