[RFC] Build ANDROID_HOST=y on mac

[RFC] Build ANDROID_HOST=y on mac

[William Roberts]

>From commit 35d702 on https://github.com/williamcroberts/selinux/tree/fix-mac

I have a branch that is building on my elcapitan mac, requesting any
comments anyone
wishes to make, before I send them out.

If you wish to test, this is the procedure

1. Build libsepol (assumes at root of tree)
    a,  cd libsepol
    b. make
2. Build libselinux
    a. cd libselinux (assumes at root of tree)
    b. make ANDROID_HOST=y

This essentially gets us to where a build server/maintainer
can test patches quickly on mac, with some assurance
it's not busted without downloading all of Android.

It's still wise to check in an Android tree if possible IMHO.

-- 
Respectfully,

William C Roberts

Re: [RFC] Build ANDROID_HOST=y on mac

[Joshua Brindle]

William Roberts wrote:
>  From commit 35d702 on https://github.com/williamcroberts/selinux/tree/fix-mac
>
> I have a branch that is building on my elcapitan mac, requesting any
> comments anyone
> wishes to make, before I send them out.
>
> If you wish to test, this is the procedure
>
> 1. Build libsepol (assumes at root of tree)
>      a,  cd libsepol
>      b. make
> 2. Build libselinux
>      a. cd libselinux (assumes at root of tree)
>      b. make ANDROID_HOST=y
>

This works for me.


> This essentially gets us to where a build server/maintainer
> can test patches quickly on mac, with some assurance
> it's not busted without downloading all of Android.
>
> It's still wise to check in an Android tree if possible IMHO.
>

Re: [RFC] Build ANDROID_HOST=y on mac

[Joshua Brindle]

Joshua Brindle wrote:
> William Roberts wrote:
>> From commit 35d702 on
>> https://github.com/williamcroberts/selinux/tree/fix-mac
>>
>> I have a branch that is building on my elcapitan mac, requesting any
>> comments anyone
>> wishes to make, before I send them out.
>>
>> If you wish to test, this is the procedure
>>
>> 1. Build libsepol (assumes at root of tree)
>> a, cd libsepol
>> b. make
>> 2. Build libselinux
>> a. cd libselinux (assumes at root of tree)
>> b. make ANDROID_HOST=y
>>
>
> This works for me.

make install DESTDIR=/tmp/someidr mostly works, Mac ln does not support 
--relative so that fails. ANDROID_HOST also needs to be set in the top 
level makefile so that it propagates down:

ANDROID_HOST ?= n

>
>
>> This essentially gets us to where a build server/maintainer
>> can test patches quickly on mac, with some assurance
>> it's not busted without downloading all of Android.
>>
>> It's still wise to check in an Android tree if possible IMHO.
>>
>

Re: [RFC] Build ANDROID_HOST=y on mac

[William Roberts]

[-- Attachment #1: Type: text/plain, Size: 1388 bytes --]

On Sep 28, 2016 16:54, "Joshua Brindle" <brindle@quarksecurity.com> wrote:
>
> Joshua Brindle wrote:
>>
>> William Roberts wrote:
>>>
>>> From commit 35d702 on
>>> https://github.com/williamcroberts/selinux/tree/fix-mac
>>>
>>> I have a branch that is building on my elcapitan mac, requesting any
>>> comments anyone
>>> wishes to make, before I send them out.
>>>
>>> If you wish to test, this is the procedure
>>>
>>> 1. Build libsepol (assumes at root of tree)
>>> a, cd libsepol
>>> b. make
>>> 2. Build libselinux
>>> a. cd libselinux (assumes at root of tree)
>>> b. make ANDROID_HOST=y
>>>
>>
>> This works for me.
>
>
> make install DESTDIR=/tmp/someidr mostly works, Mac ln does not support
--relative so that fails. ANDROID_HOST also needs to be set in the top
level makefile so that it propagates down:
>
> ANDROID_HOST ?= n

Yeah install doesn't work on Mac, that's why for Darwin we just set the
path to the libsepol location for sefcontext_compile.

As for ANDROID_HOST, why does it need to go higher? It's only used in
libselinux and is declared and used just like DISABLE_SETRANS...I'm not
following you?

>
>
>>
>>
>>> This essentially gets us to where a build server/maintainer
>>> can test patches quickly on mac, with some assurance
>>> it's not busted without downloading all of Android.
>>>
>>> It's still wise to check in an Android tree if possible IMHO.
>>>
>>
>

[-- Attachment #2: Type: text/html, Size: 2063 bytes --]

Re: [RFC] Build ANDROID_HOST=y on mac

[Joshua Brindle]

William Roberts wrote:
> On Sep 28, 2016 16:54, "Joshua Brindle"<brindle@quarksecurity.com>  wrote:
>> Joshua Brindle wrote:
>>> William Roberts wrote:
>>>>  From commit 35d702 on
>>>> https://github.com/williamcroberts/selinux/tree/fix-mac
>>>>
>>>> I have a branch that is building on my elcapitan mac, requesting any
>>>> comments anyone
>>>> wishes to make, before I send them out.
>>>>
>>>> If you wish to test, this is the procedure
>>>>
>>>> 1. Build libsepol (assumes at root of tree)
>>>> a, cd libsepol
>>>> b. make
>>>> 2. Build libselinux
>>>> a. cd libselinux (assumes at root of tree)
>>>> b. make ANDROID_HOST=y
>>>>
>>> This works for me.
>>
>> make install DESTDIR=/tmp/someidr mostly works, Mac ln does not support
> --relative so that fails. ANDROID_HOST also needs to be set in the top
> level makefile so that it propagates down:
>> ANDROID_HOST ?= n
>
> Yeah install doesn't work on Mac, that's why for Darwin we just set the
> path to the libsepol location for sefcontext_compile.
>
> As for ANDROID_HOST, why does it need to go higher? It's only used in
> libselinux and is declared and used just like DISABLE_SETRANS...I'm not
> following you?
>

Because I was building from the top, basically seeing if I could get a 
usable toolchain out of it, but that looks like it would require a bit 
more work.

>>
>>>
>>>> This essentially gets us to where a build server/maintainer
>>>> can test patches quickly on mac, with some assurance
>>>> it's not busted without downloading all of Android.
>>>>
>>>> It's still wise to check in an Android tree if possible IMHO.
>>>>
>

Re: [RFC] Build ANDROID_HOST=y on mac

[William Roberts]

[-- Attachment #1: Type: text/plain, Size: 1809 bytes --]

On Sep 28, 2016 17:07, "Joshua Brindle" <brindle@quarksecurity.com> wrote:
>
> William Roberts wrote:
>>
>> On Sep 28, 2016 16:54, "Joshua Brindle"<brindle@quarksecurity.com>
wrote:
>>>
>>> Joshua Brindle wrote:
>>>>
>>>> William Roberts wrote:
>>>>>
>>>>>  From commit 35d702 on
>>>>> https://github.com/williamcroberts/selinux/tree/fix-mac
>>>>>
>>>>> I have a branch that is building on my elcapitan mac, requesting any
>>>>> comments anyone
>>>>> wishes to make, before I send them out.
>>>>>
>>>>> If you wish to test, this is the procedure
>>>>>
>>>>> 1. Build libsepol (assumes at root of tree)
>>>>> a, cd libsepol
>>>>> b. make
>>>>> 2. Build libselinux
>>>>> a. cd libselinux (assumes at root of tree)
>>>>> b. make ANDROID_HOST=y
>>>>>
>>>> This works for me.
>>>
>>>
>>> make install DESTDIR=/tmp/someidr mostly works, Mac ln does not support
>>
>> --relative so that fails. ANDROID_HOST also needs to be set in the top
>> level makefile so that it propagates down:
>>>
>>> ANDROID_HOST ?= n
>>
>>
>> Yeah install doesn't work on Mac, that's why for Darwin we just set the
>> path to the libsepol location for sefcontext_compile.
>>
>> As for ANDROID_HOST, why does it need to go higher? It's only used in
>> libselinux and is declared and used just like DISABLE_SETRANS...I'm not
>> following you?
>>
>
> Because I was building from the top, basically seeing if I could get a
usable toolchain out of it, but that looks like it would require a bit more
work.

Ahh yeah, that won't work yet, but that would be nice.

>
>
>>>
>>>>
>>>>> This essentially gets us to where a build server/maintainer
>>>>> can test patches quickly on mac, with some assurance
>>>>> it's not busted without downloading all of Android.
>>>>>
>>>>> It's still wise to check in an Android tree if possible IMHO.
>>>>>
>>
>

[-- Attachment #2: Type: text/html, Size: 2963 bytes --]

Re: [RFC] Build ANDROID_HOST=y on mac

[Joshua Brindle]

William Roberts wrote:
> On Sep 28, 2016 17:07, "Joshua Brindle"<brindle@quarksecurity.com>  wrote:
>> William Roberts wrote:
>>> On Sep 28, 2016 16:54, "Joshua Brindle"<brindle@quarksecurity.com>
> wrote:
>>>> Joshua Brindle wrote:
>>>>> William Roberts wrote:
>>>>>>    From commit 35d702 on
>>>>>> https://github.com/williamcroberts/selinux/tree/fix-mac
>>>>>>
>>>>>> I have a branch that is building on my elcapitan mac, requesting any
>>>>>> comments anyone
>>>>>> wishes to make, before I send them out.
>>>>>>
>>>>>> If you wish to test, this is the procedure
>>>>>>
>>>>>> 1. Build libsepol (assumes at root of tree)
>>>>>> a, cd libsepol
>>>>>> b. make
>>>>>> 2. Build libselinux
>>>>>> a. cd libselinux (assumes at root of tree)
>>>>>> b. make ANDROID_HOST=y
>>>>>>
>>>>> This works for me.
>>>>
>>>> make install DESTDIR=/tmp/someidr mostly works, Mac ln does not support
>>> --relative so that fails. ANDROID_HOST also needs to be set in the top
>>> level makefile so that it propagates down:
>>>> ANDROID_HOST ?= n
>>>
>>> Yeah install doesn't work on Mac, that's why for Darwin we just set the
>>> path to the libsepol location for sefcontext_compile.
>>>
>>> As for ANDROID_HOST, why does it need to go higher? It's only used in
>>> libselinux and is declared and used just like DISABLE_SETRANS...I'm not
>>> following you?
>>>
>> Because I was building from the top, basically seeing if I could get a
> usable toolchain out of it, but that looks like it would require a bit more
> work.
>
> Ahh yeah, that won't work yet, but that would be nice.

I got checkpolicy and checkmodule out. No cilc stuff and dispol/dismod 
needs byteswap.h and endian.h.

>
>>
>>>>>> This essentially gets us to where a build server/maintainer
>>>>>> can test patches quickly on mac, with some assurance
>>>>>> it's not busted without downloading all of Android.
>>>>>>
>>>>>> It's still wise to check in an Android tree if possible IMHO.
>>>>>>
>