On Tue, Aug 28, 2018 at 11:39:28PM +0200, marcandre.lureau(a)redhat.com wrote:
From: Marc-André Lureau <marcandre.lureau(a)redhat.com>
Similar to the qemu_tpm.c, add a unit with a few functions to
start/stop and setup the cgroup of the external vhost-user-gpu
process. See function documentation.
Signed-off-by: Marc-André Lureau <marcandre.lureau(a)redhat.com>
---
src/conf/device_conf.h | 1 +
src/qemu/Makefile.inc.am | 2 +
src/qemu/qemu_vhost_user_gpu.c | 318 +++++++++++++++++++++++++++++++++
src/qemu/qemu_vhost_user_gpu.h | 48 +++++
4 files changed, 369 insertions(+)
create mode 100644 src/qemu/qemu_vhost_user_gpu.c
create mode 100644 src/qemu/qemu_vhost_user_gpu.h
diff --git a/src/conf/device_conf.h b/src/conf/device_conf.h
index ff7d6c9d5f..79a7ea9fe2 100644
--- a/src/conf/device_conf.h
+++ b/src/conf/device_conf.h
@@ -175,6 +175,7 @@ struct _virDomainDeviceInfo {
* cases we might want to prevent that from happening by
* locking the isolation group */
bool isolationGroupLocked;
+ int vhost_user_fd;
};
int virDomainDeviceInfoCopy(virDomainDeviceInfoPtr dst,
diff --git a/src/qemu/Makefile.inc.am b/src/qemu/Makefile.inc.am
index 2afa67f195..28daf9d426 100644
--- a/src/qemu/Makefile.inc.am
+++ b/src/qemu/Makefile.inc.am
@@ -56,6 +56,8 @@ QEMU_DRIVER_SOURCES = \
qemu/qemu_qapi.h \
qemu/qemu_tpm.c \
qemu/qemu_tpm.h \
+ qemu/qemu_vhost_user_gpu.c \
+ qemu/qemu_vhost_user_gpu.h \
$(NULL)
diff --git a/src/qemu/qemu_vhost_user_gpu.c b/src/qemu/qemu_vhost_user_gpu.c
new file mode 100644
index 0000000000..9007614020
--- /dev/null
+++ b/src/qemu/qemu_vhost_user_gpu.c
@@ -0,0 +1,318 @@
+/*
+ * qemu_vhost_user_gpu.c: QEMU vhost-user GPU support
+ *
+ * Copyright (C) 2018 Red Hat, Inc.
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library. If not, see
+ * <
http://www.gnu.org/licenses/>.
+ *
+ * Author: Marc-André Lureau <marcandre.lureau(a)redhat.com>
+ */
+
+#include <config.h>
+
+#include "qemu_extdevice.h"
+#include "qemu_domain.h"
+#include "qemu_security.h"
+
+#include "conf/domain_conf.h"
+#include "vircommand.h"
+#include "viralloc.h"
+#include "virlog.h"
+#include "virutil.h"
+#include "virfile.h"
+#include "virstring.h"
+#include "virtime.h"
+#include "virpidfile.h"
+#include "qemu_vhost_user_gpu.h"
+
+#define VIR_FROM_THIS VIR_FROM_NONE
+
+VIR_LOG_INIT("qemu.vhost-user-gpu")
+
+/*
+ * Look up the vhost-user-gpu executable; to be found on the host
+ */
+static char *vhost_user_gpu_path;
+
+static int
+qemuVhostUserGPUInit(void)
+{
+ if (!vhost_user_gpu_path) {
+ vhost_user_gpu_path = virFindFileInPath("vhost-user-gpu");
+ if (!vhost_user_gpu_path) {
+ virReportSystemError(ENOENT, "%s",
+ _("Unable to find 'vhost-user-gpu' binary
in $PATH"));
+ return -1;
+ }
+ if (!virFileIsExecutable(vhost_user_gpu_path)) {
+ virReportError(VIR_ERR_INTERNAL_ERROR,
+ _("vhost-user-gpu %s is not an executable"),
+ vhost_user_gpu_path);
+ VIR_FREE(vhost_user_gpu_path);
+ return -1;
+ }
+ }
+
+ return 0;
+}
+
+
+static char *
+qemuVhostUserGPUCreatePidFilename(const char *stateDir,
+ const char *shortName,
+ const char *alias)
+{
+ char *pidfile = NULL;
+ char *devicename = NULL;
+
+ if (virAsprintf(&devicename, "%s-%s-vhost-user-gpu", shortName, alias)
< 0)
+ return NULL;
+
+ pidfile = virPidFileBuildPath(stateDir, devicename);
+
+ VIR_FREE(devicename);
+
+ return pidfile;
+}
+
+
+/*
+ * qemuVhostUserGPUGetPid
+ *
+ * @stateDir: the directory where vhost-user-gpu writes the pidfile into
+ * @shortName: short name of the domain
+ * @alias: video device alias
+ * @pid: pointer to pid
+ *
+ * Return -errno upon error, or zero on successful reading of the pidfile.
+ * If the PID was not still alive, zero will be returned, and @pid will be
+ * set to -1;
+ */
+static int
+qemuVhostUserGPUGetPid(const char *stateDir,
+ const char *shortName,
+ const char *alias,
+ pid_t *pid)
+{
+ int ret;
+ char *pidfile = qemuVhostUserGPUCreatePidFilename(stateDir, shortName, alias);
+ if (!pidfile)
+ return -ENOMEM;
+
+ ret = virPidFileReadPathIfAlive(pidfile, pid, vhost_user_gpu_path);
+
+ VIR_FREE(pidfile);
+
+ return ret;
+}
+
+
+/*
+ * qemuExtVhostUserGPUStart:
+ *
+ * @driver: QEMU driver
+ * @def: domain definition
+ * @video: the video device
+ * @logCtxt: log context
+ *
+ * Start the external vhost-user-gpu process:
+ * - open a socketpair for vhost-user communication
+ * - have the command line built
+ * - start the external process and sync with it before QEMU start
+ */
+int qemuExtVhostUserGPUStart(virQEMUDriverPtr driver,
+ virDomainDefPtr def,
+ virDomainVideoDefPtr video,
+ qemuDomainLogContextPtr logCtxt)
+{
+ int ret = -1;
+ virCommandPtr cmd = NULL;
+ int exitstatus = 0;
+ char *errbuf = NULL;
+ virQEMUDriverConfigPtr cfg;
+ char *pidfile, *shortName = virDomainDefGetShortName(def);
+ virTimeBackOffVar timebackoff;
+ const unsigned long long timeout = 500000; /* ms */
500 * 1000 /* ms */ is easier to read
+ int cmdret = 0, rc;
+ int pair[2] = { -1, -1 };
+
+ pid_t pid;
+
+ if (!shortName)
+ return -1;
+
+ cfg = virQEMUDriverGetConfig(driver);
+
+ /* stop any left-over for this VM */
+ qemuExtVhostUserGPUStop(driver, def, video);
+
+ if (!(pidfile = qemuVhostUserGPUCreatePidFilename(
+ cfg->stateDir, shortName, video->info.alias)))
+ goto error;
+
+ if (socketpair(AF_UNIX, SOCK_STREAM, 0, pair) < 0) {
+ virReportSystemError(errno, "%s", _("failed to create
socket"));
+ goto error;
+ }
This socket will inherit the label from libvirtd and it cannot be changed
afterwards, see:
https://bugzilla.redhat.com/show_bug.cgi?id=999926
We have SecuritySetSocketLabel and SecurityClearSocketLabel that change
the context temporarily.
Having them on-disk would let us change the label afterwards, but for
FD passing, they should be labeled before the fork (unlike all the other
paths which we label after forking)
Jano