Commit 67bdd7b0 authored by Nam Cao's avatar Nam Cao Committed by Paul Walmsley
Browse files

riscv: Split out measure_cycles() for reuse 



Byte cycle measurement and word cycle measurement of scalar misaligned
access are very similar. Split these parts out into a common
measure_cycles() function to avoid duplication.

This function will also be reused for vector misaligned access probe in a
follow-up commit.

Signed-off-by: default avatarNam Cao <namcao@linutronix.de>
Link: https://patch.msgid.link/50d0598e45acc56c95176e52fbbe56e1f4becc84.1770830596.git.namcao@linutronix.de


Signed-off-by: default avatarPaul Walmsley <pjw@kernel.org>
parent 6455c6c1

arch/riscv/kernel/unaligned_access_speed.c

+33 −36
Original line number Diff line number Diff line
@@ -31,30 +31,15 @@ static long unaligned_vector_speed_param = RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNO 
static cpumask_t fast_misaligned_access;



#ifdef CONFIG_RISCV_PROBE_UNALIGNED_ACCESS

static int check_unaligned_access(void *param)

static u64 measure_cycles(void (*func)(void *dst, const void *src, size_t len),

			  void *dst, void *src, size_t len)

{

	int cpu = smp_processor_id();

	u64 start_cycles, end_cycles;

	u64 word_cycles;

	u64 byte_cycles;

	u64 start_cycles, end_cycles, cycles = -1ULL;

	u64 start_ns;

	int ratio;

	struct page *page = param;

	void *dst;

	void *src;

	long speed = RISCV_HWPROBE_MISALIGNED_SCALAR_SLOW;



	if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN)

		return 0;



	/* Make an unaligned destination buffer. */

	dst = (void *)((unsigned long)page_address(page) | 0x1);

	/* Unalign src as well, but differently (off by 1 + 2 = 3). */

	src = dst + (MISALIGNED_BUFFER_SIZE / 2);

	src += 2;

	word_cycles = -1ULL;

	/* Do a warmup. */

	__riscv_copy_words_unaligned(dst, src, MISALIGNED_COPY_SIZE);

	func(dst, src, len);



	preempt_disable();



	/*
@@ -66,29 +51,41 @@ static int check_unaligned_access(void *param) 
		start_cycles = get_cycles64();

		/* Ensure the CSR read can't reorder WRT to the copy. */

		mb();

		__riscv_copy_words_unaligned(dst, src, MISALIGNED_COPY_SIZE);

		func(dst, src, len);

		/* Ensure the copy ends before the end time is snapped. */

		mb();

		end_cycles = get_cycles64();

		if ((end_cycles - start_cycles) < word_cycles)

			word_cycles = end_cycles - start_cycles;

		if ((end_cycles - start_cycles) < cycles)

			cycles = end_cycles - start_cycles;

	}



	byte_cycles = -1ULL;

	__riscv_copy_bytes_unaligned(dst, src, MISALIGNED_COPY_SIZE);

	preempt_enable();



	start_ns = ktime_get_mono_fast_ns();

	while (ktime_get_mono_fast_ns() < start_ns + MISALIGNED_ACCESS_NS) {

		start_cycles = get_cycles64();

		mb();

		__riscv_copy_bytes_unaligned(dst, src, MISALIGNED_COPY_SIZE);

		mb();

		end_cycles = get_cycles64();

		if ((end_cycles - start_cycles) < byte_cycles)

			byte_cycles = end_cycles - start_cycles;

	return cycles;

}



	preempt_enable();

static int check_unaligned_access(void *param)

{

	int cpu = smp_processor_id();

	u64 word_cycles;

	u64 byte_cycles;

	int ratio;

	struct page *page = param;

	void *dst;

	void *src;

	long speed = RISCV_HWPROBE_MISALIGNED_SCALAR_SLOW;



	if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN)

		return 0;



	/* Make an unaligned destination buffer. */

	dst = (void *)((unsigned long)page_address(page) | 0x1);

	/* Unalign src as well, but differently (off by 1 + 2 = 3). */

	src = dst + (MISALIGNED_BUFFER_SIZE / 2);

	src += 2;



	word_cycles = measure_cycles(__riscv_copy_words_unaligned, dst, src, MISALIGNED_COPY_SIZE);

	byte_cycles = measure_cycles(__riscv_copy_bytes_unaligned, dst, src, MISALIGNED_COPY_SIZE);



	/* Don't divide by zero. */

	if (!word_cycles || !byte_cycles) {