On Fri, 17 Oct 2003, David Fox wrote:

>I think that the wisest approach is, instead of suggesting a kernel 
>module to the XFree86 folks, you do two things.  First, suggest a kernel 
>module to the Linux folks that implements a protocol for accessing the 
>resource you are trying to use.  Then you go to  the XFree86 folks and 
>suggest a module to utilize that protocol in the X server.

That's not any better.  Unless someone comes up with a real 
problem that isn't just theoretical, and a real solution which 
requires or will work best being in kernel, and then implements 
it, and puts their money where their mouth is and proves that the 
solution not only works, but solves a real problem / really does 
improve performance, etc. they're not going to be taken 
seriously.

You just aren't going to get anywhere with random hypothetical 
guesses as to what will be a magic performance boost until you 
crack out the tools to find performance hits, and write the code 
to fix them.  If that can be done in XFree86 itself - great.  If 
it can be done in the kernel, fine.  If it can be done in XFree86 
without requiring the kernel, and done as well or better in 
XFree86, even better, as then reliance on a random OS's kernel 
module is avoided.

The kernel folk are going to tell you the same thing - that you 
dont just go and implement random code in the kernel and hope it 
fixes something.  You find a problem, and you investigate 
potential solutions to it.  If that involves the kernel - fine.  
Then you come to both the X developers and the kernel developers 
(of the particular OS kernel) and say something to the effect of:

"My following attached patch that implements foo in the Linux (or
BSD or whatever)  kernel, and an appropriate patch for XFree86
which uses this functionality optionally is attached.  I've done 
performance tests on <foobedoo> in X and have determined it isn't 
possible to improve the performance of foobedoo without an 
additional kernel interface.  My kernel interface implements 
this, and does so as cleanly as I could devise. The performance 
gain in fooblahblah applications is n% so this is definitely 
worth it and not just a negligible gain.  I'm interested on 
hearing what other developers think of the patch I have created, 
and what feedback they can give so I can improve it, or try 
alternate methods."

Or something to that effect.  As I said before, making random 
suggestions to use kernel modules abstractly like it's a godsend 
for performance isn't going to get anyone anywhere, wether their 
intentions are extremely wll intentioned or not.

So to take the whole topic out of the pie-in-the-sky land, and 
put it on the concrete ground:

What _specific_ area of XFree86 performance are you (or anyone 
else) thinking needs improvement, what solutions have you 
investigated or even thought about which could improve this 
performance by modifying XFree86 itself, a driver, Mesa, or other 
userland code?  If you do think the kernel might help for this 
problem, what steps have you taken to determine if that is truely 
reasonable, and have you tested your theory?  Have you discussed 
that one small idea with other developers to see what they think 
about the alleged problem, wether it even really is a problem at 
all, how important it is, what other solutions there might be, 
etc. etc. etc.

All of this "lets stuff things in the kernel, because kernel code 
is automatically 20000 times faster right?" stuff gets boring 
fast.

Show me the code.

-- 
Mike A. Harris

_______________________________________________
Devel mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/devel

Reply via email to