From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:57313) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aWmsD-0004go-Pk for qemu-devel@nongnu.org; Fri, 19 Feb 2016 10:21:41 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1aWms9-0005rY-Or for qemu-devel@nongnu.org; Fri, 19 Feb 2016 10:21:37 -0500 References: <1453960195-15181-1-git-send-email-bharata@linux.vnet.ibm.com> <1453960195-15181-6-git-send-email-bharata@linux.vnet.ibm.com> From: Thomas Huth Message-ID: <56C732F1.9080003@redhat.com> Date: Fri, 19 Feb 2016 16:21:21 +0100 MIME-Version: 1.0 In-Reply-To: <1453960195-15181-6-git-send-email-bharata@linux.vnet.ibm.com> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH v7 05/13] cpu: Reclaim vCPU objects List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Bharata B Rao , qemu-devel@nongnu.org Cc: mjrosato@linux.vnet.ibm.com, Zhu Guihua , ehabkost@redhat.com, imammedo@redhat.com, aik@ozlabs.ru, agraf@suse.de, mdroth@linux.vnet.ibm.com, Gu Zheng , qemu-ppc@nongnu.org, tyreld@linux.vnet.ibm.com, nfont@linux.vnet.ibm.com, Chen Fan , pbonzini@redhat.com, afaerber@suse.de, david@gibson.dropbear.id.au On 28.01.2016 06:49, Bharata B Rao wrote: > From: Gu Zheng >=20 > In order to deal well with the kvm vcpus (which can not be removed with= out any > protection), we do not close KVM vcpu fd, just record and mark it as st= opped > into a list, so that we can reuse it for the appending cpu hot-add requ= est if > possible. It is also the approach that kvm guys suggested: > https://www.mail-archive.com/kvm@vger.kernel.org/msg102839.html >=20 > Signed-off-by: Chen Fan > Signed-off-by: Gu Zheng > Signed-off-by: Zhu Guihua > Signed-off-by: Bharata B Rao > [- Explicit CPU_REMOVE() from qemu_kvm/tcg_destroy_vcpu(= ) > isn't needed as it is done from cpu_exec_exit() > - Use iothread mutex instead of global mutex during > destroy > - Don't cleanup vCPU object from vCPU thread context > but leave it to the callers (device_add/device_del)] > Reviewed-by: David Gibson > --- > cpus.c | 38 +++++++++++++++++++++++++++++++++++ > include/qom/cpu.h | 10 +++++++++ > include/sysemu/kvm.h | 1 + > kvm-all.c | 57 ++++++++++++++++++++++++++++++++++++++++++++= +++++++- > kvm-stub.c | 5 +++++ > 5 files changed, 110 insertions(+), 1 deletion(-) >=20 > diff --git a/cpus.c b/cpus.c > index 1e97cc4..c5631f0 100644 > --- a/cpus.c > +++ b/cpus.c > @@ -953,6 +953,18 @@ void async_run_on_cpu(CPUState *cpu, void (*func)(= void *data), void *data) > qemu_cpu_kick(cpu); > } > =20 > +static void qemu_kvm_destroy_vcpu(CPUState *cpu) > +{ > + if (kvm_destroy_vcpu(cpu) < 0) { > + error_report("kvm_destroy_vcpu failed"); > + exit(EXIT_FAILURE); > + } > +} > + > +static void qemu_tcg_destroy_vcpu(CPUState *cpu) > +{ Will this be populated by a later patch in your series? If not, maybe add a debugging statement here so that it is clear that there is a TODO left here? > +} > + > static void flush_queued_work(CPUState *cpu) > { > struct qemu_work_item *wi; > @@ -1053,6 +1065,11 @@ static void *qemu_kvm_cpu_thread_fn(void *arg) > } > } > qemu_kvm_wait_io_event(cpu); > + if (cpu->exit && !cpu_can_run(cpu)) { > + qemu_kvm_destroy_vcpu(cpu); > + qemu_mutex_unlock_iothread(); > + return NULL; > + } > } You could increase readability of the code by changing the condition of the loop instead - currently it is a "while (1)" ... you could turn that into a "do { ... } while (!cpu->exit || cpu_can_run(cpu))" and then destroy the cpu after the loop. > return NULL; ... > diff --git a/include/qom/cpu.h b/include/qom/cpu.h > index 2e5229d..32a2c71 100644 > --- a/include/qom/cpu.h > +++ b/include/qom/cpu.h > @@ -232,6 +232,7 @@ struct kvm_run; > * @halted: Nonzero if the CPU is in suspended state. > * @stop: Indicates a pending stop request. > * @stopped: Indicates the CPU has been artificially stopped. > + * @exit: Indicates the CPU has exited due to an unplug operation. There is also a "exit_request" member in this struct already ... maybe you could name your new variable differently to avoid confusion? Something like "remove_request" or "unplug_request" ? > * @crash_occurred: Indicates the OS reported a crash (panic) for this= CPU > * @tcg_exit_req: Set to force TCG to stop executing linked TBs for th= is > * CPU and return to its top level loop. > @@ -284,6 +285,7 @@ struct CPUState { > bool created; > bool stop; > bool stopped; > + bool exit; > bool crash_occurred; > bool exit_request; > uint32_t interrupt_request; ... Apart from that, the patch looks fine to me. Thomas