Age | Commit message (Collapse) | Author | Files |
|
The system-wide mount will no longer mount items in fstab marked
as _netdev (requires a network device). These _netdev items are
handled by the netfs daemon now.
Additionally, add support for glusterfs :)
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
This changes the currect code for console initialization to make it dynamic.
For example when boot on UML the number of consoles is 15 and not 63.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
|
|
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
Signed-off-by: Allan McRae <allan@archlinux.org>
|
|
Just convert vc/N to ttyN in all required places.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
Signed-off-by: Allan McRae <allan@archlinux.org>
|
|
Backgrounding the udev uevent triggering right after the launch
of the udev daemon allows us to do more in parallel before we wait
for uevents to complete (udevadm settle). This should speed up boot
Ref: FS#12706
Idea-from: Vahid Hamidullah <vh22@njit.edu>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
fixes #14437
|
|
<hugo@archlinux.org>
|
|
The hwclock man page recommends calling hwclock with --adjust
at start time, and periodically during system run. So let's do
that here.
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
initscripts. This requires a new udev package (at least 139-1)
|
|
This is simply confusing that we use the kernel "quiet" parameter
to mute fsck output. Removed for now - perhaps we can add a better
solution later.
Ref: FS#12928
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
- For the random-seed file location to be more "FHS compliant" is better to put in /var/lib/misc/random-seed (Variable state
information, that persists after reboot) than the /var/run/random-seed (Run-time variable data, that not necessarily persists
after reboot) for example with /var/run mounted as TMPFS.
- The random-seed file permision should be "0600" not "0644"
- The size of random-seed acording to random(4) manpage should be the size specified in "/proc/sys/kernel/random/poolsize" in
other case 512 bytes.
|
|
Don't fail and force a reboot if the user cancels an fsck
on boot.
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
This allows X terminals to write to /var/run/utmp
so that they are reported properly to commands like
who and w
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
|
|
This reverts commit 119b8df1fb1258231750309f01e747e72f382493.
These nodes are created like 10 lines above, the [ -f ... ] check would always return false anyway (-f == regular file)
|
|
Udev requires /dev/console, /dev/null, and /dev/zero to exist
before running. Create these on init if they're missing
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
Previous behavior only set the mode for vc0.
See FS#11093
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
This should aleviate some gripes
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
This is an obvious oversight on my part - this change allows
the overridden functions to work in daemon scripts and
anything else that uses /etc/rc.d/functions by itself.
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
|
|
|
|
udev/initscript changes. This patch seems to fix such issues.
|
|
|
|
Contrary to what one may think, screen blanking not only affects the
console, but also the X server. This line may lead to unwanted blank screens
although one has DPMS disabled in X. Furthermore, it adds a hardcoded,
non-configurable default setting to Arch.
This should either be made configurable (but IMO, we don't need more stuff
in rc.conf) or added to rc.local by the user if needed. As this is a
one-liner, the second solution sounds good to me.
|
|
|
|
We haven't been using initrd for a while, so these hacks are unnecessary.
The "encrypted root device hack" was obsoleted by proper udev rules long
ago. The code to umount and free the initrd is not critical, so if somebody
insists on using initrd, it can be done in rc.local.
|
|
hwlock's manpage. This is contrary to the behaviour in older initscripts.
|
|
up some comments
|
|
libusb has migrated to /dev/bus/usb, which is managed by udev and thus more
flexible. Some legacy applications (like VirtualBox) still use it. However,
since users need special permissions to access usbfs, they have to be
specified via mount options in fstab. So there are two cases:
1) a user does not need usbfs
2) a user needs it, but must add it to fstab to force the right permissions
Our old special handling in rc.sysinit is thus obsolete (and it was ugly
anyway).
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
Looks like a 'bin/' slipped in there somewhere. Whoops
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
This was missed, and caused some gross output if a user
mistakenly had a space in their timezone (we should fail
gracefully here, not let bash error).
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
These nodes are not covered by default udev, so lets make
sure they're added properly.
Ref: FS#9726
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
* There is a patch for the kernel now, so no need to workaround the bug
* Add a missing check for an empty KEYMAP
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
It seems that there's a bug in the kernel 2.6.24,
which makes the built-in defkeymap to produce incorrect composite chars.
Thus keymaps that don't have 'include "compose.latin1"' have issues (FS#9593).
While it is better to fix this bug in kernel and all affected keymaps
loading defkeymaps before other keymaps seems to not hurt anyway.
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
Fixes FS#9451
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
* Since 2.6.24 the virtual terminal is in UTF-8 mode
and the keyboard is in Unicode mode by default,
and configurable only at run-time.
* To make non-UTF-8 consoles work correctly:
- the non-UTF-8 mode is set for the virtual terminal,
- the XLATE mode is set for the keyboard.
* UTF-8 mode is still set explicitly (not relying on default),
so console works correctly even in a case when user set vt.default_utf8=0
in kernel commandline, but LOCALE in rc.conf is *.UTF-8.
* % is a special symbol for printf, so it should be written twice.
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
|
|
So we can see timing and all that jazz
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
Use the always available (in sh) printf method instead.
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
Signed-off-by: Roman Kyrylych <roman@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|
|
We need, at the very least, these three device nodes
(null, console, and zero) created before we get udev
up and running
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
|