1 |
Jason wrote: |
2 |
> ons won't be out of sync then. Gentoo has genkernel for this purpose. |
3 |
> |
4 |
>> Also, opened files and extra nodes in /dev during intiramfs phase |
5 |
>> tend to cause a headache or two... |
6 |
> |
7 |
> If your initrd follows the 3 suggestions, above, there won't be any |
8 |
> processes running to hold a file or device open. Unless you had |
9 |
> something else in mind? |
10 |
> |
11 |
> Jason. |
12 |
|
13 |
|
14 |
|
15 |
I ahve used initrd for booting from LVM2 partition in the past and have |
16 |
always had headaches with this sort of thing- and I believe exactly this |
17 |
was covered in some article on gentoo.org |
18 |
|
19 |
Point is that you need to run some programs to do some things ( why else |
20 |
have initramfs ?) before system becomes really useable for the user and |
21 |
you can't always control all aspects of programs that you have ran. Some |
22 |
might spawn some daemon with opened files or at least devices etc etc. I |
23 |
have mounted XFS on one machine and because of this two daemons run on |
24 |
my machine etc. |
25 |
|
26 |
Also, if you need autodetection and device for some PnP hardware ( USB, |
27 |
1394 etc ) you need udev and hotplug system to generate enw devices in |
28 |
dev and then recreate them in /new_root/dev. |
29 |
And if some daemon has continually opened something in /dev, after |
30 |
chroot/pivot_root your initrd might still occupy your memory. AFAIK |
31 |
initramfs should eb way better with this, but any user data about it is |
32 |
very scarce on the net... |
33 |
-- |
34 |
gentoo-amd64@l.g.o mailing list |