Note that there are still some patches queued upstream for fcoe-utils to
enable it to work with the new lldpad and to add support to fipvlan to
bring up FCoE connections without requiring fcoemon to run.
The invocations of the various tools as in this patch should be final though,
see the discussion in:
http://bugzilla.redhat.com/show_bug.cgi?id=563794
This is the second revision of this patch, which no longer adds /etc/fcoe
to the initrd as that is not needed.
lvchange and vgchange '--monitor n' will not prevent lvm from
attempting to dlopen the libdevmapper-event library.
dracut git commit 47ab3b6c5e introduced the use of '--monitor n' but
'--ignoremonitoring' is needed now that the libdevmapper-event library
isn't copied into the initramfs (ever since 0fae59d6eb)
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
On debian systems xen-detect does not resite somewhere in $PATH,
but under /usr/lib/xen-default/bin. This patch ensures that this
is searched as well when locating and installing xen-detect.
Common wisdom to enter single user on Linux is to edit command
line and add "single". This was not possible because switch_root
was always called with empty init arguments. Collect them from
command line and pass to real init when switching root.
Signed-off-by: Andrey Borzenkov <arvidjaar@mail.ru>
Adds the readonly_overlay karg for cases where the dm snapshot should be set to readonly. Use case would be a livecd that is configured to have a readonly root where filling up the dm snapshot would cause a problem.
If multipath isn't installed, don't use it. If we're in hostonly mode,
only install the multipath module if it's used for / . Otherwise, if
the user was dumb enough to install it, they get it during bootup.
- the use of sed is placeholder "hack" until lvm2 provides a proper
tool for changing lvm.conf
- lvm_scan.sh should run lvm commands with --ignorelockingfailure to
re-use lvm's existing initrd-specific logic; future lvm2 changes
will split this flag out into various new command-line switches
- no monitoring should be started from within initramfs
- NOTE: the same should apply to 90dmraid/install
- the correct types would be: '[ "blkext", 1 , "cciss0", 16 ]'
but lvm2 (>= 2.02.52) already properly supports both 'blkext' and
'cciss' (including cciss0 -> cciss7)
This patch adds support for user mode suspend to disk. It is installed
in parallel to kernel mode suspend module; either will fail if
system was not suspended using correct tool so next one can be tried.
Signed-off-by: Andrey Borzenkov <arvidjaar@mail.ru>
/lib/udev/cosole_init will load either non-unicode or unicode versions
of keyboard layout for the same value of KEYMAP depending on language
setting. The simplest solution is to install both versions in initrd;
it does not take much space.
While on it, copy some additional maps to ensure emergency shell
has the same keyboard layout as full system.
Signed-off-by: Andrey Borzenkov <arvidjaar@mail.ru>
Signed-off-by: Luca Berra <bluca@vodka.it>
for some unknown reason the emergency shell
starts with stderr closed, at first I even tought it was not working at
all, then I came up with this hack, which seems to work properly. I also
change the prompt to remind which step are we breaking to.
Install all modules that are any of:
- scsi device handler
- dm log handler
- dm path selector
- dm target
It would be nice if we could tell which log handlers and targets are
multipath related, but we really can't.
The primary source for dasd initialization script and udev rules is
now in s390utils package. The s390utils-base subpackage, that carries
the required files, is always installed on s390/s390x, because it's
part of the Core group in comps.
Signed-off-by: Dan Horák <dan@danny.cz>
New "filesystems" command line/config file option is added with the ability to
control the list of kernel filesystem modules that are included in the generic
initramfs.
Signed-off-by: Dan Horák <dan@danny.cz>
The new rd_DASD parameter allows dracut to handle multiple rd_DASD
options. One parameter per DASD. The syntax is:
rd_DASD=<device path>[,readonly=X][,erplog=X][,use_diag=X][,failfast=X]
The device path is a CCW device path, such as 0.0.0200. The optional
parameters are sysfs attributes for the DASD. The X value can be 0 or
1. Dracut will write out each of the rd_DASD settings to
/etc/dasd.conf and on bootup, the dasdconf.sh script will parse this
file and bring each DASD online with the specified attribute settings.
Some distros, including debian unstable with 2.6.30, still ship
old style ide drivers. These should be installed as well.
Sadly there are no symbols to use for nm, so a simple =ide needs
to suffice.
The manpage for dhclient-script says:
Before actually configuring the address, dhclient-script should
somehow ARP for it and exit with a nonzero status if it receives a
reply.
By using arping in dracut this is very easy, since arping has a
specific option to do just that.
This patch adds STP timeout error handling with arping. It's rather
simple since it only cares about the primary interface and blindly
assumes that if no gateway is available the root server is on the
same subnet.
The usual approach to setting mtus is to set the interface down,
set the mtu then set the interface back up again. Modern hardware
and/or drivers may support setting this on the fly, so we try
this and fall back to the old behaviour it it doesn't work.
In addition this patch only allows mtus greater than 576, this is
taken from debian/ubuntu dhclient-script.
When assembling containers + embedded arrays from mdadm.conf,
mdadm needs the /dev/md# node for the container to assemble the
arrays within the container. Stopping the udev exec queue, results in
this node not getting created and mdadm failing to online the
arrays within the container.
Not having stop / start udev exec-queue around "mdadm -As --run" should
be safe as the exact same command is run from rc.sysinit without
any queue locking.
This is a more sane solution, than ignoring subsequent "change" events.
The only danger is that we could loop, if a lvm scan triggers a broken
md partition, which triggers a broken PV and so on.
Better fix the scanning tools, not to emit change events for devices,
if no action was taken.
ifname=<interface>:<MAC>
Assign network device name <interface> (ie eth0) to the NIC
with MAC <MAC>.
Note that if you use this option you *must* specify an ifname=
argument for all interfaces used in ip= or fcoe= arguments
ifname=<interface>:<MAC>
Assign network device name <interface> (ie eth0) to the NIC with MAC <MAC>.
Note that if you use this option you *must* specify an ifname= argument
for all interfaces used in ip= or fcoe= arguments
Copy /etc/mdadm.conf to initramfs (even for non-hostonly) if
mdadmconf="yes" is set in dracut.conf or --mdadmconf is specified on the
dracut command line.
This was done, because there seems _no_ sane way to autoassemble md raid
arrays.
also moved rd_NO_MD to an udev ENV
I've looked at the LVM rules used in dracut just recently
and it needs fixing - we should react to change events only
for DM devices, so we have to skip vol_id/blkid call on ADD:
KERNEL=="dm-[0-9]*", ACTION=="add", GOTO="lvm_end"
Also, MD devices have their own rules, where vol_id/blkid
is called and where the symlinks are created (when looking
into raw initrd, this is in 64-md-raid.rules).
Also, if those rules are meant to be for DM devices only,
maybe we should skip symlink creation for the other devices
there, to keep the rules clean and straightforward. I think
we shouldn't create/recreate symlinks for non-dm devices in
LVM/DM rules (..should be in appropriate rules for that type
of device):
KERNEL!="dm-[0-9]*", GOTO="lvm_end"
Having it unconditionally pass pulls in all the networking cruft even
for systems that do not need it, and that sorta defeats the purpose of
hostonly mode.
Supported cmdline formats:
fcoe=<networkdevice>:<dcb|nodcb>
fcoe=<macaddress>:<dcb|nodcb>
Note currently only nodcb is supported, the dcb option is reserved for
future use.
Note letters in the macaddress must be lowercase!
Examples:
fcoe=eth0:nodcb
fcoe=4A:3F:4C:04:F8:D7:nodcb
This introduces filter_kernel_modules, which should be used to install
all kernel modules that match whatever criteria you want.
If running in --hostonly, filter_kernel_modules will only consider
modules that are loaded in the kernel, otherwise it will consider
all the modules installed on the system for the appropriate kernel.
This drastically reduces initramfs generation time when using --hostonly
by eliminating lots of unneeded filesystem activity.
Instead of grovelling through all the modules available for the
kernel looking for block devices, only look at the modules that are
actually loaded. This speeds things up by a rather large amount
when generating the initramfs with --hostonly.
While we are at it, only load the filesystem module that will actually
be used for the root filesystem when running in --hostonly instead
of all the filesystem modules that happen to be loaded at the time.
Since different distros may or may not use vol_id in udev, and blkid
is generally replacing vol_id, abstract them out into a function which
tries to use vol_id first and blkid second, on the assumption that
blkid can take over for vol_id if vol_id is no longer there.
This module provides syslog functionality in the initrd.
This is especially interesting when complex configuration being
used to provide access to the device the rootfs resides on.
When this module is installed into the ramfs it is triggered by
the udev event from the nic being setup (online).
Then if syslog is configured it is started and will forward all
kernel messages to the given syslog server.
The syslog implementation is detected automatically by finding the
apropriate binary with the following order:
rsyslogd
syslogd
syslog-ng
Then if detected the syslog.conf is generated and syslog is started.
Bootparameters:
syslogserver=ip Where to syslog to
sysloglevel=level What level has to be logged
syslogtype=rsyslog|syslog|syslogng
Don't auto detect syslog but set it
https://bugzilla.redhat.com/show_bug.cgi?id=515589
It ends up installing the label.so control plugin which isn't supposed
to get installed into the initrd. this makes cairo and libX11 and all sorts of
things move into the initrd that aren't supposed to.
I am not happy about this. It shouldn't be the job of dracut to do this. The initscripts should
deal with the plain /dev/.initramfs/ifcfg/ directory accordingly. Doing this for now because
notting insists upon it. We need to clean this up after we network option passing working.
If you're using a persistent overlay, you might want to reset it
at boot time if it has become corrupted somehow. Support using
reset_overlay as a command line optino to do so
The persistent overlay can be specified with an overlay= argument
on the command line. We'll probably try to move this into the
root= syntax soon, but this is the old way that works
livecd-creator previously added 'liveimg' and used root=CDLABEL=;
it's easy enough to support that old syntax for now at least
and it will make it easier to get people testing
Fedora/Red Hat live images are implemented as an ext3fs inside of
a squashfs. Writability is achieved with a device-mapper snapshot
on top of that.
This gives the basic support without a lot of things like persistent
overlays, iso md5sum checking, etc and also with a new basic syntax
that has to be specified as root=live:LABEL=...
As discussed before, it would be nice to be able to specify
the iscsi chap credentials inside the netroot=iscsi:.....
syntax, this patch implements this in a backwards compatible way, like
this:
iscsi:username:pass@127.0.0.1::3260::iqn.2009-01.com.example:testdisk
iscsi:username:pass:reverse:pass@127.0.0.1::3260::iqn.2009-01.com.example:test
The only downside is that the backwards compatibility is broken when there
is an @ in the iscsi target name (very unlikely), that can still be used,
but only like this:
iscsi:@192.168.1.100::3260::iqn.2009-01.com.example:testdi@sk
/tmp/nfs.rpc_pipefs_path can contain the path where
/var/lib/nfs/rpc_pipefs will be moved before switch_root.
This is useful if /var is a separate partition in the real root.
--kernel-only
only install kernel drivers and firmware files
--no-kernel
do not install kernel drivers and firmware files
All kernel module related install commands moved from "install"
to "installkernel".
For "--kernel-only" all installkernel scripts of the specified
modules are used, regardless of any checks, so that all modules
which might be needed by any dracut generic image are in.
The basic idea is to create two images. One image with the kernel
modules and one without. So if the kernel changes, you only have
to replace one image.
Grub and the kernel can handle multiple images, so grub entry can
look like this:
title Fedora (2.6.29.5-191.fc11.i586)
root (hd0,0)
kernel /vmlinuz-2.6.29.5-191.fc11.i586 ro rhgb quiet
initrd /initrd-20090722.img
initrd /initrd-kernel-2.6.29.5-191.fc11.i586.img
initrd /initrd-config.img
initrd-20090722.img
the image provided by the initrd rpm
one old backup version is kept like with the kernel
initrd-kernel-2.6.29.5-191.fc11.i586.img
the image provided by the kernel rpm
initrd-config.img
optional image with local configuration files
* SYNTAX
bridge=<bridgename>:<ethname>
If bridge without parameters, assume bridge=br0:eth0
* When <ethname> would be configured by network scripts, instead create a bridge named <bridgename> then add <ethname> to that bridge.
* Then $netif becomes <bridgename> instead of <ethname> and all existing scripts process netroot mount via this new $netif instead of <ethname>.
* Include a few test cases in NFS and NBD
KEYBOARDTYPE=sun|pc
will be written to /etc/sysconfig/keyboard in the initramfs
KEYTABLE=<keytable file>
will be written to /etc/sysconfig/keyboard in the initramfs
SYSFONT= Console font
will be written to /etc/sysconfig/i18n in the initramfs
SYSFONTACM= Console map.
will be written to /etc/sysconfig/i18n in the initramfs
UNIMAP= Unicode font map.
will be written to /etc/sysconfig/i18n in the initramfs
LANG=<locale>
will be written to /etc/sysconfig/i18n in the initramfs
LANG= set locale for all categories, can be any two letter ISO
language code
LVM
rd_NO_LVM
disable LVM detection
rd_LVM_VG=<volume group name>
only activate the volume groups with the given name
crypto LUKS
rd_NO_LUKS
disable crypto LUKS detection
rd_LUKS_UUID=<luks uuid>
only activate the LUKS partitions with the given UUID
MD
rd_NO_MD
disable MD RAID detection
rd_MD_UUID=<md uuid>
only activate the raid sets with the given UUID
DMRAID
rd_NO_DM
disable DM RAID detection
rd_DM_UUID=<dmraid uuid>
only activate the raid sets with the given UUID
- corrected the loglevel for warn()
- prepended with "dracut: " for kmesg to seperate from kernel messages
you can pipe to vinfo() for informational messages
Intel BIOS raid is being shifted from dmraid to mdraid because mdraid offers
more features. So if an imsm metadata capable mdadm is present use mdraid
instead of dmraid for isw_raid_member's
This patch also adds code to mdraid_start.sh so that the raidsets
inside the imsm containers get started once udev is done probing
(doing this earlier leads to potentially degraded use of the sets and
an unwanted resync).
TODO: /etc/passwd and /etc/group are not removed yet due to 90mdraid.
dledford said he'll go in and clean this up since he has the hardware
to actually test the mdmon stuff.
init now has the following points to inject scripts:
/cmdline/*.sh
scripts for command line parsing
/pre-udev/*.sh
scripts to run before udev is started
/pre-trigger/*.sh
scripts to run before the main udev trigger is pulled
/initqueue/*.sh
runs in parallel to the udev trigger
Udev events can add scripts here with /sbin/initqueue.
If /sbin/initqueue is called with the "--onetime" option, the script
will be removed after it was run.
If /initqueue/work is created and udev >= 143 then this loop can
process the jobs in parallel to the udevtrigger.
If the udev queue is empty and no root device is found or no root
filesystem was mounted, the user will be dropped to a shell after
a timeout.
Scripts can remove themselves from the initqueue by "rm $job".
/pre-mount/*.sh
scripts to run before the root filesystem is mounted
NFS is an exception, because it has no device node to be created
and mounts in the udev events
/mount/*.sh
scripts to mount the root filesystem
NFS is an exception, because it has no device node to be created
and mounts in the udev events
If the udev queue is empty and no root device is found or no root
filesystem was mounted, the user will be dropped to a shell after
a timeout.
/pre-pivot/*.sh
scripts to run before the real init is executed and the initramfs
disappears
All processes started before should be killed here.
The behaviour of the dmraid module demonstrates how to use the new
mechanism. If it detects a device which is part of a raidmember from a
udev rule, it installs a job to scan for dmraid devices, if the udev
queue is empty. After a scan, it removes itsself from the queue.
it solves the following case:
root=/dev/nfs nfsroot=server:/path
- the server could be reachable on any interface
- any interface can get an IP by dhcp
- only one IP is allowed to mount the root
Udev rules set a /dev/root symlink to the real root and add
a mount script to /mount/. This enables the proper use of pre-mount
scripts and prevents mount being killed by a udev timeout.
Jobs are no longer handled inside the udev events.
/sbin/initqueue is called with the commands to queue.
init will work on these jobs sequentially, so that we prevent jobs
from being killed by udev timeouts.
This serialization also prevents some problems introduced by
the udev event parallelization.
Talked with Debian nbdroot author and he agreed we shouldn't support their obsolete syntax in dracut.
Their root=/dev/nbd[0-9] is no longer needed. Also their syntax was very standardized requiring
a separate boot= parameter.
- Remove root=nfs nfsroot=...
- Remove root=nfs4 nfsroot=...
- Remove root=/dev/nfs4 nfsroot=...
- Legacy nfsroot= without root=/dev/nfs should be unsupported according to the nfsroot.txt.
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob;f=Documentation/filesystems/nfsroot.txt
45 root=/dev/nfs
46
47 This is necessary to enable the pseudo-NFS-device. Note that it's not a
48 real device but just a synonym to tell the kernel to use NFS instead of
49 a real device.
Pretty much everyone uses it, and things break of userspace
does not make it because of a case of mistaken identities.
/sbin/start_udev on FC11, I am looking at you.
The Bug causing dhclient to recall BIND has been identified:
dhclient-script runs with #!/bin/sh -e, causing setup_interface
to fail if no hostname is provided by dhcp as it is the last statement
in that function.
Harald Hoyer