qemu-kvm/SOURCES/kvm-main-loop-drop-spin_counter.patch

107 lines
3.8 KiB
Diff

From 473673fd0f834adf72f80caddab0d942a99dfe03 Mon Sep 17 00:00:00 2001
From: Jeffrey Cody <jcody@redhat.com>
Date: Wed, 20 Jun 2018 17:44:46 +0200
Subject: [PATCH 059/268] main-loop: drop spin_counter
RH-Author: Jeffrey Cody <jcody@redhat.com>
Message-id: <835200bb37bc622f0f6f9c471c809a1b73a08482.1529516334.git.jcody@redhat.com>
Patchwork-id: 80906
O-Subject: [RHEL-7.6 qemu-kvm-rhev PATCH v2 1/1] main-loop: drop spin_counter
Bugzilla: 1168213
RH-Acked-by: Paolo Bonzini <pbonzini@redhat.com>
RH-Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
RH-Acked-by: Markus Armbruster <armbru@redhat.com>
From: Stefan Hajnoczi <stefanha@redhat.com>
Commit d759c951f3287fad04210a52f2dc93f94cf58c7f ("replay: push
replay_mutex_lock up the call tree") removed the !timeout lock
optimization in the main loop.
The idea of the optimization was to avoid ping-pongs between threads by
keeping the Big QEMU Lock held across non-blocking (!timeout) main loop
iterations.
A warning is printed when the main loop spins without releasing BQL for
long periods of time. These warnings were supposed to aid debugging but
in practice they just alarm users. They are considered noise because
the cause of spinning is not shown and is hard to find.
Now that the lock optimization has been removed, there is no danger of
hogging the BQL. Drop the spin counter and the infamous warning.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
(cherry picked from commit 21891a5a3011608845b5d7f1f9cce60cdc2bcc62)
Note: The original patch series from Stefan makes note that the
patch can be dropped because the lock optimization changed
upstream.
However, even without that optimization change upstream, this
warning / debug message is not very useful, and is just noise that
does not aid in debugging. Go ahead and remove it anyway.
Signed-off-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>
---
tests/qemu-iotests/common.filter | 1 -
util/main-loop.c | 25 -------------------------
2 files changed, 26 deletions(-)
diff --git a/tests/qemu-iotests/common.filter b/tests/qemu-iotests/common.filter
index c5f4bcf..7acb454 100644
--- a/tests/qemu-iotests/common.filter
+++ b/tests/qemu-iotests/common.filter
@@ -77,7 +77,6 @@ _filter_qemu()
{
sed -e "s#\\(^\\|(qemu) \\)$(basename $QEMU_PROG):#\1QEMU_PROG:#" \
-e 's#^QEMU [0-9]\+\.[0-9]\+\.[0-9]\+ monitor#QEMU X.Y.Z monitor#' \
- -e '/main-loop: WARNING: I\/O thread spun for [0-9]\+ iterations/d' \
-e $'s#\r##' # QEMU monitor uses \r\n line endings
}
diff --git a/util/main-loop.c b/util/main-loop.c
index 992f9b0..affe040 100644
--- a/util/main-loop.c
+++ b/util/main-loop.c
@@ -222,36 +222,11 @@ static int os_host_main_loop_wait(int64_t timeout)
{
GMainContext *context = g_main_context_default();
int ret;
- static int spin_counter;
g_main_context_acquire(context);
glib_pollfds_fill(&timeout);
- /* If the I/O thread is very busy or we are incorrectly busy waiting in
- * the I/O thread, this can lead to starvation of the BQL such that the
- * VCPU threads never run. To make sure we can detect the later case,
- * print a message to the screen. If we run into this condition, create
- * a fake timeout in order to give the VCPU threads a chance to run.
- */
- if (!timeout && (spin_counter > MAX_MAIN_LOOP_SPIN)) {
- static bool notified;
-
- if (!notified && !qtest_enabled() && !qtest_driver()) {
- warn_report("I/O thread spun for %d iterations",
- MAX_MAIN_LOOP_SPIN);
- notified = true;
- }
-
- timeout = SCALE_MS;
- }
-
-
- if (timeout) {
- spin_counter = 0;
- } else {
- spin_counter++;
- }
qemu_mutex_unlock_iothread();
replay_mutex_unlock();
--
1.8.3.1