cmsis-dsp: restrict f16 support to targets with f16 arithmetic

CMSIS-DSP f16 sources use _Float16 arithmetic, which requires MVE float
on Cortex-M processors. So far, CONFIG_CMSIS_DSP_FLOAT16 only depended
on CONFIG_FP16, which controls the storage format and is enabled on
all ARM targets. This caused build failures on Cortex-M55 without MVE
when -mfpu=fpv5-sp-d16 is used.

Making CONFIG_CMSIS_DSP_FLOAT16 depend on the newly introduced
CONFIG_FP16_ARITHMETIC ensures that the hardware instructions for
arithmetic operations used by the DSP library are present.

Signed-off-by: Martin Jäger <martin.jaeger@a-labs.io>

Author: martinjaeger

modules/cmsis-dsp/Kconfig

@@ -308,7 +308,7 @@ config CMSIS_DSP_AUTOVECTORIZE
 config CMSIS_DSP_FLOAT16
 	bool "Half-Precision (16-bit Float) Support"
 	default y
-	depends on FP16
+	depends on FP16_ARITHMETIC
 	help
 	  This option enables the half-precision (16-bit) floating-point
 	  operations support.