2014-12-18 03:24:08

by Chunyan Zhang

Subject: [PATCH v3] mtd: test: Replace timeval with ktime_t in speedtest.c and torturetest.c

This patch changes the 32-bit time type (timeval) to the 64-bit one
(ktime_t), since 32-bit time types will break in the year 2038.

I use ktime_t instead of timeval to define 'start' and 'finish'
which are used to get the time for tow points.

This patch also changes do_gettimeofday() to ktime_get() accordingly,
since ktime_get returns a ktime_t, but do_gettimeofday returns a
struct timeval, and the other reason is that ktime_get() uses
the monotonic clock.

This patch is based on another patch which privides a millisecond
time difference function 'ktime_ms_delta' in ktime.h

http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html

Signed-off-by: Chunyan Zhang <[email protected]>
Reviewed-by: Arnd Bergmann <[email protected]>
---
Changes since v2:
- Use the new function ktime_ms_delta which is added in another
patch listed up.

Changes since v1:
- Add including <linux/ktime.h> in torturetest.c
---
drivers/mtd/tests/speedtest.c | 10 +++++-----
drivers/mtd/tests/torturetest.c | 10 +++++-----
2 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/drivers/mtd/tests/speedtest.c b/drivers/mtd/tests/speedtest.c
index 5ee9f70..87a6e25 100644
--- a/drivers/mtd/tests/speedtest.c
+++ b/drivers/mtd/tests/speedtest.c
@@ -22,6 +22,7 @@
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/init.h>
+#include <linux/ktime.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/err.h>
@@ -49,7 +50,7 @@ static int pgsize;
static int ebcnt;
static int pgcnt;
static int goodebcnt;
-static struct timeval start, finish;
+static ktime_t start, finish;

static int multiblock_erase(int ebnum, int blocks)
{
@@ -168,12 +169,12 @@ static int read_eraseblock_by_2pages(int ebnum)

static inline void start_timing(void)
{
- do_gettimeofday(&start);
+ start = ktime_get();
}

static inline void stop_timing(void)
{
- do_gettimeofday(&finish);
+ finish = ktime_get();
}

static long calc_speed(void)
@@ -181,8 +182,7 @@ static long calc_speed(void)
uint64_t k;
long ms;

- ms = (finish.tv_sec - start.tv_sec) * 1000 +
- (finish.tv_usec - start.tv_usec) / 1000;
+ ms = ktime_ms_delta(finish, start);
if (ms == 0)
return 0;
k = goodebcnt * (mtd->erasesize / 1024) * 1000;
diff --git a/drivers/mtd/tests/torturetest.c b/drivers/mtd/tests/torturetest.c
index eeab969..7e77ed4 100644
--- a/drivers/mtd/tests/torturetest.c
+++ b/drivers/mtd/tests/torturetest.c
@@ -26,6 +26,7 @@
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/init.h>
+#include <linux/ktime.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/err.h>
@@ -79,18 +80,18 @@ static unsigned char *check_buf;
static unsigned int erase_cycles;

static int pgsize;
-static struct timeval start, finish;
+static ktime_t start, finish;

static void report_corrupt(unsigned char *read, unsigned char *written);

static inline void start_timing(void)
{
- do_gettimeofday(&start);
+ start = ktime_get();
}

static inline void stop_timing(void)
{
- do_gettimeofday(&finish);
+ finish = ktime_get();
}

/*
@@ -322,8 +323,7 @@ static int __init tort_init(void)
long ms;

stop_timing();
- ms = (finish.tv_sec - start.tv_sec) * 1000 +
- (finish.tv_usec - start.tv_usec) / 1000;
+ ms = ktime_ms_delta(finish, start);
pr_info("%08u erase cycles done, took %lu "
"milliseconds (%lu seconds)\n",
erase_cycles, ms, ms / 1000);
--
1.7.9.5

2014-12-18 07:54:32

by Arnd Bergmann

Subject: Re: [PATCH v3] mtd: test: Replace timeval with ktime_t in speedtest.c and torturetest.c

On Thursday 18 December 2014 11:23:31 Chunyan Zhang wrote:
> This patch changes the 32-bit time type (timeval) to the 64-bit one
> (ktime_t), since 32-bit time types will break in the year 2038.
>
> I use ktime_t instead of timeval to define 'start' and 'finish'
> which are used to get the time for tow points.
>
> This patch also changes do_gettimeofday() to ktime_get() accordingly,
> since ktime_get returns a ktime_t, but do_gettimeofday returns a
> struct timeval, and the other reason is that ktime_get() uses
> the monotonic clock.
>
> This patch is based on another patch which privides a millisecond
> time difference function 'ktime_ms_delta' in ktime.h
>
> http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html
>
> Signed-off-by: Chunyan Zhang <[email protected]>
> Reviewed-by: Arnd Bergmann <[email protected]>
>

The new version still looks good to me, but as there is now a
dependency on another patch, I'd suggest we queue this up in the
y2038 branch together with the patch that introduces ktime_ms_delta.

David or Brian, can you provide an Ack for this, or do you have
any objections?

Arnd