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 all uses of timeval in imon.c
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 use a new function which is provided by another patch listed below
to get the millisecond time difference.
http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html
Signed-off-by: Chunyan Zhang <[email protected]>
---
drivers/media/rc/imon.c | 49 +++++++++++++----------------------------------
1 file changed, 13 insertions(+), 36 deletions(-)
diff --git a/drivers/media/rc/imon.c b/drivers/media/rc/imon.c
index b8837dd..a641139 100644
--- a/drivers/media/rc/imon.c
+++ b/drivers/media/rc/imon.c
@@ -31,6 +31,7 @@
#include <linux/errno.h>
#include <linux/init.h>
#include <linux/kernel.h>
+#include <linux/ktime.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/uaccess.h>
@@ -41,7 +42,6 @@
#include <linux/usb/input.h>
#include <media/rc-core.h>
-#include <linux/time.h>
#include <linux/timer.h>
#define MOD_AUTHOR "Jarod Wilson <[email protected]>"
@@ -1158,29 +1158,6 @@ out:
return retval;
}
-static inline int tv2int(const struct timeval *a, const struct timeval *b)
-{
- int usecs = 0;
- int sec = 0;
-
- if (b->tv_usec > a->tv_usec) {
- usecs = 1000000;
- sec--;
- }
-
- usecs += a->tv_usec - b->tv_usec;
-
- sec += a->tv_sec - b->tv_sec;
- sec *= 1000;
- usecs /= 1000;
- sec += usecs;
-
- if (sec < 0)
- sec = 1000;
-
- return sec;
-}
-
/**
* The directional pad behaves a bit differently, depending on whether this is
* one of the older ffdc devices or a newer device. Newer devices appear to
@@ -1191,16 +1168,16 @@ static inline int tv2int(const struct timeval *a, const struct timeval *b)
*/
static int stabilize(int a, int b, u16 timeout, u16 threshold)
{
- struct timeval ct;
- static struct timeval prev_time = {0, 0};
- static struct timeval hit_time = {0, 0};
+ ktime_t ct;
+ static ktime_t prev_time = {0};
+ static ktime_t hit_time = {0};
static int x, y, prev_result, hits;
int result = 0;
- int msec, msec_hit;
+ long msec, msec_hit;
- do_gettimeofday(&ct);
- msec = tv2int(&ct, &prev_time);
- msec_hit = tv2int(&ct, &hit_time);
+ ct = ktime_get();
+ msec = ktime_ms_delta(ct, prev_time);
+ msec_hit = ktime_ms_delta(ct, hit_time);
if (msec > 100) {
x = 0;
@@ -1596,9 +1573,9 @@ static void imon_incoming_packet(struct imon_context *ictx,
int i;
u64 scancode;
int press_type = 0;
- int msec;
- struct timeval t;
- static struct timeval prev_time = { 0, 0 };
+ long msec;
+ ktime_t t;
+ static ktime_t prev_time = {0};
u8 ktype;
/* filter out junk data on the older 0xffdc imon devices */
@@ -1692,10 +1669,10 @@ static void imon_incoming_packet(struct imon_context *ictx,
/* Only panel type events left to process now */
spin_lock_irqsave(&ictx->kc_lock, flags);
- do_gettimeofday(&t);
+ t = ktime_get();
/* KEY_MUTE repeats from knob need to be suppressed */
if (ictx->kc == KEY_MUTE && ictx->kc == ictx->last_keycode) {
- msec = tv2int(&t, &prev_time);
+ msec = ktime_ms_delta(t, prev_time);
if (msec < ictx->idev->rep[REP_DELAY]) {
spin_unlock_irqrestore(&ictx->kc_lock, flags);
return;
--
1.7.9.5
On Thursday 18 December 2014 11:37:13 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 all uses of timeval in imon.c
>
> 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 use a new function which is provided by another patch listed below
> to get the millisecond time difference.
The patch looks great. Just a few small details that could still be
improved:
> http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html
>
> Signed-off-by: Chunyan Zhang <[email protected]>
In general, when you give a mailing list link, use the 'Link' tag
under your Signed-off-by line, like
Link: http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html
It's not used much yet, but getting more popular and seems useful to me.
In this particular case, when you have patches that depend on one
another, you can make do it even better by sending all three patches
as a series with a [PATCH 0/3] cover letter.
If the media maintainers can provide an Ack for this patch, I would
suggest to queue it up in the y2038 branch together with your first
patch that it depends on.
> @@ -1191,16 +1168,16 @@ static inline int tv2int(const struct timeval *a, const struct timeval *b)
> */
> static int stabilize(int a, int b, u16 timeout, u16 threshold)
> {
> - struct timeval ct;
> - static struct timeval prev_time = {0, 0};
> - static struct timeval hit_time = {0, 0};
> + ktime_t ct;
> + static ktime_t prev_time = {0};
> + static ktime_t hit_time = {0};
> static int x, y, prev_result, hits;
> int result = 0;
The "= {0}" part here is redundant, since static variables are always
initialized to zero. Normally, adding the explicit initializer can
help readability, but in this case I would leave it out because it shows
implementation details of ktime_t that are better hidden from drivers.
> @@ -1596,9 +1573,9 @@ static void imon_incoming_packet(struct imon_context *ictx,
> int i;
> u64 scancode;
> int press_type = 0;
> - int msec;
> - struct timeval t;
> - static struct timeval prev_time = { 0, 0 };
> + long msec;
> + ktime_t t;
> + static ktime_t prev_time = {0};
> u8 ktype;
Same thing here of course.
Arnd
On Thu, Dec 18, 2014 at 3:50 PM, Arnd Bergmann <[email protected]> wrote:
> On Thursday 18 December 2014 11:37:13 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 all uses of timeval in imon.c
>>
>> 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 use a new function which is provided by another patch listed below
>> to get the millisecond time difference.
>
> The patch looks great. Just a few small details that could still be
> improved:
>
>> http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html
>>
>> Signed-off-by: Chunyan Zhang <[email protected]>
>
> In general, when you give a mailing list link, use the 'Link' tag
> under your Signed-off-by line, like
>
> Link: http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html
>
> It's not used much yet, but getting more popular and seems useful to me.
>
> In this particular case, when you have patches that depend on one
> another, you can make do it even better by sending all three patches
> as a series with a [PATCH 0/3] cover letter.
>
OK, I'll send a patch-set including these three patches
> If the media maintainers can provide an Ack for this patch, I would
> suggest to queue it up in the y2038 branch together with your first
> patch that it depends on.
>
>> @@ -1191,16 +1168,16 @@ static inline int tv2int(const struct timeval *a, const struct timeval *b)
>> */
>> static int stabilize(int a, int b, u16 timeout, u16 threshold)
>> {
>> - struct timeval ct;
>> - static struct timeval prev_time = {0, 0};
>> - static struct timeval hit_time = {0, 0};
>> + ktime_t ct;
>> + static ktime_t prev_time = {0};
>> + static ktime_t hit_time = {0};
>> static int x, y, prev_result, hits;
>> int result = 0;
>
> The "= {0}" part here is redundant, since static variables are always
> initialized to zero. Normally, adding the explicit initializer can
> help readability, but in this case I would leave it out because it shows
> implementation details of ktime_t that are better hidden from drivers.
>
OK, I'll modify them soon
Thanks,
Chunyan
>> @@ -1596,9 +1573,9 @@ static void imon_incoming_packet(struct imon_context *ictx,
>> int i;
>> u64 scancode;
>> int press_type = 0;
>> - int msec;
>> - struct timeval t;
>> - static struct timeval prev_time = { 0, 0 };
>> + long msec;
>> + ktime_t t;
>> + static ktime_t prev_time = {0};
>> u8 ktype;
>
> Same thing here of course.
>
> Arnd
Em Thu, 18 Dec 2014 17:38:14 +0800
Chunyan Zhang <[email protected]> escreveu:
> On Thu, Dec 18, 2014 at 3:50 PM, Arnd Bergmann <[email protected]> wrote:
> > On Thursday 18 December 2014 11:37:13 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 all uses of timeval in imon.c
> >>
> >> 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 use a new function which is provided by another patch listed below
> >> to get the millisecond time difference.
> >
> > The patch looks great. Just a few small details that could still be
> > improved:
Yes, patch looks OK. After addressing the bits pointed by Arnd:
Acked-by: Mauro Carvalho Chehab <[email protected]>
Feel free to merge via y2038 tree.
> >
> >> http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html
> >>
> >> Signed-off-by: Chunyan Zhang <[email protected]>
> >
> > In general, when you give a mailing list link, use the 'Link' tag
> > under your Signed-off-by line, like
> >
> > Link: http://lkml.iu.edu//hypermail/linux/kernel/1412.2/00625.html
> >
> > It's not used much yet, but getting more popular and seems useful to me.
> >
> > In this particular case, when you have patches that depend on one
> > another, you can make do it even better by sending all three patches
> > as a series with a [PATCH 0/3] cover letter.
> >
> OK, I'll send a patch-set including these three patches
>
> > If the media maintainers can provide an Ack for this patch, I would
> > suggest to queue it up in the y2038 branch together with your first
> > patch that it depends on.
> >
> >> @@ -1191,16 +1168,16 @@ static inline int tv2int(const struct timeval *a, const struct timeval *b)
> >> */
> >> static int stabilize(int a, int b, u16 timeout, u16 threshold)
> >> {
> >> - struct timeval ct;
> >> - static struct timeval prev_time = {0, 0};
> >> - static struct timeval hit_time = {0, 0};
> >> + ktime_t ct;
> >> + static ktime_t prev_time = {0};
> >> + static ktime_t hit_time = {0};
> >> static int x, y, prev_result, hits;
> >> int result = 0;
> >
> > The "= {0}" part here is redundant, since static variables are always
> > initialized to zero. Normally, adding the explicit initializer can
> > help readability, but in this case I would leave it out because it shows
> > implementation details of ktime_t that are better hidden from drivers.
> >
> OK, I'll modify them soon
>
> Thanks,
> Chunyan
>
> >> @@ -1596,9 +1573,9 @@ static void imon_incoming_packet(struct imon_context *ictx,
> >> int i;
> >> u64 scancode;
> >> int press_type = 0;
> >> - int msec;
> >> - struct timeval t;
> >> - static struct timeval prev_time = { 0, 0 };
> >> + long msec;
> >> + ktime_t t;
> >> + static ktime_t prev_time = {0};
> >> u8 ktype;
> >
> > Same thing here of course.
> >
> > Arnd
On Thu, Dec 18, 2014 at 7:00 PM, Mauro Carvalho Chehab
<[email protected]> wrote:
> Em Thu, 18 Dec 2014 17:38:14 +0800
> Chunyan Zhang <[email protected]> escreveu:
>
>> On Thu, Dec 18, 2014 at 3:50 PM, Arnd Bergmann <[email protected]> wrote:
>> > On Thursday 18 December 2014 11:37:13 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 all uses of timeval in imon.c
>> >>
>> >> 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 use a new function which is provided by another patch listed below
>> >> to get the millisecond time difference.
>> >
>> > The patch looks great. Just a few small details that could still be
>> > improved:
>
> Yes, patch looks OK. After addressing the bits pointed by Arnd:
>
> Acked-by: Mauro Carvalho Chehab <[email protected]>
>
> Feel free to merge via y2038 tree.
>
Ok, thank you, I'll send the updated patch-set soon.
Chunyan