[lxc-devel] [RFC] Fix for parsing various mount options

[Ryousei Takano]

Hi Guido,

On Apr 3, 2009, at 5:49 PM, Guido Trotter wrote:

> On Fri, Apr 3, 2009 at 4:25 AM, Ryousei Takano
> <takano-ryousei at aist.go.jp> wrote:
>> Hi Daniel and all,
>>
>> I am trying a fast container cloning using aufs (advanced multi layered
>> unification filesystem). That means a container rootfs is overlaid on top of
>> the template rootfs in a copy-on-write manner.
>>
>> lxc-start setups the rootfs according to $LXCPATH/test/fstab:
>>
>>        none /opt/lxc.root/rootfs.test aufs br=/opt/lxc.root/cowfs.test:/opt/lxc.root/rootfs.debian.tmpl=ro 0 0
>>
>>        /opt/lxc.root/rootfs.test /usr/local/var/lib/lxc/test/rootfs none rbind 0 0
>>
>> To do that, we need to pass the aufs specific mount option to mount(2).
>> But setup_mount() ignores this option.
>>
>> This patch fixes setup_mount() for parsing these kind of options.
>> As far as I know hasmntopt(3) cannot parse non literal options, so it adds
>> a new function called parse_mntopts() instead of using hasmntopt(3).
>>
>> Any comments and suggestions will be appreciated.
>>
>> Signed-off-by: Ryousei Takano <takano-ryousei at aist.go.jp>
>> ---
>>  src/lxc/conf.c |   93 ++++++++++++++++++++++++++++++++++++++++++++++++-------
>>  1 files changed, 81 insertions(+), 12 deletions(-)
>>
>> diff --git a/src/lxc/conf.c b/src/lxc/conf.c
>> index 6c3476a..5da9660 100644
>> --- a/src/lxc/conf.c
>> +++ b/src/lxc/conf.c
>> @@ -1026,13 +1026,82 @@ static int setup_cgroup(const char *name)
>>                                      line, MAXPATHLEN, (void *)name);
>>  }
>>
>> +static int parse_mntopts(struct mntent *mntent, unsigned long *mntflags,
>> +                        char **mntopts)
>> +{
>> +#if 1
>
> I think the
>
> #if 1 ...
> # else
>
> makes it less readable, by adding dead code...
> If you really want the other code, why not putting it in a comment,
> together with an explanation of why it wouldn't work?
>
I agree with you.  This patch is not completed.

I hope someone also needs this feature, but I am not sure it is a better
way to do that.


> Thanks,
>
> Guido
>

Best regards,
Ryousei Takano