1 |
On Sun, May 30, 2021 at 02:42:01PM -0400, Mike wrote: |
2 |
> |
3 |
> |
4 |
> On 5/30/21 2:28 PM, Ionen Wolkens wrote: |
5 |
> > On Sun, May 30, 2021 at 01:29:12PM -0400, mpagano@g.o wrote: |
6 |
> >> From: Mike Pagano <mpagano@g.o> |
7 |
> >> |
8 |
> >> As the purpose of pkg_pretend is to run sanity checks during |
9 |
> >> dependency calculation time, provide the default implementation |
10 |
> >> and perform CONFIG_CHECK within it. |
11 |
> >> |
12 |
> >> See bug #759238 |
13 |
> >> |
14 |
> > [...] |
15 |
> >> |
16 |
> >> +# @FUNCTION: linux-mod_pkg_pretend |
17 |
> >> +# @DESCRIPTION: |
18 |
> >> +# Check the CONFIG_CHECK options |
19 |
> >> +linux-mod_pkg_pretend() { |
20 |
> >> + |
21 |
> >> + debug-print-function ${FUNCNAME} $* |
22 |
> >> + # External modules use kernel symbols (bug #591832, #759238) |
23 |
> >> + CONFIG_CHECK+=" !TRIM_UNUSED_KSYMS" |
24 |
> >> +} |
25 |
> > |
26 |
> > Is this not supposed to actually run the checks? Correct me if I'm |
27 |
> > missing something but it's just setting the value. linux-mod normally |
28 |
> > run the checks through linux-info_pkg_setup and its check_extra_config. |
29 |
> |
30 |
> Thanks for the analysis, Ionen. |
31 |
> The eclass does not actually run the check as it exists today. |
32 |
> Maybe there was a reason for that when it was originally coded? |
33 |
|
34 |
It does indirectly: |
35 |
# External modules use kernel symbols (bug #591832) |
36 |
CONFIG_CHECK+=" !TRIM_UNUSED_KSYMS" |
37 |
|
38 |
linux-info_pkg_setup; |
39 |
^^^^^ checks ran here |
40 |
|
41 |
Older nvidia-drivers ebuild was kind of nasty with that, it did its |
42 |
own checks then the linux-mod.eclass' linux-info call did as well |
43 |
(duplicate messages), which I now let linux-mod handle. |
44 |
|
45 |
> |
46 |
> |
47 |
> > For nvidia-drivers, this value will also be lost for the pkg_setup test |
48 |
> > (needed to check, say.. gentoo-kernel emerged in-between) because I |
49 |
> > currently set a local CONFIG_CHECK="..." inside pkg_setup() |
50 |
> > (there's also a conditional CONFIG_CHECK, part of why not global) |
51 |
> > |
52 |
> > Some other ebuilds set CONFIG_CHECK in pkg_setup I believe. |
53 |
> |
54 |
> Maybe virtualbox-modules should do it's own CONFIG_CHECK as nvidia-drivers does. |
55 |
> As that package is the impetus of this patch. |
56 |
> |
57 |
> > Not that I can't change this for nvidia, I guess I could set a global |
58 |
> > scope CONFIG_CHECK with !FATAL-only and += the non-fatal ones in |
59 |
> > pkg_setup to avoid message duplication. |
60 |
> > |
61 |
> > Have same concerns as mgorny wrt exported pkg_pretend, plus I'd also |
62 |
> > need to add my own pkg_pretend wrapper to check MODULES_OPTIONAL_USE |
63 |
> > |
64 |
> |
65 |
|
66 |
-- |
67 |
ionen |