From: Davide Libenzi <davidel@xmailserver.org>
To: Alexander Shishkin <virtuoso@slind.org>
Cc: Thomas Gleixner <tglx@linutronix.de>,
Kyle Moffett <kyle@moffetthome.net>,
Valdis.Kletnieks@vt.edu, LKML <linux-kernel@vger.kernel.org>,
John Stultz <johnstul@us.ibm.com>,
Andrew Morton <akpm@linux-foundation.org>,
"H. Peter Anvin" <hpa@zytor.com>,
Kay Sievers <kay.sievers@vrfy.org>, Greg KH <gregkh@suse.de>,
Chris Friesen <chris.friesen@genband.com>,
Linus Torvalds <torvalds@linux-foundation.org>,
"Kirill A. Shutemov" <kirill@shutemov.name>
Subject: Re: [PATCHv6 0/7] system time changes notification
Date: Wed, 17 Nov 2010 12:42:52 -0800 (PST) [thread overview]
Message-ID: <alpine.DEB.2.00.1011171130210.1595@davide-lnx1> (raw)
In-Reply-To: <20101117190659.GB26184@shisha.kicks-ass.net>
On Wed, 17 Nov 2010, Alexander Shishkin wrote:
> > But what you folks really want for this stuff is an extension to
> > timerfd as you want to be able to poll, right?
> >
> > So what about the following:
> >
> > Add a new flag TDF_NOTIFY_CLOCK_WAS_SET to the timerfd flags. Now this
> > flag adds the timer to a separate list, which gets woken up when the
> > clock is set.
> >
> > No new syscall, just a few lines of code in fs/timerfd.c and
> > clock_was_set().
> >
> > Thoughts ?
>
> Something like this (sans ugliness)?
Oh, gosh, please. This is interface-multiplexing-a-palooza.
It should be decided if the feature makes sense, and then have proper
interface, instead of multiplexing unrelated insterfaces.
This is a sort of system-event-report pattern. What is wrong with using a
netlink-based transport for those kind of things?
> Signed-off-by: Alexander Shishkin <virtuoso@slind.org>
> ---
> fs/timerfd.c | 52 +++++++++++++++++++++++++++++++++++++++++++++-
> include/linux/timerfd.h | 3 +-
> kernel/hrtimer.c | 2 +
> 3 files changed, 54 insertions(+), 3 deletions(-)
>
> diff --git a/fs/timerfd.c b/fs/timerfd.c
> index 8c4fc14..51a033e 100644
> --- a/fs/timerfd.c
> +++ b/fs/timerfd.c
> @@ -30,8 +30,14 @@ struct timerfd_ctx {
> u64 ticks;
> int expired;
> int clockid;
> + int clock_notifier; /* list_empty()? */
> + struct list_head notifiers_list;
> };
>
> +/* TFD_NOTIFY_CLOCK_SET timers go here */
> +static DEFINE_SPINLOCK(notifiers_lock);
> +static LIST_HEAD(notifiers_list);
> +
> /*
> * This gets called when the timer event triggers. We set the "expired"
> * flag, but we do not re-arm the timer (in case it's necessary,
> @@ -51,6 +57,21 @@ static enum hrtimer_restart timerfd_tmrproc(struct hrtimer *htmr)
> return HRTIMER_NORESTART;
> }
>
> +void timerfd_clock_was_set(void)
> +{
> + struct timerfd_ctx *ctx;
> + unsigned long flags;
> +
> + spin_lock(¬ifiers_lock);
> + list_for_each_entry(ctx, ¬ifiers_list, notifiers_list) {
> + spin_lock_irqsave(&ctx->wqh.lock, flags);
> + ctx->ticks++;
> + wake_up_locked(&ctx->wqh);
> + spin_unlock_irqrestore(&ctx->wqh.lock, flags);
> + }
> + spin_unlock(¬ifiers_lock);
> +}
> +
> static ktime_t timerfd_get_remaining(struct timerfd_ctx *ctx)
> {
> ktime_t remaining;
> @@ -65,6 +86,14 @@ static void timerfd_setup(struct timerfd_ctx *ctx, int flags,
> enum hrtimer_mode htmode;
> ktime_t texp;
>
> + if (flags & TFD_NOTIFY_CLOCK_SET) {
> + ctx->clock_notifier = 1;
> + ctx->ticks = 0;
> + INIT_LIST_HEAD(&ctx->notifiers_list);
> + list_add(&ctx->notifiers_list, ¬ifiers_list);
> + return;
> + }
> +
> htmode = (flags & TFD_TIMER_ABSTIME) ?
> HRTIMER_MODE_ABS: HRTIMER_MODE_REL;
>
> @@ -83,7 +112,12 @@ static int timerfd_release(struct inode *inode, struct file *file)
> {
> struct timerfd_ctx *ctx = file->private_data;
>
> - hrtimer_cancel(&ctx->tmr);
> + if (ctx->clock_notifier) {
> + spin_lock(¬ifiers_lock);
> + list_del(&ctx->notifiers_list);
> + spin_unlock(¬ifiers_lock);
> + } else
> + hrtimer_cancel(&ctx->tmr);
> kfree(ctx);
> return 0;
> }
> @@ -113,6 +147,7 @@ static ssize_t timerfd_read(struct file *file, char __user *buf, size_t count,
>
> if (count < sizeof(ticks))
> return -EINVAL;
> +
> spin_lock_irq(&ctx->wqh.lock);
> if (file->f_flags & O_NONBLOCK)
> res = -EAGAIN;
> @@ -120,7 +155,7 @@ static ssize_t timerfd_read(struct file *file, char __user *buf, size_t count,
> res = wait_event_interruptible_locked_irq(ctx->wqh, ctx->ticks);
> if (ctx->ticks) {
> ticks = ctx->ticks;
> - if (ctx->expired && ctx->tintv.tv64) {
> + if (ctx->expired && ctx->tintv.tv64 && !ctx->clock_notifier) {
> /*
> * If tintv.tv64 != 0, this is a periodic timer that
> * needs to be re-armed. We avoid doing it in the timer
> @@ -218,13 +253,20 @@ SYSCALL_DEFINE4(timerfd_settime, int, ufd, int, flags,
> * it to the new values.
> */
> for (;;) {
> + spin_lock(¬ifiers_lock);
> spin_lock_irq(&ctx->wqh.lock);
> if (hrtimer_try_to_cancel(&ctx->tmr) >= 0)
> break;
> spin_unlock_irq(&ctx->wqh.lock);
> + spin_unlock(¬ifiers_lock);
> cpu_relax();
> }
>
> + if (ctx->clock_notifier) {
> + ctx->clock_notifier = 0;
> + list_del(&ctx->notifiers_list);
> + }
> +
> /*
> * If the timer is expired and it's periodic, we need to advance it
> * because the caller may want to know the previous expiration time.
> @@ -243,6 +285,7 @@ SYSCALL_DEFINE4(timerfd_settime, int, ufd, int, flags,
> timerfd_setup(ctx, flags, &ktmr);
>
> spin_unlock_irq(&ctx->wqh.lock);
> + spin_unlock(¬ifiers_lock);
> fput(file);
> if (otmr && copy_to_user(otmr, &kotmr, sizeof(kotmr)))
> return -EFAULT;
> @@ -262,6 +305,11 @@ SYSCALL_DEFINE2(timerfd_gettime, int, ufd, struct itimerspec __user *, otmr)
> ctx = file->private_data;
>
> spin_lock_irq(&ctx->wqh.lock);
> + if (ctx->clock_notifier) {
> + spin_unlock_irq(&ctx->wqh.lock);
> + return -EINVAL;
> + }
> +
> if (ctx->expired && ctx->tintv.tv64) {
> ctx->expired = 0;
> ctx->ticks +=
> diff --git a/include/linux/timerfd.h b/include/linux/timerfd.h
> index 2d07929..c3ddad9 100644
> --- a/include/linux/timerfd.h
> +++ b/include/linux/timerfd.h
> @@ -19,6 +19,7 @@
> * shared O_* flags.
> */
> #define TFD_TIMER_ABSTIME (1 << 0)
> +#define TFD_NOTIFY_CLOCK_SET (1 << 1)
> #define TFD_CLOEXEC O_CLOEXEC
> #define TFD_NONBLOCK O_NONBLOCK
>
> @@ -26,6 +27,6 @@
> /* Flags for timerfd_create. */
> #define TFD_CREATE_FLAGS TFD_SHARED_FCNTL_FLAGS
> /* Flags for timerfd_settime. */
> -#define TFD_SETTIME_FLAGS TFD_TIMER_ABSTIME
> +#define TFD_SETTIME_FLAGS (TFD_TIMER_ABSTIME | TFD_NOTIFY_CLOCK_SET)
>
> #endif /* _LINUX_TIMERFD_H */
> diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
> index 72206cf..a779e0f 100644
> --- a/kernel/hrtimer.c
> +++ b/kernel/hrtimer.c
> @@ -642,8 +642,10 @@ static void retrigger_next_event(void *arg)
> * resolution timer interrupts. On UP we just disable interrupts and
> * call the high resolution interrupt code.
> */
> +void timerfd_clock_was_set(clockid_t clockid);
> void clock_was_set(void)
> {
> + timerfd_clock_was_set(CLOCK_REALTIME);
> /* Retrigger the CPU local events everywhere */
> on_each_cpu(retrigger_next_event, NULL, 1);
> }
> --
> 1.7.2.1.45.gb66c2
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
- Davide
next prev parent reply other threads:[~2010-11-17 20:45 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-11-11 19:29 [PATCHv6 0/7] system time changes notification Alexander Shishkin
2010-11-11 19:29 ` [PATCHv6 1/7] notify userspace about time changes Alexander Shishkin
2010-11-11 20:55 ` Arnd Bergmann
2010-11-11 19:29 ` [PATCHv6 2/7] wire up sys_time_change_notify() on ARM Alexander Shishkin
2010-11-11 19:29 ` [PATCHv6 3/7] wire up sys_time_change_notify() on x86 Alexander Shishkin
2010-11-11 19:29 ` [PATCHv6 4/7] wire up sys_time_change_notify() on ia64 Alexander Shishkin
2010-11-11 19:30 ` [PATCHv6 5/7] wire up sys_time_change_notify() on s390 Alexander Shishkin
2010-11-11 19:30 ` [PATCHv6 6/7] wire up sys_time_change_notify() on powerpc Alexander Shishkin
2010-11-11 19:30 ` [PATCHv6 7/7] wire up sys_time_change_notify() on blackfin Alexander Shishkin
2010-11-11 20:28 ` [PATCHv6 0/7] system time changes notification Valdis.Kletnieks
2010-11-11 20:51 ` Alexander Shishkin
2010-11-11 21:16 ` Thomas Gleixner
2010-11-11 22:11 ` Kyle Moffett
2010-11-11 22:36 ` john stultz
2010-11-11 23:19 ` Kyle Moffett
2010-11-11 23:41 ` john stultz
2010-11-11 23:45 ` john stultz
2010-11-11 22:50 ` Thomas Gleixner
2010-11-12 2:35 ` Davide Libenzi
2010-11-17 19:06 ` Alexander Shishkin
2010-11-17 20:42 ` Davide Libenzi [this message]
2010-11-17 21:29 ` Alexander Shishkin
2010-11-17 21:34 ` Kay Sievers
2010-11-18 15:59 ` Alexander Shishkin
2010-11-17 21:46 ` Thomas Gleixner
2010-11-18 9:49 ` Alexander Shishkin
2010-11-18 13:08 ` Artem Bityutskiy
2010-11-12 9:25 ` Alan Cox
2010-11-12 10:53 ` Richard Cochran
2010-11-12 11:25 ` Alan Cox
2010-11-12 10:47 ` Kay Sievers
2010-11-12 12:30 ` Alexander Shishkin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=alpine.DEB.2.00.1011171130210.1595@davide-lnx1 \
--to=davidel@xmailserver.org \
--cc=Valdis.Kletnieks@vt.edu \
--cc=akpm@linux-foundation.org \
--cc=chris.friesen@genband.com \
--cc=gregkh@suse.de \
--cc=hpa@zytor.com \
--cc=johnstul@us.ibm.com \
--cc=kay.sievers@vrfy.org \
--cc=kirill@shutemov.name \
--cc=kyle@moffetthome.net \
--cc=linux-kernel@vger.kernel.org \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
--cc=virtuoso@slind.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).