Am 27. Dezember 2014 09:33:21 MEZ, schrieb Sven Joachim <svenjoac at gmx.de>:
>On 2014-12-27 02:44 +0100, Michael Biebl wrote:
>
>> It looks like we won't get initramfs-tools 0.118 with usr-mount
>support
>> in the initramfs, as this created quite a few regressions which need
>to
>> be dealt with first.
>>
>> I therefore think we need to workaround this issue in systemd and
>make
>> sure systemd-remount-fs.service doesn't fail for split-usr setups.
>> So marking this bug as RC and bumping the severity to serious.
>>
>> The patch might be as simple as the attached one. But we'll need to
>> check if it doesn't cause any regressions with dracut and
>> initramfs-tools 0.118, which actually do mount /usr in the initramfs.
>>
>> diff --git a/src/remount-fs/remount-fs.c
>b/src/remount-fs/remount-fs.c
>> index 847637a..44cc959 100644
>> --- a/src/remount-fs/remount-fs.c
>> +++ b/src/remount-fs/remount-fs.c
>> @@ -77,10 +77,9 @@ int main(int argc, char *argv[]) {
>> int k;
>> char *s;
>>
>> - /* Remount the root fs, /usr and all API VFS */
>> + /* Remount the root fs and all API VFS */
>> if (!mount_point_is_api(me->mnt_dir) &&
>> - !path_equal(me->mnt_dir, "/") &&
>> - !path_equal(me->mnt_dir, "/usr"))
>> + !path_equal(me->mnt_dir, "/"))
>> continue;
>>
>> log_debug("Remounting %s", me->mnt_dir);
>
>I haven't looked at dracut, but with initramfs-tools 0.118 /usr is
>going
>to end up being mounted readonly.
Hmm, good point. We'll need a more elaborate patch then which deals with both cases