1 |
Mike Gilbert: |
2 |
> On Sat, Mar 23, 2013 at 1:42 PM, Dan Johansson <Dan.Johansson@×××.nu> wrote: |
3 |
>> Hello, |
4 |
>> |
5 |
>> Some of my servers are running with a kernel without module-support. |
6 |
>> On these servers something has started to pull in sys-apps/kmod, which when compiled complains about missing modules-support in the kernel (as it should). |
7 |
>> Doing an "equery d sys-apps/kmod" I can see that the following two packages depends on sy-apps/kmod: |
8 |
>> sys-fs/udev-197-r8 (kmod ? >=sys-apps/kmod-12) |
9 |
>> virtual/modutils-0 (sys-apps/kmod[tools]) |
10 |
>> |
11 |
>> sys-fs/udev has "-kmod" in its USE-flags, so that should not be an issue). |
12 |
>> # emerge --verbose --pretend sys-fs/udev |
13 |
>> [ebuild R ] sys-fs/udev-197-r8 USE="acl openrc -doc -gudev -hwdb -introspection -keymap -kmod (-selinux) -static-libs" 0 kB |
14 |
>> |
15 |
>> With virtual/modutils its an other thing, here we have a circular dependency between virtual/modutils and sys-apps/kmod if the tools USE-flag is set. |
16 |
> That circular dep is interesting; sys-apps/kmod only depends on |
17 |
> virtual/modutils because it inherits linux-mod.eclass. |
18 |
> |
19 |
> Could you file about about the circular dependency please? |
20 |
I recognized another issue possibly caused by inheriting linux-mod.eclass: |
21 |
kmod is pulling in virtual/linux-sources as a dependency, which actually |
22 |
is not necessary at all (and maybe even annoying, it's at least some |
23 |
hundred megabytes). |