On Thu, Jan 18, 2018 at 17:04:36 +0100, Michal Privoznik wrote:
This is an extended definition of virStoragePRDef because it contains runtime information (like path to pr helper socket, its pid and alias). Since these are driver dependant we should have a driver specific structure instead of putting all of that into driver agnostic structure.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> --- src/qemu/qemu_domain.c | 120 +++++++++++++++++++++++++++++++++++++++++++++++++ src/qemu/qemu_domain.h | 18 ++++++++ 2 files changed, 138 insertions(+)
diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c index e8539dcab..7fa8c93b7 100644 --- a/src/qemu/qemu_domain.c +++ b/src/qemu/qemu_domain.c @@ -65,6 +65,7 @@ #endif #include <sys/time.h> #include <fcntl.h> +#include <signal.h> #if defined(HAVE_SYS_MOUNT_H) # include <sys/mount.h> #endif @@ -1829,6 +1830,9 @@ qemuDomainObjPrivateDataClear(qemuDomainObjPrivatePtr priv)
virBitmapFree(priv->migrationCaps); priv->migrationCaps = NULL; + + virHashFree(priv->prHelpers); + priv->prHelpers = NULL; }
@@ -10917,6 +10921,122 @@ qemuDomainCheckMigrationCapabilities(virQEMUDriverPtr driver, }
+static void +qemuDomainDiskPRObjectHashFree(void *payload, + const void *name) +{ + qemuDomainDiskPRObjectPtr tmp = payload; + + if (tmp->managed && + tmp->pid != (pid_t) -1) { + VIR_DEBUG("Forcibly killing pr-manager: %s", (const char *) name); + virProcessKillPainfully(tmp->pid, true);
Is this really a good idea? If you restart libvirtd gracefully, this will kill all the PR daemons. Also I'd prefer if all the process management code would be in qemu_process.c.