[platform] Fix use of fabs in constexpr context

src/modm/math/utils/misc.hpp

@@ -18,6 +18,7 @@
 #include <cstddef>
 #include <cmath>
 #include <stdint.h>
+#include <type_traits>
 
 #include <modm/architecture/utils.hpp>
 
@@ -161,6 +162,20 @@ max(const T& a, const T& b, Compare compare)
 		return a;
 }
 
+/**
+ * @brief constexpr implementation of fabs
+ */
+template <typename Float>
+    requires std::is_floating_point_v<Float>
+constexpr Float constexpr_fabs(Float number)
+{
+    if (number >= 0) {
+        return number;
+    } else
+        return -number;
+    }
+}
+
 /// @}
 
 }	// namespace modm

src/modm/platform/can/common/can_bit_timings.hpp

@@ -17,6 +17,7 @@
 
 #include <modm/architecture/interface/clock.hpp>
 #include <modm/math/units.hpp>
+#include <modm/math/utils/misc.hpp>
 #include <cmath>
 
 namespace modm
@@ -77,7 +78,7 @@ class CanBitTiming
 		for(uint8_t bs1Bs2 = minBs1Bs2; bs1Bs2 <= maxBs1Bs2; ++bs1Bs2) {
 			float idealPrescaler = float(Clk) / (Bitrate * (1 + bs1Bs2));
 			uint32_t intPrescaler = round_uint32(idealPrescaler);
-			float error = fabs(1 - intPrescaler/idealPrescaler);
+			float error = constexpr_fabs(1 - intPrescaler/idealPrescaler);
 			if(error <= minError) {
 				bestPrescaler = intPrescaler;
 				minError = error;