[arch-projects] [initscripts][PATCH] rc.sysinit: kill off VERBOSE= parameter
Upstream already supports this via the loglevel= parameter on the kernel cmdline, so we should support using this instead of our own homegrown solution. Signed-off-by: Dave Reisner <d@falconindy.com> --- This conflicts with one of the other patches I sent 2 days ago, but I guess I'd rather see this merged than the whitespace cleanup (can always cleanup the cleanup). Say goodnight to our short lived VERBOSE friend. d rc.conf | 2 -- rc.sysinit | 7 ------- 2 files changed, 0 insertions(+), 9 deletions(-) diff --git a/rc.conf b/rc.conf index 26cd695..04a93ba 100644 --- a/rc.conf +++ b/rc.conf @@ -17,7 +17,6 @@ # CONSOLEFONT: found in /usr/share/kbd/consolefonts (only needed for non-US) # CONSOLEMAP: found in /usr/share/kbd/consoletrans # USECOLOR: use ANSI color sequences in startup messages -# VERBOSE: Verbose level (from 1 to 8). man 3 syslog for level info # LOCALE="en_US.UTF-8" DAEMON_LOCALE="no" @@ -27,7 +26,6 @@ KEYMAP="us" CONSOLEFONT= CONSOLEMAP= USECOLOR="yes" -VERBOSE="3" # ----------------------------------------------------------------------- # HARDWARE diff --git a/rc.sysinit b/rc.sysinit index c87e3dc..ee77959 100755 --- a/rc.sysinit +++ b/rc.sysinit @@ -43,13 +43,6 @@ status "Mounting Root Read-Only" /bin/mount -n -o remount,ro / /sbin/minilogd /sbin/bootlogd -p /run/bootlogd.pid -# Set console verbosity -for cmdarg in $(< /proc/cmdline); do - [[ "$cmdarg" == verbose ]] && VERBOSE=8 && break - [[ "$cmdarg" =~ verbose=[1-8] ]] && VERBOSE=${BASH_REMATCH[0]-8} && break -done -/bin/dmesg -n ${VERBOSE:-3} - HWCLOCK_PARAMS="--hctosys" case $HARDWARECLOCK in UTC) HWCLOCK_PARAMS+=" --utc";; -- 1.7.5.4
On Sat, Jun 4, 2011 at 8:22 PM, Dave Reisner <d@falconindy.com> wrote:
Upstream already supports this via the loglevel= parameter on the kernel cmdline, so we should support using this instead of our own homegrown solution.
Very nice. Is this in your public repo so I can pull directly? I will (probably) not accept any patches introducing new variables to rc.conf going forward. Removing/changing them is too much of a hassle. Other things, I'd like to see go: HARDWARECLOCK: the third line in /var/lib/hwclock/adjtime contains this information, and should be moved to /etc (following upstream), but discussion needed. The value is set by "hwclock --systohc --{utc,localtime}". TIMEZONE: /etc/localtime contains this information and is not kept properly in sync with rc.conf. Would need a nice cmdline tool to update /etc/localtime. USECOLOR: not important, but we should be able to detect if color is supported... UDEV_TIMEOUT: the standard value is tweaked from time to time by upstream, so they probably know better than us what is a good default. We'll get rid of it once it is supported in udev.conf (should be a simple upstream patch, which will be accepted as it is on their TODO). NETWORK_PERSIST: once we find a way to auto-detect if this is needed. This should be simple, but we would need someone with the relevant setup to test. USE{LVM,DMRAID,BTRFS}: once they are handled by udev (don't know how difficult that will be, a task for upstream) Cheers, Tom
On Sat, Jun 04, 2011 at 08:45:25PM +0200, Tom Gundersen wrote:
On Sat, Jun 4, 2011 at 8:22 PM, Dave Reisner <d@falconindy.com> wrote:
Upstream already supports this via the loglevel= parameter on the kernel cmdline, so we should support using this instead of our own homegrown solution.
Very nice. Is this in your public repo so I can pull directly?
i just created a for-tom branch to get around the conflict i mentioned. d
I will (probably) not accept any patches introducing new variables to rc.conf going forward. Removing/changing them is too much of a hassle.
Other things, I'd like to see go:
HARDWARECLOCK: the third line in /var/lib/hwclock/adjtime contains this information, and should be moved to /etc (following upstream), but discussion needed. The value is set by "hwclock --systohc --{utc,localtime}".
There's gotta be a reason that everyone puts the adjust file in /var/lib instead of /etc. I'm really not sure what that reason is, though.
TIMEZONE: /etc/localtime contains this information and is not kept properly in sync with rc.conf. Would need a nice cmdline tool to update /etc/localtime. USECOLOR: not important, but we should be able to detect if color is supported...
Even if you can support color, you might not necessarily want it.
UDEV_TIMEOUT: the standard value is tweaked from time to time by upstream, so they probably know better than us what is a good default. We'll get rid of it once it is supported in udev.conf (should be a simple upstream patch, which will be accepted as it is on their TODO). NETWORK_PERSIST: once we find a way to auto-detect if this is needed. This should be simple, but we would need someone with the relevant setup to test.
Very hard, if not impossible, to do with a high degree of accuracy. I think this one may need to stick around. Maybe there's a magical silver bullet I don't know about. Other distros maintain a list of network filesystem and tie that into their shutdown procedure to avoid killing the network.
USE{LVM,DMRAID,BTRFS}: once they are handled by udev (don't know how difficult that will be, a task for upstream)
Cheers,
Tom
On Sat, Jun 4, 2011 at 8:54 PM, Dave Reisner <d@falconindy.com> wrote:
i just created a for-tom branch to get around the conflict i mentioned.
Thanks. I'll do the merge.
HARDWARECLOCK: the third line in /var/lib/hwclock/adjtime contains this information, and should be moved to /etc (following upstream), but discussion needed. The value is set by "hwclock --systohc --{utc,localtime}".
There's gotta be a reason that everyone puts the adjust file in /var/lib instead of /etc. I'm really not sure what that reason is, though.
adjtime contains the driftinformation, that is updated automatically and regularly, and hence really belongs in /var, it also contains the UTC/LOCALTIME info which is needed at early boot and only rarely changed by the admin, so belongs in /etc. The correct fix would be to submit a patch to upstream to split the file in two, but given the choice between the two evils, I think having it in /etc is the correct choice for now (as most people don't even use the drift information, and we can't use /etc ro anyway).
USECOLOR: not important, but we should be able to detect if color is supported...
Even if you can support color, you might not necessarily want it.
Why should anyone care? Then again, I guess I don't really care either way myself, so it can stay.
NETWORK_PERSIST: once we find a way to auto-detect if this is needed. This should be simple, but we would need someone with the relevant setup to test.
Very hard, if not impossible, to do with a high degree of accuracy. I think this one may need to stick around. Maybe there's a magical silver bullet I don't know about. Other distros maintain a list of network filesystem and tie that into their shutdown procedure to avoid killing the network.
I have forgotten, what does systemd do at the moment? They have a list? I guess that at some point the kernel must know that a given filesystem is a networked filesystem, so one would hope that this information was somehow exported in a generic way. -t
On Sat, Jun 4, 2011 at 8:22 PM, Dave Reisner <d@falconindy.com> wrote:
Upstream already supports this via the loglevel= parameter on the kernel cmdline, so we should support using this instead of our own homegrown solution.
Hi Dave, I wondering about which upstream you speak? In don't find any reference of loglevel in udev and in mkinitcpio. Regards, -- Sébastien Luttringer www.seblu.net
On Sat, Jun 04, 2011 at 11:40:04PM +0200, Seblu wrote:
On Sat, Jun 4, 2011 at 8:22 PM, Dave Reisner <d@falconindy.com> wrote:
Upstream already supports this via the loglevel= parameter on the kernel cmdline, so we should support using this instead of our own homegrown solution.
Hi Dave,
I wondering about which upstream you speak? In don't find any reference of loglevel in udev and in mkinitcpio.
Regards,
The kernel itself reads loglevel and sets its own parameters accordingly: http://www.kernel.org/doc/Documentation/kernel-parameters.txt d
participants (3)
-
Dave Reisner
-
Seblu
-
Tom Gundersen