From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:53668) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bxYPX-0002ur-Mb for qemu-devel@nongnu.org; Fri, 21 Oct 2016 07:54:56 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bxYPT-0008Tp-Rr for qemu-devel@nongnu.org; Fri, 21 Oct 2016 07:54:55 -0400 Received: from mail-qt0-x232.google.com ([2607:f8b0:400d:c0d::232]:34276) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1bxYPT-0008TV-Nd for qemu-devel@nongnu.org; Fri, 21 Oct 2016 07:54:51 -0400 Received: by mail-qt0-x232.google.com with SMTP id q7so83026901qtq.1 for ; Fri, 21 Oct 2016 04:54:51 -0700 (PDT) From: =?UTF-8?q?Alex=20Benn=C3=A9e?= Date: Fri, 21 Oct 2016 12:54:42 +0100 Message-Id: <20161021115442.18420-1-alex.bennee@linaro.org> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Subject: [Qemu-devel] [PATCH] cpus: make qemu_mutex_iothread_locked() understand co-routines List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: stefanha@redhat.com, pbonzini@redhat.com Cc: qemu-devel@nongnu.org, =?UTF-8?q?Alex=20Benn=C3=A9e?= , Peter Crosthwaite , Richard Henderson There is a slight wart when checking for the state of the BQL when using GThread base co-routines (which we keep for ThreadSanitizer runs). While the main-loop holds the BQL it is suspended until the co-routine completes however the co-routines run in a separate thread so checking the TLS variable could be wrong. We fix this by expanding the check to include qemu_in_coroutine() for GThread based builds. As it is not used for production builds I'm not overly worried about any performance impact which should be negligible anyway. Signed-off-by: Alex Bennée Cc: Stefan Hajnoczi --- configure | 3 +++ cpus.c | 13 +++++++++++++ 2 files changed, 16 insertions(+) diff --git a/configure b/configure index 91a14c1..97b89fb 100755 --- a/configure +++ b/configure @@ -5461,6 +5461,9 @@ if test "$rbd" = "yes" ; then fi echo "CONFIG_COROUTINE_BACKEND=$coroutine" >> $config_host_mak +if test "$coroutine" = "gthread" ; then + echo "CONFIG_COROUTINE_GTHREAD=1" >> $config_host_mak +fi if test "$coroutine_pool" = "yes" ; then echo "CONFIG_COROUTINE_POOL=1" >> $config_host_mak else diff --git a/cpus.c b/cpus.c index 0c18a9f..a3e189a 100644 --- a/cpus.c +++ b/cpus.c @@ -49,6 +49,10 @@ #include "hw/nmi.h" #include "sysemu/replay.h" +#ifdef CONFIG_COROUTINE_GTHREAD +#include "qemu/coroutine.h" +#endif + #ifndef _WIN32 #include "qemu/compatfd.h" #endif @@ -1422,9 +1426,18 @@ bool qemu_in_vcpu_thread(void) static __thread bool iothread_locked = false; +/* + * There is a slight wart when using gthread based co-routines. Here + * the BQL is held by the main-thread which is suspended until the + * co-routines complete. + */ bool qemu_mutex_iothread_locked(void) { +#ifdef CONFIG_COROUTINE_GTHREAD + return iothread_locked || qemu_in_coroutine(); +#else return iothread_locked; +#endif } void qemu_mutex_lock_iothread(void) -- 2.9.3