From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:34844) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1gNIgD-0002DZ-KZ for qemu-devel@nongnu.org; Thu, 15 Nov 2018 09:31:38 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1gNIg5-0003rK-IE for qemu-devel@nongnu.org; Thu, 15 Nov 2018 09:31:35 -0500 Received: from mail-wm1-x32f.google.com ([2a00:1450:4864:20::32f]:55297) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1gNIg5-0003pz-B3 for qemu-devel@nongnu.org; Thu, 15 Nov 2018 09:31:29 -0500 Received: by mail-wm1-x32f.google.com with SMTP id y139so1666740wmc.5 for ; Thu, 15 Nov 2018 06:31:27 -0800 (PST) Sender: Paolo Bonzini From: Paolo Bonzini Date: Thu, 15 Nov 2018 15:31:14 +0100 Message-Id: <20181115143124.19234-1-pbonzini@redhat.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Subject: [Qemu-devel] [PATCH for-3.2 00/10] vhost: preparation for qgraph conversion of vhost-user-test List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: lvivier@redhat.com, mst@redhat.com, jasowang@redhat.com, =?UTF-8?q?Marc-Andr=C3=A9=20Lureau?= The vhost-user-test is one of the most complex qtests in the tree, and converting it to qgraph poses some unique challenges. This series addresses them in a way that does not depend on the qgraph conversion itself, but prepares for it. First of all, vhost-user-test is slow and that is a pain when debugging it. :) So patches 1 and 2 fix this bug. Second, qgraph will be able to test vhost-user-test on pSeries and ARM virt (with virtio-mmio) machine types. Patches 3 and 4 prepares for this by compiling vhost-net even on TCG-only targets, and by supporting wrong-endianness vnet headers. Patches 5 and 6 instead make the test more robust, so that it can run on those machines too. Third, vhost-user-test is a bit strange in that it has more global state than other tests (in particular a GMainLoop, but also a directory in /tmp and some weird uses of global_qtest). Patches 7 to 10 thus remove this global state already, which simplifies the code and also will reduce the diff for the qgraph conversion. Please review and ack! Thanks, Paolo Paolo Bonzini (10): vhost-user-test: use g_cond_broadcast vhost-user-test: signal data_cond when s->rings changes vhost-net: compile it for all targets vhost-user: support cross-endian vnet headers vhost-user-test: support VHOST_USER_PROTOCOL_F_CROSS_ENDIAN vhost-user-test: skip if there is no memory at address 0 vhost-user-test: reduce usage of global_qtest vhost-user-test: create a main loop per TestServer vhost-user-test: small changes to init_hugepagefs vhost-user-test: create a temporary directory per TestServer configure | 13 +-- include/exec/poison.h | 1 - net/net.c | 2 +- net/vhost-user.c | 13 +++ tests/Makefile.include | 5 +- tests/vhost-user-test.c | 234 ++++++++++++++++++++++------------------ 6 files changed, 147 insertions(+), 121 deletions(-) -- 2.19.1