adjust cpu power for secondary threads on POWER6
in [Kernel]
|
Prev: Oops in NFS (RHEL4, but also in kernel bugzilla)
Next: Is configfs the right solution for configuration based fs?
From: Olof Johansson on 18 Jun 2008 23:10 Hi, > +static unsigned int pseries_cpu_power(int cpu, unsigned int > default_power) > +{ > + struct device_node *np; > + unsigned int thread, power; > + > + if (!cpu_has_feature(CPU_FTR_ASYM_POWER)) > + return default_power; > + > + power = default_power; Why not just NULL out the cpu_power function pointers on platforms that don't have the feature bit instead? (or the other way around: NULL by default, and set only on platforms that have imbalanced threads. It'd be even nicer to do so without using up a cpu feature bit, since it'd then only be used in determining whether to set the ppc_md function pointer or not, but I don't know if there's a convenient way to probe for this at runtime (cpu device tree node properties?). -Olof -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Olof Johansson on 18 Jun 2008 23:20
On Jun 18, 2008, at 9:58 PM, Olof Johansson wrote: > Why not just NULL out the cpu_power function pointers on platforms > that don't have the feature bit instead? (or the other way around: > NULL by default, and set only on platforms that have imbalanced > threads. Nevermind; bogus optimization. I just noticed that it's only used when setting up the sched domains, and isn't adding path length later on. :) -Olof -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ |