Re: [PATCH 1/6] staging: android: binder: Remove some funny && usage

From: Arve Hjønnevåg
Date: Fri Jun 19 2009 - 20:21:47 EST


On Fri, Jun 19, 2009 at 3:53 PM, Daniel Walker<dwalker@xxxxxxxxxx> wrote:
> On Fri, 2009-06-19 at 12:20 -0700, Brian Swetland wrote:
>> 2009/6/17 Daniel Walker <dwalker@xxxxxxxxxx>:
>> > On Wed, 2009-06-17 at 14:26 -0700, Arve Hjønnevåg wrote:
>> >> On Wed, Jun 17, 2009 at 9:31 AM, Jeremy Fitzhardinge<jeremy@xxxxxxxx> wrote:
>> >> > On 06/17/09 09:08, Daniel Walker wrote:
>> >> ...
>> >> > Also, what its usermode ABI is, how stable it is, whether its generally
>> >> > useful, does it have glibc/other library support, etc.  Would you ever want
>> >> > to use this in a non-Android context?
>> >>
>> >> You could use this in a non-android context, but the abi is not
>> >> stable. There is some documentaion of the current user space api at
>> >> http://developer.android.com/reference/android/os/IBinder.html. You
>> >> can also find more information at http://www.open-binder.org/ which is
>> >> where the api came from.
>> >
>> > Why does all this need to be done in the kernel? Couldn't any of the
>> > current IPC mechanisms be re-used to accomplish this?
>>
>> Arve can probably go into more detail here, but I believe the two
>> notable properties of the binder that are not present in existing IPC
>> mechanisms in the kernel (that I'm aware of) are:
>> - avoiding copies by having the kernel copy from the writer into a
>> ring buffer in the reader's address space directly (allocating space
>> if necessary)
>
> This sounds like a performance speed up ..
>
>> - managing the lifespan of proxied remoted userspace objects that can
>> be shared and passed between processes (upon which the userspace
>> binder library builds its remote reference counting model)
>
> The "managing the lifespan" sounds very much like part of the
> description for DBus ..  I think the main competing interface would be
> DBus. I know it's used in the software for the OpenMoko phone , and I
> think the Palm Pre uses it too.
>
> Did Google evaluate DBus at all?

Some of our user-space code have in the past used or still use dbus,
but as far as I know it has not been seriously considered as a
replacement for the binder.

> Also are there any userspace test cases
> that Google used to test the performance of this interface. Or test
> cases to compare the binder with something like sockets, or any other
> type of IPC?
>
> If Google believes the binder is the right solution for IPC, how was
> that conclusion formed?
>
> Daniel

These are mostly questions for the framework team. The binder driver
is there to support our user space code. At some point we used the
driver from www.open-binder.org, but we ran into, and fixed, a lot of
bugs (especially when processes died), so we determined it would be
faster to rewrite the driver from scratch.

--
Arve Hjønnevåg
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/