1 |
On Wed, 3 Jul 2013 16:52:37 +0300 |
2 |
Sergei Trofimovich <slyfox@g.o> wrote: |
3 |
|
4 |
> On Wed, 3 Jul 2013 15:06:07 +0200 |
5 |
> Tom Wijsman <TomWij@g.o> wrote: |
6 |
> |
7 |
> > On Tue, 2 Jul 2013 21:16:07 +0300 |
8 |
> > Sergei Trofimovich <slyfox@g.o> wrote: |
9 |
> > |
10 |
> > > udev's case: |
11 |
> > > It's _not_ the kernel which upgrade breaks user's system. Why do |
12 |
> > > you try to "fix" it? If you want to save user's box - make check |
13 |
> > > at pre-install time. Otherwise it will break. |
14 |
> > |
15 |
> > Won't work for stage3's case, where this check won't run. |
16 |
> |
17 |
> Non-issue. |
18 |
> - stage3 does not work out of the box. you can't boot or run it |
19 |
> - stage3 does not work at all in containers w/o devtmpfs kernel. |
20 |
> no matter what you try to force on newly built kernel. |
21 |
|
22 |
Make check doesn't work there. A distro default config option does. |
23 |
|
24 |
> Let's consider real-world examples: |
25 |
> |
26 |
> My friend at cloud hosting company decided to try gentoo on |
27 |
> a xen hypervisor box (3.0.6 host w/o devtmpfs). unpacked old |
28 |
> stage3, which worked; merged new stuff there and rebooted. |
29 |
> No kernel change, old xen 3.0.6. Guess what happened when he |
30 |
> upgraded udev afterwards. |
31 |
> |
32 |
> Another my friend unpacked stage3 on his handheld with android kernel. |
33 |
> 2.6.3x-something w/o accept4() syscall. Original stage3 worked, |
34 |
> upgrade broke system. |
35 |
|
36 |
Stage3 contains the upgraded udev these days. |
37 |
|
38 |
> > > kernel's case: |
39 |
> > > CONFIG_DEVTMPFS needs 'default yes', right? One-liner to |
40 |
> > > gentoo-sources. |
41 |
> > |
42 |
> > It's not as simple as a one-liner, because we need to respect people |
43 |
> > that want to build a kernel without that; not everyone who uses |
44 |
> > genpatches runs a Gentoo system, and note every other OS needs that |
45 |
> > variable to enabled. |
46 |
> |
47 |
> 'default yes' is a Kconfig option. It picks a value if user didn't |
48 |
> pick it explicitely. |
49 |
> |
50 |
> linux/drivers/base/Kconfig: |
51 |
> config DEVTMPFS |
52 |
> bool "Maintain a devtmpfs filesystem to mount at /dev" |
53 |
> depends on HOTPLUG |
54 |
> + default y |
55 |
> |
56 |
> One-liner, no? |
57 |
|
58 |
We don't want to change DEVTMPFS itself, because not everyone whom |
59 |
uses genpatches wants DEVTMPFS itself enabled by default; we instead |
60 |
plan to provide it as a distribution specific option that enables it. |
61 |
|
62 |
https://lkml.org/lkml/2012/7/13/369 |
63 |
|
64 |
> > > Why 'hide' it? It's very counterintuitive. |
65 |
> > |
66 |
> > What is being hidden? |
67 |
> |
68 |
> You plan to have it visible only when something gentoo-specific needs |
69 |
> to be disabled. |
70 |
|
71 |
No, I did not state such thing. |
72 |
|
73 |
> [from origina post] |
74 |
> > This is why I think it would be handy to add a Gentoo section to |
75 |
> > the kernel, along the lines as described by Linus. |
76 |
> > |
77 |
> > https://lkml.org/lkml/2012/7/13/369 |
78 |
> |
79 |
> AFAIU I won't allow deselecting option until you deselect |
80 |
> gentoo-specific bit. |
81 |
|
82 |
Then you disable the bit and deselect it, I don't see a problem here. |
83 |
|
84 |
> If you plan to upstream everything, then ok, no distro-specific hacks. |
85 |
|
86 |
I don't see how providing distro default options as described is a hack. |
87 |
|
88 |
> > > What will you do with all-those-required-to-boot-properly |
89 |
> > > - root filesystem |
90 |
> > > - disk controller |
91 |
> > > - USB keyboard drivers |
92 |
> > > ? |
93 |
> > > Include them all unless CONFIG_DONT_REMOVER_OR_WONT_BOOT |
94 |
> > > option? |
95 |
> > |
96 |
> > These don't fall under options that need to be enabled for everyone. |
97 |
> > |
98 |
> > From what I see on chat and forums, people often set all these fine; |
99 |
> > yet they not always enable CONFIG_DEVTMPFS and similar variables. |
100 |
> > |
101 |
> > We deal with the absolutely necessary, not spoon feed every option. |
102 |
> > |
103 |
> > > I'm afraid I don't see how it's a solution. |
104 |
> > > Suppose, tomorrow's udev will require CONFIG_foo, and glibc will |
105 |
> > > require CONFIG_bar. How will you save user with your mechanism? |
106 |
> > |
107 |
> > I don't see how we don't save them; but well, I'm not entirely sure |
108 |
> > if you're talking about the opening post mechanism or another one |
109 |
> > here. |
110 |
> |
111 |
> Yeah, the opening post one. |
112 |
> Suppose, tomorrow's udev requires CONFIG_CGROUPS=y. |
113 |
> What are your actions to save user's box? |
114 |
|
115 |
If it's along the lines of what's proposed, add it as well. |
116 |
|
117 |
> Issue new gentoo-sources and pray user will install/boot it |
118 |
> before new udev? See my first two examples. |
119 |
|
120 |
That's what news items are for; we're not trying to deal with this |
121 |
short term case, but rather with the long term case after that. |
122 |
|
123 |
-- |
124 |
With kind regards, |
125 |
|
126 |
Tom Wijsman (TomWij) |
127 |
Gentoo Developer |
128 |
|
129 |
E-mail address : TomWij@g.o |
130 |
GPG Public Key : 6D34E57D |
131 |
GPG Fingerprint : C165 AF18 AB4C 400B C3D2 ABF0 95B2 1FCD 6D34 E57D |