bin/admin-upgrade: add kexec support

Adds a new `--kexec` flag to `ostree admin upgrade` which will cause
the deployment to be loaded into kexec after the upgrade completes.
It is particularly useful in conjunction with the `--reboot` flag to
perform a reboot into the new deployment without waiting for the
(often slow) firmware initialization to take place. (And in my case,
allows me to avoid a normal reboot, which can be unreliable on my
hardware).

After an image has been loaded (using the `kexec_file_load` syscall),
the `systemctl-reboot` command (which is called when the existing
`-r` flag is included) will trigger a kexec on the loaded image
rather than a normal reboot. From `systemctl(1)`:

  If a new kernel has been loaded via kexec --load, a kexec will be
  performed instead of a reboot, unless "SYSTEMCTL_SKIP_AUTO_KEXEC=1"
  has been set. If a new root file system has been set up on
  "/run/nextroot/", a soft-reboot will be performed instead of a
  reboot, unless "SYSTEMCTL_SKIP_AUTO_SOFT_REBOOT=1" has been set.

A good in-depth technical explanation of kexec can be found here:
https://web.archive.org/web/20090505132901/http://www.ibm.com/developerworks/linux/library/l-kexec.html

My implementation uses the `kexec_file_load` syscall rather than the
older `kexec_load` syscall, which allows the kernel to verify the
signatures of the new kernel. It is supported on Linux 3.17 and
newer. I assume this probably won't be an issue, but if it is, it's
not that hard to put a preprocessor directive around the kexec stuff
to disable it for older kernels. Even RHEL is new enough now to
not be an issue :)

Closes: ostreedev#435

man/ostree-admin-upgrade.xml

@@ -111,6 +111,14 @@ License along with this library. If not, see <https://www.gnu.org/licenses/>.
                     Reboot after a successful upgrade.
                 </para></listitem>
             </varlistentry>
+
+            <varlistentry>
+                <term><option>--kexec</option>,<option>-k</option></term>
+
+                <listitem><para>
+                    Load new deployment into kexec after a successful upgrade.
+                </para></listitem>
+            </varlistentry>
 
             <varlistentry>
                 <term><option>--allow-downgrade</option></term>

src/libostree/ostree-sysroot-deploy.c

@@ -30,6 +30,7 @@
 #include <sys/poll.h>
 #include <sys/socket.h>
 #include <sys/statvfs.h>
+#include <linux/kexec.h>
 
 #ifdef HAVE_LIBMOUNT
 #include <libmount.h>
@@ -2390,6 +2391,51 @@ ostree_sysroot_write_deployments (OstreeSysroot *self, GPtrArray *new_deployment
                                                         error);
 }
 
+static gboolean
+kexec_load_kernel (OstreeSysroot *self, OstreeDeployment *deployment,
+                   GCancellable *cancellable, GError **error)
+{
+  GLNX_AUTO_PREFIX_ERROR ("Loading kernel into kexec", error);
+  OstreeBootconfigParser *bootconfig = ostree_deployment_get_bootconfig (deployment);
+  const char *kargs = ostree_bootconfig_parser_get(bootconfig, "options");
+  g_autofree char *deployment_dirpath = ostree_sysroot_get_deployment_dirpath (self, deployment);
+  glnx_autofd int deployment_dfd = -1;
+  if (!glnx_opendirat (self->sysroot_fd, deployment_dirpath, FALSE, &deployment_dfd, error))
+    return FALSE;
+
+  /* Find the kernel/initramfs in the tree */
+  g_autoptr (OstreeKernelLayout) kernel_layout = NULL;
+  if (!get_kernel_from_tree (self, deployment_dfd, &kernel_layout, cancellable, error))
+    return FALSE;
+
+  unsigned long flags = 0;
+  glnx_autofd int kernel_fd = -1;
+  glnx_autofd int initrd_fd = -1;
+
+  if (!glnx_openat_rdonly (kernel_layout->boot_dfd, kernel_layout->kernel_srcpath,
+                           TRUE, &kernel_fd, error))
+    return FALSE;
+
+  /* initramfs is optional */
+  if (kernel_layout->initramfs_srcpath)
+    {
+      if (!glnx_openat_rdonly (kernel_layout->boot_dfd, kernel_layout->initramfs_srcpath,
+                               TRUE, &initrd_fd, error))
+        {
+          return FALSE;
+        }
+    }
+  else
+    {
+      flags |= KEXEC_FILE_NO_INITRAMFS;
+    }
+
+  if (syscall (SYS_kexec_file_load, kernel_fd, initrd_fd, strlen (kargs) + 1, kargs, flags))
+    return glnx_throw_errno_prefix(error, "kexec_file_load");
+
+  return TRUE;
+}
+
 /* Handle writing out a new bootloader config. One reason this needs to be a
  * helper function is to handle wrapping it with temporarily remounting /boot
  * rw.
@@ -2994,6 +3040,9 @@ ostree_sysroot_write_deployments_with_options (OstreeSysroot *self, GPtrArray *n
         return FALSE;
     }
 
+  if (new_deployments->len && !kexec_load_kernel(self, new_deployments->pdata[0], cancellable, error))
+    return FALSE;
+
   {
     g_autofree char *msg
         = g_strdup_printf ("%s; bootconfig swap: %s; bootversion: %s, deployment count change: %i",

src/libostree/ostree-sysroot-upgrader.c

@@ -601,6 +601,10 @@ ostree_sysroot_upgrader_deploy (OstreeSysrootUpgrader *self, GCancellable *cance
   /* Experimental flag to enable staging */
   gboolean stage = (self->flags & OSTREE_SYSROOT_UPGRADER_FLAGS_STAGE) > 0
                    || getenv ("OSTREE_EX_STAGE_DEPLOYMENTS") != NULL;
+  OstreeSysrootSimpleWriteDeploymentFlags write_deployment_flags = OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_NONE;
+  if ((self->flags & OSTREE_SYSROOT_UPGRADER_FLAGS_LOAD_KEXEC) > 0) {
+    write_deployment_flags |= OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_LOAD_KEXEC;
+  }
   if (stage)
     {
       if (!ostree_sysroot_stage_tree (self->sysroot, self->osname, self->new_revision, self->origin,
@@ -616,7 +620,7 @@ ostree_sysroot_upgrader_deploy (OstreeSysrootUpgrader *self, GCancellable *cance
         return FALSE;
 
       if (!ostree_sysroot_simple_write_deployment (self->sysroot, self->osname, new_deployment,
-                                                   self->merge_deployment, 0, cancellable, error))
+                                                   self->merge_deployment, write_deployment_flags, cancellable, error))
         return FALSE;
     }
 
@@ -635,6 +639,8 @@ ostree_sysroot_upgrader_flags_get_type (void)
                 "OSTREE_SYSROOT_UPGRADER_FLAGS_IGNORE_UNCONFIGURED", "ignore-unconfigured" },
               { OSTREE_SYSROOT_UPGRADER_FLAGS_STAGE, "OSTREE_SYSROOT_UPGRADER_FLAGS_STAGE",
                 "stage" },
+              { OSTREE_SYSROOT_UPGRADER_FLAGS_LOAD_KEXEC, "OSTREE_SYSROOT_UPGRADER_FLAGS_LOAD_KEXEC",
+                "kexec" },
               { 0, NULL, NULL } };
       GType g_define_type_id
           = g_flags_register_static (g_intern_static_string ("OstreeSysrootUpgraderFlags"), values);

src/libostree/ostree-sysroot-upgrader.h

@@ -36,6 +36,7 @@ G_BEGIN_DECLS
  * unconfigured-state key
  * @OSTREE_SYSROOT_UPGRADER_FLAGS_STAGE: Enable "staging" (finalization at shutdown); recommended
  *    (Since: 2021.4)
+ * @OSTREE_SYSROOT_UPGRADER_FLAGS_LOAD_KEXEC: Enable loading the new deployment into kexec
  *
  * Flags controlling operation of an #OstreeSysrootUpgrader.
  */
@@ -44,6 +45,7 @@ typedef enum
   OSTREE_SYSROOT_UPGRADER_FLAGS_NONE = (1 << 0),
   OSTREE_SYSROOT_UPGRADER_FLAGS_IGNORE_UNCONFIGURED = (1 << 1),
   OSTREE_SYSROOT_UPGRADER_FLAGS_STAGE = (1 << 2),
+  OSTREE_SYSROOT_UPGRADER_FLAGS_LOAD_KEXEC = (1 << 3),
 } OstreeSysrootUpgraderFlags;
 
 _OSTREE_PUBLIC

src/libostree/ostree-sysroot.c

@@ -1913,6 +1913,8 @@ ostree_sysroot_simple_write_deployment (OstreeSysroot *sysroot, const char *osna
                                         OstreeSysrootSimpleWriteDeploymentFlags flags,
                                         GCancellable *cancellable, GError **error)
 {
+  const gboolean load_kexec
+      = (flags & OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_LOAD_KEXEC) > 0;
   const gboolean postclean = (flags & OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_NO_CLEAN) == 0;
   const gboolean make_default
       = !((flags & OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_NOT_DEFAULT) > 0);
@@ -1987,7 +1989,7 @@ ostree_sysroot_simple_write_deployment (OstreeSysroot *sysroot, const char *osna
   if (!added_new)
     g_ptr_array_add (new_deployments, g_object_ref (new_deployment));
 
-  OstreeSysrootWriteDeploymentsOpts write_opts = { .do_postclean = postclean };
+  OstreeSysrootWriteDeploymentsOpts write_opts = { .do_postclean = postclean, .load_kexec = load_kexec };
   if (!ostree_sysroot_write_deployments_with_options (sysroot, new_deployments, &write_opts,
                                                       cancellable, error))
     return FALSE;

src/libostree/ostree-sysroot.h

@@ -161,7 +161,8 @@ typedef struct
 {
   gboolean do_postclean;
   gboolean disable_auto_early_prune;
-  gboolean unused_bools[7];
+  gboolean load_kexec;
+  gboolean unused_bools[6];
   int unused_ints[7];
   gpointer unused_ptrs[7];
 } OstreeSysrootWriteDeploymentsOpts;
@@ -259,6 +260,7 @@ typedef enum
   OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_NO_CLEAN = (1 << 2),
   OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_RETAIN_PENDING = (1 << 3),
   OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_RETAIN_ROLLBACK = (1 << 4),
+  OSTREE_SYSROOT_SIMPLE_WRITE_DEPLOYMENT_FLAGS_LOAD_KEXEC = (1 << 5),
 } OstreeSysrootSimpleWriteDeploymentFlags;
 
 _OSTREE_PUBLIC

src/ostree/ot-admin-builtin-upgrade.c

@@ -31,6 +31,7 @@
 #include <unistd.h>
 
 static gboolean opt_reboot;
+static gboolean opt_kexec;
 static gboolean opt_allow_downgrade;
 static gboolean opt_pull_only;
 static gboolean opt_deploy_only;
@@ -42,6 +43,7 @@ static GOptionEntry options[] = {
   { "os", 0, 0, G_OPTION_ARG_STRING, &opt_osname,
     "Use a different operating system root than the current one", "OSNAME" },
   { "reboot", 'r', 0, G_OPTION_ARG_NONE, &opt_reboot, "Reboot after a successful upgrade", NULL },
+  { "kexec", 'k', 0, G_OPTION_ARG_NONE, &opt_kexec, "Stage new kernel in kexec", NULL },
   { "allow-downgrade", 0, 0, G_OPTION_ARG_NONE, &opt_allow_downgrade,
     "Permit deployment of chronologically older trees", NULL },
   { "override-commit", 0, 0, G_OPTION_ARG_STRING, &opt_override_commit,
@@ -72,16 +74,18 @@ ot_admin_builtin_upgrade (int argc, char **argv, OstreeCommandInvocation *invoca
                    "Cannot simultaneously specify --pull-only and --deploy-only");
       return FALSE;
     }
-  else if (opt_pull_only && opt_reboot)
+  else if (opt_pull_only && (opt_reboot || opt_kexec))
     {
       g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED,
-                   "Cannot simultaneously specify --pull-only and --reboot");
+                   "Cannot simultaneously specify --pull-only and --reboot or --kexec");
       return FALSE;
     }
 
   OstreeSysrootUpgraderFlags flags = 0;
   if (opt_stage)
     flags |= OSTREE_SYSROOT_UPGRADER_FLAGS_STAGE;
+  if (opt_kexec)
+    flags |= OSTREE_SYSROOT_UPGRADER_FLAGS_LOAD_KEXEC;
 
   g_autoptr (OstreeSysrootUpgrader) upgrader = ostree_sysroot_upgrader_new_for_os_with_flags (
       sysroot, opt_osname, flags, cancellable, error);