From: nadav.amit@gmail.com
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: kvm@vger.kernel.org, Nadav Amit <nadav.amit@gmail.com>,
Krish Sadhukhan <krish.sadhukhan@oracle.com>
Subject: [PATCH v2] x86: Some cleanup of delay() and io_delay()
Date: Fri, 3 May 2019 04:13:07 -0700 [thread overview]
Message-ID: <20190503111307.10716-1-nadav.amit@gmail.com> (raw)
From: Nadav Amit <nadav.amit@gmail.com>
There is no guarantee that a self-IPI would be delivered immediately.
In eventinj, io_delay() is called after self-IPI is generated but does
nothing.
In general, there is mess in regard to delay() and io_delay(). There are
two definitions of delay() and they do not really look on the timestamp
counter and instead count invocations of "pause" (or even "nop"), which
might be different on different CPUs/setups, for example due to
different pause-loop-exiting configurations.
To address these issues change io_delay() to really do a delay, based on
timestamp counter, and move common functions into delay.[hc].
Cc: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Signed-off-by: Nadav Amit <nadav.amit@gmail.com>
---
lib/x86/delay.c | 9 ++++++---
lib/x86/delay.h | 7 +++++++
x86/eventinj.c | 5 +----
x86/ioapic.c | 8 +-------
4 files changed, 15 insertions(+), 14 deletions(-)
diff --git a/lib/x86/delay.c b/lib/x86/delay.c
index 595ad24..e7d2717 100644
--- a/lib/x86/delay.c
+++ b/lib/x86/delay.c
@@ -1,8 +1,11 @@
#include "delay.h"
+#include "processor.h"
void delay(u64 count)
{
- while (count--)
- asm volatile("pause");
-}
+ u64 start = rdtsc();
+ do {
+ pause();
+ } while (rdtsc() - start < count);
+}
diff --git a/lib/x86/delay.h b/lib/x86/delay.h
index a9bf894..a51eb34 100644
--- a/lib/x86/delay.h
+++ b/lib/x86/delay.h
@@ -3,6 +3,13 @@
#include "libcflat.h"
+#define IPI_DELAY 1000000
+
void delay(u64 count);
+static inline void io_delay(void)
+{
+ delay(IPI_DELAY);
+}
+
#endif
diff --git a/x86/eventinj.c b/x86/eventinj.c
index d2dfc40..901b9db 100644
--- a/x86/eventinj.c
+++ b/x86/eventinj.c
@@ -7,6 +7,7 @@
#include "apic-defs.h"
#include "vmalloc.h"
#include "alloc_page.h"
+#include "delay.h"
#ifdef __x86_64__
# define R "r"
@@ -16,10 +17,6 @@
void do_pf_tss(void);
-static inline void io_delay(void)
-{
-}
-
static void apic_self_ipi(u8 v)
{
apic_icr_write(APIC_DEST_SELF | APIC_DEST_PHYSICAL | APIC_DM_FIXED |
diff --git a/x86/ioapic.c b/x86/ioapic.c
index 2ac4ac6..c32dabd 100644
--- a/x86/ioapic.c
+++ b/x86/ioapic.c
@@ -4,6 +4,7 @@
#include "smp.h"
#include "desc.h"
#include "isr.h"
+#include "delay.h"
static void toggle_irq_line(unsigned line)
{
@@ -165,13 +166,6 @@ static void test_ioapic_level_tmr(bool expected_tmr_before)
expected_tmr_before ? "true" : "false");
}
-#define IPI_DELAY 1000000
-
-static void delay(int count)
-{
- while(count--) asm("");
-}
-
static void toggle_irq_line_0x0e(void *data)
{
irq_disable();
--
2.17.1
next reply other threads:[~2019-05-03 18:36 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-05-03 11:13 nadav.amit [this message]
2019-05-03 18:36 ` [PATCH v2] x86: Some cleanup of delay() and io_delay() Nadav Amit
2019-05-03 18:57 ` Krish Sadhukhan
2019-05-03 19:00 ` Nadav Amit
2019-05-20 14:15 ` 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=20190503111307.10716-1-nadav.amit@gmail.com \
--to=nadav.amit@gmail.com \
--cc=krish.sadhukhan@oracle.com \
--cc=kvm@vger.kernel.org \
--cc=pbonzini@redhat.com \
/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