1 |
Rich Freeman posted on Sun, 21 Sep 2014 22:34:23 -0400 as excerpted: |
2 |
|
3 |
> On Sun, Sep 21, 2014 at 10:02 PM, Frank Peters |
4 |
> <frank.peters@×××××××.net> wrote: |
5 |
|
6 |
>> There are things which are not system calls that could easily be |
7 |
>> changed. It is not too far fetched to consider a time if and when |
8 |
>> systemd became so popular and entrenched that the kernel would be |
9 |
>> hard-coded to pass control only to systemd and nothing else. |
10 |
> |
11 |
> That seems extremely unlikely. How many people ran anything other than |
12 |
> sysvinit as their init for the 15 years or so before upstart came along? |
13 |
> Making the kernel dependent on systemd would defeat the whole purpose |
14 |
> of having a separation between userspace and kernelspace. |
15 |
|
16 |
Agreed. There's far too many and too broad usages of the Linux kernel |
17 |
for that sort of hard-coding, at least without at least a kconfig option |
18 |
for it. |
19 |
|
20 |
Is android suddenly going to switch to systemd? Unlikely, and it's |
21 |
generally acknowledged to be the biggest usage of the Linux kernel out |
22 |
there these days, so hard-coding android breakage isn't going to happen. |
23 |
|
24 |
Plus even if it did, we're dealing with open source here and Google would |
25 |
simply patch that out and their own solution in as they do with a bunch |
26 |
of other stuff. And if google could do that, so could anyone else. |
27 |
|
28 |
Then there's the tivos and the embedded medical devices and the multiple |
29 |
automotive systems likely running their own little embedded Linux |
30 |
kernels. Hard-coding systemd for all of that? Not going to happen. |
31 |
|
32 |
As for the loss of the usb static device nodes, did you (Frank) file a |
33 |
bug about it breaking your userspace? That's one of Linus' most firm |
34 |
kernel rules -- you do *NOT* change the userspace/kernelspace API/ABI and |
35 |
break userspace. However, there's a known exception. Rather like the |
36 |
old philosophical question as to whether if a tree falls in the forest |
37 |
and nobody hears/sees it, did it actually fall at all, if nobody notices |
38 |
the userspace/kernelspace ABI breaking, did it really break at all? |
39 |
|
40 |
Unfortunately, for support for stuff like the big databases, etc, the big |
41 |
users all tend to be on enterprise distros with years-old kernels and |
42 |
sometimes the changes that break that don't get noticed for years simply |
43 |
because nobody running those apps is running anything close to current |
44 |
kernels, or if they do, they aren't reporting the problem. Bu the time |
45 |
the breakage is actually noticed and reported two years later, other |
46 |
userspace may depend on the new behavior and it can become a choice of |
47 |
which userspace to break, the newer stuff depending on the new behavior |
48 |
or the older stuff that was broken but that nobody noticed or reported |
49 |
for years. That can cause problems, particularly when those old and now |
50 |
broken userspace programs are big-dollar enterprise users, but sometimes |
51 |
it happens. |
52 |
|
53 |
And Linus and the other kernel devs are constantly pointing out that if |
54 |
they break userspace, report it as soon as possible so it can be fixed. |
55 |
Those who fail to do so, unfortunately very occasionally have to live |
56 |
with the resulting breakage, at least to some extent, tho they still go |
57 |
to rather extreme lengths to finesse things if and when they can. |
58 |
|
59 |
So if your userspace breaks due to a kernel change, report it as soon as |
60 |
you detect it and ask that it be fixed. Linus is very likely to make |
61 |
sure it happens. If you didn't do that, well... |
62 |
|
63 |
-- |
64 |
Duncan - List replies preferred. No HTML msgs. |
65 |
"Every nonfree program has a lord, a master -- |
66 |
and if you use the program, he is your master." Richard Stallman |