There ought to be hdparm support in the initscripts, setting the right things on your hard drive can make a huge difference. Search the gentoo wiki, theres a great article there.
]]>Ive followed this thread to increase my boot speed and have done everything that you have recommended, but I have a really strange boot..
can you please look at my bootchart image.. the kjournald takes way to long to initiate.. I have checked plockerys image but he doesnt have kjournald, but khelper.. I dont have khelper.. what is khelper?
http://www.badongo.com/pic.php?file=boo … tchart.png
cheers
jinn
phrakture wrote:Well rock on, 10 seconds is great. As for shutting it down permanently, the kernel boot line does that for now, though kernel 2.6.13 has devfs removed altogether, so that should do it. 8)
Which means with kernel 2.6.13 in use i could kick the devfsd pkg in the trashbin?
Yes. If you use udev, you don't need devfsd.
The devfsd packages is still in the repo for people still running kernel 2.4 or using an older 2.6 kernel with devfs.
Well rock on, 10 seconds is great. As for shutting it down permanently, the kernel boot line does that for now, though kernel 2.6.13 has devfs removed altogether, so that should do it. 8)
Which means with kernel 2.6.13 in use i could kick the devfsd pkg in the trashbin?
]]>and no, as far as I can tell, usb agent shouldn't be doing much on boot - I'm pretty sure mine doesn't but I'll have to boot chart my machine again when I get a chance....
Been away for a couple of days, I have taken bootcharts with no usb modules at all (see first link below) and only uhci_hcd (which loads usbcore) loaded (see second link).
http://img157.imageshack.us/img157/8776 … rt35id.png
http://img157.imageshack.us/img157/9306 … rt43hb.png
It looks like uhci_hcd causes all the usb.agent activity, since there is none without it.
Anyway boottime is now quite acceptable, so thanks for sticking with me the whole time (and to others who have chimed in from time to time).
It has beeen a great help and I have learn't a lot in the process!!
good luck
~jnengland77
PS: Sorry about my crappy english on forums...
Sorry I don't know what you mean.
By usb keys do you mean a usbstick or pen drive which is what I have?
Yeah, that's just my jive-talk... I've always called them "keys" because they were all made like keychains..
i gather your opinion is that these modules should not be causing all this usbagent activity in the chart.
What is usbagent anyway?
usb agent is part of hotplug. specifically the usb portion of it. it does the same thing as the rest of hotplu, but the details are different for USB stuff.
and no, as far as I can tell, usb agent shouldn't be doing much on boot - I'm pretty sure mine doesn't but I'll have to boot chart my machine again when I get a chance....
]]>Not really sure. You mentioned USB keys - can you try booting without that? If it's the usb drive causing the activity, there should be a way to load all the required modules before hand, so usb.agent shouldn't have o scan so much....
Sorry I don't know what you mean.
By usb keys do you mean a usbstick or pen drive which is what I have?
However, I have no usb storage connected by default at all.
My p2 simply has 2 usb 1.1 ports in the rear.
I load autofs (automount) so that I can plug in a usbstick into /media/usbstick if I need to.
My auto.media file usb line reads like this:
usbstick -fstype=auto,sync,nodev,nosuid,umask=000 :/dev/sda1
But If I am reading the chart right all this usbagent stuff happens before autofs is loaded.
Other than that uhci_hcd loads usbcore which is also loaded from rc.sysinit.
i gather your opinion is that these modules should not be causing all this usbagent activity in the chart.
What is usbagent anyway?
Peter
]]>Phrakture wrote:
Hrrrmm... I see udevd *and* devfsd starting up there. Make sure you append "devfs=nomount" to your kernel boot line, so that devfsd doesn't start up (devfs = dead)
OK I stopped devfs as you suggest via append.
What a difference! It took another 10 seconds or more off boot time.
(I noticed during boot that now the message says "starting udev daemon" whereas it always used to say "starting devfs daemon".)I am wondering, therefore, why Arch loads both udev and devfs but runs devfs by default.
Is there a way to stop devfs permanently without the append line eg in rc.sysinit?
Well rock on, 10 seconds is great. As for shutting it down permanently, the kernel boot line does that for now, though kernel 2.6.13 has devfs removed altogether, so that should do it. 8)
I loaded usb-storage but it only seemed to make the problem you refer to even worse.
See the two bootcharts below.
Both charts have been taken with the new append line in lilo. [Interestingly boot time is shown as almost identical to the original chart and yet it is 10 seconds less for me overall!!]
...
What is the problem from your point of view?
Not really sure. You mentioned USB keys - can you try booting without that? If it's the usb drive causing the activity, there should be a way to load all the required modules before hand, so usb.agent shouldn't have o scan so much....
Still, a 30s boot on a P2 is great.
On a side note, do you have your hostname in /etc/hosts along with localhost? This makes X boot a shade faster as it doesn't need to check to see if your "hostname" is something external.
]]>Hrrrmm... I see udevd *and* devfsd starting up there. Make sure you append "devfs=nomount" to your kernel boot line, so that devfsd doesn't start up (devfs = dead)
OK I stopped devfs as you suggest via append.
What a difference! It took another 10 seconds or more off boot time.
(I noticed during boot that now the message says "starting udev daemon" whereas it always used to say "starting devfs daemon".)
I am wondering, therefore, why Arch loads both udev and devfs but runs devfs by default.
Is there a way to stop devfs permanently without the append line eg in rc.sysinit?
Phrakture wrote:
Also, it seems your udev is kicking off alot more hotplug activity than mine does... I'll have to look at it, but that usb.agent line with all the blue is kinda bad... (try adding usb-storage to your MODULES array, so hotplug may not have to scan for it...)
I loaded usb-storage but it only seemed to make the problem you refer to even worse.
See the two bootcharts below.
Both charts have been taken with the new append line in lilo. [Interestingly boot time is shown as almost identical to the original chart and yet it is 10 seconds less for me overall!!]
The first link is without usb-storage loaded.
The second is with usb-storage loaded.
You will see that the usbagent activity to which you refer is much greater with the second.
http://img159.imageshack.us/img159/883/ … rt17uw.png
http://img159.imageshack.us/img159/7541 … rt26wg.png
I just assumed that the activity of usb agent shown in the chart is the loading of the appropriate drivers for uhci_hcd. Hence why it is greater when loading usb-storage as well.
What is the problem from your point of view?
Peter
]]>But this nuts-and-bolts method seems much lighter
]]>