Fixing devPrivates

Tiago Vignatti tiago.vignatti at nokia.com
Thu Apr 29 07:34:47 PDT 2010


Hi,

On Thu, Apr 29, 2010 at 08:59:06AM +0200, ext Keith Packard wrote:
> 
> What's the problem with devPrivates?

Reviewing your proposal made me think if we really need devPrivates mechanism
at all.

It only exists to not change ABI all the time on data structures. But hey, is
this a _real_ problem? I mean, for who cares about ABI, we have a way to track
the control changes just going to xf86Module.h and bumping ABI_*. 

There's also the readability of the code which is really really really
terrifying using such mechanism.


So why do not remove entirely devPrivates?


> What I don't have yet is any kind of memory usage or performance
> comparison with the current devPrivates implementation. I should
> probably do that just to document the benefit in more concrete terms,
> but I'd rather not bother and just get this merged fairly soon.

I started one standalone Xorg and here are the results:

- with current devPrivates: 5008 kB
- with new proposed devPrivates: 5032 kB


I found pretty weird because I was waiting to decrease or, in the worst case,
keep the same amount of memory used with your proposed work. Anyway, that's a
gross measurement, just calling a standalone Xorg, i.e. without any client or
whatever interacting. Maybe all those data structures keeping devPrivates will
get swapped to physical memory only when actually some interactions happen? I
don't know.


Cheers,

                            Tiago


More information about the xorg-devel mailing list