diff --git a/Makefile b/Makefile
index 65026da..8b21431 100644
--- a/Makefile
+++ b/Makefile
@@ -1003,6 +1003,9 @@
 # We could use PULLED_IN_LIBS, but that would be ugly.
 FEATURE_LIBS += $(call debug_shell,grep -q "NEEDLIBZ := yes" .libdeps && printf "%s" "-lz")
 
+FEATURE_CFLAGS += $(call debug_shell,grep -q "CLOCK_GETTIME := yes" .features && printf "%s" "-D'HAVE_CLOCK_GETTIME=1'")
+FEATURE_LIBS += $(call debug_shell,grep -q "CLOCK_GETTIME := yes" .features && printf "%s" "-lrt")
+
 LIBFLASHROM_OBJS = $(CHIP_OBJS) $(PROGRAMMER_OBJS) $(LIB_OBJS)
 OBJS = $(CLI_OBJS) $(LIBFLASHROM_OBJS)
 
@@ -1299,6 +1302,18 @@
 endef
 export LINUX_I2C_TEST
 
+define CLOCK_GETTIME_TEST
+#include <time.h>
+
+int main(int argc, char **argv)
+{
+	struct timespec res;
+	clock_gettime(CLOCK_REALTIME, &res);
+	return 0;
+}
+endef
+export CLOCK_GETTIME_TEST
+
 features: compiler
 	@echo "FEATURES := yes" > .features.tmp
 ifneq ($(NEED_LIBFTDI), )
@@ -1341,6 +1356,13 @@
 	@ { $(CC) $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) .featuretest.c -o .featuretest$(EXEC_SUFFIX) >&2 && \
 		( echo "found."; echo "UTSNAME := yes" >> .features.tmp ) ||	\
 		( echo "not found."; echo "UTSNAME := no" >> .features.tmp ) } 2>>$(BUILD_DETAILS_FILE) | tee -a $(BUILD_DETAILS_FILE)
+	@printf "Checking for clock_gettime support... " | tee -a $(BUILD_DETAILS_FILE)
+	@echo "$$CLOCK_GETTIME_TEST" >.featuretest.c
+	@printf "\nexec: %s\n" "$(CC) $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) -lrt .featuretest.c -o .featuretest$(EXEC_SUFFIX)" >>$(BUILD_DETAILS_FILE)
+	@ { $(CC) $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) -lrt .featuretest.c -o .featuretest$(EXEC_SUFFIX) >&2 && \
+		( echo "found."; echo "CLOCK_GETTIME := yes" >>.features.tmp ) || \
+		( echo "not found."; echo "CLOCK_GETTIME := no" >>.features.tmp ) } \
+		2>>$(BUILD_DETAILS_FILE) | tee -a $(BUILD_DETAILS_FILE)
 	@$(DIFF) -q .features.tmp .features >/dev/null 2>&1 && rm .features.tmp || mv .features.tmp .features
 	@rm -f .featuretest.c .featuretest$(EXEC_SUFFIX)
 
diff --git a/udelay.c b/udelay.c
index 7c6961d..3a44c16 100644
--- a/udelay.c
+++ b/udelay.c
@@ -21,13 +21,64 @@
 
 #ifndef __LIBPAYLOAD__
 
+#include <stdbool.h>
 #include <unistd.h>
+#include <errno.h>
 #include <time.h>
 #include <sys/time.h>
 #include <stdlib.h>
 #include <limits.h>
 #include "flash.h"
 
+static bool use_clock_gettime = false;
+
+#if HAVE_CLOCK_GETTIME == 1
+
+#ifdef _POSIX_MONOTONIC_CLOCK
+static clockid_t clock_id = CLOCK_MONOTONIC;
+#else
+static clockid_t clock_id = CLOCK_REALTIME;
+#endif
+
+static void clock_usec_delay(int usecs)
+{
+	struct timespec now;
+	clock_gettime(clock_id, &now);
+
+	const long end_nsec = now.tv_nsec + usecs * 1000L;
+	const struct timespec end = {
+		end_nsec / (1000 * 1000 * 1000) + now.tv_sec,
+		end_nsec % (1000 * 1000 * 1000)
+	};
+	do {
+		clock_gettime(clock_id, &now);
+	} while (now.tv_sec < end.tv_sec || (now.tv_sec == end.tv_sec && now.tv_nsec < end.tv_nsec));
+}
+
+static int clock_check_res(void)
+{
+	struct timespec res;
+	if (!clock_getres(clock_id, &res)) {
+		if (res.tv_sec == 0 && res.tv_nsec <= 100) {
+			msg_pinfo("Using clock_gettime for delay loops (clk_id: %d, resolution: %ldns).\n",
+				  (int)clock_id, res.tv_nsec);
+			use_clock_gettime = true;
+			return 1;
+		}
+	} else if (clock_id != CLOCK_REALTIME && errno == EINVAL) {
+		/* Try again with CLOCK_REALTIME. */
+		clock_id = CLOCK_REALTIME;
+		return clock_check_res();
+	}
+	return 0;
+}
+#else
+
+static inline void clock_usec_delay(int usecs) {}
+static inline int clock_check_res(void) { return 0; }
+
+#endif /* HAVE_CLOCK_GETTIME == 1 */
+
 /* loops per microsecond */
 static unsigned long micro = 1;
 
@@ -87,6 +138,9 @@
 
 void myusec_calibrate_delay(void)
 {
+	if (clock_check_res())
+		return;
+
 	unsigned long count = 1000;
 	unsigned long timeusec, resolution;
 	int i, tries = 0;
@@ -189,6 +243,8 @@
 	/* If the delay is >1 s, use internal_sleep because timing does not need to be so precise. */
 	if (usecs > 1000000) {
 		internal_sleep(usecs);
+	} else if (use_clock_gettime) {
+		clock_usec_delay(usecs);
 	} else {
 		myusec_delay(usecs);
 	}