qemu: Add support for overriding max threads per process limit

Some VM configurations may result in a large number of threads created by
the associated qemu process which can exceed the system default limit. The
maximum number of threads allowed per process is controlled by the pids
cgroup controller and is set to 16k when creating VMs with systemd's
machined service. The maximum number of threads per process is recorded
in the pids.max file under the machine's pids controller cgroup hierarchy,
e.g.

$cgrp-mnt/pids/machine.slice/machine-qemu\\x2d1\\x2dtest.scope/pids.max

Maximum threads per process is controlled with the TasksMax property of
the systemd scope for the machine. This patch adds an option to qemu.conf
which can be used to override the maximum number of threads allowed per
qemu process. If the value of option is greater than zero, it will be set
in the TasksMax property of the machine's scope after creating the machine.
Signed-off-by: NJim Fehlig <jfehlig@suse.com>
Reviewed-by: NDaniel P. Berrangé <berrange@redhat.com>

src/lxc/lxc_cgroup.c

@@ -455,6 +455,7 @@ virCgroupPtr virLXCCgroupCreate(virDomainDefPtr def,
                             nnicindexes, nicindexes,
                             def->resource->partition,
                             -1,
+                            0,
                             &cgroup) < 0)
         goto cleanup;
 

src/qemu/libvirtd_qemu.aug

@@ -95,6 +95,7 @@ module Libvirtd_qemu =
                  | limits_entry "max_core"
                  | bool_entry "dump_guest_core"
                  | str_entry "stdio_handler"
+                 | int_entry "max_threads_per_process"
 
    let device_entry = bool_entry "mac_filter"
                  | bool_entry "relaxed_acs_check"

src/qemu/qemu.conf

@@ -613,6 +613,16 @@
 #max_processes = 0
 #max_files = 0
 
+# If max_threads_per_process is set to a positive integer, libvirt
+# will use it to set the maximum number of threads that can be
+# created by a qemu process. Some VM configurations can result in
+# qemu processes with tens of thousands of threads. systemd-based
+# systems typically limit the number of threads per process to
+# 16k. max_threads_per_process can be used to override default
+# limits in the host OS.
+#
+#max_threads_per_process = 0
+
 # If max_core is set to a non-zero integer, then QEMU will be
 # permitted to create core dumps when it crashes, provided its
 # RAM size is smaller than the limit set.

src/qemu/qemu_cgroup.c

@@ -929,6 +929,7 @@ qemuInitCgroup(virDomainObjPtr vm,
                             nnicindexes, nicindexes,
                             vm->def->resource->partition,
                             cfg->cgroupControllers,
+                            cfg->maxThreadsPerProc,
                             &priv->cgroup) < 0) {
         if (virCgroupNewIgnoreError())
             goto done;

src/qemu/qemu_conf.c

@@ -670,6 +670,8 @@ virQEMUDriverConfigLoadProcessEntry(virQEMUDriverConfigPtr cfg,
         return -1;
     if (virConfGetValueUInt(conf, "max_files", &cfg->maxFiles) < 0)
         return -1;
+    if (virConfGetValueUInt(conf, "max_threads_per_process", &cfg->maxThreadsPerProc) < 0)
+        return -1;
 
     if (virConfGetValueType(conf, "max_core") == VIR_CONF_STRING) {
         if (virConfGetValueString(conf, "max_core", &corestr) < 0)

src/qemu/qemu_conf.h

@@ -162,6 +162,7 @@ struct _virQEMUDriverConfig {
 
     unsigned int maxProcesses;
     unsigned int maxFiles;
+    unsigned int maxThreadsPerProc;
     unsigned long long maxCore;
     bool dumpGuestCore;
 

src/qemu/test_libvirtd_qemu.aug.in

@@ -76,6 +76,7 @@ module Test_libvirtd_qemu =
 { "set_process_name" = "1" }
 { "max_processes" = "0" }
 { "max_files" = "0" }
+{ "max_threads_per_process" = "0" }
 { "max_core" = "unlimited" }
 { "dump_guest_core" = "1" }
 { "mac_filter" = "1" }

src/util/vircgroup.c

@@ -1118,6 +1118,7 @@ virCgroupNewMachineSystemd(const char *name,
                            int *nicindexes,
                            const char *partition,
                            int controllers,
+                           unsigned int maxthreads,
                            virCgroupPtr *group)
 {
     int rv;
@@ -1134,7 +1135,8 @@ virCgroupNewMachineSystemd(const char *name,
                                       isContainer,
                                       nnicindexes,
                                       nicindexes,
-                                      partition)) < 0)
+                                      partition,
+                                      maxthreads)) < 0)
         return rv;
 
     if (controllers != -1)
@@ -1246,6 +1248,7 @@ virCgroupNewMachine(const char *name,
                     int *nicindexes,
                     const char *partition,
                     int controllers,
+                    unsigned int maxthreads,
                     virCgroupPtr *group)
 {
     int rv;
@@ -1262,6 +1265,7 @@ virCgroupNewMachine(const char *name,
                                          nicindexes,
                                          partition,
                                          controllers,
+                                         maxthreads,
                                          group)) == 0)
         return 0;
 

src/util/vircgroup.h

@@ -98,6 +98,7 @@ int virCgroupNewMachine(const char *name,
                         int *nicindexes,
                         const char *partition,
                         int controllers,
+                        unsigned int maxthreads,
                         virCgroupPtr *group)
     ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2)
     ATTRIBUTE_NONNULL(3);

src/util/virsystemd.c

@@ -252,12 +252,14 @@ int virSystemdCreateMachine(const char *name,
                             bool iscontainer,
                             size_t nnicindexes,
                             int *nicindexes,
-                            const char *partition)
+                            const char *partition,
+                            unsigned int maxthreads)
 {
     int ret;
     DBusConnection *conn;
     char *creatorname = NULL;
     char *slicename = NULL;
+    char *scopename = NULL;
     static int hasCreateWithNetwork = 1;
 
     if ((ret = virSystemdHasMachined()) < 0)
@@ -403,11 +405,31 @@ int virSystemdCreateMachine(const char *name,
             goto cleanup;
     }
 
+    if (maxthreads > 0) {
+        if (!(scopename = virSystemdMakeScopeName(name, drivername, false)))
+            goto cleanup;
+
+        if (virDBusCallMethod(conn,
+                              NULL,
+                              NULL,
+                              "org.freedesktop.systemd1",
+                              "/org/freedesktop/systemd1",
+                              "org.freedesktop.systemd1.Manager",
+                              "SetUnitProperties",
+                              "sba(sv)",
+                              scopename,
+                              true,
+                              1,
+                              "TasksMax", "t", (uint64_t)maxthreads) < 0)
+            goto cleanup;
+    }
+
     ret = 0;
 
  cleanup:
     VIR_FREE(creatorname);
     VIR_FREE(slicename);
+    VIR_FREE(scopename);
     return ret;
 }
 

src/util/virsystemd.h

@@ -50,7 +50,8 @@ int virSystemdCreateMachine(const char *name,
                             bool iscontainer,
                             size_t nnicindexes,
                             int *nicindexes,
-                            const char *partition);
+                            const char *partition,
+                            unsigned int maxthreads);
 
 int virSystemdTerminateMachine(const char *name);
 

tests/virsystemdtest.c

@@ -175,7 +175,7 @@ static int testCreateContainer(const void *opaque ATTRIBUTE_UNUSED)
                                 123,
                                 true,
                                 0, NULL,
-                                "highpriority.slice") < 0) {
+                                "highpriority.slice", 0) < 0) {
         fprintf(stderr, "%s", "Failed to create LXC machine\n");
         return -1;
     }
@@ -208,7 +208,7 @@ static int testCreateMachine(const void *opaque ATTRIBUTE_UNUSED)
                                 123,
                                 false,
                                 0, NULL,
-                                NULL) < 0) {
+                                NULL, 0) < 0) {
         fprintf(stderr, "%s", "Failed to create KVM machine\n");
         return -1;
     }
@@ -245,7 +245,7 @@ static int testCreateNoSystemd(const void *opaque ATTRIBUTE_UNUSED)
                                       123,
                                       false,
                                       0, NULL,
-                                      NULL)) == 0) {
+                                      NULL, 0)) == 0) {
         unsetenv("FAIL_NO_SERVICE");
         fprintf(stderr, "%s", "Unexpected create machine success\n");
         return -1;
@@ -279,7 +279,7 @@ static int testCreateSystemdNotRunning(const void *opaque ATTRIBUTE_UNUSED)
                                       123,
                                       false,
                                       0, NULL,
-                                      NULL)) == 0) {
+                                      NULL, 0)) == 0) {
         unsetenv("FAIL_NOT_REGISTERED");
         fprintf(stderr, "%s", "Unexpected create machine success\n");
         return -1;
@@ -313,7 +313,7 @@ static int testCreateBadSystemd(const void *opaque ATTRIBUTE_UNUSED)
                                       123,
                                       false,
                                       0, NULL,
-                                      NULL)) == 0) {
+                                      NULL, 0)) == 0) {
         unsetenv("FAIL_BAD_SERVICE");
         fprintf(stderr, "%s", "Unexpected create machine success\n");
         return -1;
@@ -348,7 +348,7 @@ static int testCreateNetwork(const void *opaque ATTRIBUTE_UNUSED)
                                 123,
                                 true,
                                 nnicindexes, nicindexes,
-                                "highpriority.slice") < 0) {
+                                "highpriority.slice", 0) < 0) {
         fprintf(stderr, "%s", "Failed to create LXC machine\n");
         return -1;
     }