On Fri, 28 Mar 2003, Joshua Andrews wrote:

>I am curious about the direction of the Red Hat kernel code and 
>incompatibility with xfree86-kernel code.

Care to elaborate on that?  What incompatibility?

>Is Red Hat going to merge it's kernel drm into it's xfree86?

That makes no sense.  Kernel DRM is kernel source code for kernel 
modules.  All Red Hat kernel modules are provided by the Red Hat 
kernel package, and that is intentional and not likely to change 
ever due to the support nightmares caused by the exponential 
matrix of complexity that would result with zero gain.


>Is Red Hat planning on merging "GATOS" into it's xfree86 for
>those of us with ATI video cards who have depended on ati.2?

No.  I _personally_ have an interest in GATOS, but my interest is 
more in "the functionality" that GATOS provides rather than the 
exact source code of the GATOS project itself.

The licensing of the GATOS code for quite some time now has been 
confusing at best.  The website mentioned GPL in a few places, 
but didn't really say _what_ was GPL.  The sourceforge site says 
"Other/proprietary" and the actual source code doesn't contain 
any document at all as to what license the code is under.  As 
such, while I have compiled and used GATOS a fair number of times 
in the last couple of years, I haven't done much more than that 
with it as I assumed the license to be GPL and/or something else 
not compatible with the MIT license X is licensed under.

Very recently (a couple weeks ago actually), someone came to me 
in IRC, and within a few hours all of the licensing issues were 
clarified by the project maintainer.  GATOS video driver code is 
under the same MIT license that XFree86 is under, and that 
removed one wall that stood in the way of me delving any deeper.

Since then, I have investigated what all is involved in patching 
GATOS into XFree86 4.3.0.  In short, it would be a _lot_ of work, 
and would essentially be forking the driver(s) in a major way 
unless the code were accepted directly into XFree86 CVS.  The 
only way the code will get into XFree86 CVS is if Kevin Martin 
checks it into CVS, or if it gets into DRI CVS and later makes it 
into XFree86 CVS.

Both Kevin Martin, and the DRI project also believe it to be a 
large task to do this and do it properly, and there are also some 
technical issues that prevent it from being a no brainer to do.  

In order for this to happen, it will require someone to begin 
looking into all of the technical details and open up discussion, 
solve the problems that exist, and then begin integration into 
either DRI-CVS or XFree86 CVS.  That requires either:

1) A volunteer to dedicate their own personal free time to do so

or

2) Someone being paid to do it, and it still requires acceptance 
   of the relevant X developers in order to be completed.


>Is Red Hat counting on the large number of users with ATI multimedia 
>video cards to push the development of third party, redhat9-specific 
>ati.2 drivers?

No.  When it comes to what Red Hat is doing with GATOS, I am the
one who decides that.  Red Hat is not a video card vendor, and
does not make money off of the sale of video cards.  Red Hat is
also not a video driver house.  The video support present in Red
Hat Linux is what is provided by XFree86.org, and is enhanced by
patches that I've written and that others have written that fix
bugs and some of which add features or support for new hardware.

GATOS is not a small patch.  There is also no way that we are 
going to ship 2 drivers for the same piece of hardware and then 
handle all of the technical, logistical, and support problems 
that that incurs as well as modifying our tools to handle 
multiple drivers per card.

GATOS requires special DRM modules, which would mean that we
would have to either ship XFree86's driver, or ship GATOS's 
driver, or ship both and have a big complex mess to deal with.  
It just is not feasible by any far stretch of the imagination.

The proper thing to do, is for GATOS, or parts of GATOS to be 
integrated directly into XFree86 or DRI-CVS, and make it into a 
proper XFree86 release.  Until someone has the time to do this 
work, and for it to be accepted into XFree86, GATOS will not be a 
part of Red Hat Linux.  If we were to ship GATOS at this point in 
any manner - we then would be taking on the responsibility of 
maintaining and supporting that, and we would receive zero help 
from XFree86.org or elsewhere except perhaps from the GATOS 
maintainer himself.

XFree86 is a huge complicated piece of code, and pulling code 
from 19 different CVS repositories is not good use of my time.

I have produced a unified diff of GATOS to apply to our sources, 
but there are parts of the GATOS code that contain gratuitous 
whitespace and/or formatting changes, and there are parts of the 
code which overlap other patches we are applying.  Also, the 
GATOS code backs out some bug fixes that are present in XFree86 I 
noticed, and it is possible it backs out other bug fixes that I 
didn't happen to notice in the 500Kb patch.

I do want to see the GATOS functionality incorporated into Red 
Hat Linux, and I will poke at it in my rather limited spare time, 
however before I even begin to consider adding the majority of 
functionality into our XFree86 packaging, it will require some 
level of commitment from another developer to help get it into 
shape and put it into upstream CVS as there's no way I'm 
maintaining all of that (mostly unknown) code myself.

>Will I be able to run redhat9 with a 2.4.18xx kernel so that I
>can enjoy the mutimedia functionality of my ATI video card?

Not unless you download the GATOS drivers and add them to your 
system yourself.

If this upsets you, then join the [EMAIL PROTECTED] mailing list
and complain about it there.  I'm doing my part (or I will be
before long) to attempt to get the ball rolling on GATOS, but to
date, the relevant people that need to be involved to get things
going are doing so as unpaid volunteers, and that includes me.  I 
can't see any viable profitable business reason for Red Hat to 
pay me to participate in this.  As such, my efforts are volunteer 
driven on my own personal weekend time, and aren't likely to 
occur for months at best.

Another thing that is somewhat of a roadblock for me to work on 
this, is that it is something that needs to be done very 
carefully and in chunks.  It isn't clear yet how easy it will be 
to break it down into chunks, and it'll require someone with CVS 
access to commit the time to review and commit things to CVS in a 
timely manner.  I'm not about to work on a patch for 2 days and 
submit it to XFree86.org for it to sit in the patch queue for 9 
months and end up irrelevant.

While my response may be long winded, I hope it answers every 
possible question that anyone might have about GATOS and it's 
place in Red Hat Linux.  For those of you who prefer the Reader's 
Digest condensed version:  "No, and it wont happen until GATOS is 
included in XFree86 CVS".

<snicker>



-- 
Mike A. Harris     ftp://people.redhat.com/mharris
OS Systems Engineer - XFree86 maintainer - Red Hat



-- 
Phoebe-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/phoebe-list

Reply via email to