From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linaro.org; s=google; h=content-transfer-encoding:mime-version:message-id:in-reply-to:date :subject:cc:to:from:user-agent:references:from:to:cc:subject:date :message-id:reply-to; bh=VWDnWtBs+9ZgC5TTdDxLxwi0yw71TmWvtpz30R9yoNc=; b=ftn0rqBSymuxDxeV2qbNJF7peQNPd21kWxAq9SjM8dwj7UsgKYzQByRVsvhwjxg4qG odOO3TU5PALW+RS1A5ubRIBCIfdEw/1h7lu7ApjuSm6RQlMj1+qpnHq06+0fBQoLvem7 c0kqHqr1PnctbvhQUPI9amFkMZfd0A48f3EP+nyQI5NQ4B5kS+P9qN+VK4npOBqx1olB vX3bDmUhkq/QelgAncm9LpVtegkfcFP3tBBJxweEmc+E3GX6rBFzsCsQQINjtpp9GdNC WL1OCv8f6q4JAgOxzN7eLeGTJPmTdZzGkyvuc8+XGv6hijDqbazfOLVJFvyISc+T1xfE 8xoQ== References: <20221010172813.204597-1-mst@redhat.com> <20221010172813.204597-8-mst@redhat.com> <3a002a6a-9135-ad45-3fed-7dd270450d0b@linux.ibm.com> From: Alex =?utf-8?Q?Benn=C3=A9e?= Date: Fri, 14 Oct 2022 09:37:00 +0100 In-reply-to: <3a002a6a-9135-ad45-3fed-7dd270450d0b@linux.ibm.com> Message-ID: <87pmeuol2i.fsf@linaro.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Virtio-fs] Regression save/restore of vsock: (was [PULL 07/55] hw/virtio: move vm_running check to virtio_device_started) List-Id: Development discussions about virtio-fs List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Christian Borntraeger Cc: "Michael S. Tsirkin" , qemu-devel@nongnu.org, Peter Maydell , "Dr. David Alan Gilbert" , Stefan Hajnoczi , Mathieu Poirier , virtio-fs@redhat.com Christian Borntraeger writes: > Am 10.10.22 um 19:29 schrieb Michael S. Tsirkin: >> From: Alex Benn=C3=A9e >> All the boilerplate virtio code does the same thing (or should at >> least) of checking to see if the VM is running before attempting to >> start VirtIO. Push the logic up to the common function to avoid >> getting a copy and paste wrong. >> Signed-off-by: Alex Benn=C3=A9e >> Message-Id: <20220802095010.3330793-11-alex.bennee@linaro.org> >> Reviewed-by: Michael S. Tsirkin >> Signed-off-by: Michael S. Tsirkin > > This results in a regression for our s390x CI when doing save/restore of = guests with vsock: > > > #1 0x000003ff9a248580 raise (libc.so.6 + 0x48580) > #2 0x000003ff9a22b5c0 abort (libc.so.6 + 0x2b5c0) > #3 0x000003ff9a2409da __assert_fail_base (libc.so.6 + 0x= 409da) > #4 0x000003ff9a240a4e __assert_fail (libc.so.6 + 0x40a4e= ) > #5 0x000002aa2d69a066 vhost_vsock_common_pre_save (qemu-= system-s390x + 0x39a066) > #6 0x000002aa2d55570e vmstate_save_state_v (qemu-system-= s390x + 0x25570e) > #7 0x000002aa2d556218 vmstate_save_state (qemu-system-s3= 90x + 0x256218) > #8 0x000002aa2d570ba4 > qemu_savevm_state_complete_precopy_non_iterable (qemu-system-s390x + > 0x270ba4) > #9 0x000002aa2d5710b6 qemu_savevm_state_complete_precopy= (qemu-system-s390x + 0x2710b6) > #10 0x000002aa2d564d0e migration_completion (qemu-system-= s390x + 0x264d0e) > #11 0x000002aa2d8db25c qemu_thread_start (qemu-system-s39= 0x + 0x5db25c) > #12 0x000003ff9a296248 start_thread (libc.so.6 + 0x96248) > #13 0x000003ff9a31183e thread_start (libc.so.6 + 0x11183e= ) Which test does this break? Looking at the change the only thing I can think of is there is a subtle change in the order of checks because if the device is set as use_started we return the result regardless of vm or config state: if (vdev->use_started) { return vdev->started; } Could some printfs confirm that? --=20 Alex Benn=C3=A9e