On Wed, Jan 08, 2020 at 05:20:41PM +0000, Mark Rutland wrote:

On Tue, Jan 07, 2020 at 06:35:39PM +0000, Mark Brown wrote:

quoted

+	/* Add multiple values to mirror the generic code. */
+	for (i = 0; i < 16; i++)
+		if (__arm64_rndr(&val))
+			add_device_randomness(&val, sizeof(val));
+}

Given the next patch also needs to check the I reckon it is worth
factoring the ID register check into a helper:

/* Used on the boot CPU before the CPU feature framework is up */
static inline bool __init __early_cpu_has_rndr(void)
{
	unsigned long ftr = read_sysreg_s(SYS_ID_AA64ISAR0_EL1);
	return (ftr >> ID_AA64ISAR0_RNDR_SHIFT) & 0xf;
}

There was also some suggestion about making a generic helper that takes
the register and bitfield to look at.

At some point I do also start to question if we're really winning with
all this stuff static inline...