Messages in this thread | | | Date | Fri, 20 Sep 2013 18:34:34 +0200 | Subject | Re: Regression on cpufreq in v3.12-rc1 | From | Linus Walleij <> |
| |
On Fri, Sep 20, 2013 at 5:40 PM, Linus Walleij <linus.walleij@linaro.org> wrote: > On Fri, Sep 20, 2013 at 5:32 PM, Srivatsa S. Bhat > <srivatsa.bhat@linux.vnet.ibm.com> wrote: >> On 09/20/2013 08:51 PM, Linus Walleij wrote: > >>> sa11x0_pcmcia_init() which starts this chain of events is called as >>> an fs_initcall(), see drivers/pcmcia/sa1100_generic.c >> >> But fs_initcall() comes after arch_initcall() right? (looking at >> include/linux/init.h). Since the corresponding cpufreq-driver (drivers/ >> cpufreq/sa1100-cpufreq.c) uses arch_initcall() to register itself, it >> is all the more surprising how this could happen... > > Hm it does happen if the cpufreq driver does not probe *at all* > right? I will look closer at this, maybe it just doesn't register > properly for some SA1100 variants...
Aha yeah this is it. The h3600 has an SA1110 and uses drivers/cpufreq/sa1110-cpufreq.c.
This driver needs to know the RAM mounted on the board, if it is this or that type, to perform properly.
This is passed from the command line or a default is selected from the machine type:
if (!name[0]) { if (machine_is_assabet()) name = "TC59SM716-CL3"; if (machine_is_pt_system3()) name = "K4S641632D"; if (machine_is_h3100()) name = "KM416S4030CT"; if (machine_is_jornada720()) name = "K4S281632B-1H"; if (machine_is_nanoengine()) name = "MT48LC8M16A2TG-75"; }
In this case we have neither. So the cpufreq driver fails to register. But the kernel used to survive in any case, as you could call cpufreq_get() without a cpufreq driver registered.
I think your suggested patch fixes the issue.
As a proper fix I'd like to add a default RAM type for the h3600 but that is for later...
Yours, Linus Walleij
| |