On Sun, May 13, 2007 at 05:56:48PM +0300, Roman Kyrylych wrote: > 2007/5/13, Marc Deop i Argemí <[EMAIL PROTECTED]>: > > On Sunday 13 May 2007, Roman Kyrylych wrote: > > > I'm thinking about kernel26laptop with all fine-tuning for laptops > > > (more than just kernel26suspend2) > > > > Wow, good idea! If it's serious i'm willing to help! > > I didn't said I'm going to do this. :-) > I could but I don't have a laptop. :-( > I'm just saying such kernel would be nice, as I see that most > kernel-related issues are with laptops. > Enabling some features, like those mentioned in this thread is not a > good idea for default kernel, > but might help on laptops, where speed is less important than power saving\ > (those who want speed could always use default kernel). > I think such kernel could be based on kernel26suspend2 + additional > patches & config options. > to be honest. I think it's a stupid idea.
So far, all you have lined up, are two, very very dubious config changes. We'll start with the CONFIG_USB_SUSPEND: - It breaks some drivers. - It can cause problems with suspend, and can cause a laptop to wake up from s2ram - 1w is nothing. It's not worth saving given the above downsides, that could even cost more battery - You could save the aforementioned 1w, just by unloading your usb driver. Far better solution than a kludgy config option. As for CONFIG_TIMER_STATS - The kernel developers state this has a significant overhead. When they say significant, they mean significant. - All config_timer_stats adds is extra debuggin info for power top, a once off only, analysis tool. Powertop saves no power, it *uses* power. Powertop only serves to find where the power is going. - Given the above, you don't need to run power top often. Maybe monthly, most likely less frequently, if ever. Is this really worth building and packaging a second kernel and modules (18 at my last count) that uses more power, and reduces battery life? Put a kernel only with an adapted config in the AUR. Any more than that is overkill. If you have other *reasonable* patches, bug tracker them and request inclusion in the stock kernel. They usually get included, and the other kernels should track them too. Don't second guess the kernel developers, they know more than you. If they strongly suggest you don't enable something, or say that it has overhead, maybe you ought to listen. James _______________________________________________ arch mailing list [email protected] http://archlinux.org/mailman/listinfo/arch
