Re: Aureal Vortex issues

Re: Aureal Vortex issues

[Manuel Jander]

Hi Maarten,

On Sun, 2005-02-06 at 23:19 +0100, Alien wrote:
> well, i finally got it working...
> 
> it seems that since you changed the mmio address to a pointer to unsigned int, 
> made the hwwrite wrong since it did a >>2 (on my 64bit platform that should 
> be a >>3 ofcourse... i disabled some other code and it works

AFAIK shifting byte offsets to make them dword offsets should just work
on any platform. Its just plain PCI memory mapped I/O after all. By the
way, the addresses defined in the header files are byte addresses.

Have you tried a newer kernel ? Try a vanilla kernel with minimal patch
contaminants if possible. 
I read that post regarding your emu10k1 oopsing too. AFAIK the emu10k1
uses port i/o instead of mmio. That makes me think that if this oopses
are linked somehow, the problem is either a widespread common mistake or
somewhere deeper in the PCI i/o layer of the kernel. 

> i'll give you the changed files, you do with it what you want:
> 
> not that it's better to define the mmio as a void* so that additions involving 
> that do not have to be >>2 or >>3 or whatever.
> 
> i disabled the a3d routines and the vortex3D (since vortex3D was already 
> disabled with an #if 0 ...

What "#if 0" do you mean ? In the current Alsa version in kernel 2.6.9
the 3D sources are accesible and working for au8830's through the "a3d"
PCM device.
The 3D support code is very well tested. It is not considered unstable.
The problem is that there is no library taking advantage of it (and
coordinate mapping functions are missing).

Anyway, it makes me glad to know that you got working. I'll take at a
look at your code changes ASAP and commit possible patches to the ALSA
list, with credits to you of course.

Best Regards, thanks.
-- 
Manuel Jander
Electronic Engineer



-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: Re: Aureal Vortex issues

[Alien]

[-- Attachment #1: Type: text/plain, Size: 3414 bytes --]

Op maandag 7 februari 2005 18:55, schreef Manuel Jander:
> Hi Maarten,
>
> On Sun, 2005-02-06 at 23:19 +0100, Alien wrote:
> > well, i finally got it working...
> >
> > it seems that since you changed the mmio address to a pointer to unsigned
> > int, made the hwwrite wrong since it did a >>2 (on my 64bit platform that
> > should be a >>3 ofcourse... i disabled some other code and it works
>
> AFAIK shifting byte offsets to make them dword offsets should just work
> on any platform. Its just plain PCI memory mapped I/O after all. By the
> way, the addresses defined in the header files are byte addresses.
>
> Have you tried a newer kernel ? Try a vanilla kernel with minimal patch
> contaminants if possible.
> I read that post regarding your emu10k1 oopsing too. AFAIK the emu10k1
> uses port i/o instead of mmio. That makes me think that if this oopses
> are linked somehow, the problem is either a widespread common mistake or
> somewhere deeper in the PCI i/o layer of the kernel.

those are two different oopses, the last one happens with with the snd-emu10k1 
as well as the snd-au8810 because it's the snd-hwdep that oopses, and that 
happens when only when i get rid of all debug stuff (something about bad 
scheduling...?)

> > i'll give you the changed files, you do with it what you want:
> >
> > not that it's better to define the mmio as a void* so that additions
> > involving that do not have to be >>2 or >>3 or whatever.
> >
> > i disabled the a3d routines and the vortex3D (since vortex3D was already
> > disabled with an #if 0 ...
>
> What "#if 0" do you mean ? In the current Alsa version in kernel 2.6.9
> the 3D sources are accesible and working for au8830's through the "a3d"
> PCM device.
> The 3D support code is very well tested. It is not considered unstable.
> The problem is that there is no library taking advantage of it (and
> coordinate mapping functions are missing).

i'm using alsa-1.0.8

the vortex_vort3D implementations are within an #if 0 so i commented the calls 
to those functions as well. additionally, there is one a3d function that has 
problems, so i commented that function call to it as well...

these _fixes_ is not what you should commit.

the mmio type i have changed to "void __iomem *" from "unsigned long __iomem 
*", since if you'd to mmio+1 it would really mean mmio+sizeof(unsigned long) 
which it countertargetted by having the addition in the hwwrite >>2 (only 
works if unsigned long is 32bits of course ... those two changes, you can 
commit.

there could be more problems related to 64bit, but i only have one machine and 
it's a server, so i can't really afford to test on it, with kernel oopses 
etc...

> Anyway, it makes me glad to know that you got working. I'll take at a
> look at your code changes ASAP and commit possible patches to the ALSA
> list, with credits to you of course.
>
> Best Regards, thanks.

the mmio typecast and changing of the hwwrite and hwread macro is the 
important part, the a3d function i commented somewhere in the init has some 
64bit issues, i'd be happy too look at it, but i can't really test, unless 
i'm sure it'll work.

best regards

PS: the most unfortunate thing is that i do not have the time to get into this 
deeper, i have lotsa other things that need to be programmed.
-- 
Alien is my name and head-biting is my game.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]