linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Thomas Gleixner <tglx@linutronix.de>
To: LKML <linux-kernel@vger.kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>, Ingo Molnar <mingo@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	John Stultz <john.stultz@linaro.org>,
	Frederic Weisbecker <fweisbec@gmail.com>,
	Anna-Maria Behnsen <anna-maria@linutronix.de>
Subject: [patch 27/44] posix-cpu-timers: Provide array based access to expiry cache
Date: Mon, 19 Aug 2019 16:32:08 +0200	[thread overview]
Message-ID: <20190819143803.961800814@linutronix.de> (raw)
In-Reply-To: 20190819143141.221906747@linutronix.de

Using struct task_cputime for the expiry cache is a pretty odd choice and
comes with magic defines to rename the fields for usage in the expiry
cache.

struct task_cputime is basically a u64 array with 3 members, but it has
distinct members.

The expiry cache content is different than the content of task_cputime
because

  expiry[PROF]  = task_cputime.stime + task_cputime.utime
  expiry[VIRT]  = task_cputime.utime
  expiry[SCHED] = task_cputime.sum_exec_runtime

So there is no direct mapping between task_cputime and the expiry cache and
the #define based remapping is just a horrible hack.

Having the expiry cache array based allows further simplification of the
expiry code.

To avoid an all in one cleanup which is hard to review add a temporary
anonymous union into struct task_cputime which allows array based access to
it. That requires to reorder the members. Add a build time sanity check to
validate that the members are at the same place.

The union and the build time checks will be removed after conversion.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
 include/linux/posix-timers.h   |   26 +++++++++++++++++++-------
 include/linux/sched/types.h    |    4 ++--
 kernel/time/posix-cpu-timers.c |   10 ++++++++++
 3 files changed, 31 insertions(+), 9 deletions(-)

--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -65,27 +65,39 @@ static inline int clockid_to_fd(const cl
 /*
  * Alternate field names for struct task_cputime when used on cache
  * expirations. Will go away soon.
+ *
+ * stime corresponds to CLOCKCPU_PROF
+ * utime corresponds to CLOCKCPU_VIRT
+ * sum_exex_runtime corresponds to CLOCKCPU_SCHED
+ *
+ * The ordering is currently enforced so struct task_cputime and the
+ * expiries array in struct posix_cputimers are equivalent.
  */
-#define virt_exp			utime
 #define prof_exp			stime
+#define virt_exp			utime
 #define sched_exp			sum_exec_runtime
 
 #ifdef CONFIG_POSIX_TIMERS
 /**
  * posix_cputimers - Container for posix CPU timer related data
- * @cputime_expires:	Earliest-expiration cache
+ * @cputime_expires:	Earliest-expiration cache task_cputime based
+ * @expiries:		Earliest-expiration cache array based
  * @cpu_timers:		List heads to queue posix CPU timers
  *
  * Used in task_struct and signal_struct
  */
 struct posix_cputimers {
-	struct task_cputime	cputime_expires;
-	struct list_head	cpu_timers[CPUCLOCK_MAX];
+	/* Temporary union until all users are cleaned up */
+	union {
+		struct task_cputime	cputime_expires;
+		u64			expiries[CPUCLOCK_MAX];
+	};
+	struct list_head		cpu_timers[CPUCLOCK_MAX];
 };
 
 static inline void posix_cputimers_init(struct posix_cputimers *pct)
 {
-	memset(&pct->cputime_expires, 0, sizeof(pct->cputime_expires));
+	memset(&pct->expiries, 0, sizeof(pct->expiries));
 	INIT_LIST_HEAD(&pct->cpu_timers[0]);
 	INIT_LIST_HEAD(&pct->cpu_timers[1]);
 	INIT_LIST_HEAD(&pct->cpu_timers[2]);
@@ -96,13 +108,13 @@ static inline void posix_cputimers_group
 {
 	posix_cputimers_init(pct);
 	if (cpu_limit != RLIM_INFINITY)
-		pct->cputime_expires.prof_exp = cpu_limit * NSEC_PER_SEC;
+		pct->expiries[CPUCLOCK_PROF] = cpu_limit * NSEC_PER_SEC;
 }
 
 static inline void posix_cputimers_rt_watchdog(struct posix_cputimers *pct,
 					       u64 runtime)
 {
-	pct->cputime_expires.sched_exp = runtime;
+	pct->expiries[CPUCLOCK_SCHED] = runtime;
 }
 
 /* Init task static initializer */
--- a/include/linux/sched/types.h
+++ b/include/linux/sched/types.h
@@ -4,8 +4,8 @@
 
 /**
  * struct task_cputime - collected CPU time counts
- * @utime:		time spent in user mode, in nanoseconds
  * @stime:		time spent in kernel mode, in nanoseconds
+ * @utime:		time spent in user mode, in nanoseconds
  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
  *
  * This structure groups together three kinds of CPU time that are tracked for
@@ -13,8 +13,8 @@
  * these counts together and treat all three of them in parallel.
  */
 struct task_cputime {
-	u64				utime;
 	u64				stime;
+	u64				utime;
 	unsigned long long		sum_exec_runtime;
 };
 
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -18,6 +18,16 @@
 
 #include "posix-timers.h"
 
+static inline void temporary_check(void)
+{
+	BUILD_BUG_ON(offsetof(struct task_cputime, stime) !=
+		     CPUCLOCK_PROF * sizeof(u64));
+	BUILD_BUG_ON(offsetof(struct task_cputime, utime) !=
+		     CPUCLOCK_VIRT * sizeof(u64));
+	BUILD_BUG_ON(offsetof(struct task_cputime, sum_exec_runtime) !=
+		     CPUCLOCK_SCHED * sizeof(u64));
+}
+
 static void posix_cpu_timer_rearm(struct k_itimer *timer);
 
 /*



  parent reply	other threads:[~2019-08-19 15:46 UTC|newest]

Thread overview: 83+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-19 14:31 [patch 00/44] posix-cpu-timers: Cleanup and consolidation Thomas Gleixner
2019-08-19 14:31 ` [patch 01/44] posix-timers: Cleanup forward declarations and includes Thomas Gleixner
2019-08-20 12:20   ` Frederic Weisbecker
2019-08-20 13:03     ` Thomas Gleixner
2019-08-20 13:48   ` Frederic Weisbecker
2019-08-23  2:12   ` [tip: timers/core] " tip-bot2 for Thomas Gleixner
2019-08-19 14:31 ` [patch 02/44] alarmtimers: Avoid rtc.h include Thomas Gleixner
2019-08-20 13:49   ` Frederic Weisbecker
2019-08-23  2:12   ` [tip: timers/core] " tip-bot2 for Thomas Gleixner
2019-08-19 14:31 ` [patch 03/44] posix-timer: Use a callback for cancel synchronization Thomas Gleixner
2019-08-20  2:21   ` Christoph Hellwig
2019-08-20 13:08     ` Thomas Gleixner
2019-08-20 13:59   ` Frederic Weisbecker
2019-08-23  2:12   ` [tip: timers/core] posix-timers: Use a callback for cancel synchronization on PREEMPT_RT tip-bot2 for Thomas Gleixner
2019-08-19 14:31 ` [patch 04/44] posix-cpu-timers: Fixup stale comment Thomas Gleixner
2019-08-20 14:26   ` Frederic Weisbecker
2019-08-20 17:57     ` Thomas Gleixner
2019-08-20 20:48       ` Frederic Weisbecker
2019-08-20 21:43         ` Thomas Gleixner
2019-08-20 22:56           ` Frederic Weisbecker
2019-08-21 13:31             ` Thomas Gleixner
2019-08-21 15:51               ` Frederic Weisbecker
2019-08-23  2:12   ` [tip: timers/core] " tip-bot2 for Thomas Gleixner
2019-08-19 14:31 ` [patch 05/44] posix-cpu-timers: Sanitize bogus WARNONS Thomas Gleixner
2019-08-21 11:34   ` Frederic Weisbecker
2019-08-23  2:12   ` [tip: timers/core] " tip-bot2 for Thomas Gleixner
2019-08-19 14:31 ` [patch 06/44] posix-cpu-timers: Remove tsk argument from run_posix_cpu_timers() Thomas Gleixner
2019-08-21 11:36   ` Frederic Weisbecker
2019-08-23  2:12   ` [tip: timers/core] " tip-bot2 for Thomas Gleixner
2019-08-19 14:31 ` [patch 07/44] posix-cpu-timers: Simplify sighand locking in run_posix_cpu_timers() Thomas Gleixner
2019-08-21 12:09   ` Frederic Weisbecker
2019-08-21 13:25     ` Thomas Gleixner
2019-08-21 15:42       ` Frederic Weisbecker
2019-08-21 17:13         ` Thomas Gleixner
2019-08-19 14:31 ` [patch 08/44] posix-cpu-timers: Provide task validation functions Thomas Gleixner
2019-08-19 14:31 ` [patch 09/44] posix-cpu-timers: Use common permission check in posix_cpu_clock_get() Thomas Gleixner
2019-08-19 14:31 ` [patch 10/44] posix-cpu-timers: Use common permission check in posix_cpu_timer_create() Thomas Gleixner
2019-08-19 14:31 ` [patch 11/44] posix-cpu-timers: Provide quick sample function for itimer Thomas Gleixner
2019-08-19 14:31 ` [patch 12/44] itimers: Use quick sample function Thomas Gleixner
2019-08-19 14:31 ` [patch 13/44] posix-cpu-timers: Sample directly in timer check Thomas Gleixner
2019-08-19 14:31 ` [patch 14/44] posix-cpu-timers: Rename thread_group_cputimer() and make it static Thomas Gleixner
2019-08-19 14:31 ` [patch 15/44] posix-cpu-timer: Comsolidate thread group sample code Thomas Gleixner
2019-08-19 19:07   ` Ingo Molnar
2019-08-19 14:31 ` [patch 16/44] posix-cpu-timers: Use clock ID in posix_cpu_timer_set() Thomas Gleixner
2019-08-19 14:31 ` [patch 17/44] posix-cpu-timers: Use clock ID in posix_cpu_timer_get() Thomas Gleixner
2019-08-19 14:31 ` [patch 18/44] posix-cpu-timers: Use clock ID in posix_cpu_timer_rearm() Thomas Gleixner
2019-08-19 14:32 ` [patch 19/44] posix-cpu-timer: Remove pointless return value check Thomas Gleixner
2019-08-19 14:32 ` [patch 20/44] posix-cpu-timers: Simplify sample functions Thomas Gleixner
2019-08-19 14:32 ` [patch 21/44] posix-cpu-timers: Get rid of pointer indirection Thomas Gleixner
2019-08-19 14:32 ` [patch 22/44] posix-cpu-timers: Sample task times once in expiry check Thomas Gleixner
2019-08-19 14:32 ` [patch 23/44] posix-cpu-timers: Move prof/virt_ticks into caller Thomas Gleixner
2019-08-19 14:32 ` [patch 24/44] posix-cpu-timers: Create a container struct Thomas Gleixner
2019-08-19 14:32 ` [patch 25/44] sched: Move struct task_cputime to types.h Thomas Gleixner
2019-08-19 14:32 ` [patch 26/44] posix-cpu-timers: Move expiry cache into struct posix_cputimers Thomas Gleixner
2019-08-19 14:32 ` Thomas Gleixner [this message]
2019-08-19 19:32   ` [patch 27/44] posix-cpu-timers: Provide array based access to expiry cache Ingo Molnar
2019-08-20 20:22     ` Thomas Gleixner
2019-08-19 14:32 ` [patch 28/44] posix-cpu-timers: Simplify timer queueing Thomas Gleixner
2019-08-19 14:32 ` [patch 29/44] posix-cpu-timers: Simplify set_process_cpu_timer() Thomas Gleixner
2019-08-19 14:32 ` [patch 30/44] posix-cpu-timers: Switch check_*_timers() to array cache Thomas Gleixner
2019-08-19 14:32 ` [patch 31/44] posix-cpu-timers: Remove the odd field rename defines Thomas Gleixner
2019-08-19 14:32 ` [patch 32/44] posix-cpu-timers: Provide array based sample functions Thomas Gleixner
2019-08-19 14:32 ` [patch 33/44] posix-cpu-timers: Make expiry checks array based Thomas Gleixner
2019-08-19 14:32 ` [patch 34/44] posix-cpu-timers: Remove cputime_expires Thomas Gleixner
2019-08-19 14:32 ` [patch 35/44] posix-cpu-timers: Switch thread group sampling to array Thomas Gleixner
2019-08-19 14:32 ` [patch 36/44] posix-cpu-timers: Get rid of zero checks Thomas Gleixner
2019-08-19 14:32 ` [patch 37/44] posix-cpu-timers: Consolidate timer expiry further Thomas Gleixner
2019-08-19 14:32 ` [patch 38/44] posix-cpu-timers: Respect INFINITY for hard RTTIME limit Thomas Gleixner
2019-08-19 20:06   ` Ingo Molnar
2019-08-19 20:29     ` Thomas Gleixner
2019-08-19 14:32 ` [patch 39/44] posix-cpu-timers: Get rid of 64bit divisions Thomas Gleixner
2019-08-19 14:32 ` [patch 40/44] posix-cpu-timers: Remove pointless comparisions Thomas Gleixner
2019-08-19 20:10   ` Ingo Molnar
2019-08-19 14:32 ` [patch 41/44] posix-cpu-timers: Deduplicate rlimit handling Thomas Gleixner
2019-08-19 14:32 ` [patch 42/44] posix-cpu-timers: Move state tracking to struct posix_cputimers Thomas Gleixner
2019-08-19 19:13   ` Ingo Molnar
2019-08-19 20:29     ` Thomas Gleixner
2019-08-19 14:32 ` [patch 43/44] posix-cpu-timers: Utilize timerqueue for storage Thomas Gleixner
2019-08-19 14:32 ` [patch 44/44] posix-cpu-timers: Expire timers directly Thomas Gleixner
2019-08-19 19:09   ` Ingo Molnar
2019-08-20 13:07   ` Thomas Gleixner
2019-08-20  2:18 ` [patch 00/44] posix-cpu-timers: Cleanup and consolidation Christoph Hellwig
2019-08-20 13:09   ` Thomas Gleixner

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=20190819143803.961800814@linutronix.de \
    --to=tglx@linutronix.de \
    --cc=anna-maria@linutronix.de \
    --cc=fweisbec@gmail.com \
    --cc=john.stultz@linaro.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=oleg@redhat.com \
    --cc=peterz@infradead.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).