Add support for KVM, monitoring bbot diagnostics

4 files changed, 68 insertions, 4 deletions

diff --git a/bootstrap b/bootstrap
index ca7cffe..cfbf59a 100755
--- a/bootstrap
+++ b/bootstrap
@@ -155,6 +155,8 @@ if [ "$stage" -eq "1" ]; then
   pkgs+=",bzip2,xz-utils"
   pkgs+=",less,nano"
 
+  pkgs+=",qemu-kvm,socat"
+
   pkgs+=",g++,pkg-config"
 
   sudo debootstrap \
@@ -293,6 +295,7 @@ set -x
 #
 addgroup --gid 2000 build
 adduser --uid 2000 --gid 2000 --home /build --gecos "" --disabled-password build
+adduser build kvm
 echo "build ALL=(ALL) NOPASSWD:ALL" >/etc/sudoers.d/build
 chmod 0440 /etc/sudoers.d/build
 
@@ -415,6 +418,8 @@ EOF
   subvol_snapshot -r "$root" "$root-5"
 fi
 
+exit 0
+
 # Test.
 #
 if [ ! -e /tmp/buildos-state ]; then
diff --git a/buildos b/buildos
index 0461e68..6c3e008 100755
--- a/buildos
+++ b/buildos
@@ -77,6 +77,14 @@ for v in "${cmdline[@]}"; do
   fi
 done
 
+# Make sure /dev/kvm has correct permissions (while all the udev scripts seem
+# to be there, it does not work for some reason).
+#
+if sudo test -c /dev/kvm; then
+  sudo chgrp kvm /dev/kvm
+  sudo chmod 0660 /dev/kvm
+fi
+
 hname="$(hostname)"
 
 # Get the build id.
@@ -721,6 +729,7 @@ function bbot_start () # <toolchain-name> <toolchain-index>
 # start their bbot agents.
 #
 toolchain_boots=()
+declare -A toolchain_cursors # Latest systemd journal cursor.
 
 # Monitoring loop.
 #
@@ -844,7 +853,38 @@ EOF
 	    sudo systemctl reset-failed "bbot-agent@$tn" 2>&1 | \
 	      tee -a "$tr/bbot-$count.log" 1>&2
 	  else
-	    continue # Nothing to do.
+	    # See if there is any diagnostics in the systemd journal. We
+	    # notify about warning and up.
+	    #
+	    # The old versions journalctl behavior is to not output anything
+	    # (not even the cursor) if there are no new entries. The new
+	    # versions output the old cursor.
+	    #
+	    c=(sudo journalctl --unit "bbot-agent@$tn")
+
+	    # Get the last cursor if any.
+	    #
+	    oc="${toolchain_cursors["$tn"]}"
+	    if [ -n "$oc" ]; then
+	      c+=("--after-cursor" "$oc")
+	    fi
+
+	    nc="$("${c[@]}" --no-pager --quiet --priority 4 --show-cursor | \
+sed -n -re 's/^-- cursor: (.+)$/\1/p')"
+
+	    # If we have no new entries, then nothing to do.
+	    #
+	    if [ -n "$nc" -a "$nc" != "$oc" ]; then
+	      s="bbot-agent@$tn service issued new diagnostics"
+
+	      info "$s"
+	      email "$s" <<EOF
+$tn.bbot_cmd: ssh build@$hname ${c[@]}
+EOF
+	      toolchain_cursors["$tn"]="$nc"
+	    fi
+
+	    continue
 	  fi
 	  ;;
 	1)
diff --git a/doc/manual.cli b/doc/manual.cli
index b91f6ae..2d275aa 100644
--- a/doc/manual.cli
+++ b/doc/manual.cli
@@ -199,7 +199,7 @@ While any suitable filesystem can be used, \c{ext4} is a good choice, with
 journaling disabled if used on a flash drive/card. For example:
 
 \
-mkfs.ext4 -L buildos.machines -O ^has_journal /dev/sdX
+mkfs.ext4 -L buildos.state -O ^has_journal /dev/sdX
 \
 
 \h2#config-storage-machines|Machines|
@@ -423,7 +423,7 @@ build$ cd /build/machines/default/linux-gcc_6
 # Make user build the owner of the machine subvolume.
 #
 build$ sudo btrfs property set -ts linux-gcc_6-1.0 ro false
-build$ sudo chown build:build linux-gcc_6-1.0
+build$ sudo chown build:build linux-gcc_6-1.0 linux-gcc_6-1.0/*
 build$ btrfs property set -ts linux-gcc_6-1.0 ro true
 
 # Make the subvolume the current machine.
@@ -450,7 +450,7 @@ build$ cd /build/machines/default/linux-gcc_6
 # Make user build the owner of the new machine subvolume.
 #
 build$ sudo btrfs property set -ts linux-gcc_6-1.1 ro false
-build$ sudo chown build:build linux-gcc_6-1.1
+build$ sudo chown build:build linux-gcc_6-1.1 linux-gcc_6-1.1/*
 build$ btrfs property set -ts linux-gcc_6-1.1 ro true
 
 # Switch the current machine atomically.
@@ -505,4 +505,17 @@ Note also that on reboot the Build OS monitor examines and cleans up
 machine directories of any stray subvolumes. As a result, an alternative
 approach would be to remove the current machine symlink and reboot the
 build host.
+
+\h#machines-login|Logging into a Machine|
+
+A running QEMU/KVM machine (that is, one being bootstrapped or used for
+building) can be accessed with a VNC client. @@ TODO
+
+\
+echo cont | ssh build@build socat - UNIX-CONNECT:/tmp/<toolchain>-monitor
+\
+
+\
+ssh -f -L 5901:127.0.0.1:5901 build@build sleep 1 && vncviewer 127.0.0.1:5901
+\
 "
diff --git a/init b/init
index 694dcb0..18a6b0b 100755
--- a/init
+++ b/init
@@ -66,6 +66,12 @@ SYSTEMD_LOG_LEVEL=info /lib/systemd/systemd-udevd --daemon --resolve-names=never
 udevadm trigger --action=add
 udevadm settle || true
 
+# Initialize KVM.
+#
+#if ! (modprobe kvm_intel || modprobe kvm_amd); then
+#  error "no virtualization support available (is it disabled in BIOS?)"
+#fi
+
 # Parse the kernel command line. This is complicated by the fact that the
 # values can be quoted, for example:
 #