Problem booting generic kernel and another problem with fluxbox - libpng warning
 

I made a post before that booting was slow but I realized I been booting the huge kernel that is used by default all this time. I been trying to use the generic kernel instead and read the mkinitrd man page and README.initrd in /boot and tried it but had no luck.

My partition scheme looks like:

As root, I did:

And this created a initrd.gz in /boot.

This was my original /etc/lilo.conf

but I changed it to:

I typed "Lilo" and updated the new configuration. I rebooted and it had absolutely no change than before.
__________________________

Also I have problem with fluxbox. Although I can set the background wallpaper with ~/.fluxbox/init,

I'm constantly getting:

I made a post before about my computer constantly freezing and the culprit was setting the background on fluxbox using the overlay. I'm able to replicate this freezing setting the wallpaper using overlay so I want to avoid using that method.

I'm curring using (doing fbsetbg -i)

to set my wallpaper.

What's the "boot=" set to in lilo.conf.

How do you know you are booting the huge kernel.

Also, try setting "compact" in lilo.conf, as without that all kernels appear to load slowly.

Cheers.

I have boot=/dev/sda

When I reinstalled slackware yesterday and loaded the huge.s kernel, it's the same as what's loading when I boot this desktop which is slow.

I already did the compact in lilo.conf but that's not the slow part since the entire .... gets loaded in 2 seconds.

I solved the fluxbox problem by using feh from slackbuilds to set my wallpaper.
The other error along is all solved since I noticed fluxbox was trying to load xmodmap on start but I removed that. The linpng warning was because of something was wrong with the image and I replaced it and the warning disappeared.

I can't seem to solve the mkinitrd problem though. According to the README.initrd, the method I posted should be perfect.

When I installed slackware, I did a full install so the requirement that's listed in the README should be fulfilled. When I installed it, I remember seeing something like "generic-kernel-2.6.29.6 (requires mkinitrd)" so I think I'm still loading the huge kernel since the mkinitrd doesn't seem to be working.

I'm not sure if this matters but doing "uname -a" gives me:

mkinitrd -c -k 2.6.29.6 -m mbcache:jbd:ext4 -f ext4 -r /dev/sda2

You don't have the modules for your chipset.

Thank you. I realized my mistake and did a search and found Alien BOB's script to figure this out.

I cd to /boot and did:
My cd /boot looks like:

I edited my lilo.conf just like how I did it before in my first post in this thread.

I typed lilo after and rebooted but there was no change.

I doubled checked if I have the requirement listed in the README and it looked fine.

Still no chipset module, those are only for usb and ext4 - actually, passing -m ext4 will automaticially pull in crc16 , mbcache, and jbd2.

For my PC which has an Intel P45 chipset with ICH10 controller, also has a JMicron controller, but the hard drives are attached to the ICH10 controller, I would include -

-m shpchp:ahci:ext4

You want to do lspci -k to get the modules in use for anything that says SATA, or IDE interface - which ever your boot drive is connected to. Out of habbit I always include the PCI bridge (shpchp). USB is not needed for an intrid unless you are booting from an USB drive. An initrd should only contain the modules neccessary to get / mounted. Once / is mounted the kernel will auto load the modules on demand as needed.

If you post a complete lspci -k we'll help you with an initrd.

Thank you very much. :)

Try this for the -m section

Thank you once again. Just making sure since I noticed something strange,

My original mkinitrd is:I was looking at other peoples mkinitrd and it seems like I'm the only with the "-o /boot/initrd.gz" at the end. Should I add that in so it's something like:

or just should be good?

I was wondering about this because I have a seperate /boot and /root partition and maybe this could cause a problem but I noticed you said /root so /boot has nothing to do with this?

The -o options tells mkinitrd where to drop it's file at. This helps when running more than one kernel or more than one initrd. So you can -o /boot/test1.gz -o /boot/initrd2.gz -o /some/where/something.gz

The intitrd should go into /boot to keep things sane. Kernels, configs .... keep them in once place - just makes things easier By default, not passing an -o option, mkinitrd makes the file as /boot/initrd.gz. So, -o /boot/initrd.gz is the default :)

I see, thank you very much! Testing it out now :)

I finished testing it and nothing changed. I already had a initrd.gz in /boot so maybe I should rm initrd-tree/ and initrd.gz from /boot then redo it again?

What is not changing?

edit -------- re-re-read the entire post again.

Changing from Huge to generic kernel will not make a big difference in boot speed. Changing these kernels will only effect the initial decompression of the kernel. Maybe 2-4 seconds depending on your PC.

Things to speed up boot time -

add noatime to your fstab. The part that says defaults - change that to noatime.

chmod -x non essestial services in /etc/rc.d

edit /etc/rc.d/rc.M to stop some of the gtk updating and ldconfig. NOTE these serve a purpose and are a good idea, but if you're impatient they can save a few more seconds. Make a backup copy first.

Thanks. It's still loading the exact same kernel (huge kernel) even though it should be loading the generic kernel instead.

Edited my post while you were posting.

Ah I see. I been reading other thread but I'll just go with this quote I found there:

Until I become competent enough so I know exactly what I'm doing. I rather not do something stupid and break my system. 20 seconds or so boot time isn't so bad anyway.

You can't fix it until it's broken ... wait it's ... Why fix what isn't broken :)


Try qemu-kvm, VMware, or VirtualBox, much less devastating to break a virtual machine than a live system.