Age | Commit message (Collapse) | Author | Files | Lines |
|
Newer qemu warns about a depcrecated readonly statement:
warning: short-form boolean option 'readonly' deprecated
Please use readonly=on instead
Fix this by using the correct readonly=on statement.
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
|
|
When running virtme natively on a aarch64 host, we should use the same
version of GIC as the host. We also don't need to emulate a CPU. Instead
we can run on the host one.
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
|
|
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
virtme-configkernel: Fix --allnoconfig option
|
|
run.py: introduce '--cpus' argument
|
|
Introduce a --cpu argument to virtme-run, so we can set qemu's -smp commandline
parameter.
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
|
|
Trying to switch to any non-root user from a virtme environment returns
the following error:
su: Authentication failure
Populate dummy entries in /etc/shadow to allow switching to any user
defined in the system.
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
|
|
Kernel commit 911a91c39cab ("kconfig: rename silentoldconfig to
syncconfig"), so also rename the updatetarget in virtme to reflect the
change.
Signed-off-by: Marcos Paulo de Souza <mpdesouza@suse.com>
|
|
'foo %s bar' % (baz) doesn't need parentheses.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
run.py: Add blk-disks argument
|
|
This new argument will create a new virtio-blk device, instead of the
scsi-blk that can be created by using --disk argument, accepting the
same argument of --disk.
By using --disk-blk "disk2=file1" will create
/dev/disk/by-id/virtio-disk.
Signed-off-by: Marcos Paulo de Souza <mpdesouza@suse.com>
|
|
This function will be used in the next patch.
Signed-off-by: Marcos Paulo de Souza <mpdesouza@suse.com>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Apparently having a virtfs export that spans multiple host
filesystems can cause problems in the guest. QEMU 4.2+ will warn
about this and supports a workaround: multidevs=remap. Probe for
new enough QEMU and pass multidevs=remap on 4.2+.
See #60.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
remove duplicated check from virtme-prep-kdir-mods
|
|
Make scripts PATH friendly for searching python3
|
|
Per #34, this fixes udev.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Dmitry Yakunin <zeil@yandex-team.ru>
|
|
This causes Xen's console output to be visible. It's a horrible
layering violation.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Existence of modules.order is checked twice.
Signed-off-by: Pavel Tatashin <pasha.tatashin@soleen.com>
|
|
This should fix #53.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This factors the logic out from find_busybox()
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Now the algorithm prefers the 'static' versions regardless of path,
and it searches PATH first (for native) as advertised.
See issue #52.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
|
|
mkinitramfs.py: Search for busybox{.,-}static first
|
|
In currentl Tumbleweed, using busybox dynamic linked does not work. For
now let's check if we have busybox static first.
Fix: #51
Signed-off-by: Marcos Paulo de Souza <mpdesouza@suse.com>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Now --save-initramfs uses the actual filesystem path for QEMU's -initrd.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
run: Add a --cwd command to change the working directory
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Marcos Paulo de Souza <mpdsouza@suse.com>
|
|
This will make it possible to iterate over all supported arches to
do clever things like detecting the architecture.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Linux removed CONFIG_CROSS_COMPILE:
commit f1089c92da791034af73478159626007cba7f092
Author: Masahiro Yamada <yamada.masahiro@socionext.com>
Date: Mon May 28 18:21:39 2018 +0900
kbuild: remove CONFIG_CROSS_COMPILE support
Update virtme-configkernel to give instructions that actually work.
Cross-compiling a kernel is annoying enough now that I think virtme
should add some kind of helper to do the dirty work.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This expects a kernel with the plan 9 stuff supported (not yet in
defconfig) and a new QEMU (as described in the README). I'm also not
100% sure it's working, as I'm getting
/bin/sh: exec: line 1: /run/virtme/guesttools/virtme-init: not found
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
|
|
It's unlikely that --installed-kernel actually works with --mods=none,
but I think that we should try it and possibly fail to boot rather
than just ignoring the --mods setting.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
configkernel: Work in kernel build directories and not just source dirs
|
|
virtme-prep-kdir-mods: Check if depmod is in the user's PATH
|
|
Remove not used imports
|
|
Remove not used virtme-loadmods script
|
|
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
This script is not being used since 865d8717c806 ("virtme-init: Turn off virtme-loadmods").
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
openSUSE doesn't have /usr/sbin in the user's default path, so depmod
cannot be found. This change makes virtme-run --mods=auto to work on
openSUSE.
Fixes: 3a65e4b514c0 ("Add virtme-prep-kdir-mods and support it")
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
mypy --check-untyped-defs -p virtme passes now
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
The output was getting too messy.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Make virtme-mkinitramfs great again
|
|
This is hopefully good enough for master now.
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
|
|
Fix module-finding in $KDIR/.virtme_mods
|
|
Miscellaneous small fixes
|
|
commands/configkernel.py: Check KBUILD_OUTPUT environment variable
|
|
More potential paths for busybox executable
|
|
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
If KBUILD_OUTPUT is defined, and if it's a directory, write the .config file in location defined.
Fixes: #24
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
With this commit, virtme-mkinitramfs is installed like virtme-run and
virtme-configkernel.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
This new argument waits for a string where the new initrd file will be
placed. If not informed, the file is printed in the stdout, which is the
current behavior.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
Following the same pattern of run.py.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
When creating the initramfs, busybox is needed, so search for it in the
root filesystem.
Suggested-by: Andy Lutomirsky <luto@kernel.org>
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
Is a shell script, very similar to virtme-run, that call
mkinitramfs.main function to create an initram image.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
Now that mkinitramfs.py lives inside commands, we can safely do relative
imports, as run.py already does.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
Later patches will fix mkinitramfs.py and make use of it.
Suggested-by: Andy Lutomirsky <luto@kernel.org>
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
|
|
Introduce new argument --mods none|use|auto to be used in conjunction
with --kdir to determine how virtme's kernel modules directory is
generated/updated inside a compiled kernel source directory.
The following arguments are currently supported:
- none: ignore kernel modules
- use: asks user to refresh virtme's kernel modules directory by
running virtme-prep-kdir-mods
- auto: automatically refreshes virtme's kernel modules directory
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
|
|
Make sure to refresh the kernel modules' directory when modules.order
changes.
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
|
|
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Void Linux's busybox-static package, for example, installs it as
'busybox.static'.
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Call virtme-prep-kdir-mods from virtme-run to automatically setup
virtme's modules directory when possible.
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
|
|
Get the list of kernel modules from modules.order, otherwise we may miss
some of them in the temporary virtme module directory.
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
|
|
This is still experimental. With this patch, you can run
virtme-prep-kdir-mods, and then virtme-run --kdir will pick up
modules from the kernel directory.
Known issues:
- There are no docs.
- This is unlikely to work well if forced module signing is on. An
option to virtme-prep-kdir-mods could be added to work around
that, but it would run slower.
- There is no intelligent handling for the case where the kernel
was modified and rebuilt but virtme-prep-kdir-mods was re-run.
Because it uses symlinks, it will only load modules that still
have their .ko files around, but virtme-run won't notice the
discrepency. Maybe file timestamps could be used to try to
detect this.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
|
|
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Using udhcpc default retries
|
|
virtme-init: correctly initialize resolv.conf under ubuntu
|
|
When --kdir is provided there's not an easy and automated way to use
kernel modules from the guest.
A possible solution could be to install all the modules inside the
kernel build directory (argument of --kdir) inside a temporary directory
.tmp_moddir/ as following:
$ make modules_install INSTALL_MOD_DIR=.tmp_moddir
In this way if .tmp_moddir is found the guest can use the kernel modules
from this directory, otherwise fallback to the old behavior and do not
provide an automated way to access the kernel modules.
TODO: document this .tmp_moddir/ directory somewhere.
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
|
|
In recent distro (i.e., Ubuntu) resolv.conf is a symlink to
/run/systemd/resolve/stub-resolv.conf that is not visible by the guest.
This results in a failure when trying to initialize this file.
Make sure to resolve the symlink first and initialize the actual file
with an empty file before trying to rewrite its content.
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
|
|
I've re-used the bridge of libvirtd. A vm gets it lease by the 3rd dhcp requests.
Log of udhcpc without this patch.
udhcpc: sending discover
udhcpc: no lease, failing
Log of udhcpc with this patch.
udhcpc: started, v1.27.2
udhcpc: sending discover
udhcpc: sending discover
udhcpc: sending select for 192.168.123.76
udhcpc: lease of 192.168.123.76 obtained, lease time 3600
This patch removes the '-t' arg to use the default retries which is 3.
Signed-off-by: Jian Wen <wenjianhn@gmail.com>
|
|
Signed-off-by: Haggai Eran <haggai.eran@gmail.com>
|
|
openSUSE and SLE don't have modprobe in the user's PATH, so check
/usr/sbin too. With this patch applied, an ordinary user can execute
virtme-run --installed-kernel
Without the need to be root in openSUSE/SLE.
Signed-off-by: Marcos Paulo de Souza <mpdesouza@suse.com>
|
|
when running scripts.
Signed-off-by: James Yonan <james@openvpn.net>
|
|
This isn't obviously terribly useful without careful configuration
of the host.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
-net is outdated. Use -device and -netdev as recommended by the QEMU manpage.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Respect --rw flag in initramfs
|
|
Mirror the --rw flag from virtme-run.
Signed-off-by: Lorenz Bauer <lmb@cloudflare.com>
|
|
Make the initramfs mount the root filesystems read-write if
the user requested it via the --rw flag.
Signed-off-by: Lorenz Bauer <lmb@cloudflare.com>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
PR #26
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
In my environment, /dev is available via the 9p root mount, and
therefore virtme-init does not mount a fresh devtmpfs. This means
that udevd later is unable to generate new entries for /dev/vport*.
Using --script-sh or --script-exec therefore fails.
Instead, use findmnt to check whether /dev is automounted by the
current kernel.
Signed-off-by: Lorenz Bauer <i@lmb.io>
|
|
Using a shared folder like --rwdir=/run/output=/path/on/host fails,
since the mountpoint under /run does not exist. Create the mountpoint
if it doesn't exist. Also abort the script if mounting fails.
Signed-off-by: Lorenz Bauer <i@lmb.io>
|
|
This new flag sets the "-m" flag of QEMU, letting the user to user the
amount of memory the guest will have.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
Message-Id: <20190328022819.1143715-1-marcos.souza.org@gmail.com>
|
|
Automatically mount cgroup v2 filesystem under /sys/fs/cgroup/unified as
it seems a pretty standard place and tools relying on cgroupv2 are
expecting to find this filesystem mounted there.
Signed-off-by: Andrea Righi <righi.andrea@gmail.com>
|
|
The guest tools are not always package resources.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
We're not doing ourselves any favors by being installed as a zip file --
the guest tools need to get extracted to do anything. It works, but still.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
I don't like autodiscovering files in the filesytem.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
The distutils.setup doesn't preserve package data file permission,
which makes it problematic when installing virtme-guest scripts.
Get rid of it, and instead require setuptools.
|
|
setup.py doesn't copy the necessary bash files from virtme/guest folder
to the installed folder.
This can be solved if we transform virtme/guest folder, which only have bash
scripts, into a python package by add a __init__.py file and instructing
setup.py that virtme.guest is a package and we will need its contents as
package data.
This idea was based on the setuptools docs advice:
https://setuptools.readthedocs.io/en/latest/setuptools.html#non-package-data-files
|
|
On Fedora, /usr/lib/modules contains vmlinuz files. Since /boot access
is always a little bit dicey, try /usr/lib/modules first.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Replace the mentions to prereqs.config file with
the virtme-configkernel tool.
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
|
|
|
|
virtme-udhcpc-script: make resolv.conf world-readable.
Arguably a nicer solution would be to name the file /run/virtme/resolv.conf, but that's a different issue entirely.
|
|
Sets hostname in guest init and passes it to qemu's '-name' flag for
easier identification.
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
'mktemp' creates files that are only accessible to the current user
(and doesn't have any options to change this), but in this case the
file we're creating is going to end up as resolv.conf, which should be
world-readable. So here we manually chmod it as such after creating it.
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Minor changes for Debian/Ubuntu and OpenSUSE
|
|
On OpenSUSE, /usr/bin/busybox is a dynamic binary, while
/usr/bin/busybox-static is static. We need the static version, so add
it.
Signed-off-by: Frank Zago <fzago@cray.com>
|
|
On Ubuntu, the real resolv.conf is in /run/resolvconf, not
/run/NetworkManager, so that create that path in the VM as well.
Signed-off-by: Frank Zago <fzago@cray.com>
|
|
|
|
--rw/--rwdir and --script-sh/--script-exec provide these features now.
(Also tack on a newline at EOF.)
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
|
|
This removes some steps in using a bunch of testing tools.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
bash, for example, wants /dev/fd to exist for implementing process
substitution (e.g. 'cat <(echo hello)'). Might as well provide the
usual three stdio links while we're there.
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
As of version 4.4 bash now warns about null bytes in command
substitutions. Sidestep it here by using sed to strip the null
terminator off the contents of 9pnet_virtio's mount_tag sysfs
attributes.
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Signed-off-by: James Yonan <james@openvpn.net>
|
|
This improves script mode.
|
|
virtme-run: Handle non-tty stdout
|
|
Catch and ignore ENOTTY exceptions from os.get_terminal_size().
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Doesn't make any real difference to the guest, but avoids trying to
put things in / when unpacking the cpio archive for debugging purposes.
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
This allows passing multiple semicolon-delimited shell commands.
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
The <> redirection defaults to stdin, not stdout, so redirect file
descriptors 1 and 2 to 0.
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
The _INIT script in mkinitramfs.py mounts a tmpfs on /run; mounting
another one over it hides /run/virtme, breaking --script-sh and
--script-exec (which execute a script from /run/virtme/data).
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This is a silly workaround for /etc/resolv.conf being a symlink
to /var/run/NetworkManager/resolv.conf.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Modern systemd renames it, so probe to find its new name.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This makes dbus a bit more reliable, which improves the (low) chance
that a full graphical session works correctly inside virtme.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
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>
|
|
When I added '--' support, I put user-specified kopts on the wrong
side of the '--'.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
|
|
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
We no longer require guest tools to exist in the rootfs to avoid
initramfs, nor do we even try to use it if it's there. Instead we
pull it in using virtfs and boot from it using the '--' kernel
command line feature to embed a scriptlet.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This is how systemd does it, so it's probably a good idea.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Everything works better if HOME points somewhere writable.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
A bunch of service depend on /run being fully set up, and
systemd-tmpfiles is a common way to set it up. Invoke it.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
It's slow, and it triggers a lockup under some circumstances.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
We can, and there's no need to treat x86 as special.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This will allow things like building sparc32 using a sparc64 toolchain.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
All current arches have qemuname matching the arch name, but this may
change for future arches.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This also fixes a silly bug in which the rootfs had a useless "ro"
string option.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This makes arm in particular much easier to use because it picks up the
dtb file automatically.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This is enough to get some console output.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
The versatilepb thing wasn't really working, and vexpress-a15 is modern
and functional.
This will remain awkward to use until I add generic device tree support.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
CentOS 6 is mostly compatible with virtme and has a fully functional QEMU,
but it's missing /run. Use the fallback tmpfs if that happens.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This will improve compatibility with CentOS 6, according to GitHub
issue 5.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
Signed-off-by: Andy Lutomirski <luto@kernel.org>
|
|
This fixes crashes when using --kimg on a kernel image with a version
that matches an installed kernel.
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Noticed by Tom H. Thanks.
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
This behaves considerably better.
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
For Python 3.2 support, virtme would need to adjust how it uses
os.uname() and it would need a replacement for shutil.which().
The latter is annoying enough that I don't really want to do it.
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
It doesn't really work, and there's a fancier version out for Fedora review.
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
I suspect that most users will want a terminal most of the time.
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|
|
|
|
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
|