Hi,
I want to ask, if you have any guide or documentation, how to write a real time app on l4linux. As you know, l4llinux doesn't guarantee whether the application .which run on top of it, will be close to real time or not. In this case, should we write an separate l4 app and communicate with l4linux over IPC? If yes, again, how can I implement this kind of an app. For instance, if we take the server-client app under examples folder, can we apply the same scenario? How can I communicate with L4Linux?
Sincerely,
Cem Akpolat
Hi,
On Tue Dec 08, 2015 at 14:18:17 +0100, cem akpolat wrote:
I want to ask, if you have any guide or documentation, how to write a real time app on l4linux. As you know, l4llinux doesn't guarantee whether the application .which run on top of it, will be close to real time or not. In this case, should we write an separate l4 app and communicate with l4linux over IPC? If yes, again, how can I implement this kind of an app. For instance, if we take the server-client app under examples folder, can we apply the same scenario? How can I communicate with L4Linux?
L4Linux is as real-time as any other virtualized Linux is, so yes, you probably want to have real-time work separate from Linux. L4Linux has a server interface which can be used to implement services. Look at arch/l4/server for examples. It's a bit more complicated because of the mixture of C and C++ but look at the input-srv file as the most simple example.
Adam
Hi Adam,
with the real time I was talking about the short response times such as if an operation takes in linux 1 second, my expectation from microkernel, is to have less then 1 second. Especially, for the embedded system, where the microkernel can undertake a critical role, the duration of response time is extremely important. If you mean, the example will be running separately will give the same performance, then it is needless to implement it and in this case my logic is totally wrong! Because with the real time, i was comparing l4linux with rtlinux.
Thanks for your response!
Cem Akpolat
On Thu, Dec 10, 2015 at 11:18 PM, Adam Lackorzynski < adam@os.inf.tu-dresden.de> wrote:
Hi,
On Tue Dec 08, 2015 at 14:18:17 +0100, cem akpolat wrote:
I want to ask, if you have any guide or documentation, how to write a
real
time app on l4linux. As you know, l4llinux doesn't guarantee whether the application .which run on top of it, will be close to real time or not.
In
this case, should we write an separate l4 app and communicate with
l4linux
over IPC? If yes, again, how can I implement this kind of an app. For instance, if we take the server-client app under examples folder, can we apply the same scenario? How can I communicate with L4Linux?
L4Linux is as real-time as any other virtualized Linux is, so yes, you probably want to have real-time work separate from Linux. L4Linux has a server interface which can be used to implement services. Look at arch/l4/server for examples. It's a bit more complicated because of the mixture of C and C++ but look at the input-srv file as the most simple example.
Adam
Adam adam@os.inf.tu-dresden.de Lackorzynski http://os.inf.tu-dresden.de/~adam/
l4-hackers mailing list l4-hackers@os.inf.tu-dresden.de http://os.inf.tu-dresden.de/mailman/listinfo/l4-hackers
Hi,
On Fri Dec 11, 2015 at 00:26:51 +0100, cem akpolat wrote:
with the real time I was talking about the short response times such as if an operation takes in linux 1 second, my expectation from microkernel, is to have less then 1 second. Especially, for the embedded system, where the microkernel can undertake a critical role, the duration of response time is extremely important. If you mean, the example will be running separately will give the same performance, then it is needless to implement it and in this case my logic is totally wrong! Because with the real time, i was comparing l4linux with rtlinux.
There's probably some misunderstanding. If some piece of work needs 1 second to do something then it needs this second, no matter where it runs. If you want it to complete faster then you have to work on the algorithm or use faster hardware or similar approaches. When you run some work on the microkernel directly you have more control over it's execution behavior than you typically have in Linux but a smaller kernel does not make your code run faster!
Adam
l4-hackers@os.inf.tu-dresden.de