From: Andy Lutomirski <luto@kernel.org>
To: x86@kernel.org
Cc: Borislav Petkov <bp@suse.de>,
Peter Zijlstra <peterz@infradead.org>,
John Stultz <john.stultz@linaro.org>,
linux-kernel@vger.kernel.org, Len Brown <lenb@kernel.org>,
Huang Rui <ray.huang@amd.com>,
Denys Vlasenko <dvlasenk@redhat.com>,
kvm@vger.kernel.org, Ralf Baechle <ralf@linux-mips.org>,
Andy Lutomirski <luto@kernel.org>
Subject: [PATCH v3 14/18] x86: Add rdtsc_ordered() and use it in trivial call sites
Date: Tue, 16 Jun 2015 17:36:02 -0700 [thread overview]
Message-ID: <dddbf98a2af53312e9aa73a5a2b1622fe5d6f52b.1434501121.git.luto@kernel.org> (raw)
In-Reply-To: <cover.1434501120.git.luto@kernel.org>
In-Reply-To: <cover.1434501120.git.luto@kernel.org>
rdtsc_barrier(); rdtsc() is an unnecessary mouthful and requires
more thought than should be necessary. Add an rdtsc_ordered()
helper and replace the trivial call sites with it.
This should not change generated code. The duplication of the fence
asm is temporary.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
arch/x86/entry/vdso/vclock_gettime.c | 16 ++--------------
arch/x86/include/asm/msr.h | 26 ++++++++++++++++++++++++++
arch/x86/kernel/trace_clock.c | 7 +------
arch/x86/kvm/x86.c | 16 ++--------------
arch/x86/lib/delay.c | 9 +++------
5 files changed, 34 insertions(+), 40 deletions(-)
diff --git a/arch/x86/entry/vdso/vclock_gettime.c b/arch/x86/entry/vdso/vclock_gettime.c
index 0340d93c18ca..ca94fa649251 100644
--- a/arch/x86/entry/vdso/vclock_gettime.c
+++ b/arch/x86/entry/vdso/vclock_gettime.c
@@ -175,20 +175,8 @@ static notrace cycle_t vread_pvclock(int *mode)
notrace static cycle_t vread_tsc(void)
{
- cycle_t ret;
- u64 last;
-
- /*
- * Empirically, a fence (of type that depends on the CPU)
- * before rdtsc is enough to ensure that rdtsc is ordered
- * with respect to loads. The various CPU manuals are unclear
- * as to whether rdtsc can be reordered with later loads,
- * but no one has ever seen it happen.
- */
- rdtsc_barrier();
- ret = (cycle_t)rdtsc();
-
- last = gtod->cycle_last;
+ cycle_t ret = (cycle_t)rdtsc_ordered();
+ u64 last = gtod->cycle_last;
if (likely(ret >= last))
return ret;
diff --git a/arch/x86/include/asm/msr.h b/arch/x86/include/asm/msr.h
index ff0c120dafe5..02bdd6c65017 100644
--- a/arch/x86/include/asm/msr.h
+++ b/arch/x86/include/asm/msr.h
@@ -127,6 +127,32 @@ static __always_inline unsigned long long rdtsc(void)
return EAX_EDX_VAL(val, low, high);
}
+/**
+ * rdtsc_ordered() - read the current TSC in program order
+ *
+ * rdtsc_ordered() returns the result of RDTSC as a 64-bit integer.
+ * It is ordered like a load to a global in-memory counter. It should
+ * be impossible to observe non-monotonic rdtsc_unordered() behavior
+ * across multiple CPUs as long as the TSC is synced.
+ */
+static __always_inline unsigned long long rdtsc_ordered(void)
+{
+ /*
+ * The RDTSC instruction is not ordered relative to memory
+ * access. The Intel SDM and the AMD APM are both vague on this
+ * point, but empirically an RDTSC instruction can be
+ * speculatively executed before prior loads. An RDTSC
+ * immediately after an appropriate barrier appears to be
+ * ordered as a normal load, that is, it provides the same
+ * ordering guarantees as reading from a global memory location
+ * that some other imaginary CPU is updating continuously with a
+ * time stamp.
+ */
+ alternative_2("", "mfence", X86_FEATURE_MFENCE_RDTSC,
+ "lfence", X86_FEATURE_LFENCE_RDTSC);
+ return rdtsc();
+}
+
static inline unsigned long long native_read_pmc(int counter)
{
DECLARE_ARGS(val, low, high);
diff --git a/arch/x86/kernel/trace_clock.c b/arch/x86/kernel/trace_clock.c
index 67efb8c96fc4..80bb24d9b880 100644
--- a/arch/x86/kernel/trace_clock.c
+++ b/arch/x86/kernel/trace_clock.c
@@ -12,10 +12,5 @@
*/
u64 notrace trace_clock_x86_tsc(void)
{
- u64 ret;
-
- rdtsc_barrier();
- ret = rdtsc();
-
- return ret;
+ return rdtsc_ordered();
}
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index b0afdc74c28a..dfccaf2f2e00 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -1419,20 +1419,8 @@ EXPORT_SYMBOL_GPL(kvm_write_tsc);
static cycle_t read_tsc(void)
{
- cycle_t ret;
- u64 last;
-
- /*
- * Empirically, a fence (of type that depends on the CPU)
- * before rdtsc is enough to ensure that rdtsc is ordered
- * with respect to loads. The various CPU manuals are unclear
- * as to whether rdtsc can be reordered with later loads,
- * but no one has ever seen it happen.
- */
- rdtsc_barrier();
- ret = (cycle_t)rdtsc();
-
- last = pvclock_gtod_data.clock.cycle_last;
+ cycle_t ret = (cycle_t)rdtsc_ordered();
+ u64 last = pvclock_gtod_data.clock.cycle_last;
if (likely(ret >= last))
return ret;
diff --git a/arch/x86/lib/delay.c b/arch/x86/lib/delay.c
index f24bc59ab0a0..4453d52a143d 100644
--- a/arch/x86/lib/delay.c
+++ b/arch/x86/lib/delay.c
@@ -54,11 +54,9 @@ static void delay_tsc(unsigned long __loops)
preempt_disable();
cpu = smp_processor_id();
- rdtsc_barrier();
- bclock = rdtsc();
+ bclock = rdtsc_ordered();
for (;;) {
- rdtsc_barrier();
- now = rdtsc();
+ now = rdtsc_ordered();
if ((now - bclock) >= loops)
break;
@@ -79,8 +77,7 @@ static void delay_tsc(unsigned long __loops)
if (unlikely(cpu != smp_processor_id())) {
loops -= (now - bclock);
cpu = smp_processor_id();
- rdtsc_barrier();
- bclock = rdtsc();
+ bclock = rdtsc_ordered();
}
}
preempt_enable();
--
2.4.2
next prev parent reply other threads:[~2015-06-17 0:37 UTC|newest]
Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-06-17 0:35 [PATCH v3 00/18] x86/tsc: Clean up rdtsc helpers Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 01/18] x86/tsc: Inline native_read_tsc and remove __native_read_tsc Andy Lutomirski
2015-06-17 9:26 ` Borislav Petkov
2015-07-06 15:39 ` [tip:x86/asm] x86/asm/tsc: Inline native_read_tsc() and remove __native_read_tsc() tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 02/18] x86/msr/kvm: Remove vget_cycles() Andy Lutomirski
2015-06-17 9:42 ` Borislav Petkov
2015-06-17 13:34 ` Paolo Bonzini
2015-07-06 15:40 ` [tip:x86/asm] x86/asm/tsc, kvm: " tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 03/18] x86/tsc/paravirt: Remove the read_tsc and read_tscp paravirt hooks Andy Lutomirski
2015-06-17 9:56 ` Borislav Petkov
2015-06-19 15:32 ` Borislav Petkov
2015-06-19 16:14 ` Andy Lutomirski
2015-06-19 17:13 ` Borislav Petkov
2015-07-06 15:40 ` [tip:x86/asm] x86/asm/tsc, x86/paravirt: Remove read_tsc() and read_tscp() " tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 04/18] x86/tsc: Replace rdtscll with native_read_tsc Andy Lutomirski
2015-06-17 10:03 ` Borislav Petkov
2015-07-06 15:40 ` [tip:x86/asm] x86/asm/tsc: Replace rdtscll() with native_read_tsc () tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 05/18] x86/tsc: Remove the rdtscp and rdtscpll macros Andy Lutomirski
2015-07-06 15:41 ` [tip:x86/asm] x86/asm/tsc: Remove the rdtscp() and rdtscpll() macros tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 06/18] x86/tsc: Use the full 64-bit tsc in tsc_delay Andy Lutomirski
2015-07-06 15:41 ` [tip:x86/asm] x86/asm/tsc: Use the full 64-bit TSC in delay_tsc() tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 07/18] x86/cpu/amd: Use the full 64-bit TSC to detect the 2.6.2 bug Andy Lutomirski
2015-07-06 15:41 ` [tip:x86/asm] x86/asm/tsc, " tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 08/18] baycom_epp: Replace rdtscl() with native_read_tsc() Andy Lutomirski
2015-06-17 0:49 ` Thomas Sailer
2015-06-20 13:54 ` walter harms
2015-06-20 14:14 ` Thomas Gleixner
2015-06-20 14:26 ` Andy Lutomirski
2015-06-20 16:30 ` Thomas Gleixner
2015-07-06 15:42 ` [tip:x86/asm] x86/asm/tsc, drivers/net/hamradio/baycom_epp: " tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 09/18] staging/lirc_serial: Remove TSC-based timing Andy Lutomirski
2015-07-06 15:42 ` [tip:x86/asm] x86/asm/tsc, " tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 10/18] input/joystick/analog: Switch from rdtscl() to native_read_tsc() Andy Lutomirski
2015-07-06 15:42 ` [tip:x86/asm] x86/asm/tsc, " tip-bot for Andy Lutomirski
2015-06-17 0:35 ` [PATCH v3 11/18] drivers/input/gameport: Replace rdtscl() with native_read_tsc() Andy Lutomirski
2015-07-06 15:43 ` [tip:x86/asm] x86/asm/tsc, drivers/input/gameport: Replace rdtscl () " tip-bot for Andy Lutomirski
2015-06-17 0:36 ` [PATCH v3 12/18] x86/tsc: Remove rdtscl() Andy Lutomirski
2015-07-06 15:43 ` [tip:x86/asm] x86/asm/tsc: " tip-bot for Andy Lutomirski
2015-06-17 0:36 ` [PATCH v3 13/18] x86/tsc: Rename native_read_tsc() to rdtsc() Andy Lutomirski
2015-06-24 21:38 ` Borislav Petkov
2015-07-06 15:43 ` [tip:x86/asm] x86/asm/tsc: " tip-bot for Andy Lutomirski
2015-06-17 0:36 ` Andy Lutomirski [this message]
2015-07-06 15:44 ` [tip:x86/asm] x86/asm/tsc: Add rdtsc_ordered() and use it in trivial call sites tip-bot for Andy Lutomirski
2015-08-21 7:45 ` [tip:x86/asm] x86/asm/tsc: Add rdtscll() merge helper tip-bot for Ingo Molnar
2015-06-17 0:36 ` [PATCH v3 15/18] x86/tsc: Use rdtsc_ordered() in check_tsc_warp() and drop extra barriers Andy Lutomirski
2015-07-06 15:44 ` [tip:x86/asm] x86/asm/tsc/sync: " tip-bot for Andy Lutomirski
2015-06-17 0:36 ` [PATCH v3 16/18] x86/tsc: In read_tsc, use rdtsc_ordered() instead of get_cycles() Andy Lutomirski
2015-07-06 15:44 ` [tip:x86/asm] x86/asm/tsc: Use rdtsc_ordered() in read_tsc() " tip-bot for Andy Lutomirski
2015-06-17 0:36 ` [PATCH v3 17/18] x86/kvm/tsc: Drop extra barrier and use rdtsc_ordered in kvmclock Andy Lutomirski
2015-06-17 7:47 ` Paolo Bonzini
2015-06-17 13:31 ` Paolo Bonzini
2015-06-20 21:50 ` Borislav Petkov
2015-07-06 15:45 ` [tip:x86/asm] x86/asm/tsc, x86/kvm: Drop open-coded barrier and use rdtsc_ordered() " tip-bot for Andy Lutomirski
2015-06-17 0:36 ` [PATCH v3 18/18] x86/tsc: Remove rdtsc_barrier() Andy Lutomirski
2015-07-06 15:45 ` [tip:x86/asm] x86/asm/tsc: " tip-bot for Andy Lutomirski
2015-06-17 11:11 ` [PATCH v3 00/18] x86/tsc: Clean up rdtsc helpers Borislav Petkov
2015-06-17 13:37 ` Paolo Bonzini
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=dddbf98a2af53312e9aa73a5a2b1622fe5d6f52b.1434501121.git.luto@kernel.org \
--to=luto@kernel.org \
--cc=bp@suse.de \
--cc=dvlasenk@redhat.com \
--cc=john.stultz@linaro.org \
--cc=kvm@vger.kernel.org \
--cc=lenb@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=peterz@infradead.org \
--cc=ralf@linux-mips.org \
--cc=ray.huang@amd.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).