Thanks for the inputs. After removing -n from restorecon, I see that file
context is changed as desired. But, when I try restorecon from *.rc file,
it doesn't seem to work!


On Fri, Jan 3, 2014 at 12:57 PM, William Roberts
<[email protected]>wrote:

> Oh sri said the ext4 is not created at build. My guess its similar to efs.
> On Jan 3, 2014 12:56 PM, [email protected] wrote:
>
>> Ha I forgot to tell him to remove n... Doh
>> On Jan 3, 2014 11:57 AM, "Stephen Smalley" <[email protected]>
>> wrote:
>>
>>> The -n option to restorecon means "Do Not change", i.e. do not set the
>>> context.  When combined with the -v option (verbose), it shows what it
>>> would set but does not apply the change.  So it isn't surprising that
>>> restorecon -nv did not change anything.  Remove the -n option if you
>>> want to set manually.
>>> What kind of filesystem is on the partition (e.g. ext4), and how is it
>>> initially created?
>>> You have to pass in the right options to make_ext4fs if you want to
>>> label the files based on file_contexts; the build Makefiles have been
>>> modified to do this automatically for the system image when it is
>>> built.  Or if the partition starts empty, you can just restorecon it
>>> from init.*.rc as is done for /data and /cache in the init.rc file.
>>>
>>>
>>> On Fri, Jan 3, 2014 at 2:32 PM, sri linux <[email protected]> wrote:
>>> > Adding mailing list back to the thread - In one of my reply, got
>>> removed as
>>> > I hit just "Reply" instead of  "Reply to all"!
>>> >
>>> >
>>> > On Fri, Jan 3, 2014 at 11:04 AM, sri linux <[email protected]>
>>> wrote:
>>> >>
>>> >> In one of the TE file, I have defined file_type as
>>> >> type <partition_name>_data_file, file_type;
>>> >>
>>> >> And in file_contexts, I have defined below to switch context of the
>>> file
>>> >> /<partition_name>/<dir_name>(/.*)?
>>> >> u:object_r:<partition_name>_data_file:s0
>>> >>
>>> >> When I tried restorecon from the shell, I see below log, which seem
>>> to be
>>> >> trying to change the context of the file, but, when I cross check
>>> using "ls
>>> >> -lZ" I still see default context.
>>> >> restorecon -nv /<partition_name>/<dir_name>/<file_name>
>>> >> Relabeling /<partition_name>/<dir_name>/<file_name> from
>>> >> u:object_r:unlabeled:s0 to u:object_r:<partition_name>_data_file:s0.
>>> >>
>>> >> Just wondering what Am I missing here!
>>> >>
>>> >>
>>> >> On Fri, Jan 3, 2014 at 9:25 AM, William Roberts <
>>> [email protected]>
>>> >> wrote:
>>> >>>
>>> >>> ok then a restorecon should do the trick. Depending on your version
>>> >>> you may need to restorecon each file and you need to make sure the
>>> >>> file_context is located either in the ramdisk or data. Data will only
>>> >>> work if this filesystem is mounted AFTER data, so the safest option
>>> is
>>> >>> ensuring it is in /file_contexts which is located in the ramdisk.
>>> >>>
>>> >>> in the init.rc after the filesystem in question is mounted you will
>>> >>> need restorecon entries. Based on the version you are running you may
>>> >>> need to restorecon each file. I would recommend just verifying the
>>> >>> file_contexts, and manually shelling into the device and
>>> transitioning
>>> >>> to su. You can do this by: adb shell followed by su 0. Once in as
>>> >>> super user, you can issue the restorecon command. if an ls -Z file
>>> >>> still shows the invalid label. I would try restorecon -nv to get some
>>> >>> debugging info as to why its failing. Perhaps your expression in the
>>> >>> file_contexts entry is not proper and is not matching anything.
>>> >>>
>>> >>> If you need to restorecon the whole filesystem, after the mount in
>>> the
>>> >>> init.rc you can do restorecon_recursive if your init supports it.
>>> AOSP
>>> >>> now has this feature. It was added here:
>>> >>> https://android-review.googlesource.com/#/c/67580/
>>> >>>
>>> >>> At this point, I think to get more specific help, you will need to
>>> >>> provide more specifics to your problem. Can you give us the snippet
>>> of
>>> >>> your file_contexts, and other commands, etc. Diffs are really helpful
>>> >>> here. You can change the data to keep it anonymous if needed.
>>> >>>
>>> >>> Keep in mind that adb root; adb shell; will result in a root terminal
>>> >>> running in the shell domain which may not have any permissions to do
>>> >>> what you need. It is very important that you invoke su to transition
>>> >>> into the su domain, which is unconfined.
>>> >>>
>>> >>> Bill
>>> >>>
>>> >>> On Fri, Jan 3, 2014 at 8:01 AM, sri linux <[email protected]>
>>> wrote:
>>> >>> > Yes, filesystem is mounted as read & write.
>>> >>> >
>>> >>> >
>>> >>> > On Thu, Jan 2, 2014 at 6:52 PM, William Roberts
>>> >>> > <[email protected]>
>>> >>> > wrote:
>>> >>> >>
>>> >>> >> Ok is the filesystem mounted writable? If not, then a restorecon
>>> would
>>> >>> >> fail.
>>> >>> >>
>>> >>> >> On Thu, Jan 2, 2014 at 4:18 PM, sri linux <[email protected]>
>>> wrote:
>>> >>> >> > As mentioned already, I tried restorecon, which doesn't have any
>>> >>> >> > affect.
>>> >>> >> > File is created at build time in one of the partition
>>> >>> >> > (separate/different
>>> >>> >> > partition and not boot/system/data) and filesystem is ext4 type.
>>> >>> >> > Filesystem
>>> >>> >> > is mounted as read & write.
>>> >>> >> >
>>> >>> >> >
>>> >>> >> > On Mon, Dec 30, 2013 at 12:14 PM, William Roberts
>>> >>> >> > <[email protected]>
>>> >>>
>>> >>> >> > wrote:
>>> >>> >> >>
>>> >>> >> >> And what type of filesystem is it?
>>> >>> >> >>
>>> >>> >> >> On Mon, Dec 30, 2013 at 11:28 AM, Stephen Smalley
>>> >>> >> >> <[email protected]> wrote:
>>> >>> >> >> > Which partition?  And how is the file created?
>>> >>> >> >> >
>>> >>> >> >> > On Fri, Dec 27, 2013 at 7:05 PM, sri linux <
>>> [email protected]>
>>> >>> >> >> > wrote:
>>> >>> >> >> >> Dear Experts,
>>> >>> >> >> >>
>>> >>> >> >> >> I have a file in one of the partition, for which, I see
>>> below as
>>> >>> >> >> >> a
>>> >>> >> >> >> default
>>> >>> >> >> >> context:
>>> >>> >> >> >> -rw-rw-r-- system   root
>>>  u:object_r:unlabeled:s0
>>> >>> >> >> >> file_xyz
>>> >>> >> >> >>
>>> >>> >> >> >> I tried to change the context using type_transition, which
>>> >>> >> >> >> doesn't
>>> >>> >> >> >> seem
>>> >>> >> >> >> to
>>> >>> >> >> >> be working - I still see unlabeled in the context.
>>> >>> >> >> >>
>>> >>> >> >> >> Can someone suggest the best way to change the context of
>>> the
>>> >>> >> >> >> file
>>> >>> >> >> >> which is
>>> >>> >> >> >> unlabeled?
>>> >>> >> >> >>
>>> >>> >> >> >>
>>> >>> >> >> >>
>>> >>> >> >> >> _______________________________________________
>>> >>> >> >> >> Seandroid-list mailing list
>>> >>> >> >> >> [email protected]
>>> >>> >> >> >> To unsubscribe, send email to
>>> >>> >> >> >> [email protected].
>>> >>> >> >> >> To get help, send an email containing "help" to
>>> >>> >> >> >> [email protected].
>>> >>> >> >> >>
>>> >>> >> >> > _______________________________________________
>>> >>> >> >> > Seandroid-list mailing list
>>> >>> >> >> > [email protected]
>>> >>> >> >> > To unsubscribe, send email to
>>> [email protected].
>>> >>> >> >> > To get help, send an email containing "help" to
>>> >>> >> >> > [email protected].
>>> >>> >> >>
>>> >>> >> >>
>>> >>> >> >>
>>> >>> >> >> --
>>> >>> >> >> Respectfully,
>>> >>> >> >>
>>> >>> >> >> William C Roberts
>>> >>> >> >
>>> >>> >> >
>>> >>> >>
>>> >>> >>
>>> >>> >>
>>> >>> >> --
>>> >>> >> Respectfully,
>>> >>> >>
>>> >>> >> William C Roberts
>>> >>> >
>>> >>> >
>>> >>>
>>> >>>
>>> >>>
>>> >>> --
>>> >>> Respectfully,
>>> >>>
>>> >>> William C Roberts
>>> >>
>>> >>
>>> >
>>> >
>>> > _______________________________________________
>>> > Seandroid-list mailing list
>>> > [email protected]
>>> > To unsubscribe, send email to [email protected].
>>> > To get help, send an email containing "help" to
>>> > [email protected].
>>> >
>>>
>>
_______________________________________________
Seandroid-list mailing list
[email protected]
To unsubscribe, send email to [email protected].
To get help, send an email containing "help" to 
[email protected].

Reply via email to