This adds a new API call guestfs_find0, which is like guestfs_find
but mainly doesn't suffer from the protocol limit of the earlier
command. The earlier command is not deprecated because it is
still very useful.
guestfs_find0 uses a FileOut parameter and writes the results to
an external file. The filenames in the output are separated by
ASCII NUL characters (so a bit like "find -print0").
There is also the addition of a regression test for this command.
guestfs_find has to send the complete list of files in a single
protocol message (hence, limited to 2-4 MB). Unfortunately on
a typical Linux guest, guestfs_find ("/") will exceed this limit
resulting in an error.
Therefore we should add an unlimited version of this call in a
future commit. This commit just documents the current limit.
This adds a new tool call virt-tar which is a general purpose
archive and uploading tool. It doesn't add any functionality
which wasn't previously possible using guestfish, but makes it
simpler to access for some users.
Examples:
virt-tar -zx myguest /home home.tar.gz
virt-tar -zu myguest uploadstuff.tar.gz /tmp
This moves the tool programs into a single directory:
cat/* -> tools/virt-cat
df/* -> tools/virt-df
edit/* -> tools/virt-edit
rescue/* -> tools/virt-rescue
This in itself simplifies the build process because we only need
one Makefile and one copy of 'run-locally'.
'run-*-locally' has become just 'run-locally' and takes an extra
parameter which is the name of the tool, eg:
run-locally cat [virt-cat params...]
virt-inspector stays in its own directory, because this contains
more than just a single Perl script.
This commit adds a generic mechanism for deriving language bindings
for virt-inspector, and implements one concrete binding, for OCaml.
The bindings are generated from the RELAX NG schema (virt-inspector.rng)
which is supposed to be a correct and always up to date description
of the XML that the virt-inspector program can generate.
From the RNG we generate a set of types to describe the output of
virt-inspector for the language, plus an XML parser, plus some
glue code to actually run an external instance of virt-inspector
and parse the resulting XML.
At runtime, an external 'virt-inspector --xml <name>' command runs
and the XML is parsed into language-specific structures.
This has been tested on the four example files (inspector/example?.xml)
The only particular difficulty about the OCaml binding is the use of
Obj.magic, which is naughty but works because of the isomorphism
between the representation of tuples and records in OCaml. This
seems to cause no problems in my test program. Apart from this, the
OCaml binding is straightforward and could be adapted easily for any
other languages that want type-safe virt-inspector bindings.
It's important to keep virt-inspector.rng up to date with changes
to virt-inspector's XML output format.
A parallel build could fail due to the use in ocaml/examples
of ocaml/guestfs.cmi before it was built.
* Makefile.am (SUBDIRS): Add both ocaml and ocaml/examples,
to ensure they're built in this order, and not in parallel.
* ocaml/Makefile.am (SUBDIRS): Don't define.
Link demo scripts with just-build library, not the installed one.
* ocaml/examples/Makefile.am (OCAMLFINDFLAGS): New variable.
(lvs, viewer): Use it.
This fix is based on a suggestion from Rich Jones.
This addresses RHBZ#526917.
If $os->{kernels} wasn't defined, virt-inspector would exit with
an error, leaving partial XML output.
Change the code so it doesn't die in this case, instead just
leaves out the <kernels> section.
This is an example of how to write an app which uses libguestfs
and libvirt, and has a responsive user interface (using threads).
It is a Gtk-based "graphical virt-df".
Read the top of the 'viewer.ml' file first for instructions on
how to compile. This program is not compiled by default.
"make syntax-check" was failing. This fixes it.
* HACKING: Indent with spaces, not TABs.
* configure.ac: Likewise.
* rescue/virt-rescue: Likewise.
* src/generator.ml: Likewise.
Edit any file in a guest. This was possibly previously
using guestfish, but having a separate command makes it
simpler.
The usage is simply:
virt-edit mydomain /some/file
It runs $EDITOR or vi on the file, and if the user changes
it, uploads the result back to the VM.
This command runs a "rescue appliance" against a virtual machine
or disk image. This is useful for making ad-hoc interactive
changes to virtual machines.
$ virt-rescue --ro /dev/vg_trick/F11x64
Welcome to virt-rescue, the libguestfs rescue shell.
Note: The contents of / are the rescue appliance.
You have to mount the guest's partitions under /sysroot
before you will be able to examine them.
bash: cannot set terminal process group (-1): Inappropriate ioctl for device
bash: no job control in this shell
><rescue> mount /dev/vg_f11x64/lv_root /sysroot
EXT4-fs (dm-0): barriers enabled
kjournald2 starting: pid 269, dev dm-0:8, commit interval 5 seconds
EXT4-fs (dm-0): internal journal on dm-0:8
EXT4-fs (dm-0): delayed allocation enabled
EXT4-fs: file extents enabled
EXT4-fs: mballoc enabled
EXT4-fs (dm-0): mounted filesystem with ordered data mode
><rescue> ls /sysroot/
bin dev home lib64 media opt root selinux sys usr
boot etc lib lost+found mnt proc sbin srv tmp var
><rescue> exit
When the g->direct flag is set, the appliance stdin/stdout
are not connected to the library. Instead they inherit the
stdin/stdout of the caller.
This is used to implement virt-rescue.
vmchannel is no longer required, so we shouldn't test for it.
However we should test instead for user mode networking support.
Also fix up the documentation / error messages.
Always test for vmchannel and user mode networking support in
qemu. This gives us more troubleshooting information if people
report bugs.
This commit removes the requirement for vmchannel, although retaining
support for it.
In this configuration, known as "null vmchannel", the library
listens on a random loopback port. It passes the number of this
port to the appliance (guestfs_vmchannel=tcp:10.0.2.2:12345), and
the daemon then connects back. The library, instead of connecting,
listens and accepts the connection during guestfs_launch.
QEMU SLIRP (user mode networking) is still required to make this
work: SLIRP forwards the TCP connection transparently (instead of
explicitly as with guestfwd) to 127.0.0.1:<port>
There is a window in which any other local process on the machine
could see the port number in the qemu command line and try to
connect to it. This would be a devastating security hole, because
any local process could pretend to be the daemon and feed back
malicious replies to our remote procedure calls. To prevent this,
we check the UID of the other side of the TCP connection. If
the UID is different from the library's EUID, then we reject the
connection. To do this we have to parse /proc/net/tcp. (On Solaris
we could use getsockopt (SO_PEERCRED), but this doesn't work on
Linux TCP sockets).
Other vmchannel(s) are still supported. This is important, because
we can't in general be sure the qemu will always support SLIRP.
In particular, in recent versions of qemu it is possible to compile
out SLIRP.
This reimplements parts of commit da0a4f8d1f
in a different, but compatible way.
We pass guestfs_vmchannel=tcp:<ip>:<port> on the command line. This
is intended to be used as follows (now and in future versions):
tcp:10.0.2.4:6666 for guestfwd vmchannel
tcp:10.0.2.2:<port> for future "no vmchannel" implementation
/dev/vcon4 for future virtio-console vmchannel*
It also accepts the old-style guestfs=10.0.2.4:6666 parameter which
is sent by older libraries, and turns this transparently into the
correct format above.
If no guestfs_vmchannel is passed, then this defaults to the guestfwd
vmchannel which older libraries would expect.
* Maybe this last one should be dev:/dev/vcon4 or file:/dev/vcon4, but
we don't need to decide that now.
This just tidies up the main() function in the daemon. There is
no longer a huge fixed-sized buffer used for the kernel command
line. Variables are moved closer to where they are used. Some
local variables turned out to be unused - these are removed. Finally
the part that reads the kernel command line is moved into a separate
function.
There is only minimal functional change here (it now prints out
the kernel command line).
Literal '4' replaced with 'sizeof' expression, at Jim Meyering's
suggestion during patch review.
