All of lore.kernel.org
 help / color / mirror / Atom feed
From: Pavel Tatashin <pasha.tatashin@oracle.com>
To: steven.sistare@oracle.com, daniel.m.jordan@oracle.com,
	linux@armlinux.org.uk, schwidefsky@de.ibm.com,
	heiko.carstens@de.ibm.com, john.stultz@linaro.org,
	sboyd@codeaurora.org, x86@kernel.org,
	linux-kernel@vger.kernel.org, mingo@redhat.com,
	tglx@linutronix.de, hpa@zytor.com, douly.fnst@cn.fujitsu.com,
	peterz@infradead.org, prarit@redhat.com, feng.tang@intel.com,
	pmladek@suse.com, gnomes@lxorguk.ukuu.org.uk,
	linux-s390@vger.kernel.org
Subject: [PATCH v12 09/11] x86/tsc: prepare for early sched_clock
Date: Thu, 21 Jun 2018 17:25:16 -0400	[thread overview]
Message-ID: <20180621212518.19914-10-pasha.tatashin@oracle.com> (raw)
In-Reply-To: <20180621212518.19914-1-pasha.tatashin@oracle.com>

We will change sched_clock() to be called early. But, during boot
sched_clock() changes its output without notifying us about the change of
clock source.

This happens in tsc_init(), when static_branch_enable(&__use_tsc) is
called.

native_sched_clock() changes from outputing jiffies to reading tsc, but
sched is not notified in anyway. So, to preserve the continoutity in this
place we add the offset of sched_clock() to the calculation of cyc2ns.

Without this change, the output would look like this:

[    0.004000] ..TIMER: vector=0x30 apic1=0 pin1=2 apic2=-1 pin2=-1
[    0.009000] tsc: Fast TSC calibration using PIT
[    0.010000] tsc: Detected 3192.137 MHz processor
[    0.011000] clocksource: tsc-early: mask: 0xffffffffffffffff max_cycles: 0x2e03465ceb2,    max_idle_ns: 440795259855 ns

static_branch_enable(__use_tsc) is called, and timestamps became precise
but reduced:

[    0.002233] Calibrating delay loop (skipped), value calculated using timer frequency..     6384.27 BogoMIPS (lpj=3192137)
[    0.002516] pid_max: default: 32768 minimum: 301

Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>
---
 arch/x86/kernel/tsc.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index 186395041725..654a01cc0358 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -133,7 +133,9 @@ static inline unsigned long long cycles_2_ns(unsigned long long cyc)
 	return ns;
 }
 
-static void set_cyc2ns_scale(unsigned long khz, int cpu, unsigned long long tsc_now)
+static void set_cyc2ns_scale(unsigned long khz, int cpu,
+			     unsigned long long tsc_now,
+			     unsigned long long sched_now)
 {
 	unsigned long long ns_now;
 	struct cyc2ns_data data;
@@ -146,7 +148,7 @@ static void set_cyc2ns_scale(unsigned long khz, int cpu, unsigned long long tsc_
 	if (!khz)
 		goto done;
 
-	ns_now = cycles_2_ns(tsc_now);
+	ns_now = cycles_2_ns(tsc_now) + sched_now;
 
 	/*
 	 * Compute a new multiplier as per the above comment and ensure our
@@ -936,7 +938,7 @@ static int time_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
 		if (!(freq->flags & CPUFREQ_CONST_LOOPS))
 			mark_tsc_unstable("cpufreq changes");
 
-		set_cyc2ns_scale(tsc_khz, freq->cpu, rdtsc());
+		set_cyc2ns_scale(tsc_khz, freq->cpu, rdtsc(), 0);
 	}
 
 	return 0;
@@ -1285,7 +1287,7 @@ static void tsc_refine_calibration_work(struct work_struct *work)
 
 	/* Update the sched_clock() rate to match the clocksource one */
 	for_each_possible_cpu(cpu)
-		set_cyc2ns_scale(tsc_khz, cpu, tsc_stop);
+		set_cyc2ns_scale(tsc_khz, cpu, tsc_stop, 0);
 
 out:
 	if (tsc_unstable)
@@ -1356,7 +1358,7 @@ void __init tsc_early_delay_calibrate(void)
 
 void __init tsc_init(void)
 {
-	u64 lpj, cyc;
+	u64 lpj, cyc, sch;
 	int cpu;
 
 	if (!boot_cpu_has(X86_FEATURE_TSC)) {
@@ -1403,9 +1405,10 @@ void __init tsc_init(void)
 	 * up if their speed diverges)
 	 */
 	cyc = rdtsc();
+	sch = local_clock();
 	for_each_possible_cpu(cpu) {
 		cyc2ns_init(cpu);
-		set_cyc2ns_scale(tsc_khz, cpu, cyc);
+		set_cyc2ns_scale(tsc_khz, cpu, cyc, sch);
 	}
 
 	static_branch_enable(&__use_tsc);
-- 
2.17.1


  parent reply	other threads:[~2018-06-21 21:27 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-06-21 21:25 [PATCH v12 00/11] Early boot time stamps for x86 Pavel Tatashin
2018-06-21 21:25 ` [PATCH v12 01/11] x86: text_poke() may access uninitialized struct pages Pavel Tatashin
2018-06-21 21:37   ` Randy Dunlap
2018-06-25  8:14   ` Peter Zijlstra
2018-06-25  8:39     ` Thomas Gleixner
2018-06-25  9:09       ` Peter Zijlstra
2018-06-25  9:18         ` Thomas Gleixner
2018-06-25  9:22           ` Peter Zijlstra
2018-06-25 12:32             ` Pavel Tatashin
2018-06-25 13:48               ` Peter Zijlstra
2018-06-25 14:06                 ` Pavel Tatashin
2018-06-21 21:25 ` [PATCH v12 02/11] x86: initialize static branching early Pavel Tatashin
2018-06-23  9:16   ` Borislav Petkov
2018-06-23 13:11     ` Pavel Tatashin
2018-06-21 21:25 ` [PATCH v12 03/11] x86/tsc: redefine notsc to behave as tsc=unstable Pavel Tatashin
2018-06-23 13:32   ` Thomas Gleixner
2018-06-21 21:25 ` [PATCH v12 04/11] kvm/x86: remove kvm memblock dependency Pavel Tatashin
2018-06-23 13:36   ` Thomas Gleixner
2018-07-05 16:12   ` Paolo Bonzini
2018-07-06  9:24     ` Thomas Gleixner
2018-07-06  9:36       ` Paolo Bonzini
2018-07-06  9:45         ` Thomas Gleixner
2018-07-06 10:08           ` Paolo Bonzini
2018-07-06 10:44             ` Thomas Gleixner
2018-07-06 10:50               ` Thomas Gleixner
2018-07-06 15:03                 ` Pavel Tatashin
2018-07-06 15:09                   ` Paolo Bonzini
2018-06-21 21:25 ` [PATCH v12 05/11] s390/time: add read_persistent_wall_and_boot_offset() Pavel Tatashin
2018-06-25  7:07   ` Martin Schwidefsky
2018-06-25 12:45     ` Pavel Tatashin
2018-06-21 21:25 ` [PATCH v12 06/11] time: replace read_boot_clock64() with read_persistent_wall_and_boot_offset() Pavel Tatashin
2018-06-23 13:49   ` Thomas Gleixner
2018-06-21 21:25 ` [PATCH v12 07/11] s390/time: remove read_boot_clock64() Pavel Tatashin
2018-06-21 21:25 ` [PATCH v12 08/11] ARM/time: " Pavel Tatashin
2018-06-23 13:52   ` Thomas Gleixner
2018-06-21 21:25 ` Pavel Tatashin [this message]
2018-06-23 16:50   ` [PATCH v12 09/11] x86/tsc: prepare for early sched_clock Thomas Gleixner
2018-06-23 18:49     ` Pavel Tatashin
2018-06-23 20:11     ` Thomas Gleixner
2018-06-23 21:29       ` Pavel Tatashin
2018-06-23 23:38         ` Thomas Gleixner
2018-06-24  2:43           ` Pavel Tatashin
2018-06-24  7:30             ` Thomas Gleixner
2018-06-26 15:42           ` Thomas Gleixner
2018-06-26 18:42             ` Pavel Tatashin
2018-06-26 19:47               ` Pavel Tatashin
2018-06-28  7:31               ` Thomas Gleixner
2018-06-28 10:43                 ` Thomas Gleixner
2018-06-28 11:46                   ` Peter Zijlstra
2018-06-28 12:27                     ` Thomas Gleixner
2018-06-28 19:42                   ` Pavel Tatashin
2018-06-29  7:30                     ` Thomas Gleixner
2018-06-29  8:57                       ` Pavel Tatashin
2018-07-03 20:59                         ` Thomas Gleixner
2018-07-02 17:18                       ` Konrad Rzeszutek Wilk
2018-06-29 14:30                   ` Andy Shevchenko
2018-06-29 17:50                     ` Andy Shevchenko
2018-07-09 23:16                   ` Boris Ostrovsky
2018-06-21 21:25 ` [PATCH v12 10/11] sched: early boot clock Pavel Tatashin
2018-06-25  8:55   ` Peter Zijlstra
2018-06-25 12:44     ` Pavel Tatashin
2018-06-25 19:23     ` Pavel Tatashin
2018-06-26  9:00       ` Peter Zijlstra
2018-06-26 11:27         ` Pavel Tatashin
2018-06-26 11:51           ` Pavel Tatashin
2018-06-26 15:07           ` Peter Zijlstra
2018-06-21 21:25 ` [PATCH v12 11/11] x86/tsc: use tsc early Pavel Tatashin
2018-06-23 16:56   ` Thomas Gleixner
2018-06-23 21:38     ` Pavel Tatashin

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=20180621212518.19914-10-pasha.tatashin@oracle.com \
    --to=pasha.tatashin@oracle.com \
    --cc=daniel.m.jordan@oracle.com \
    --cc=douly.fnst@cn.fujitsu.com \
    --cc=feng.tang@intel.com \
    --cc=gnomes@lxorguk.ukuu.org.uk \
    --cc=heiko.carstens@de.ibm.com \
    --cc=hpa@zytor.com \
    --cc=john.stultz@linaro.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-s390@vger.kernel.org \
    --cc=linux@armlinux.org.uk \
    --cc=mingo@redhat.com \
    --cc=peterz@infradead.org \
    --cc=pmladek@suse.com \
    --cc=prarit@redhat.com \
    --cc=sboyd@codeaurora.org \
    --cc=schwidefsky@de.ibm.com \
    --cc=steven.sistare@oracle.com \
    --cc=tglx@linutronix.de \
    --cc=x86@kernel.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.