Dear Rong,
I have gone through your implementation of the binder driver and your
approach is very neat.
However, I am curious to know why you chose to do a completely new
implementation?
Why didnt you replace the binder_lock in the old implementation itself? You
mentioned that it is impossible
Hi Rong,
Do you have any numbers which show that the below implementation is
better than the current implementation?
Regards,
Seenu.
On Wednesday, 25 January 2012 19:11:20 UTC+5:30, Rong wrote:
>
> Hi Folks,
>
> I've just finished a fresh implementation of the android binder
> driver, would
Hi Rong,
I started leaning binder driver (and protocol) and I came accoss your
post and tried to use client/server test on a regular driver on a 3.4 kernel
I have 3 issues to discuss.
First question is connected to your RFC , I do not understand why binder
uses global lock, I see it prote
On Wednesday, April 11, 2012 9:54:01 PM UTC-4, Brooke wrote:
>
> This may seem naive, but what exactly does /dev/binder do?
It's the device file for accessing a kernel driver which provides a special
inter-process communications scheme optimized for use by object-oriented
languages and where e
Hi,
This may seem naive, but what exactly does /dev/binder do? I have some
Android code that I'm compiling off target that used to complain about
/dev/binder and now it fails since there is no /dev/binder on teh Linux PC.
Question I have is whats it doing, why is it needed, and why just Android
Thanks, will definitely do it. About deadlock, I do little digging Its seem
caused when binder holding mmap_sem sometimes never release it, don't know
how, I don't dig much futher.
--
unsubscribe: android-kernel+unsubscr...@googlegroups.com
website: http://groups.google.com/group/android-kernel
On Feb 10, 12:01 am, dh wrote:
> Hi,
>
> > I had thought about before too. From the implementation point of view,
> > you need someone to send you either a binder or handler (like
> > invitation) to be able to create a kernel level reference (numbered
> > locally), which you can then use to star
Yes, I'd like to know something about the security of the binder
implementation.
Can I just forge an IPC by talking to /dev/binder?
-Earlence
On Feb 9, 2:01 pm, dh wrote:
> Hi,
>
> > I had thought about before too. From the implementation point of view,
> > you need someone to send you either a
Hi,
> I had thought about before too. From the implementation point of view,
> you need someone to send you either a binder or handler (like
> invitation) to be able to create a kernel level reference (numbered
> locally), which you can then use to start transaction with. As the
> reference n
Thanks for the encouraging. It's indeed quite hard to get a big
picture on how it works as there are layers of encapsulation on it,
and no really up-to-date documents can you refer to. If you have any
questions on the existing binder driver, you could try me, not
guaranteed to answer but certainly
> From a security pov, is it possible to guess binder reference numbers of
> binder nodes, thus by-passing the service manager to initiate IPC
> 'directly' from one application to another? Is that possible in the
> traditional Binder implementation? If yes, did you regard this in your
> project?
Hi,
sounds like a great piece of work. I'm not into the details of that
low-level kernel stuff, but there's one question that sticks around in my
mind:
>From a security pov, is it possible to guess binder reference numbers of
binder nodes, thus by-passing the service manager to initiate IPC
'
Very interesting, I have deadlock problem with current binder
implementation on mainline kernel (3.3-rc1), I look into the codes and it
quite messy. Trying to find alternative implementation and found yours, its
so exciting that someone finally working on it. Hopefull it can become full
binder
On Jan 26, 5:05 am, Tim Bird wrote:
> On 01/25/2012 05:41 AM, rong wrote:
>
> > Hi Folks,
>
> > I've just finished a fresh implementation of the android binder
> > driver, would love to see some suggestions or comments on the code as
> > well as the whole binder IPC idea.
>
> This is really inte
14 matches
Mail list logo
