diff options
| author | Andy Lutomirski <luto@kernel.org> | 2016-04-30 17:38:13 -0700 |
|---|---|---|
| committer | Andy Lutomirski <luto@kernel.org> | 2016-04-30 17:38:13 -0700 |
| commit | ea58d4232b5c0a77a30a244280c35e6a18525f0a (patch) | |
| tree | 2e21e57d9faa888c28ebc37df63b7a7f187bb074 | |
| parent | a449fc204bd23d8a601d55f50c3cba4ff5ed591f (diff) | |
| download | virtme-ea58d4232b5c0a77a30a244280c35e6a18525f0a.tar.gz | |
virtme-run: Don't panic if an unrecognized argument is passed to -a
The kernel interprets:
init=/bin/sh foobar -- baz quux
as a request to run /bin/sh foobar baz quux, unless foobar is a
known kernel parameter. As a result, passing '-a foobar' to
virtme-run will panic the guest. I think this is a bug or design
error in the kernel argument parser.
Fortunately, there's another counterbalancing quirk:
foobar init=/bin/sh -- baz quux
will cause foobar to be ignored because init=/bin/sh resets
accumulated init arguments.
Reorder our argument generation to take advantage of this quirk and
avoid panicking if a bad argument is passed.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
| -rw-r--r-- | virtme/commands/run.py | 8 |
1 files changed, 6 insertions, 2 deletions
diff --git a/virtme/commands/run.py b/virtme/commands/run.py index 51d9c11..8a2a3d7 100644 --- a/virtme/commands/run.py +++ b/virtme/commands/run.py @@ -197,8 +197,6 @@ def main(): export_virtfs(qemu, arch, qemuargs, guest_tools_path, 'virtme.guesttools') - kernelargs.append('init=/bin/sh') - initargs = ['-c', 'mount -t tmpfs run /run;mkdir -p /run/virtme/guesttools;/bin/mount -n -t 9p -o ro,version=9p2000.L,trans=virtio,access=any virtme.guesttools /run/virtme/guesttools;exec /run/virtme/guesttools/virtme-init'] # Map modules @@ -395,6 +393,12 @@ def main(): # Now that we're done setting up kernelargs, append user-specified args # and then initargs kernelargs.extend(args.kopt) + + # Unknown options get turned into arguments to init, which is annoying + # because we're explicitly passing '--' to set the arguments directly. + # Fortunately, 'init=' will clear any arguments parsed so far, so make + # sure that 'init=' appears directly before '--'. + kernelargs.append('init=/bin/sh') kernelargs.append('--') kernelargs.extend(initargs) |