Originally posted by gens
View Post
Announcement
Collapse
No announcement yet.
"PulseVideo" Coming To Complement PulseAudio?
Collapse
X
-
-
Originally posted by gens View Postand thats why you have a protocol
client -> server: "i wrote the first period"
Originally posted by gens View Postmemfd locks can only be set once
Originally posted by gens View Postway to use them in this would be to create a memfd, seal it then send it and then free it
and so on for every 4k of sent data (it has to be 4k)
it can be of any size, here it will be one frame, i.e. 30 kdbus messages per second or so
Comment
-
Originally posted by gens View Postonly if PA runs under the same UID as the program
that is not how distributions pack it
and there is no security reason for it (app can only write to the buffer, regardless if the UID is same or not)
Comment
-
Originally posted by Ancurio View Post
Can you elaborate on what you mean by "this would trash the cpu cache"? From my limited understanding, I would assume that the server or client constantly allocating and freeing same-sized memfd's would result in the allocator reusing the same memory regions over and over again (same as an OpenGL driver would during repeated buffer orphaning).
in the meantime the client can make another one and another one and so on
a simple ring buffer is also better since you don't have to even make a system call to write a msg
(other then to send a msg over a socket, that you have to do anyway)
allocating memory is not that simple, i looked into it and there are still plenty things the kernel MM does i don't know about or am sure how they work
Comment
-
Originally posted by pal666 View Postclient truncate mapping and crash server
which is on purpose, because it makes crashes impossible
did i told you that you are imbecile ?
it can be of any size, here it will be one frame, i.e. 30 kdbus messages per second or so
that is 4k on x86 under linux
(see man 2 mmap)
PS preventing a shm file being truncated is easy
edit:
it doesn't even matter since it won't reflect on the servers fd
wow, that you made me think is a new one
il have to play with truncating nowLast edited by gens; 30 June 2015, 03:45 PM.
Comment
-
Originally posted by gens View Posti already expanded on that in that i dont like having prejudice against people
PS this was the context, of which MoonMoon has taken the statement out of
Originally posted by gens View Postand i already stated that i would apologize if my prediction of this being as bad as PA was proven false
as a man should, because someone who never admits of being wrong will never learn
Comment
-
Originally posted by gens View Posti would really like to see some code from you, please
it is to satisfy my curiosity in you programming knowledge
you will get to see the project im working on in my free time, that i offer in return
i don't usually ask for this nor like asking it, but you 3 have just brought insults to this thread
i am working on code in my non-free time, so i can't show it to random internet morons
but you could do your promised apology and stop making fool of yourself as first comment in future thread
Comment
-
Originally posted by gens View Post
shared memory can NOT be smaller then a page
you tried to argue that it can't be more than one page( "and so on for every 4k of sent data (it has to be 4k)" ), now you are trying to make some cheap excuse
but it can be of any size, just wasting part of last page.
Comment
-
Originally posted by psychoticmeow View PostWhy do you seem to hate so badly everything Poettering has been involved in and resort to weak "technical" arguments that you refuse to acknowledge have been refuted?
I see. We should show you our code because that has what bearing exactly on the quality of your posting?
you didn't provide anything technical, just insults and out of context quotes
its kind of hard to estimate if you know what you are insulting about when you don't actually say anything concrete
you three need to learn that criticism is a Good Thing
and stop crying about it
Comment
Comment