Why call calibrate_delay() in smp.c: secondary_start_kernel()

2011/1/18 Santosh Shilimkar [off-list ref]:

+config ARCH_SKIP_SECONDARY_CALIBRATE
+ ? ? ? bool "Skip secondary CPU calibration"
+ ? ? ? depends on SMP
+ ? ? ? help
+ ? ? ? ? On some architectures, secondary cores shares clock with
primiary
+ ? ? ? ? core and hence scale together. Hence secondary core lpj
calibration
+ ? ? ? ? is not necessary and can be skipped to save considerable time.
+
+ ? ? ? ? If unsure, say n.

Why do you want to select that manually? Surely the kernel
writer should know whether the cores are clocked together
or not. Just:

config ARCH_HAS_COMMON_CORES_CLOCK
      bool
      depends on SMP

This will be default "n" so archs that want to flag to the
build system that the core clock is common can e.g.:

config ARCH_U8500
        bool "ST-Ericsson U8500 Series"
        select CPU_V7
+       select ARCH_HAS_COMMON_CORES_CLOCK

Looks reasonable? (And feel free to add that to U8500
as part of the patch.)

?/*
+ * Skip the secondary calibration on architectures sharing clock
+ * with primary cpu. Archs can use ARCH_SKIP_SECONDARY_CALIBRATE
+ * for this.
+ */
+static inline int skip_secondary_calibrate(void)

bool? Returning an error code seem overdesigned for a function
named like that.

Also the name is misleading, if you look at how you
use it, it should be named "do_not_skip_secondary_calibrate()".

+{
+#ifdef CONFIG_ARCH_SKIP_SECONDARY_CALIBRATE
+ ? ? ? return 0;

return false;

+#else
+ ? ? ? return -ENXIO;

return true;

+#endif
+}

Change sematics of that function and
if (!skip_secondary_calibrate())
  ...

? ? ? ?smp_store_cpu_info(cpu);

...and thanks for driving this!
Linus Walleij