From: Thomas Gleixner <tglx@linutronix.de>
To: LKML <linux-kernel@vger.kernel.org>
Cc: Andy Lutomirski <luto@kernel.org>,
x86@kernel.org, Peter Zijlstra <peterz@infradead.org>,
Matt Rickard <matt@softrans.com.au>,
Stephen Boyd <sboyd@kernel.org>,
John Stultz <john.stultz@linaro.org>,
Florian Weimer <fweimer@redhat.com>,
"K. Y. Srinivasan" <kys@microsoft.com>,
Vitaly Kuznetsov <vkuznets@redhat.com>,
devel@linuxdriverproject.org,
virtualization@lists.linux-foundation.org,
Paolo Bonzini <pbonzini@redhat.com>,
Arnd Bergmann <arnd@arndb.de>, Juergen Gross <jgross@suse.com>
Subject: [patch 10/11] x86/vdso: Move cycle_last handling into the caller
Date: Fri, 14 Sep 2018 14:50:16 +0200 [thread overview]
Message-ID: <20180914125118.998589817@linutronix.de> (raw)
In-Reply-To: 20180914125006.349747096@linutronix.de
[-- Attachment #1: x86-vdso--Move-cycle_last-handling-into-the-caller.patch --]
[-- Type: text/plain, Size: 2778 bytes --]
Dereferencing gtod->cycle_last all over the place and foing the cycles <
last comparison in the vclock read functions generates horrible code. Doing
it at the call site is much better and gains a few cycles both for TSC and
pvclock.
Caveat: This adds the comparison to the hyperv vclock as well, but I have
no way to test that.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
arch/x86/entry/vdso/vclock_gettime.c | 39 ++++++-----------------------------
1 file changed, 7 insertions(+), 32 deletions(-)
--- a/arch/x86/entry/vdso/vclock_gettime.c
+++ b/arch/x86/entry/vdso/vclock_gettime.c
@@ -76,9 +76,8 @@ static notrace const struct pvclock_vsys
static notrace u64 vread_pvclock(void)
{
const struct pvclock_vcpu_time_info *pvti = &get_pvti0()->pvti;
- u64 ret;
- u64 last;
u32 version;
+ u64 ret;
/*
* Note: The kernel and hypervisor must guarantee that cpu ID
@@ -111,13 +110,7 @@ static notrace u64 vread_pvclock(void)
ret = __pvclock_read_cycles(pvti, rdtsc_ordered());
} while (pvclock_read_retry(pvti, version));
- /* refer to vread_tsc() comment for rationale */
- last = gtod->cycle_last;
-
- if (likely(ret >= last))
- return ret;
-
- return last;
+ return ret;
}
#endif
#ifdef CONFIG_HYPERV_TSCPAGE
@@ -130,30 +123,10 @@ static notrace u64 vread_hvclock(void)
}
#endif
-notrace static u64 vread_tsc(void)
-{
- u64 ret = (u64)rdtsc_ordered();
- u64 last = gtod->cycle_last;
-
- if (likely(ret >= last))
- return ret;
-
- /*
- * GCC likes to generate cmov here, but this branch is extremely
- * predictable (it's just a function of time and the likely is
- * very likely) and there's a data dependence, so force GCC
- * to generate a branch instead. I don't barrier() because
- * we don't actually need a barrier, and if this function
- * ever gets inlined it will generate worse code.
- */
- asm volatile ("");
- return last;
-}
-
notrace static inline u64 vgetcyc(int mode)
{
if (mode == VCLOCK_TSC)
- return vread_tsc();
+ return (u64)rdtsc_ordered();
#ifdef CONFIG_PARAVIRT_CLOCK
else if (mode == VCLOCK_PVCLOCK)
return vread_pvclock();
@@ -168,17 +141,19 @@ notrace static inline u64 vgetcyc(int mo
notrace static int do_hres(clockid_t clk, struct timespec *ts)
{
struct vgtod_ts *base = >od->basetime[clk];
+ u64 cycles, last, ns;
unsigned int seq;
- u64 cycles, ns;
do {
seq = gtod_read_begin(gtod);
ts->tv_sec = base->sec;
ns = base->nsec;
+ last = gtod->cycle_last;
cycles = vgetcyc(gtod->vclock_mode);
if (unlikely((s64)cycles < 0))
return vdso_fallback_gettime(clk, ts);
- ns += (cycles - gtod->cycle_last) * gtod->mult;
+ if (cycles > last)
+ ns += (cycles - last) * gtod->mult;
ns >>= gtod->shift;
} while (unlikely(gtod_read_retry(gtod, seq)));
next prev parent reply other threads:[~2018-09-14 12:52 UTC|newest]
Thread overview: 81+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-09-14 12:50 [patch 00/11] x86/vdso: Cleanups, simmplifications and CLOCK_TAI support Thomas Gleixner
2018-09-14 12:50 ` [patch 01/11] clocksource: Provide clocksource_arch_init() Thomas Gleixner
2018-09-14 12:50 ` [patch 02/11] x86/time: Implement clocksource_arch_init() Thomas Gleixner
2018-09-14 15:45 ` Vitaly Kuznetsov
2018-09-15 6:05 ` Thomas Gleixner
2018-09-14 12:50 ` [patch 03/11] x86/vdso: Enforce 64bit clocksource Thomas Gleixner
2018-09-14 12:50 ` [patch 04/11] x86/vdso: Use unsigned int consistently for vsyscall_gtod_data::seq Thomas Gleixner
2018-09-14 12:50 ` [patch 05/11] x86/vdso: Introduce and use vgtod_ts Thomas Gleixner
2018-09-14 12:50 ` [patch 06/11] x86/vdso: Collapse high resolution functions Thomas Gleixner
2018-09-14 12:50 ` [patch 07/11] x86/vdso: Collapse coarse functions Thomas Gleixner
2018-09-14 12:50 ` [patch 08/11] x86/vdso: Replace the clockid switch case Thomas Gleixner
2018-09-14 12:50 ` [patch 09/11] x86/vdso: Simplify the invalid vclock case Thomas Gleixner
2018-09-17 19:25 ` Andy Lutomirski
2018-09-17 20:12 ` John Stultz
2018-09-18 7:52 ` Thomas Gleixner
2018-09-18 8:30 ` Peter Zijlstra
2018-09-18 8:52 ` Thomas Gleixner
2018-09-18 10:06 ` Thomas Gleixner
2018-09-18 10:41 ` Thomas Gleixner
2018-09-18 12:48 ` Peter Zijlstra
2018-09-18 13:23 ` Thomas Gleixner
2018-09-18 13:38 ` Peter Zijlstra
2018-09-18 15:52 ` Thomas Gleixner
2018-09-27 14:41 ` Thomas Gleixner
2018-09-18 14:01 ` Andy Lutomirski
2018-09-18 22:46 ` Thomas Gleixner
2018-09-18 23:03 ` Andy Lutomirski
2018-09-18 23:16 ` Thomas Gleixner
2018-09-27 14:36 ` Thomas Gleixner
2018-09-27 14:39 ` Andy Lutomirski
2018-09-19 9:08 ` Rasmus Villemoes
2018-09-19 13:29 ` Thomas Gleixner
2018-09-14 12:50 ` Thomas Gleixner [this message]
2018-09-14 15:26 ` [patch 10/11] x86/vdso: Move cycle_last handling into the caller Vitaly Kuznetsov
2018-09-14 12:50 ` [patch 11/11] x66/vdso: Add CLOCK_TAI support Thomas Gleixner
2018-09-14 14:04 ` Andy Lutomirski
2018-09-14 14:27 ` Thomas Gleixner
2018-09-14 14:59 ` Andy Lutomirski
2018-09-16 9:39 ` Thomas Gleixner
2018-09-14 12:56 ` [patch 00/11] x86/vdso: Cleanups, simmplifications and " Florian Weimer
2018-09-14 13:05 ` Thomas Gleixner
2018-09-14 13:06 ` Florian Weimer
2018-09-14 13:19 ` Thomas Gleixner
2018-09-14 13:09 ` Peter Zijlstra
2018-09-14 14:22 ` Arnd Bergmann
2018-09-17 13:00 ` Thomas Gleixner
2018-09-24 21:08 ` Arnd Bergmann
2018-10-03 5:15 ` Andy Lutomirski
2018-10-03 9:22 ` Vitaly Kuznetsov
2018-10-03 10:20 ` Andy Lutomirski
2018-10-03 12:01 ` Vitaly Kuznetsov
2018-10-03 14:20 ` Andy Lutomirski
2018-10-03 15:10 ` Thomas Gleixner
2018-10-03 16:18 ` Andy Lutomirski
2018-10-03 19:06 ` Marcelo Tosatti
2018-10-04 7:54 ` Vitaly Kuznetsov
2018-10-04 8:11 ` Peter Zijlstra
2018-10-04 14:00 ` Andy Lutomirski
2018-10-04 19:31 ` Peter Zijlstra
2018-10-04 20:05 ` Andy Lutomirski
2018-10-04 22:15 ` Andy Lutomirski
2018-10-06 20:27 ` Marcelo Tosatti
2018-10-06 22:28 ` Andy Lutomirski
2018-10-08 15:26 ` Marcelo Tosatti
2018-10-08 17:38 ` Andy Lutomirski
2018-10-08 19:36 ` Marcelo Tosatti
2018-10-09 20:09 ` Andy Lutomirski
2018-10-11 22:27 ` Marcelo Tosatti
2018-10-11 23:00 ` Andy Lutomirski
2018-10-15 13:39 ` Marcelo Tosatti
2018-10-06 20:49 ` Marcelo Tosatti
2018-10-04 12:00 ` Paolo Bonzini
2018-10-04 14:04 ` Andy Lutomirski
2018-10-05 21:18 ` Marcelo Tosatti
2018-10-03 19:00 ` Marcelo Tosatti
2018-10-03 19:05 ` [patch 00/11] x86/vdso: Cleanups, simmplifications and CLOCK_TAI support\ Marcelo Tosatti
2018-10-03 22:32 ` [patch 00/11] x86/vdso: Cleanups, simmplifications and CLOCK_TAI support Andy Lutomirski
2018-10-04 16:37 ` Marcelo Tosatti
2018-10-04 17:08 ` Andy Lutomirski
2018-10-04 17:28 ` Vitaly Kuznetsov
2018-10-04 20:32 ` Andy Lutomirski
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=20180914125118.998589817@linutronix.de \
--to=tglx@linutronix.de \
--cc=arnd@arndb.de \
--cc=devel@linuxdriverproject.org \
--cc=fweimer@redhat.com \
--cc=jgross@suse.com \
--cc=john.stultz@linaro.org \
--cc=kys@microsoft.com \
--cc=linux-kernel@vger.kernel.org \
--cc=luto@kernel.org \
--cc=matt@softrans.com.au \
--cc=pbonzini@redhat.com \
--cc=peterz@infradead.org \
--cc=sboyd@kernel.org \
--cc=virtualization@lists.linux-foundation.org \
--cc=vkuznets@redhat.com \
--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 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).