Patchwork [2/7] systemd: don't recommend IPv6 kernel module

login
register
mail settings
Submitter Ross Burton
Date Feb. 27, 2013, 2:17 p.m.
Message ID <a4ec8dc8fcdaa379c4abe77debb003ad49f1115d.1361974597.git.ross.burton@intel.com>
Download mbox | patch
Permalink /patch/45203/
State New
Headers show

Comments

Ross Burton - Feb. 27, 2013, 2:17 p.m.
systemd doesn't need ipv6 support directly, so don't pull it.

Signed-off-by: Ross Burton <ross.burton@intel.com>
---
 meta/recipes-core/systemd/systemd_197.bb |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Koen Kooi - Feb. 27, 2013, 2:40 p.m.
On Wed, 2013-02-27 at 14:17 +0000, Ross Burton wrote:
> systemd doesn't need ipv6 support directly, so don't pull it.

It will trigger an error on startup, which seems to scare most of my
users :(
Ross Burton - Feb. 27, 2013, 2:53 p.m.
On 27 February 2013 14:40, Koen Kooi <koen@dominion.thruhere.net> wrote:
> On Wed, 2013-02-27 at 14:17 +0000, Ross Burton wrote:
>> systemd doesn't need ipv6 support directly, so don't pull it.
>
> It will trigger an error on startup, which seems to scare most of my
> users :(

Well that's silly of systemd isn't it.  Just noticed it didn't cause a
problem here because the BSP I'm testing on appears to put ipv6 into
the kernel

Can you point to where the log is coming from - silencing that warning
seems like a sensible thing to do, as no ipv6 is a legitimate choice
(albeit rapidly not a sane choice...).

Ross
Koen Kooi - Feb. 27, 2013, 4:07 p.m.
On Wed, 2013-02-27 at 14:53 +0000, Burton, Ross wrote:
> On 27 February 2013 14:40, Koen Kooi <koen@dominion.thruhere.net> wrote:
> > On Wed, 2013-02-27 at 14:17 +0000, Ross Burton wrote:
> >> systemd doesn't need ipv6 support directly, so don't pull it.
> >
> > It will trigger an error on startup, which seems to scare most of my
> > users :(
> 
> Well that's silly of systemd isn't it.  Just noticed it didn't cause a
> problem here because the BSP I'm testing on appears to put ipv6 into
> the kernel
> 
> Can you point to where the log is coming from - silencing that warning
> seems like a sensible thing to do, as no ipv6 is a legitimate choice
> (albeit rapidly not a sane choice...)

src/core/kmod-setup.c:

static const KModule kmod_table[] = {
        { "autofs4",  "/sys/class/misc/autofs",    NULL } ,
        { "ipv6",     "/sys/module/ipv6",          NULL },
        { "unix",     "/proc/net/unix",            NULL } ,
};

[..]

                err = kmod_module_probe_insert_module(mod,
KMOD_PROBE_APPLY_BLACKLIST, NULL, NULL, NULL, NULL);
                if (err == 0)
                        log_info("Inserted module '%s'",
kmod_module_get_name(mod));
                else if (err == KMOD_PROBE_APPLY_BLACKLIST)
                        log_info("Module '%s' is blacklisted",
kmod_module_get_name(mod));
                else
                        log_error("Failed to insert module '%s'",
kmod_module_get_name(mod));

And looking at the other code you'll need to load ipv6 before starting
the sockets otherwise socket activation over ipv6 won't work.

regards,

Koen

Patch

diff --git a/meta/recipes-core/systemd/systemd_197.bb b/meta/recipes-core/systemd/systemd_197.bb
index 880fbcd..c871bb5 100644
--- a/meta/recipes-core/systemd/systemd_197.bb
+++ b/meta/recipes-core/systemd/systemd_197.bb
@@ -182,7 +182,7 @@  RRECOMMENDS_${PN} += "systemd-serialgetty \
                       util-linux-swaponoff \
                       util-linux-fsck e2fsprogs-e2fsck \
                       util-linux-mount util-linux-umount \
-                      kernel-module-autofs4 kernel-module-unix kernel-module-ipv6 \
+                      kernel-module-autofs4 kernel-module-unix \
 "
 
 PACKAGES =+ "udev-dbg udev udev-consolekit udev-utils udev-systemd"