1 |
On 2012-12-23, Michael Mol wrote: |
2 |
|
3 |
> On Dec 23, 2012 12:46 PM, "Nuno J. Silva" <nunojsilva@×××××××.pt> wrote: |
4 |
>> |
5 |
>> On 2012-12-23, Alan Mackenzie wrote: |
6 |
>> |
7 |
>> > On Sun, Dec 23, 2012 at 07:03:25PM +0200, Nuno J. Silva wrote: |
8 |
>> >> On 2012-12-23, Alan McKinnon wrote: |
9 |
>> > |
10 |
>> >> > On Sun, 23 Dec 2012 12:22:24 +0200 |
11 |
>> >> > nunojsilva@×××××××.pt (Nuno J. Silva) wrote: |
12 |
>> > |
13 |
>> >> >> On 2012-12-18, Alan McKinnon wrote: |
14 |
>> > |
15 |
>> >> >> > On Tue, 18 Dec 2012 09:08:53 -0500 |
16 |
>> >> >> > Michael Mol <mikemol@×××××.com> wrote: |
17 |
>> > |
18 |
>> > |
19 |
>> >> >> > This sentence summarizes my understanding of your post nicely: |
20 |
>> > |
21 |
>> >> >> >> Now, why is /usr special? It's because it contains executable |
22 |
> code |
23 |
>> >> >> >> the system might require while launching. |
24 |
>> > |
25 |
>> >> >> > Now there are only two approaches that could solve that problem: |
26 |
>> > |
27 |
>> >> >> > 1. Avoid it entirely |
28 |
>> >> >> > 2. Deal with it using any of a variety of bootstrap techniques |
29 |
>> > |
30 |
>> >> >> > #1 is handled by policy, whereby any code the system might require |
31 |
>> >> >> > while launching is not in /usr. |
32 |
>> > |
33 |
>> >> >> > #2 already has a solution, it's called an init*. Other solutions |
34 |
>> >> >> > exist but none are as elegant as a throwaway temporary filesystem |
35 |
>> >> >> > in RAM. |
36 |
>> > |
37 |
>> >> >> What about just mounting /usr as soon as the system boots? |
38 |
>> > |
39 |
>> > |
40 |
>> >> > Please read the thread next time. The topic under discussion is |
41 |
>> >> > solutions to the problem of not being able to do exactly that. |
42 |
>> > |
43 |
>> >> Then I suppose you can surely explain in a nutshell why can't init |
44 |
>> >> scripts simply do that? |
45 |
>> > |
46 |
>> > Because certain people with influence have rearranged the filesystem so |
47 |
>> > that programs within /usr are absolutely necessary for booting; they are |
48 |
>> > needed _before_ init has a chance to mount /usr. So either /usr has to |
49 |
>> > be in the root partition, or crazy kludges need to be used to mount /usr |
50 |
>> > before the kernel runs init. |
51 |
>> |
52 |
>> I surely don't know the udev architecture well enough, but if this is |
53 |
>> all done by the udev daemon, can't we just "mount /usr" before the |
54 |
>> daemon is started? The only needed things should be mount (which is |
55 |
>> under /bin here) and /etc/fstab. |
56 |
>> |
57 |
>> Or is something outside udev needing stuff under /usr? |
58 |
> |
59 |
> Yes. That's the pivot of the problem. |
60 |
|
61 |
What is it? |
62 |
|
63 |
I tried and I was able to mount a filesystem other than / shortly after |
64 |
linux has passed control to init, in fact, with no udev stuff running. |
65 |
|
66 |
-- |
67 |
Nuno Silva (aka njsg) |
68 |
http://njsg.sdf-eu.org/ |