On Wed, 27 May 2009 10:38:37 -0600, Matthew Wilcox wrote: > On Wed, May 27, 2009 at 06:30:05PM +0200, Jean Delvare wrote: > > > > +#if defined CONFIG_X86 && (defined CONFIG_HWMON || defined CONFIG_HWMON_MODULE) > > > > > > I don't like this. It goes against the principle that enabling a module > > > shouldn't cause the base kernel to get rebuilt. Is there a reason that > > > the hardware monitoring code can't contain this quirk instead of the > > > generic PCI code? > > > > No reason other than the fact that putting it in pci/quirks.c was > > easier. I can certainly move the quirk to hwmon/hwmon.c, but as far as > > I can see you have the exact same problem there: enabling or disabling > > PCI support will cause the base hwmon module to be rebuilt. If hwmon is > > built into the kernel then you have to relink the kernel. > > Trust me, if you enable or disable PCI support, you already have to > relink the kernel ;-) I can't really disagree... I'll send an updated patch, to be merged through the hwmon tree this time. -- Jean Delvare
