From: Peter Xu <peterx@redhat.com>
To: qemu-devel@nongnu.org
Cc: "Paolo Bonzini" <pbonzini@redhat.com>,
"Richard Henderson" <rth@twiddle.net>,
"Philippe Mathieu-Daudé" <philmd@redhat.com>,
peterx@redhat.com, "David Hildenbrand" <david@redhat.com>
Subject: [PATCH v3 7/8] memory: Assert on no ongoing memory transaction before release BQL
Date: Wed, 28 Jul 2021 14:31:50 -0400 [thread overview]
Message-ID: <20210728183151.195139-8-peterx@redhat.com> (raw)
In-Reply-To: <20210728183151.195139-1-peterx@redhat.com>
Firstly, add a "prepare" function before unlocking BQL. There're only three
places that can release the BQL: unlock(), cond_wait() or cond_timedwait().
Make sure we don't have any more ongoing memory transaction when releasing the
BQL. This will trigger an abort if we misuse the QEMU memory model, e.g., when
calling run_on_cpu() during a memory commit.
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
---
include/exec/memory-internal.h | 1 +
softmmu/cpus.c | 9 +++++++++
softmmu/memory.c | 6 ++++++
3 files changed, 16 insertions(+)
diff --git a/include/exec/memory-internal.h b/include/exec/memory-internal.h
index 9fcc2af25c..3124b91c4b 100644
--- a/include/exec/memory-internal.h
+++ b/include/exec/memory-internal.h
@@ -35,6 +35,7 @@ static inline AddressSpaceDispatch *address_space_to_dispatch(AddressSpace *as)
FlatView *address_space_get_flatview(AddressSpace *as);
void flatview_unref(FlatView *view);
+bool memory_region_has_pending_transaction(void);
extern const MemoryRegionOps unassigned_mem_ops;
diff --git a/softmmu/cpus.c b/softmmu/cpus.c
index 9154cd7e78..4d190f9076 100644
--- a/softmmu/cpus.c
+++ b/softmmu/cpus.c
@@ -31,6 +31,7 @@
#include "qapi/qapi-events-run-state.h"
#include "qapi/qmp/qerror.h"
#include "exec/gdbstub.h"
+#include "exec/memory-internal.h"
#include "sysemu/hw_accel.h"
#include "exec/exec-all.h"
#include "qemu/thread.h"
@@ -66,6 +67,11 @@
static QemuMutex qemu_global_mutex;
+static void qemu_mutex_unlock_iothread_prepare(void)
+{
+ assert(!memory_region_has_pending_transaction());
+}
+
bool cpu_is_stopped(CPUState *cpu)
{
return cpu->stopped || !runstate_is_running();
@@ -520,16 +526,19 @@ void qemu_mutex_unlock_iothread(void)
{
g_assert(qemu_mutex_iothread_locked());
iothread_locked = false;
+ qemu_mutex_unlock_iothread_prepare();
qemu_mutex_unlock(&qemu_global_mutex);
}
void qemu_cond_wait_iothread(QemuCond *cond)
{
+ qemu_mutex_unlock_iothread_prepare();
qemu_cond_wait(cond, &qemu_global_mutex);
}
void qemu_cond_timedwait_iothread(QemuCond *cond, int ms)
{
+ qemu_mutex_unlock_iothread_prepare();
qemu_cond_timedwait(cond, &qemu_global_mutex, ms);
}
diff --git a/softmmu/memory.c b/softmmu/memory.c
index 35b2568fc2..62ec00b52d 100644
--- a/softmmu/memory.c
+++ b/softmmu/memory.c
@@ -176,6 +176,12 @@ static bool memory_region_has_pending_update(void)
return memory_region_update_pending || ioeventfd_update_pending;
}
+bool memory_region_has_pending_transaction(void)
+{
+ return memory_region_transaction_depth ||
+ memory_region_has_pending_update();
+}
+
static bool memory_region_ioeventfd_before(MemoryRegionIoeventfd *a,
MemoryRegionIoeventfd *b)
{
--
2.31.1
next prev parent reply other threads:[~2021-07-28 18:38 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-07-28 18:31 [PATCH v3 0/8] memory: Sanity checks memory transaction when releasing BQL Peter Xu
2021-07-28 18:31 ` [PATCH v3 1/8] cpus: Export queue work related fields to cpu.h Peter Xu
2021-07-28 18:40 ` Philippe Mathieu-Daudé
2021-07-28 18:31 ` [PATCH v3 2/8] cpus: Move do_run_on_cpu into softmmu/cpus.c Peter Xu
2021-07-28 18:35 ` Philippe Mathieu-Daudé
2021-07-28 18:31 ` [PATCH v3 3/8] memory: Introduce memory_region_transaction_depth_{inc|dec}() Peter Xu
2021-07-28 20:06 ` David Hildenbrand
2021-07-28 18:31 ` [PATCH v3 4/8] memory: Don't do topology update in memory finalize() Peter Xu
2021-07-28 18:31 ` [PATCH v3 5/8] cpus: Use qemu_cond_wait_iothread() where proper Peter Xu
2021-07-28 18:38 ` Philippe Mathieu-Daudé
2021-07-28 18:31 ` [PATCH v3 6/8] cpus: Remove the mutex parameter from do_run_on_cpu() Peter Xu
2021-07-28 18:39 ` Philippe Mathieu-Daudé
2021-07-28 18:31 ` Peter Xu [this message]
2021-07-28 18:31 ` [PATCH v3 8/8] memory: Delay the transaction pop() until commit completed Peter Xu
2021-09-08 15:30 ` [PATCH v3 0/8] memory: Sanity checks memory transaction when releasing BQL Peter Xu
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=20210728183151.195139-8-peterx@redhat.com \
--to=peterx@redhat.com \
--cc=david@redhat.com \
--cc=pbonzini@redhat.com \
--cc=philmd@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=rth@twiddle.net \
/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).