nixpkgs/pkgs/build-support/vm/windows/controller/default.nix

264 lines
6.8 KiB
Nix
Raw Normal View History

{ stdenv, writeScript, vmTools, makeInitrd
, samba, vde2, openssh, socat, netcat-gnu, coreutils, gnugrep, gzip
, runtimeShell
}:
{ sshKey
, qemuArgs ? []
, command ? "sync"
, suspendTo ? null
, resumeFrom ? null
, installMode ? false
}:
with stdenv.lib;
let
preInitScript = writeScript "preinit.sh" ''
#!${vmTools.initrdUtils}/bin/ash -e
export PATH=${vmTools.initrdUtils}/bin
mount -t proc none /proc
mount -t sysfs none /sys
for arg in $(cat /proc/cmdline); do
if [ "x''${arg#command=}" != "x$arg" ]; then
command="''${arg#command=}"
fi
done
for i in $(cat ${modulesClosure}/insmod-list); do
insmod $i
done
mkdir -p /dev /fs
mount -t tmpfs none /dev
mknod /dev/null c 1 3
mknod /dev/zero c 1 5
mknod /dev/random c 1 8
mknod /dev/urandom c 1 9
mknod /dev/tty c 5 0
ifconfig lo up
ifconfig eth0 up 192.168.0.2
mount -t tmpfs none /fs
mkdir -p /fs/nix/store /fs/xchg /fs/dev /fs/sys /fs/proc /fs/etc /fs/tmp
mount -o bind /dev /fs/dev
mount -t sysfs none /fs/sys
mount -t proc none /fs/proc
mount -t 9p \
nixos/vm-tests: Remove msize mount option This seems to be the root cause of the random page allocation failures and @wizeman did a very good job on not only finding the root problem but also giving a detailed explanation of it in #10828. Here is an excerpt: The problem here is that the kernel is trying to allocate a contiguous section of 2^7=128 pages, which is 512 KB. This is way too much: kernel pages tend to get fragmented over time and kernel developers often go to great lengths to try allocating at most only 1 contiguous page at a time whenever they can. From the error message, it looks like the culprit is unionfs, but this is misleading: unionfs is the name of the userspace process that was running when the system ran out of memory, but it wasn't unionfs who was allocating the memory: it was the kernel; specifically it was the v9fs_dir_readdir_dotl() function, which is the code for handling the readdir() function in the 9p filesystem (the filesystem that is used to share a directory structure between a qemu host and its VM). If you look at the code, here's what it's doing at the moment it tries to allocate memory: buflen = fid->clnt->msize - P9_IOHDRSZ; rdir = v9fs_alloc_rdir_buf(file, buflen); If you look into v9fs_alloc_rdir_buf(), you will see that it will try to allocate a contiguous buffer of memory (using kzalloc(), which is a wrapper around kmalloc()) of size buflen + 8 bytes or so. So in reality, this code actually allocates a buffer of size proportional to fid->clnt->msize. What is this msize? If you follow the definition of the structures, you will see that it's the negotiated buffer transfer size between 9p client and 9p server. On the client side, it can be controlled with the msize mount option. What this all means is that, the reason for running out of memory is that the code (which we can't easily change) tries to allocate a contiguous buffer of size more or less equal to "negotiated 9p protocol buffer size", which seems to be way too big (in our NixOS tests, at least). After that initial finding, @lethalman tested the gnome3 gdm test without setting the msize parameter at all and it seems to have resolved the problem. The reason why I'm committing this without testing against all of the NixOS VM test is basically that I think we can only go better but not worse than the current state. Signed-off-by: aszlig <aszlig@redmoonstudios.org>
2015-12-14 17:26:24 +01:00
-o trans=virtio,version=9p2000.L,cache=loose \
store /fs/nix/store
mount -t 9p \
nixos/vm-tests: Remove msize mount option This seems to be the root cause of the random page allocation failures and @wizeman did a very good job on not only finding the root problem but also giving a detailed explanation of it in #10828. Here is an excerpt: The problem here is that the kernel is trying to allocate a contiguous section of 2^7=128 pages, which is 512 KB. This is way too much: kernel pages tend to get fragmented over time and kernel developers often go to great lengths to try allocating at most only 1 contiguous page at a time whenever they can. From the error message, it looks like the culprit is unionfs, but this is misleading: unionfs is the name of the userspace process that was running when the system ran out of memory, but it wasn't unionfs who was allocating the memory: it was the kernel; specifically it was the v9fs_dir_readdir_dotl() function, which is the code for handling the readdir() function in the 9p filesystem (the filesystem that is used to share a directory structure between a qemu host and its VM). If you look at the code, here's what it's doing at the moment it tries to allocate memory: buflen = fid->clnt->msize - P9_IOHDRSZ; rdir = v9fs_alloc_rdir_buf(file, buflen); If you look into v9fs_alloc_rdir_buf(), you will see that it will try to allocate a contiguous buffer of memory (using kzalloc(), which is a wrapper around kmalloc()) of size buflen + 8 bytes or so. So in reality, this code actually allocates a buffer of size proportional to fid->clnt->msize. What is this msize? If you follow the definition of the structures, you will see that it's the negotiated buffer transfer size between 9p client and 9p server. On the client side, it can be controlled with the msize mount option. What this all means is that, the reason for running out of memory is that the code (which we can't easily change) tries to allocate a contiguous buffer of size more or less equal to "negotiated 9p protocol buffer size", which seems to be way too big (in our NixOS tests, at least). After that initial finding, @lethalman tested the gnome3 gdm test without setting the msize parameter at all and it seems to have resolved the problem. The reason why I'm committing this without testing against all of the NixOS VM test is basically that I think we can only go better but not worse than the current state. Signed-off-by: aszlig <aszlig@redmoonstudios.org>
2015-12-14 17:26:24 +01:00
-o trans=virtio,version=9p2000.L,cache=loose \
xchg /fs/xchg
echo root:x:0:0::/root:/bin/false > /fs/etc/passwd
set +e
chroot /fs $command $out
echo $? > /fs/xchg/in-vm-exit
poweroff -f
'';
initrd = makeInitrd {
contents = singleton {
object = preInitScript;
symlink = "/init";
};
};
loopForever = "while :; do ${coreutils}/bin/sleep 1; done";
initScript = writeScript "init.sh" (''
#!${runtimeShell}
${coreutils}/bin/cp -L "${sshKey}" /ssh.key
${coreutils}/bin/chmod 600 /ssh.key
'' + (if installMode then ''
echo -n "Waiting for Windows installation to finish..."
while ! ${netcat-gnu}/bin/netcat -z 192.168.0.1 22; do
echo -n .
# Print a dot every 10 seconds only to shorten line length.
${coreutils}/bin/sleep 10
done
${coreutils}/bin/touch /xchg/waiting_done
echo " success."
# Loop forever, because this VM is going to be killed.
${loopForever}
'' else ''
${coreutils}/bin/mkdir -p /etc/samba /etc/samba/private \
/var/lib/samba /var/log /var/run
${coreutils}/bin/cat > /etc/samba/smb.conf <<CONFIG
[global]
security = user
map to guest = Bad User
guest account = root
workgroup = cygwin
netbios name = controller
server string = %h
log level = 1
max log size = 1000
log file = /var/log/samba.log
[nixstore]
path = /nix/store
writable = yes
guest ok = yes
[xchg]
path = /xchg
writable = yes
guest ok = yes
CONFIG
${samba}/sbin/nmbd -D
${samba}/sbin/smbd -D
echo -n "Waiting for Windows VM to become available..."
while ! ${netcat-gnu}/bin/netcat -z 192.168.0.1 22; do
echo -n .
${coreutils}/bin/sleep 1
done
${coreutils}/bin/touch /xchg/waiting_done
echo " success."
${openssh}/bin/ssh \
-o UserKnownHostsFile=/dev/null \
-o StrictHostKeyChecking=no \
-i /ssh.key \
-l Administrator \
2016-06-12 19:11:37 +02:00
192.168.0.1 -- ${lib.escapeShellArg command}
'') + optionalString (suspendTo != null) ''
${coreutils}/bin/touch /xchg/suspend_now
${loopForever}
'');
kernelAppend = concatStringsSep " " [
"panic=1"
"loglevel=4"
"console=tty1"
"console=ttyS0"
"command=${initScript}"
];
controllerQemuArgs = concatStringsSep " " (maybeKvm64 ++ [
"-pidfile $CTRLVM_PIDFILE"
"-nographic"
"-no-reboot"
"-virtfs local,path=/nix/store,security_model=none,mount_tag=store"
"-virtfs local,path=$XCHG_DIR,security_model=none,mount_tag=xchg"
"-kernel ${modulesClosure.kernel}/bzImage"
"-initrd ${initrd}/initrd"
"-append \"${kernelAppend}\""
"-net nic,vlan=0,macaddr=52:54:00:12:01:02,model=virtio"
"-net vde,vlan=0,sock=$QEMU_VDE_SOCKET"
]);
maybeKvm64 = optional (stdenv.hostPlatform.system == "x86_64-linux") "-cpu kvm64";
cygwinQemuArgs = concatStringsSep " " (maybeKvm64 ++ [
"-monitor unix:$MONITOR_SOCKET,server,nowait"
"-pidfile $WINVM_PIDFILE"
"-nographic"
"-net nic,vlan=0,macaddr=52:54:00:12:01:01"
"-net vde,vlan=0,sock=$QEMU_VDE_SOCKET"
"-rtc base=2010-01-01,clock=vm"
] ++ qemuArgs ++ optionals (resumeFrom != null) [
"-incoming 'exec: ${gzip}/bin/gzip -c -d \"${resumeFrom}\"'"
]);
modulesClosure = overrideDerivation vmTools.modulesClosure (o: {
rootModules = o.rootModules ++ singleton "virtio_net";
});
preVM = ''
(set; declare -p) > saved-env
XCHG_DIR="$(${coreutils}/bin/mktemp -d nix-vm.XXXXXXXXXX --tmpdir)"
${coreutils}/bin/mv saved-env "$XCHG_DIR/"
eval "$preVM"
QEMU_VDE_SOCKET="$(pwd)/vde.ctl"
MONITOR_SOCKET="$(pwd)/monitor"
WINVM_PIDFILE="$(pwd)/winvm.pid"
CTRLVM_PIDFILE="$(pwd)/ctrlvm.pid"
${vde2}/bin/vde_switch -s "$QEMU_VDE_SOCKET" --dirmode 0700 &
echo 'alive?' | ${socat}/bin/socat - \
UNIX-CONNECT:$QEMU_VDE_SOCKET/ctl,retry=20
'';
vmExec = ''
${vmTools.qemuProg} ${controllerQemuArgs} &
${vmTools.qemuProg} ${cygwinQemuArgs} &
echo -n "Waiting for VMs to start up..."
timeout=60
while ! test -e "$WINVM_PIDFILE" -a -e "$CTRLVM_PIDFILE"; do
timeout=$(($timeout - 1))
echo -n .
if test $timeout -le 0; then
echo " timed out."
exit 1
fi
${coreutils}/bin/sleep 1
done
echo " done."
'';
checkDropOut = ''
if ! test -e "$XCHG_DIR/waiting_done" &&
! kill -0 $(< "$WINVM_PIDFILE"); then
echo "Windows VM has dropped out early, bailing out!" >&2
exit 1
fi
'';
toMonitor = "${socat}/bin/socat - UNIX-CONNECT:$MONITOR_SOCKET";
postVM = if suspendTo != null then ''
while ! test -e "$XCHG_DIR/suspend_now"; do
${checkDropOut}
${coreutils}/bin/sleep 1
done
${toMonitor} <<CMD
stop
migrate_set_speed 4095m
migrate "exec:${gzip}/bin/gzip -c > '${suspendTo}'"
CMD
echo -n "Waiting for memory dump to finish..."
while ! echo info migrate | ${toMonitor} | \
${gnugrep}/bin/grep -qi '^migration *status: *complete'; do
${coreutils}/bin/sleep 1
echo -n .
done
echo " done."
echo quit | ${toMonitor}
wait $(< "$WINVM_PIDFILE")
eval "$postVM"
exit 0
'' else if installMode then ''
wait $(< "$WINVM_PIDFILE")
eval "$postVM"
exit 0
'' else ''
while kill -0 $(< "$CTRLVM_PIDFILE"); do
${checkDropOut}
done
if ! test -e "$XCHG_DIR/in-vm-exit"; then
echo "Virtual machine didn't produce an exit code."
exit 1
fi
eval "$postVM"
exit $(< "$XCHG_DIR/in-vm-exit")
'';
in writeScript "run-cygwin-vm.sh" ''
#!${stdenv.shell} -e
${preVM}
${vmExec}
${postVM}
''