Subject: Re: Yellow Dog Linux 2.1 and Power Mac G4 w/Radeon - Absolutenewbie
From: Juan M. Palacios (jmpalacios@mac.com)
Date: Wed Jan 16 2002 - 09:49:28 MST
Hello Aram, sorry for taking so long to write back but I've also been very full
of work lately. University
subjects brought many christmas presents along, and only is now when I starting
to feel the heat!
I don't know if by now you already solved the problem, so just in case here
go my thoughts. I'm going to have to disappoint you a bit but the error you
are experiencing, as stated by dmesg, has me a bit lost as well. From the
"dmesg"output it seems that radeon driver was compiled into the kernel rather
than as a
module in the stock release (anyone please correct me if I'm wrong!!) but, as
you very wisely realized,
it's having problem doing what it supposed to do. Coincidentally I'm having a
similar problem and haven't been able to fix it yet. The video card inside
your mac works over an agp port, just as the one in my PowerBook, so my
thoughts are that the kernel is having trouble handling that port. And that
would be why the radeaon driver cannot reserve the "fb" (which stands for
"frame buffer") region on the card.
With this scenario you'll never be able to get a fully working and accelerated
X environment over the
radeon driver until you compile a new kernel with proper agp support (either as
module or into the
kernel).
Just the same, with my PowerBook I haven't been able to get the r128 driver
for my video card working because every time I try to load it I get unresolved
symbols that point me towards a module that isn't either properly loaded or
compiled. And this would be the one that provides the agp support. Bare in mind
that I'm guessing in the dark here and this are my conclusions because I've
already ruled out other
options. I must note that, just the same, I'm using the stock kernel in the 2.1
release, so if any one has
more knowledge than me about these similar issues then please illuminate us. Is
the culprit of these
problems the agp support? From what I've tried it seems to me that "agpgart"
was compiled into the
kernel, so why can't the radeon and the r128 driver access the cards properly?
So as for a temporary fix I suggest you to use the "fbdev" option in your
XFree86Config-4 file. This will
get you 32 (which is no more than 24) color bit depth in X at practically any
resolution you want, but it will not give you accelerated video. Ask your Linux
friend to set this up for you. If you can't nail it down I'll tell you later
because now I have to go. Please write back to tell me you r experiences and
your findings.
Sincerely,...
Juan.
P.S.: I recently heard that 2.1 didn't have perfect support for some radeon
cards yet, so this may also be
what's bugging you. Again, if there is anyone more informed about this than me
then please speak up.
Thank you all!
Aram Kudurshian wrote:
> Hello Juan,
>
> Thanks again for your e-mail; I'm here with my Linux friend and we've gone
> to go through what you've suggested. However, it seems there is something
> 'funky' with the driver as my techno geek friend will explain:
>
> I've examined the dmesg output and I found a line that explains the whole
> situation, or at least most of it:
> "radeonfb: cannot reserve FB region"
> So it seems like the driver loads but cannot take control of the card. This
> is all using the kernel included with YDL 2.1 (2.4.10-r1) as I didn't have
> time to try the 2.4.18-pre3 version.
>
> Well, any help would be greatly appreciate and thanks again,
>
> Aram K.
>
> > Hello Aram. So I'm going to make the assumption that your friend
> > knew what he/she was doing and got you a newly home brewed, compiled and
> > working kernel. Were the radeon drivers compiled INTO the kernel or as
> > MODULES? I they were compiled as modules you have to make sure they are
> > being loaded as needed (most probably upon boot up). So probably a line
> > for this would have to included in a magical file called modules.conf
> > located at the directory /etc. So the full path to the file would be
> > /etc/modules.conf. It would have to be edited to include the line that
> > would load it upon boot up under an aliased name. But unfortunately I
> > don't know the alias for that module since I don't use it, maybe someone
> > else on this list can help us with that by telling us if it DOES need an
> > alias in modules.conf in the first place, and if so which one.
> >
> > If the radeon support was compiled into the kernel then it does not
> > need anything special to load, it comes with the kernel. So, I move onto
> > the next step.
> >
> > Lets suppose that either the support was built into the kernel or
> > that someone provides us with the correct method of loading the module,
> > now we have to review your booting arguments. First you could double
> > check if you are running the kernel you built by issuing the command
> > "uname -r" which returns the active kernel version, and it should match
> > the one you built. If not then go to /boot and make the proper symlinks
> > so that yaboot can find the correct kernel image (we'll go over this if
> > "uname -r" does not match the kernel release you think you're running).
> > On to your boot arguments. They have to be changed a bit to either one
> > of the following:
> >
> > append="video=radeonfb:vmode:x@y,cmode:32"
> > video=radeonfb:vmode:x@y,cmode:32
> > append="video=radeonfb:vmode:x@y,32"
> > video=radeonfb:vmode:x@y,32
> >
> > Either one should work, but that is the correct syntax and not the
> > one you were using. The "x" stands for the number of the proper video
> > mode for your screen, and that is found by looking at the output of the
> > command "man vmode". You'll find a number of modes that stand for
> > specific resolutions at specific refresh rates, so replacing the "x"
> > with the mode number that most suits your case should give you quality
> > video. The "y" WOULD stand for the refresh rate but I seriously doubt
> > it'll be needed at all, since each vmode includes a specific rate. So
> > the proper number will force the adequate rate. Specifying the "y" with
> > the same rate that a specific mode provides would only be redundant, and
> > giving a different value will most probably freak your kernel (but I'm
> > only guessing with this last assertion, maybe it just overrides). I
> > think we'll be much better off by finding the correct vmode and ditching
> > the "@y" altogether.
> >
> > So, give that a try and let me know how it goes. I take for granted
> > that you know that those arguments have to be entered into
> > /etc/yaboot.conf, and that the file has to be edited by root. If it
> > doesn't work then tell me exactly what you did and the error output.
> >
> > Have fun and enjoy your Linux learning. Regards,...
> >
> >
> > Juan.
> >
> > On Thursday, January 10, 2002, at 04:48 PM, Aram Kudurshian wrote:
> >
> >> Hello Juan,
> >>
> >> Now that I've reread what I wrote, I can actually see how little info it
> >> gave. I've experimented a bit more but it hasn't fixed anything (nor
> >> changed anything it seems). Here is what I did since I installed YDL.
> >>
> >> 1) Used the 'make menuconfig' command to enable the support for Radeon
> >> drivers + turned on SMP.
> >> 2) Rebuilt the kernel (?) and enabled it (this was done with the aid
> >> of a
> >> friend so it worked out fine)
> >> 3) Made a new yaboot set (is this how you call it) that used my new
> >> kernel.
> >> The code in this set is the same as the normal one (I just copied it
> >> out)
> >> except with the addition of this line:
> >>
> >> (tab)append="video=radeonfb:1600x1200-32@75"
> >>
> >> 4) Saved the yaboot.config file and enabled it with the ybin command
> >> (again
> >> I was under direction).
> >>
> >> This is as far as I've gotten. I am not sure what you mean by the
> >> XFree86config-4 file? Is this similar to the menuconfig system?
> >>
> >> Here is the information from the 'lsmod' command:
> >>
> >> Module Size
> >> printer 7056
> >>
> >> OK, I hope this helps you narrow down the problem and sorry for the
> >> delay,
> >> I've been busy w/ school work all day long!
> >>
> >> Aram
>
> ____________________________________________________________________________
> Aram Kudurshian aramk@mac.com
> Student/Programmer http://homepage.mac.com/aramk/
> AIM: aramk622 http://www.htmlcreator.net
This archive was generated by hypermail 2a24 : Thu Jan 17 2002 - 13:51:17 MST