1 |
On 01/11/2016 00:27, John Covici wrote: |
2 |
> On Mon, 31 Oct 2016 18:06:07 -0400, |
3 |
> Alan McKinnon wrote: |
4 |
>> |
5 |
>> On 31/10/2016 23:57, John Covici wrote: |
6 |
>>> On Mon, 31 Oct 2016 13:37:50 -0400, |
7 |
>>> Meino.Cramer@×××.de wrote: |
8 |
>>>> |
9 |
>>>> Dale <rdalek1967@×××××.com> [16-10-31 18:36]: |
10 |
>>>>> Meino.Cramer@×××.de wrote: |
11 |
>>>>>> Hi, |
12 |
>>>>>> |
13 |
>>>>>> from ancient times ;) I remember, that it is not advisable |
14 |
>>>>>> to compile a linux kernel with more than one cpu core. |
15 |
>>>>>> |
16 |
>>>>>> Is that still true, or is it save to compile it with |
17 |
>>>>>> "all you can eat" ::)) ? |
18 |
>>>>>> |
19 |
>>>>>> Cheers |
20 |
>>>>>> Meino |
21 |
>>>>>> |
22 |
>>>>>> |
23 |
>>>>> |
24 |
>>>>> While I don't upgrade my kernel very often, since I don't reboot often |
25 |
>>>>> either, I've used -j 6 for my 4 core CPU for a long time. I'm pretty |
26 |
>>>>> sure I've done that ever since I built this rig. So far, no problems |
27 |
>>>>> that I have seen. |
28 |
>>>>> |
29 |
>>>>> Dale |
30 |
>>>>> |
31 |
>>>>> :-) :-) |
32 |
>>>>> |
33 |
>>>> |
34 |
>>>> Great! Currently rebuilding the kernel with '-j7 ' :) |
35 |
>>> |
36 |
>>> I wonder, how does it make sure that a dependency is always compiled |
37 |
>>> before what depends on it, so the link and all that works? |
38 |
>>> |
39 |
>> |
40 |
>> The kernel is special - it's a completely self-contained body of code |
41 |
>> with no external dependencies, so it is internally consistency almost as |
42 |
>> a matter of course. |
43 |
>> |
44 |
>> As for compiling the right things in the right order, the Makefiles take |
45 |
>> care of that just like all other software with a build system. How does |
46 |
>> the Makefile get made? Well, when you do make <something>config, you see |
47 |
>> all those deps like how iwlwifi depends on 80211 which depends on |
48 |
>> networking etc. That tells you the order things must be built in. Shove |
49 |
>> that into a Makefile maker, and voila, Bob's your auntie. Build issues |
50 |
>> tends to get fixed during the dev period so when Linux releases a kernel |
51 |
>> there's an excellent chance it will build correctly. |
52 |
>> |
53 |
>> FWIW, none of this is a specially difficult problem. It's the kind of |
54 |
>> thing I'd expect bright CS students to be able to do at the end of the |
55 |
>> first year |
56 |
> |
57 |
> OK, I will maybe try something, but I have had gentoo problems only |
58 |
> solved by -j1 due to such things, I guess. |
59 |
> |
60 |
|
61 |
|
62 |
|
63 |
With the kernel? Very very highly unlikely. If you had that problem, |
64 |
there is something wrong with your setup and I'd be very reluctant jump |
65 |
at the kernel Makefiles being the problem. |
66 |
|
67 |
Now if you said you had that problem with libreoffice, I'd agree 100%. |
68 |
The kernel? No. |
69 |
|
70 |
|
71 |
and btw, emerge problems that are solved using -j1 are |
72 |
a) not gentoo problems at all |
73 |
b) neither are they compiler problems |
74 |
c) they are always build problems caused by the package's own shitty |
75 |
build system and by upstream devs that never heard of this idea called |
76 |
"test your stuff to make sure it works in the real world" |
77 |
|
78 |
-- |
79 |
Alan McKinnon |
80 |
alan.mckinnon@×××××.com |