Andreas Sandberg has uploaded this change for review. (
https://gem5-review.googlesource.com/c/public/gem5/+/50409 )
Change subject: cpu-kvm: Reinitialize threads on drainResume
......................................................................
cpu-kvm: Reinitialize threads on drainResume
Event queue service threads may have been re-created while the
simulator was drained. We therefore need to initialize the new thread
by setting correct signal masks and re-attaching performance counters.
Change-Id: Ic0dab80543928327021cade037770c917e73a47f
Signed-off-by: Andreas Sandberg <andreas.sandb...@arm.com>
---
M src/cpu/kvm/base.cc
M src/cpu/kvm/base.hh
2 files changed, 21 insertions(+), 8 deletions(-)
diff --git a/src/cpu/kvm/base.cc b/src/cpu/kvm/base.cc
index c7c72a8..ea43595 100644
--- a/src/cpu/kvm/base.cc
+++ b/src/cpu/kvm/base.cc
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2012, 2015, 2017 ARM Limited
+ * Copyright (c) 2012, 2015, 2017, 2021 ARM Limited
* All rights reserved
*
* The license below extends only to copyright in the software and shall
@@ -154,9 +154,9 @@
inform("KVM: Coalesced not supported by host OS\n");
}
- Event *startupEvent(
- new EventFunctionWrapper([this]{ startupThread(); }, name(),
true));
- schedule(startupEvent, curTick());
+ schedule(new EventFunctionWrapper([this]{
+ restartEqThread();
+ }, name(), true), curTick());
}
BaseKvmCPU::Status
@@ -228,7 +228,7 @@
}
void
-BaseKvmCPU::startupThread()
+BaseKvmCPU::restartEqThread()
{
// Do thread-specific initialization. We need to setup signal
// delivery for counters and timers from within the thread that
@@ -381,6 +381,13 @@
{
assert(!tickEvent.scheduled());
+ /* The simulator may have terminated the threads servicing event
+ * queues. In that case, we need to re-initialize the new
+ * threads. */
+ schedule(new EventFunctionWrapper([this]{
+ restartEqThread();
+ }, name(), true), curTick());
+
// We might have been switched out. In that case, we don't need to
// do anything.
if (switchedOut())
@@ -1275,6 +1282,11 @@
.samplePeriod(42);
}
+ // We might be re-attaching counters due threads being
+ // re-initialised after fork.
+ if (hwCycles.attached())
+ hwCycles.detach();
+
hwCycles.attach(cfgCycles,
0); // TID (0 => currentThread)
diff --git a/src/cpu/kvm/base.hh b/src/cpu/kvm/base.hh
index e5b047e..4f40064 100644
--- a/src/cpu/kvm/base.hh
+++ b/src/cpu/kvm/base.hh
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2012 ARM Limited
+ * Copyright (c) 2012, 2021 ARM Limited
* All rights reserved
*
* The license below extends only to copyright in the software and shall
@@ -682,11 +682,12 @@
* example, when setting up timers, we need to know the TID of the
* thread executing in KVM in order to deliver the timer signal to
* that thread. This method is called as the first event in this
- * SimObject's event queue.
+ * SimObject's event queue and after drainResume to handle changes
+ * to event queue service threads.
*
* @see startup
*/
- void startupThread();
+ void restartEqThread();
/** Try to drain the CPU if a drain is pending */
bool tryDrain();
--
To view, visit https://gem5-review.googlesource.com/c/public/gem5/+/50409
To unsubscribe, or for help writing mail filters, visit
https://gem5-review.googlesource.com/settings
Gerrit-Project: public/gem5
Gerrit-Branch: develop
Gerrit-Change-Id: Ic0dab80543928327021cade037770c917e73a47f
Gerrit-Change-Number: 50409
Gerrit-PatchSet: 1
Gerrit-Owner: Andreas Sandberg <andreas.sandb...@arm.com>
Gerrit-MessageType: newchange
_______________________________________________
gem5-dev mailing list -- gem5-dev@gem5.org
To unsubscribe send an email to gem5-dev-le...@gem5.org
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
