QemuEvent is currently broken on ARM due to missing memory barriers
after qatomic_*().  Apart from adding the memory barrier, a closer look
reveals some unpaired memory barriers too.  Document more clearly what
is going on.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

FreeBSD implements pthread headers using TSA (thread safety analysis)
annotations, therefore when an application is compiled with
-Wthread-safety there are some locking/annotation requirements that the
user of the pthread API has to follow.

This will also be the case in QEMU, since util/qemu-thread-posix.c uses
the pthread API. Therefore when building it with -Wthread-safety, the
compiler will throw warnings because the functions are not properly
annotated. We need TSA to be enabled because it ensures that the
critical sections of an annotated variable are properly locked.

In order to make the compiler happy and avoid adding all the necessary
macros to all callers (lock functions should use TSA_ACQUIRE, while
unlock TSA_RELEASE, and this applies to all users of pthread_mutex_lock
and pthread_mutex_unlock), simply use TSA_NO_TSA to supppress such
warnings.

Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20230117135203.3049709-2-eesposit@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

Make use of pthread_set_name_np() to be able to set the threads name
on OpenBSD.

Signed-off-by: Brad Smith <brad@comstyle.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <Y57NrCmPTVSXLWC4@humpty.home.comstyle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Usually, we let upper layers handle CPU pinning, because
pthread_setaffinity_np() (-> sched_setaffinity()) is blocked via
seccomp when starting QEMU with
    -sandbox enable=on,resourcecontrol=deny

However, we want to configure and observe the CPU affinity of threads
from QEMU directly in some cases when the sandbox option is either not
enabled or not active yet.

So let's add a way to configure CPU pinning via
qemu_thread_set_affinity() and obtain CPU affinity via
qemu_thread_get_affinity() and implement them under POSIX using
pthread_setaffinity_np() + pthread_getaffinity_np().

Implementation under Windows is possible using SetProcessAffinityMask()
+ GetProcessAffinityMask(), however, that is left as future work.

Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Message-Id: <20221014134720.168738-3-david@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>

In this case there is no need to call pthread_cond_timedwait; the
function is just a trywait and waiting on the condition variable would
always time out.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Now that QemuSemaphore is implemented through pthread_cond_t only, we can use
QemuCond and QemuMutex to make the code smaller.  Features such as mutex
tracing and CLOCK_MONOTONIC timedwait are supported in qemu-sem naturally.

Signed-off-by: Longpeng(Mike) <longpeng2@huawei.com>
Message-Id: <20220222090507.2028-4-longpeng2@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Use CLOCK_MONOTONIC, so the timeout isn't affected by changes to
the system time. It depends on the pthread_condattr_setclock(),
while some systems(e.g. mac os) does not support it, so the behavior
won't change in these systems.

Signed-off-by: Longpeng(Mike) <longpeng2@huawei.com>
Message-Id: <20220222090507.2028-3-longpeng2@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

POSIX specifies an absolute time for sem_timedwait(), it would be
affected if the system time is changing, but there is not a relative
time or monotonic clock version of sem_timedwait, so we cannot gain
from POSIX semaphore any more.

An alternative way is to use sem_trywait + usleep, maybe we can
remove CONFIG_SEM_TIMEDWAIT in this way? No, because some systems
(e.g. mac os) mark the sem_xxx API as deprecated.

So maybe remove the usage of POSIX semaphore and turn to use the
pthread variant for all systems looks better.

Signed-off-by: Longpeng(Mike) <longpeng2@huawei.com>
Message-Id: <20220222090507.2028-2-longpeng2@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

This makes the pthreads check dead in configure, so remove it
as well.

Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20211007130829.632254-9-pbonzini@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

This seems to be either a glibc or gcc bug, but the code
appears to be fine with the warning suppressed.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210803211907.150525-1-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

We will shortly convert lockable.h to _Generic, and we cannot
have two compatible types in the same expansion.  Wrap QemuMutex
in a struct, and unwrap in qemu-thread-posix.c.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210614233143.1221879-6-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Create macros for file+line expansion in qemu_rec_mutex_unlock
like we have for qemu_mutex_unlock.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210614233143.1221879-5-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Move the declarations from thread-win32.h into thread.h
and remove the macro redirection from thread-posix.h.
This will be required by following cleanups.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210614233143.1221879-4-richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

We will shortly convert lockable.h to _Generic, and we cannot
have two compatible types in the same expansion.  Wrap QemuMutex
in a struct, and unwrap in qemu-thread-posix.c.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20210614233143.1221879-6-richard.henderson@linaro.org>

Create macros for file+line expansion in qemu_rec_mutex_unlock
like we have for qemu_mutex_unlock.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20210614233143.1221879-5-richard.henderson@linaro.org>

Move the declarations from thread-win32.h into thread.h
and remove the macro redirection from thread-posix.h.
This will be required by following cleanups.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20210614233143.1221879-4-richard.henderson@linaro.org>

clang's C11 atomic_fetch_*() functions only take a C11 atomic type
pointer argument. QEMU uses direct types (int, etc) and this causes a
compiler error when a QEMU code calls these functions in a source file
that also included <stdatomic.h> via a system header file:

  $ CC=clang CXX=clang++ ./configure ... && make
  ../util/async.c:79:17: error: address argument to atomic operation must be a pointer to _Atomic type ('unsigned int *' invalid)

Avoid using atomic_*() names in QEMU's atomic.h since that namespace is
used by <stdatomic.h>. Prefix QEMU's APIs with 'q' so that atomic.h
and <stdatomic.h> can co-exist. I checked /usr/include on my machine and
searched GitHub for existing "qatomic_" users but there seem to be none.

This patch was generated using:

  $ git grep -h -o '\<atomic\(64\)\?_[a-z0-9_]\+' include/qemu/atomic.h | \
    sort -u >/tmp/changed_identifiers
  $ for identifier in $(</tmp/changed_identifiers); do
        sed -i "s%\<$identifier\>%q$identifier%g" \
            $(git grep -I -l "\<$identifier\>")
    done

I manually fixed line-wrap issues and misaligned rST tables.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200923105646.47864-1-stefanha@redhat.com>

This allows us to see the name of the thread in tsan
warning reports such as this:

  Thread T7 'CPU 1/TCG' (tid=24317, running) created by main thread at:

Signed-off-by: Robert Foley <robert.foley@linaro.org>
Reviewed-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20200609200738.445-12-robert.foley@linaro.org>
Message-Id: <20200612190237.30436-15-alex.bennee@linaro.org>

The new function is needed to implement conditional sleep for CPU
throttling. It's possible to reuse qemu_sem_timedwait, but it's more
difficult than just add qemu_cond_timedwait.

Also moved compute_abs_deadline function up the code to reuse it in
qemu_cond_timedwait_impl win32.

Signed-off-by: Yury Kotov <yury-kotov@yandex-team.ru>
Message-Id: <20190909131335.16848-2-yury-kotov@yandex-team.ru>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

If any of these signals happen on macOS, they are not delivered to other
threads and signalfd_compat receives nothing. Indeed, POSIX reference
and sigprocmask(2) note that an attempt to block the signals results in
undefined behaviour. SEGV and FPE can't also be received by signalfd(2)
on Linux.

An ability to retrieve SIGBUS via signalfd(2) is used by QEMU for
memory preallocation therefore we can't unblock it without consequences.
But it's important to leave a remark that the signal is lost on macOS.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

macOS provides pthread_setname_np that doesn't have thread id argument.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Our current implementation of qemu_thread_atexit* is broken on OSX.
This is because it works by cerating a piece of thread-specific
data with pthread_key_create() and using the destructor function
for that data to run the notifier function passed to it by
the caller of qemu_thread_atexit_add(). The expected use case
is that the caller uses a __thread variable as the notifier,
and uses the callback to clean up information that it is
keeping per-thread in __thread variables.

Unfortunately, on OSX this does not work, because on OSX
a __thread variable may be destroyed (freed) before the
pthread_key_create() destructor runs. (POSIX imposes no
ordering constraint here; the OSX implementation happens
to implement __thread variables in terms of pthread_key_create((),
whereas Linux uses different mechanisms that mean the __thread
variables will still be present when the pthread_key_create()
destructor is run.)

Fix this by switching to a scheme similar to the one qemu-thread-win32
uses for qemu_thread_atexit: keep the thread's notifiers on a
__thread variable, and run the notifiers on calls to
qemu_thread_exit() and on return from the start routine passed
to qemu_thread_start(). We do this with the pthread_cleanup_push()
API.

We take advantage of the qemu_thread_atexit_add() API
permission not to run thread notifiers on process exit to
avoid having to special case the main thread.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20181105135538.28025-3-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Introduce some hooks for the shared part of qemu thread between POSIX
and Windows implementations.  Note that in qemu_mutex_unlock_impl() we
moved the call before unlock operation which should make more sense.
And we don't need qemu_mutex_post_unlock() hook.

Put all these shared hooks into the header files.  It should be internal
to qemu-thread but not for qemu-thread users, hence put into util/
directory.

Reviewed-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20180425025459.5258-3-peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

We will conditionally have a wrapper layer depending on whether the host
has the PTHREAD_SETNAME capability.  It complicates stuff.  Let's keep
the wrapper there; we opt out the pthread_setname_np() call only.

Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20180412053444.17801-1-peterx@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

If we create a thread with QEMU_THREAD_DETACHED mode, QEMU may get a segfault with low probability.

The backtrace is:
   #0  0x00007f46c60291d7 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
   #1  0x00007f46c602a8c8 in __GI_abort () at abort.c:90
   #2  0x00000000008543c9 in PAT_abort ()
   #3  0x000000000085140d in patchIllInsHandler ()
   #4  <signal handler called>
   #5  pthread_detach (th=139933037614848) at pthread_detach.c:50
   #6  0x0000000000829759 in qemu_thread_create (thread=thread@entry=0x7ffdaa8205e0, name=name@entry=0x94d94a "io-task-worker", start_routine=start_routine@entry=0x7eb9a0 <qio_task_thread_worker>,
       arg=arg@entry=0x3f5cf70, mode=mode@entry=1) at util/qemu_thread_posix.c:512
   #7  0x00000000007ebc96 in qio_task_run_in_thread (task=0x31db2c0, worker=worker@entry=0x7e7e40 <qio_channel_socket_connect_worker>, opaque=0xcd23380, destroy=0x7f1180 <qapi_free_SocketAddress>)
       at io/task.c:141
   #8  0x00000000007e7f33 in qio_channel_socket_connect_async (ioc=ioc@entry=0x626c0b0, addr=<optimized out>, callback=callback@entry=0x55e080 <qemu_chr_socket_connected>, opaque=opaque@entry=0x42862c0,
       destroy=destroy@entry=0x0) at io/channel_socket.c:194
   #9  0x000000000055bdd1 in socket_reconnect_timeout (opaque=0x42862c0) at qemu_char.c:4744
   #10 0x00007f46c72483b3 in g_timeout_dispatch () from /usr/lib64/libglib-2.0.so.0
   #11 0x00007f46c724799a in g_main_context_dispatch () from /usr/lib64/libglib-2.0.so.0
   #12 0x000000000076c646 in glib_pollfds_poll () at main_loop.c:228
   #13 0x000000000076c6eb in os_host_main_loop_wait (timeout=348000000) at main_loop.c:273
   #14 0x000000000076c815 in main_loop_wait (nonblocking=nonblocking@entry=0) at main_loop.c:521
   #15 0x000000000056a511 in main_loop () at vl.c:2076
   #16 0x0000000000420705 in main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:4940

The cause of this problem is a glibc bug; for more information, see
https://sourceware.org/bugzilla/show_bug.cgi?id=19951

.
The solution for this bug is to use pthread_attr_setdetachstate.

There is a similar issue with pthread_setname_np, which is moved
from creating thread to created thread.

Signed-off-by: linzhecheng <linzhecheng@huawei.com>
Message-Id: <20171128044656.10592-1-linzhecheng@huawei.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
[Simplify the code by removing qemu_thread_set_name, and free the arguments
 before invoking the start routine. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

In qemu-thread-posix.c we have two implementations of the
various qemu_sem_* functions, one of which uses native POSIX
sem_* and the other of which emulates them with pthread conditions.
This is necessary because not all our host OSes support
sem_timedwait().

Instead of a hard-coded list of OSes which don't implement
sem_timedwait(), which gets out of date, make configure
test for the presence of the function and set a new
CONFIG_HAVE_SEM_TIMEDWAIT appropriately.

In particular, newer NetBSDs have sem_timedwait(), so this
commit will switch them over to using it. OSX still does
not have an implementation.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Kamil Rytarowski <n54@gmx.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

Not all platforms check whether a lock is initialized before used.  In
particular Linux seems to be more permissive than OSX.

Check initialization state explicitly in our code to catch such bugs
earlier.

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170704122325.25634-1-famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

These trace events were very useful to help me to understand and find a
reordering issue in vfio, for example:

qemu_mutex_lock locked mutex 0x10905ad8
  vfio_region_write  (0001:03:00.0:region1+0xc0, 0x2020c, 4)
qemu_mutex_unlock unlocked mutex 0x10905ad8
qemu_mutex_lock locked mutex 0x10905ad8
  vfio_region_write  (0001:03:00.0:region1+0xc4, 0xa0000, 4)
qemu_mutex_unlock unlocked mutex 0x10905ad8

that also helped me to see the desired result after the fix:

qemu_mutex_lock locked mutex 0x10905ad8
  vfio_region_write  (0001:03:00.0:region1+0xc0, 0x2000c, 4)
  vfio_region_write  (0001:03:00.0:region1+0xc4, 0xb0000, 4)
qemu_mutex_unlock unlocked mutex 0x10905ad8

So it could be a good idea to have these traces implemented. It's worth
mentioning that they should be surgically enabled during the debugging,
otherwise it can flood the trace logs with lock/unlock messages.

How to use it:
trace-event qemu_mutex_lock on|off
trace-event qemu_mutex_unlock on|off
or
trace-event qemu_mutex* on|off

Signed-off-by: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com>
Message-Id: <1493054398-26013-1-git-send-email-joserz@linux.vnet.ibm.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
[Also handle trylock, cond_wait and win32; trace "unlocked" while still
 in the critical section, so that "unlocked" always comes before the
 next "locked" tracepoint. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

QEMU will crash with the follow backtrace if the new created thread exited before
we call qemu_thread_set_name() for it.

  (gdb) bt
  #0 0x00007f9a68b095d7 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
  #1 0x00007f9a68b0acc8 in __GI_abort () at abort.c:90
  #2 0x00007f9a69cda389 in PAT_abort () from /usr/lib64/libuvpuserhotfix.so
  #3 0x00007f9a69cdda0d in patchIllInsHandler () from /usr/lib64/libuvpuserhotfix.so
  #4 <signal handler called>
  #5 pthread_setname_np (th=140298470549248, name=name@entry=0x8cc74a "io-task-worker") at ../nptl/sysdeps/unix/sysv/linux/pthread_setname.c:49
  #6 0x00000000007f5f20 in qemu_thread_set_name (thread=thread@entry=0x7ffd2ac09680, name=name@entry=0x8cc74a "io-task-worker") at util/qemu_thread_posix.c:459
  #7 0x00000000007f679e in qemu_thread_create (thread=thread@entry=0x7ffd2ac09680, name=name@entry=0x8cc74a "io-task-worker",start_routine=start_routine@entry=0x7c1300 <qio_task_thread_worker>, arg=arg@entry=0x7f99b8001720, mode=mode@entry=1) at util/qemu_thread_posix.c:498
  #8 0x00000000007c15b6 in qio_task_run_in_thread (task=task@entry=0x7f99b80033d0, worker=worker@entry=0x7bd920 <qio_channel_socket_connect_worker>, opaque=0x7f99b8003370, destroy=0x7c6220 <qapi_free_SocketAddress>) at io/task.c:133
  #9 0x00000000007bda04 in qio_channel_socket_connect_async (ioc=0x7f99b80014c0, addr=0x37235d0, callback=callback@entry=0x54ad00 <qemu_chr_socket_connected>, opaque=opaque@entry=0x38118b0, destroy=destroy@entry=0x0) at io/channel_socket.c:191
  #10 0x00000000005487f6 in socket_reconnect_timeout (opaque=0x38118b0) at qemu_char.c:4402
  #11 0x00007f9a6a1533b3 in g_timeout_dispatch () from /usr/lib64/libglib-2.0.so.0
  #12 0x00007f9a6a15299a in g_main_context_dispatch () from /usr/lib64/libglib-2.0.so.0
  #13 0x0000000000747386 in glib_pollfds_poll () at main_loop.c:227
  #14 0x0000000000747424 in os_host_main_loop_wait (timeout=404000000) at main_loop.c:272
  #15 0x0000000000747575 in main_loop_wait (nonblocking=nonblocking@entry=0) at main_loop.c:520
  #16 0x0000000000557d31 in main_loop () at vl.c:2170
  #17 0x000000000041c8b7 in main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:5083

Let's detach the new thread after calling qemu_thread_set_name().

Signed-off-by: Caoxinhua <caoxinhua@huawei.com>
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Message-Id: <1483493521-9604-1-git-send-email-zhang.zhanghailiang@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

This is complex, but I think it is reasonably documented in the source.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20170112180800.21085-5-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

GRecMutex is new in glib 2.32, so we cannot use it.  Introduce
a recursive mutex in qemu-thread instead, which will be used
instead of RFifoLock.

Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1477565348-5458-20-git-send-email-pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>

Do not use the somewhat mysterious atomic_mb_read/atomic_mb_set,
instead make sure that the operations on QemuEvent are annotated
with the desired acquire and release semantics.

In particular, qemu_event_set wakes up the waiting thread, so it must
be a release from the POV of the waker (compare with qemu_mutex_unlock).
And it actually needs a full barrier, because that's the only thing that
provides something like a "load-release".

Use smp_mb_acquire until we have atomic_load_acquire and
atomic_store_release in atomic.h.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Clean up includes so that osdep.h is included first and headers
which it implies are not included manually.

This commit was created with scripts/clean-includes.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1454089805-5470-6-git-send-email-peter.maydell@linaro.org

Signed-off-by: Veres Lajos <vlajos@gmail.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1440375847-17603-12-git-send-email-cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

PTHREAD_MUTEX_ERRORCHECK is completely broken with respect to fork.
The way to safely do fork is to bring all threads to a quiescent
state by acquiring locks (either in callers---as we do for the
iothread mutex---or using pthread_atfork's prepare callbacks)
and then release them in the child.

The problem is that releasing error-checking locks in the child
fails under glibc with EPERM, because the mutex stores a different
owner tid than the duplicated thread in the child process.  We
could make it work for locks acquired via pthread_atfork, by
recreating the mutex in the child instead of unlocking it
(we know that there are no other threads that could have taken
the mutex; but when the lock is acquired in fork's caller
that would not be possible.

The simplest solution is just to forgo error checking.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

This had a possible deadlock that was visible with rcutorture.

    qemu_event_set                    qemu_event_wait
    ----------------------------------------------------------------
                                      cmpxchg reads FREE, writes BUSY
                                      futex_wait: pthread_mutex_lock
                                      futex_wait: value == BUSY
    xchg reads BUSY, writes SET
    futex_wake: pthread_cond_broadcast
                                      futex_wait: pthread_cond_wait
                                      <deadlock>

The fix is simply to avoid condvar tricks and do the obvious locking
around pthread_cond_broadcast:

    qemu_event_set        qemu_event_wait
    ----------------------------------------------------------------
                                      cmpxchg reads FREE, writes BUSY
                                      futex_wait: pthread_mutex_lock
                                      futex_wait: value == BUSY
    xchg reads BUSY, writes SET
    futex_wake: pthread_mutex_lock
    (blocks)
                                      futex_wait: pthread_cond_wait
    (mutex unlocked)
    futex_wake: pthread_cond_broadcast
    futex_wake: pthread_mutex_unlock
                                      futex_wait: pthread_mutex_unlock

Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Destructors are the main additional feature of pthread TLS compared
to __thread.  If we were using C++ (hint, hint!) we could have used
thread-local objects with a destructor.  Since we are not, instead,
we add a simple Notifier-based API.

Note that the notifier must be per-thread as well.  We can add a
global list as well later, perhaps.

The Win32 implementation has some complications because a) detached
threads used not to have a QemuThreadData; b) the main thread does
not go through win32_start_routine, so we have to use atexit too.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1417518350-6167-3-git-send-email-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Warn if no way of setting thread name is available.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>