From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753696AbZBHUrO (ORCPT ); Sun, 8 Feb 2009 15:47:14 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753368AbZBHUq6 (ORCPT ); Sun, 8 Feb 2009 15:46:58 -0500 Received: from x35.xmailserver.org ([64.71.152.41]:56667 "EHLO x35.xmailserver.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753348AbZBHUq5 (ORCPT ); Sun, 8 Feb 2009 15:46:57 -0500 X-AuthUser: davidel@xmailserver.org Date: Sun, 8 Feb 2009 12:46:53 -0800 (PST) From: Davide Libenzi X-X-Sender: davide@alien.or.mcafeemobile.com To: Michael Kerrisk cc: "linux-man@vger.kernel.org" , lkml , Thomas Gleixner Subject: Re: Why does timerfd() only support CLOCK_REALTIME and CLOCK_MONOTONIC? In-Reply-To: Message-ID: References: User-Agent: Alpine 1.10 (DEB 962 2008-03-14) X-GPG-FINGRPRINT: CFAE 5BEE FD36 F65E E640 56FE 0974 BF23 270F 474E X-GPG-PUBLIC_KEY: http://www.xmailserver.org/davidel.asc MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sun, 8 Feb 2009, Davide Libenzi wrote: > On Mon, 9 Feb 2009, Michael Kerrisk wrote: > > > Hi Davide, > > > > At the moment I'm looking into writing man pages for timer_create(2) > > and friends. (Somewhat bizarrely, these pages do not yet exist.) As > > I looked into the source code of timer_create(), etc., and did some > > tests, I saw that timer_create() supports the following clocks: > > > > TIMER_REALTIME > > TIMER_MONOTONIC > > TIMER_PROCESS_CPUTIME_ID > > TIMER_THREAD_CPUTIME_ID > > clockid obtained from clock_getcpuclockid(3) > > clockid obtained from pthread_getcpuclockid(3) > > > > On the other hand, timerfd() only permits the first two of these. > > What's the reason for that limitation of timerfd()? (It may be worth > > adding something to the man page on this point.) > > No particular reason I can think of. If Thomas makes invalid_clockid() > available, we could allow timerfd() to support all time timers > timer_create() supports. > Do you see any reason why this won't work Thomas? Patch, not tested, below. - Davide --- fs/timerfd.c | 10 ++++------ include/linux/posix-timers.h | 1 + include/linux/timerfd.h | 13 ++++++++++--- kernel/posix-timers.c | 2 +- 4 files changed, 16 insertions(+), 10 deletions(-) Index: linux-2.6.mod/fs/timerfd.c =================================================================== --- linux-2.6.mod.orig/fs/timerfd.c 2009-02-08 12:32:25.000000000 -0800 +++ linux-2.6.mod/fs/timerfd.c 2009-02-08 12:43:57.000000000 -0800 @@ -18,6 +18,7 @@ #include #include #include +#include #include #include #include @@ -186,12 +187,9 @@ SYSCALL_DEFINE2(timerfd_create, int, clo BUILD_BUG_ON(TFD_CLOEXEC != O_CLOEXEC); BUILD_BUG_ON(TFD_NONBLOCK != O_NONBLOCK); - if (flags & ~(TFD_CLOEXEC | TFD_NONBLOCK)) + if ((flags & ~TFD_FLAGS_SET) || + invalid_clockid(clockid)) return -EINVAL; - if (clockid != CLOCK_MONOTONIC && - clockid != CLOCK_REALTIME) - return -EINVAL; - ctx = kzalloc(sizeof(*ctx), GFP_KERNEL); if (!ctx) return -ENOMEM; @@ -201,7 +199,7 @@ SYSCALL_DEFINE2(timerfd_create, int, clo hrtimer_init(&ctx->tmr, clockid, HRTIMER_MODE_ABS); ufd = anon_inode_getfd("[timerfd]", &timerfd_fops, ctx, - flags & (O_CLOEXEC | O_NONBLOCK)); + flags & TFD_SHARED_FCNTL_FLAGS); if (ufd < 0) kfree(ctx); Index: linux-2.6.mod/include/linux/posix-timers.h =================================================================== --- linux-2.6.mod.orig/include/linux/posix-timers.h 2009-02-08 12:32:25.000000000 -0800 +++ linux-2.6.mod/include/linux/posix-timers.h 2009-02-08 12:33:12.000000000 -0800 @@ -84,6 +84,7 @@ struct k_clock { struct itimerspec * cur_setting); }; +int invalid_clockid(const clockid_t which_clock); void register_posix_clock(const clockid_t clock_id, struct k_clock *new_clock); /* error handlers for timer_create, nanosleep and settime */ Index: linux-2.6.mod/kernel/posix-timers.c =================================================================== --- linux-2.6.mod.orig/kernel/posix-timers.c 2009-02-08 12:32:25.000000000 -0800 +++ linux-2.6.mod/kernel/posix-timers.c 2009-02-08 12:32:57.000000000 -0800 @@ -205,7 +205,7 @@ static int no_timer_create(struct k_itim /* * Return nonzero if we know a priori this clockid_t value is bogus. */ -static inline int invalid_clockid(const clockid_t which_clock) +int invalid_clockid(const clockid_t which_clock) { if (which_clock < 0) /* CPU clock, posix_cpu_* will check it */ return 0; Index: linux-2.6.mod/include/linux/timerfd.h =================================================================== --- linux-2.6.mod.orig/include/linux/timerfd.h 2009-02-08 12:34:25.000000000 -0800 +++ linux-2.6.mod/include/linux/timerfd.h 2009-02-08 12:36:09.000000000 -0800 @@ -11,13 +11,20 @@ /* For O_CLOEXEC and O_NONBLOCK */ #include -/* Flags for timerfd_settime. */ +/* + * CAREFUL: Check include/asm-generic/fcntl.h when defining + * new flags, since they might collide with O_* ones. We want + * to re-use O_* flags that couldn't possibly have a meaning + * from eventfd, in order to leave a free define-space for + * shared O_* flags. + */ #define TFD_TIMER_ABSTIME (1 << 0) - -/* Flags for timerfd_create. */ #define TFD_CLOEXEC O_CLOEXEC #define TFD_NONBLOCK O_NONBLOCK +#define TFD_SHARED_FCNTL_FLAGS (O_CLOEXEC | O_NONBLOCK) +#define TFD_FLAGS_SET (TFD_SHARED_FCNTL_FLAGS | TFD_TIMER_ABSTIME) + #endif /* _LINUX_TIMERFD_H */