On Wed, Oct 9, 2019 at 9:56 AM Stephen Smalley <sds@xxxxxxxxxxxxx> wrote
> On 10/8/19 5:43 PM, Paul Moore wrote:
> > On Mon, Oct 7, 2019 at 12:35 PM Richard Haines
> > <richard_c_haines@xxxxxxxxxxxxxx> wrote:
> >> On Mon, 2019-10-07 at 16:17 +0100, Richard Haines wrote:
> >>> On Mon, 2019-10-07 at 10:28 -0400, Stephen Smalley wrote:
> >>>> On 10/6/19 4:51 AM, Richard Haines wrote:
> >>>>> Kernel 5.4 commit ca2864c6e8965c37df97f11e6f99e83e09806b1c
> >>>>> ("binder: Add
> >>>>> default binder devices through binderfs when configured"),
> >>>>> changed
> >>>>> the way
> >>>>> the binder device is initialised and no longer automatically
> >>>>> generates
> >>>>> /dev/binder when CONFIG_ANDROID_BINDERFS=y.
> >>>>
> >>>> This seems like a userspace ABI break, no? Same kernel config
> >>>> before
> >>>> and after this commit yields different behavior for
> >>>> /dev/binder. I
> >>>> suppose one might argue that one would only enable
> >>>> CONFIG_ANDROID_BINDERFS if one wanted to use it instead of
> >>>> /dev/binder
> >>>> but the original commit that introduced binderfs specifically said
> >>>> that
> >>>> backward compatibility was preserved.
> >>> I'll need to check this further, but from what I've seen so far, is
> >>> that the /dev/binder is not available until you mount binderfs etc.
> >>> that's why Paul had the failure on 5.4 as before then is was
> >>> available
> >>> when the binder driver first initialised.
> >>
> >> To confirm tests using kernel 5.4-rc1
> >>
> >> Test 1 config:
> >> CONFIG_ANDROID=y
> >> CONFIG_ANDROID_BINDER_IPC=y
> >> CONFIG_ANDROID_BINDERFS=y
> >> CONFIG_ANDROID_BINDER_DEVICES="binder"
> >>
> >> On boot no /dev/binder
> >>
> >> To get this you have to:
> >> mkdir /dev/binderfs 2>/dev/null
> >> mount -t binder binder /dev/binderfs -o
> >> context=system_u:object_r:device_t:s0 2>/dev/null
> >>
> >> You then have devs:
> >> binder and binder-control
> >>
> >> Test 2 config:
> >> CONFIG_ANDROID=y
> >> CONFIG_ANDROID_BINDER_IPC=y
> >> # CONFIG_ANDROID_BINDERFS is not set
> >> CONFIG_ANDROID_BINDER_DEVICES="binder"
> >>
> >> On boot you have /dev/binder
> >
> > Disabling binderfs during build is probably not the smart thing to do
> > considering where the world is at with namespaces/containers, whatever
> > we do we should make sure the tests work with
> > CONFIG_ANDROID_BINDERFS=y.
>
> Yes, I think the question is just whether we want to have the tests use
> binderfs for kernel >= 5.0 (i.e. the point at which binderfs was first
> introduced) or for kernel >= 5.4 (i.e. the point at which binderfs usage
> became mandatory if you enable it in your config because /dev/binder is
> no longer automatically created). I'm fine either way.
Agreed in that it probably doesn't matter all that much. I might be
tempted to start with v5.4 over v5.0 since the old way still worked on
v5.0 through v5.3 and it might be nice to notice if that changes in a
v5.{0..3}.Z release.
--
paul moore
www.paul-moore.com
