gatos-devel Mailing List for GATOS

On Sat, 30 Nov 2002, Vladimir Dergachev wrote:

>
>
> On Sat, 30 Nov 2002, Nikolai Zhubr wrote:
>
> > Hi,
> > Friday, 29 November, 2002, 6:58:53, Vladimir Dergachev wrote:
> > > My preference is to work the other way around - have a mathematical
> > > abstraction to which hardware dependent code acts as a client.
> > This is ok, too. But with this approach I might need slightly different
> > abstract functions or maybe additional ones which wouldn't be needed
> > for the main project at all. Would you then reserve some?
>
> My goal is to have something general enough so I can try to persuade v4l
> folks to, at least, include something similar in their code.
>
> Which capabilities do you want ?

I forgot to add that it is mainly the data structures and their access
protocols that I am worried about. With typical field buffers taking up
300K we can well afford to have convenience functions with almost
redundant actions.

So in a sense what I am trying to do is to untangle KM_STRUCT and make it
suitable for more things. For example, I working on a VBI interface - the
necessary code is already in ati.2 CVS (radeons only for now) so km_api
needs modifications to allow several DMA transfer streams, and then, of
course, avview needs extending to test this.

                       best regards

                            Vladimir Dergachev

On Sat, 30 Nov 2002, Nikolai Zhubr wrote:

> Hi,
> Friday, 29 November, 2002, 6:58:53, Vladimir Dergachev wrote:
> > My preference is to work the other way around - have a mathematical
> > abstraction to which hardware dependent code acts as a client.
> This is ok, too. But with this approach I might need slightly different
> abstract functions or maybe additional ones which wouldn't be needed
> for the main project at all. Would you then reserve some?

My goal is to have something general enough so I can try to persuade v4l
folks to, at least, include something similar in their code.

Which capabilities do you want ?

                          Vladimir Dergachev

> --
> Best regards,
>  Nikolai Zhubr
>
> >                         best
>
> >                           Vladimir Dergachev
>
>

Hi,
Friday, 29 November, 2002, 6:58:53, Vladimir Dergachev wrote:
> My preference is to work the other way around - have a mathematical
> abstraction to which hardware dependent code acts as a client.
This is ok, too. But with this approach I might need slightly different
abstract functions or maybe additional ones which wouldn't be needed
for the main project at all. Would you then reserve some?
-- 
Best regards,
 Nikolai Zhubr

>                         best

>                           Vladimir Dergachev

On Sat, 30 Nov 2002, Andrew S. Johnson wrote:

> I finally bit the bullet and upgraded my box to Slackware-current (as of
> 29-Nov-2002) from 8.1.  I upgraded the kernel to 2.4.20, and downgraded
> automake from 1.7 to 1.6.3 for avview.  ALSA is 0.9rc6 and gcc is 3.2.1 for
> both.

AFAIK you don't have to downgrade - you can have both installed at the
same time and instead of automake use automake-1.6.3.

>
> The DRI program in question is tuxracer, which works with XFree86 4.2.1 stock
> drivers, but avview doesn't work.  I have XFree86 4.2.1.1 compiled from CVS
> and patched with the ati.2 drivers from CVS, and avview works.  However, with
> this configuration, in tuxracer I get music, and a blue intro screen, with no
> graphics.  The only key that works is escape, so at least the box isn't
> locking up.
>
> Yes, I have two X11R6 trees in /usr so I can test drivers, and I switch
> between the two using a symlink.  As I am sure many of us have learned,
> mixing kernel dri modules and XFree86 drivers from different trees is a bad
> thing, usually resulting in lockups or reboots when starting X.
>
> Maybe I'm a little too far out on the bleeding edge, but if the GATOS CVS
> drivers are going to get tested, someone has to be there.
>
> Any ideas?

First of all which card do you have ? Secondly, there was some breakage
that is unfixed because one really needs to rewrite the radeon MESA driver
to accomodate changes needed by GATOS for video capture. With major
changes going on in DRI CVS this did not seem reasonable. Or we could wait
for 2.5.x kernels to become stable so we can request memory in a specific
address range (namely with physical addresses greater than 64M).

However this breakage typically manifests in system lockup, not
malfunction of 3d rendering.

What could be happening is that you have software rendering running and
tuxracer simply cannot paint everything fast enough (and the music is
played in another thread).

                            Vladimir Dergachev

>
> Andy Johnson
>
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Get the new Palm Tungsten T
> handheld. Power & Color in a compact size!
> http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> _______________________________________________
> Gatos-devel mailing list
> Gat...@li...
> https://lists.sourceforge.net/lists/listinfo/gatos-devel
>

On Sat, 30 Nov 2002, Roel de Cock wrote:

> Hi,
>
> The GATOS drivers completely hang my SMP Athlon with Hercules 8500DV
> AIW card whenever I try to run glxgears. Glxinfo and XFree86.0.log
> report that the DRI driver loaded ok, but as soon as I start glxgears
> the machine locks up just after displaying the new window's
> frame. Judging by power consumption, at least 1 of the processors is
> fully occupied from that moment (groovy way to measure CPU
> usage, huh? ;-).
>
> I went through the checks in the DRI troubleshooting page and all
> versions seemed ok (DRM-kernel and radeon_drv etc. from Gatos, libGL*
> from Xfree86-Mesa as delivered with Redhat 8). Tried both Gatos
> binaries and home-compiled drivers, single-processor and SMP.

3d acceleration will not work for 8500 cards. Try using DRI CVS for this.

                       best

                         Vladimir Dergachev

>
> Infamous last words in XFree logfile are always:
>
>   (EE) RADEON(0): FIFO timed out, resetting engine...
>
> Running X under strace reveals that, when the machine seems to be
> hanging, the following signals get received by the X process over and
> over, tens of times per second:
>
>   16:10:56 --- SIGALRM (Alarm clock) ---
>   16:10:56 --- SIGALRM (Alarm clock) ---
>   16:10:56 --- SIGALRM (Alarm clock) ---
>   16:10:56 --- SIGALRM (Alarm clock) ---
>   16:10:56 --- SIGIO (I/O possible) ---
>   16:10:56 --- SIGIO (I/O possible) ---
>   16:10:56 --- SIGIO (I/O possible) ---
>   16:10:56 --- SIGALRM (Alarm clock) ---
>   16:10:56 --- SIGALRM (Alarm clock) ---
>   16:10:56 --- SIGALRM (Alarm clock) ---
>   16:10:57 --- SIGALRM (Alarm clock) ---
>   16:10:57 --- SIGALRM (Alarm clock) ---
>   16:10:57 --- SIGALRM (Alarm clock) ---
>   16:10:57 --- SIGALRM (Alarm clock) ---
>
> Etcetera, ad infinitum or reboot.
>
> Finally, recompiling the Radeon driver with RADEON_DEBUG set (this
> requires some tweaking of rogue macro calls) reveals a tiny bit more:
>
> # This is during usual operation:
>   (**) RADEON(0): WaitForIdle (entering): 64 entries, stat=0x00000140
>   (**) RADEON(0): WaitForIdle (entering): 64 entries, stat=0x00000140
>   (**) RADEON(0): WaitForIdle (entering): 48 entries, stat=0x8030c130
> # This is just before and during lockup:
>   (**) RADEON(0): FIFO timed out: 48 entries, stat=0x8030c130
>   (EE) RADEON(0): FIFO timed out, resetting engine...
>   (**) RADEON(0): EngineRestore (32/32)
>   (**) RADEON(0): WaitForIdle (entering): 48 entries, stat=0x8006c130
>   (**) RADEON(0): WaitForIdle (entering): 64 entries, stat=0x00000140
>
> Anyone any ideas what could be the cause? Help much appreciated!
>
> Roel
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Get the new Palm Tungsten T
> handheld. Power & Color in a compact size!
> http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> _______________________________________________
> Gatos-devel mailing list
> Gat...@li...
> https://lists.sourceforge.net/lists/listinfo/gatos-devel
>

I finally bit the bullet and upgraded my box to Slackware-current (as of=20
29-Nov-2002) from 8.1.  I upgraded the kernel to 2.4.20, and downgraded=20
automake from 1.7 to 1.6.3 for avview.  ALSA is 0.9rc6 and gcc is 3.2.1 f=
or=20
both.

The DRI program in question is tuxracer, which works with XFree86 4.2.1 s=
tock=20
drivers, but avview doesn't work.  I have XFree86 4.2.1.1 compiled from C=
VS=20
and patched with the ati.2 drivers from CVS, and avview works.  However, =
with=20
this configuration, in tuxracer I get music, and a blue intro screen, wit=
h no=20
graphics.  The only key that works is escape, so at least the box isn't=20
locking up.

Yes, I have two X11R6 trees in /usr so I can test drivers, and I switch=20
between the two using a symlink.  As I am sure many of us have learned,=20
mixing kernel dri modules and XFree86 drivers from different trees is a b=
ad=20
thing, usually resulting in lockups or reboots when starting X.

Maybe I'm a little too far out on the bleeding edge, but if the GATOS CVS=
=20
drivers are going to get tested, someone has to be there.

Any ideas?

Andy Johnson

Hi,

The GATOS drivers completely hang my SMP Athlon with Hercules 8500DV
AIW card whenever I try to run glxgears. Glxinfo and XFree86.0.log
report that the DRI driver loaded ok, but as soon as I start glxgears
the machine locks up just after displaying the new window's
frame. Judging by power consumption, at least 1 of the processors is
fully occupied from that moment (groovy way to measure CPU 
usage, huh? ;-).

I went through the checks in the DRI troubleshooting page and all
versions seemed ok (DRM-kernel and radeon_drv etc. from Gatos, libGL*
from Xfree86-Mesa as delivered with Redhat 8). Tried both Gatos
binaries and home-compiled drivers, single-processor and SMP.

Infamous last words in XFree logfile are always:

  (EE) RADEON(0): FIFO timed out, resetting engine...

Running X under strace reveals that, when the machine seems to be
hanging, the following signals get received by the X process over and
over, tens of times per second:

  16:10:56 --- SIGALRM (Alarm clock) ---
  16:10:56 --- SIGALRM (Alarm clock) ---
  16:10:56 --- SIGALRM (Alarm clock) ---
  16:10:56 --- SIGALRM (Alarm clock) ---
  16:10:56 --- SIGIO (I/O possible) ---
  16:10:56 --- SIGIO (I/O possible) ---
  16:10:56 --- SIGIO (I/O possible) ---
  16:10:56 --- SIGALRM (Alarm clock) ---
  16:10:56 --- SIGALRM (Alarm clock) ---
  16:10:56 --- SIGALRM (Alarm clock) ---
  16:10:57 --- SIGALRM (Alarm clock) ---
  16:10:57 --- SIGALRM (Alarm clock) ---
  16:10:57 --- SIGALRM (Alarm clock) ---
  16:10:57 --- SIGALRM (Alarm clock) ---

Etcetera, ad infinitum or reboot.

Finally, recompiling the Radeon driver with RADEON_DEBUG set (this
requires some tweaking of rogue macro calls) reveals a tiny bit more:

# This is during usual operation:
  (**) RADEON(0): WaitForIdle (entering): 64 entries, stat=0x00000140
  (**) RADEON(0): WaitForIdle (entering): 64 entries, stat=0x00000140
  (**) RADEON(0): WaitForIdle (entering): 48 entries, stat=0x8030c130
# This is just before and during lockup: 
  (**) RADEON(0): FIFO timed out: 48 entries, stat=0x8030c130
  (EE) RADEON(0): FIFO timed out, resetting engine...
  (**) RADEON(0): EngineRestore (32/32)
  (**) RADEON(0): WaitForIdle (entering): 48 entries, stat=0x8006c130
  (**) RADEON(0): WaitForIdle (entering): 64 entries, stat=0x00000140

Anyone any ideas what could be the cause? Help much appreciated!

Roel

On Sat, 30 Nov 2002, Jeremy Utley wrote:

> Hello!
>
> I'm having some difficulty with the Gatos km driver, and was hoping
> that someone could help me out.  I've got ati.2 loaded and working,
> and xawtv works just fine.  Now, I'm trying to enable the v4l
> capture interface via km.  The km module loads without error, but
> looking in /proc/video/dev shows no v4l devices.  I've got v4l
> support compiled into the kernel, and I've got the following modules
> loaded.
>
> Module                  Size  Used by    Not tainted
> km_drv                 27392   0 (unused)
> km_api                  1704   0 [km_drv]
> radeon                 86264   1
>
>
>
> Does anyone have any ideas on how I can get the km driver to work
> properly?

a) take a look at http://gatos.sf.net/km.php
b) take a look at output of dmesg

                      best

                        Vladimir Dergachev

>
> Thanks for any help you can give me!
>
> Jeremy
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Get the new Palm Tungsten T
> handheld. Power & Color in a compact size!
> http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> _______________________________________________
> Gatos-devel mailing list
> Gat...@li...
> https://lists.sourceforge.net/lists/listinfo/gatos-devel
>

Hello!

I'm having some difficulty with the Gatos km driver, and was hoping
that someone could help me out.  I've got ati.2 loaded and working,
and xawtv works just fine.  Now, I'm trying to enable the v4l
capture interface via km.  The km module loads without error, but
looking in /proc/video/dev shows no v4l devices.  I've got v4l
support compiled into the kernel, and I've got the following modules
loaded.

Module                  Size  Used by    Not tainted
km_drv                 27392   0 (unused)
km_api                  1704   0 [km_drv]
radeon                 86264   1



Does anyone have any ideas on how I can get the km driver to work
properly?

Thanks for any help you can give me!

Jeremy

Nikolai Zhubr wrote:
> Hi,
> Wednesday, 27 November, 2002, 13:44:36, Michal Bukovjan wrote:
> [...]
> 
>>I downloaded and successfully compiled from http://zhubr.tamb.ru/gatos.html
> 
> 
>>the following:
> 
> 
>>mp1e-cvs-20021112.tgz
>>km-cvs-hackB.tgz
> 
> 
> You also need to download mp1e-patch9.tgz and apply it manually.
> (There is instruction and some examples included inside). If any
> of examples fails to start then please send me its output.

OK, applied, tried, and it seems to work fine.

Video fine, audio fine, A/V sync seems to be fine (tried only 5 mins, 
though).

Great work! Finally, I am able to grab video on my Radeon AIW (heck, I 
own it for more than a year now :-/

I hope the mp1e backend gets integrated into avview some day.

Thanks,

Michal

On Fri, 29 Nov 2002, Jim Lill wrote:

>
> what does the avview FREQ value
>

It is MHZ/16

                 Vladimir Dergachev

>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Get the new Palm Tungsten T
> handheld. Power & Color in a compact size!
> http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> _______________________________________________
> Gatos-devel mailing list
> Gat...@li...
> https://lists.sourceforge.net/lists/listinfo/gatos-devel
>

what does the avview FREQ value

>
> OK, I compiled those and they work fine, however, i rebooted and my
> /dev/v4l dir and /dev/video0 are no longer existent.

It is likely you are using devfs and forgot to modprobe km and videodev

>
> On a good note avview works fine now, but my tv display is not ..proper,
> it's all um...it looks like when you change to a PPV channel or
> something of the like that you can't get, but you still can see a bit of
> it from the wavy lines  .. looks like someone put a magnet to the
> screen, but i think if my /dev problem is solved then it will work (I
> Hope!)

If this is what I think it is it can only be cured by poweroff and restart
(just rebooting won't help).

                         Vladimir Dergachev

>
> I'm using Gentoo and devfs if that helps...
>
> Thanks,
>
> -richard
>
> >
> >                        best
> >
> >                          Vladimir Dergachev
> >
> > > output to the console:
> > >
> > > X Error of failed request:  BadMatch (invalid parameter attributes)
> > >   Major opcode of failed request:  141 (XVideo)
> > >   Minor opcode of failed request:  14 ()
> > >   Serial number of failed request:  2476
> > >   Current serial number in output stream:  2476
> > >
> > >
> > > Also for more information, here's what I get running v4l-conf
> > >
> > > $ v4l-conf
> > > v4l-conf: using X11 display :0.0
> > > dga: version 2.0
> > > mode: 1280x960, depth=24, bpp=32, bpl=5120, base=0xf4000000
> > > can't open /dev/video0: No data available
> > >
> > > And XawTV :
> > >
> > > richard richard $ xawtv
> > > This is xawtv-3.73, running on Linux/i686 (2.4.19-gentoo-r10)
> > > can't open /dev/v4l/video0: No data available
> > > v4l-conf had some trouble, trying to continue anyway
> > > v4l2: open /dev/v4l/video0: No data available
> > > v4l: open /dev/v4l/video0: No data available
> > > no video grabber device available
> > >
> > > Thanks in advance if you help me :-)
> > >
> > > -richard
> > > ric...@co...
> > >
> > >
> > >
> > > -------------------------------------------------------
> > > This SF.net email is sponsored by: Get the new Palm Tungsten T
> > > handheld. Power & Color in a compact size!
> > > http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> > > _______________________________________________
> > > Gatos-devel mailing list
> > > Gat...@li...
> > > https://lists.sourceforge.net/lists/listinfo/gatos-devel
> > >
> >
> >
> > -------------------------------------------------------
> > This SF.net email is sponsored by: Get the new Palm Tungsten T
> > handheld. Power & Color in a compact size!
> > http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> > _______________________________________________
> > Gatos-devel mailing list
> > Gat...@li...
> > https://lists.sourceforge.net/lists/listinfo/gatos-devel
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Get the new Palm Tungsten T
> handheld. Power & Color in a compact size!
> http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> _______________________________________________
> Gatos-devel mailing list
> Gat...@li...
> https://lists.sourceforge.net/lists/listinfo/gatos-devel
>

On Thu, Nov 28, 2002 at 05:24:46PM -0500, Vladimir Dergachev wrote:
> 
> 
> On Thu, 28 Nov 2002, Richard Joyce wrote:
> 
> > I'm using the km driver (as well as ati.2) for my ATI Rage 128 Pro, and
> > I'm trying to get the video-in to work.
> >
> > Heres what I get when i run Avview:
> >
> > $ start_avview
> > libavcodec: CPU flags: mmx mmxext sse
> > control_screen=
> > setup.f.pane1.f.canvas.f.optionmenu0.menu
> > NUM_ADAPTORS=1
> > Adaptor 0: ATI Rage128 Video Overlay
> >                 type input image
> >                 ports 59
> > 59 ATI Rage128 Video Overlay
> >
> > And at this point it loads up and says: It appears you have not run
> > AVview 0.10.9 before,etc.  When i say OK to this, it crashes, with this
> 
> You need to compile ati.2 CVS drivers in order to use AVview from CVS.


OK, I compiled those and they work fine, however, i rebooted and my
/dev/v4l dir and /dev/video0 are no longer existent.

On a good note avview works fine now, but my tv display is not ..proper,
it's all um...it looks like when you change to a PPV channel or
something of the like that you can't get, but you still can see a bit of
it from the wavy lines  .. looks like someone put a magnet to the
screen, but i think if my /dev problem is solved then it will work (I
Hope!)

I'm using Gentoo and devfs if that helps...

Thanks,

-richard

> 
>                        best
> 
>                          Vladimir Dergachev
> 
> > output to the console:
> >
> > X Error of failed request:  BadMatch (invalid parameter attributes)
> >   Major opcode of failed request:  141 (XVideo)
> >   Minor opcode of failed request:  14 ()
> >   Serial number of failed request:  2476
> >   Current serial number in output stream:  2476
> >
> >
> > Also for more information, here's what I get running v4l-conf
> >
> > $ v4l-conf
> > v4l-conf: using X11 display :0.0
> > dga: version 2.0
> > mode: 1280x960, depth=24, bpp=32, bpl=5120, base=0xf4000000
> > can't open /dev/video0: No data available
> >
> > And XawTV :
> >
> > richard richard $ xawtv
> > This is xawtv-3.73, running on Linux/i686 (2.4.19-gentoo-r10)
> > can't open /dev/v4l/video0: No data available
> > v4l-conf had some trouble, trying to continue anyway
> > v4l2: open /dev/v4l/video0: No data available
> > v4l: open /dev/v4l/video0: No data available
> > no video grabber device available
> >
> > Thanks in advance if you help me :-)
> >
> > -richard
> > ric...@co...
> >
> >
> >
> > -------------------------------------------------------
> > This SF.net email is sponsored by: Get the new Palm Tungsten T
> > handheld. Power & Color in a compact size!
> > http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> > _______________________________________________
> > Gatos-devel mailing list
> > Gat...@li...
> > https://lists.sourceforge.net/lists/listinfo/gatos-devel
> >
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by: Get the new Palm Tungsten T 
> handheld. Power & Color in a compact size! 
> http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> _______________________________________________
> Gatos-devel mailing list
> Gat...@li...
> https://lists.sourceforge.net/lists/listinfo/gatos-devel

On Fri, 29 Nov 2002, Nikolai Zhubr wrote:

> Hi,
> Thursday, 28 November, 2002, 23:57:48, Vladimir Dergachev wrote:
> > I don't think separate irq handlers are wrong. However, I do plan some
> > abstraction for dma transfers and events like vsync/field capture/etc.
> It isn't wrong of course, but if you keep them separate then you'll
> have to always bother about all 3 copies because probably you might
> not be sure now what (abstract) features exactly you will want to
> add later etc.
> Hardware features, on the contrary, are probably quite well known
> already, so it seems possible to do some cleanup/abstraction and
> just freeze this code then (imo). Btw it would be nice to finally fix
> mach64 part along with :)

My preference is to work the other way around - have a mathematical
abstraction to which hardware dependent code acts as a client.

The idea is that the code that implements the abstraction can be tested
using general principle and hardware-dependent part is then small and
isolated.

The other way to say this is that I would rather maintain 3 one-screen
hardware-dependent functions than one which behaviour depends on a bunch
of other source files.

Also we do not know which hardware interfaces ATI is going to come up
with. For example they seem to have moved sound processing functions onto
Rage Theatre 200 for All-in-Wonder 9700 card.

Lastly, in practice we will probably find out that your and mine points of
view converge and actual solution will have qualities of both.

                        best

                          Vladimir Dergachev

> --
> Best regards,
>  Nikolai Zhubr
>
> > One problem here is that one cannot separate out dma start code from
> > interrupt handler by using wait queues as scheduler may not start the
> > woken up code on time. So we have to call that code direct one way or
> > another.
>
> > Also having all code in one function makes it a little faster.
>
> > I am still thinking about a good solution.
>
> >                           Vladimir Dergachev
>
>

<snip/>
> It locks up both if try recording from the AIW or if I try recording
> from the web cam after loading the km module (before I can read from it
> fine)

This is no longer true. With the experimental version I can read form
the webcam with the km module loaded. It's just recording from the aiw
that locks things up.

> > then try also an experimental version http://zhubr.tamb.ru/km-cvs-hackB.tgz
> > Let me know if it makes any difference.

<snip/>
-- 
rob <ro...@pa...>

On Thu, 2002-11-28 at 23:39, Nikolai Zhubr wrote: 
> Hi,
> Friday, 29 November, 2002, 1:24:44, rob wrote:
> > the wrong way round (the pictures correct it's just the window). But X
> > locks up hard if I try recording or if try to use my web cam (a logitech
> If you mean you try recording from your AIW (through km) and it locks up

It locks up both if try recording from the AIW or if I try recording
from the web cam after loading the km module (before I can read from it
fine)

> then try also an experimental version http://zhubr.tamb.ru/km-cvs-hackB.tgz
> Let me know if it makes any difference.

It gets worse it now locks up the moment I open avview. (This is without
the web cam drivers loaded).
The tv picture carrys on displaying quit happily but nothing else works
event telneting in and giving a sudo killall -KILL dosen't get rid of X.

This is what I have in the log

Nov 29 00:20:22 samantha kernel: Linux video capture interface: v1.00
Nov 29 00:20:42 samantha kernel: Kmultimedia API module version
alpha-3.0 loaded
Nov 29 00:20:42 samantha kernel: Kmultimedia module version
alpha-2.0-unofficial_hack_B loaded
Nov 29 00:20:42 samantha kernel: Device ATI Technologies Inc Rage 128
RF/SG AGP 01:00.0 (0x1002:0x5246) corresponds to /dev/video0

Then I get this repeated over and over and over

Nov 29 00:21:44 samantha kernel: GUI_STAT           = 0x80020040
Nov 29 00:21:44 samantha kernel: PM4_STAT           = 0x80020040
Nov 29 00:21:44 samantha kernel: PM4_BUFFER_DL_WPTR = 0x0000040e
Nov 29 00:21:44 samantha kernel: PM4_BUFFER_DL_RPTR = 0x0000040e
Nov 29 00:21:44 samantha kernel: PM4_MICRO_CNTL     = 0x40000000
Nov 29 00:21:44 samantha kernel: PM4_BUFFER_CNTL    = 0x80000011

> > Also if I could get capturing working can it capture losslessly?
> Probably not at the moment. Though mpeg1 intra-only at high bitrate
> is somewhat close to, imho.
But the card is capable of it? 

thanks for your help



-- 

rob <ro...@pa...>

Hi,
Friday, 29 November, 2002, 1:24:44, rob wrote:
> the wrong way round (the pictures correct it's just the window). But X
> locks up hard if I try recording or if try to use my web cam (a logitech
If you mean you try recording from your AIW (through km) and it locks up
then try also an experimental version http://zhubr.tamb.ru/km-cvs-hackB.tgz
Let me know if it makes any difference.

> Also if I could get capturing working can it capture losslessly?
Probably not at the moment. Though mpeg1 intra-only at high bitrate
is somewhat close to, imho.
-- 
Best regards,
 Nikolai Zhubr

Hi,
Thursday, 28 November, 2002, 23:57:48, Vladimir Dergachev wrote:
> I don't think separate irq handlers are wrong. However, I do plan some
> abstraction for dma transfers and events like vsync/field capture/etc.
It isn't wrong of course, but if you keep them separate then you'll
have to always bother about all 3 copies because probably you might
not be sure now what (abstract) features exactly you will want to
add later etc.
Hardware features, on the contrary, are probably quite well known
already, so it seems possible to do some cleanup/abstraction and
just freeze this code then (imo). Btw it would be nice to finally fix
mach64 part along with :)
-- 
Best regards,
 Nikolai Zhubr

> One problem here is that one cannot separate out dma start code from
> interrupt handler by using wait queues as scheduler may not start the
> woken up code on time. So we have to call that code direct one way or
> another.

> Also having all code in one function makes it a little faster.

> I am still thinking about a good solution.

>                           Vladimir Dergachev

On November 28, 2002 12:29 pm, Vladimir Dergachev wrote:
> On Thu, 28 Nov 2002 je...@te... wrote:
> > Now on to the question.  I'm trying to run my multimedia computer
> > without a VGA monitor with all the graphics going to the TV.  Right
> > now, I have a Rage 128 Pro card (the one that has Video In, and
> > Video/S-Video out).  I couldn't get the framebuffer console nor X to
> > display on the TV.  I didn't have the CVS devel branch of ati.2, so I
> > will be trying that.  I'm hopeful that I'll then have TV-Out for X.=20
> > But has anyone here had success getting TV-Out on a framebuffer
> > console?  It works just fine on the regular console, but not with
> > framebuffer support.
>
> I do not believe this would work - r128 framebuffer does not know how t=
o
> setup TV-out. Try using VESA framebuffer.

I'll second that (using the VESA framebuffer).

I've got a Radeon 7500 here in my machine and the only way I was able to=20
reliably get TV-Out working was to use the VESA modes.

--=20
Graham TerMarsch

On Thu, 28 Nov 2002, Richard Joyce wrote:

> I'm using the km driver (as well as ati.2) for my ATI Rage 128 Pro, and
> I'm trying to get the video-in to work.
>
> Heres what I get when i run Avview:
>
> $ start_avview
> libavcodec: CPU flags: mmx mmxext sse
> control_screen=
> setup.f.pane1.f.canvas.f.optionmenu0.menu
> NUM_ADAPTORS=1
> Adaptor 0: ATI Rage128 Video Overlay
>                 type input image
>                 ports 59
> 59 ATI Rage128 Video Overlay
>
> And at this point it loads up and says: It appears you have not run
> AVview 0.10.9 before,etc.  When i say OK to this, it crashes, with this

You need to compile ati.2 CVS drivers in order to use AVview from CVS.

                       best

                         Vladimir Dergachev

> output to the console:
>
> X Error of failed request:  BadMatch (invalid parameter attributes)
>   Major opcode of failed request:  141 (XVideo)
>   Minor opcode of failed request:  14 ()
>   Serial number of failed request:  2476
>   Current serial number in output stream:  2476
>
>
> Also for more information, here's what I get running v4l-conf
>
> $ v4l-conf
> v4l-conf: using X11 display :0.0
> dga: version 2.0
> mode: 1280x960, depth=24, bpp=32, bpl=5120, base=0xf4000000
> can't open /dev/video0: No data available
>
> And XawTV :
>
> richard richard $ xawtv
> This is xawtv-3.73, running on Linux/i686 (2.4.19-gentoo-r10)
> can't open /dev/v4l/video0: No data available
> v4l-conf had some trouble, trying to continue anyway
> v4l2: open /dev/v4l/video0: No data available
> v4l: open /dev/v4l/video0: No data available
> no video grabber device available
>
> Thanks in advance if you help me :-)
>
> -richard
> ric...@co...
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Get the new Palm Tungsten T
> handheld. Power & Color in a compact size!
> http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> _______________________________________________
> Gatos-devel mailing list
> Gat...@li...
> https://lists.sourceforge.net/lists/listinfo/gatos-devel
>

Hi,

I've got an all in wonder 128. Running on debian sid. With 2.4.19 + low
latency patch. Just installed the ati.2 drivers. Tv works fine from
xawtv and kinda ok in aaview except it looks like the window is drawn
the wrong way round (the pictures correct it's just the window). But X
locks up hard if I try recording or if try to use my web cam (a logitech
quickcam express) with gnomemeeting and only the reset switch can rescue
it:(.

All the ati stuff is the latest tarballs from your website.

Any ideas on what I can do? If there's any other info you need just let
me know.

Also if I could get capturing working can it capture losslessly? 

Thanks 

-- 
rob <ro...@pa...>

I'm using the km driver (as well as ati.2) for my ATI Rage 128 Pro, and
I'm trying to get the video-in to work.

Heres what I get when i run Avview:

$ start_avview
libavcodec: CPU flags: mmx mmxext sse
control_screen=
setup.f.pane1.f.canvas.f.optionmenu0.menu
NUM_ADAPTORS=1
Adaptor 0: ATI Rage128 Video Overlay
                type input image
                ports 59
59 ATI Rage128 Video Overlay

And at this point it loads up and says: It appears you have not run
AVview 0.10.9 before,etc.  When i say OK to this, it crashes, with this
output to the console:

X Error of failed request:  BadMatch (invalid parameter attributes)
  Major opcode of failed request:  141 (XVideo)
  Minor opcode of failed request:  14 ()
  Serial number of failed request:  2476
  Current serial number in output stream:  2476


Also for more information, here's what I get running v4l-conf

$ v4l-conf 
v4l-conf: using X11 display :0.0
dga: version 2.0
mode: 1280x960, depth=24, bpp=32, bpl=5120, base=0xf4000000
can't open /dev/video0: No data available

And XawTV :

richard richard $ xawtv
This is xawtv-3.73, running on Linux/i686 (2.4.19-gentoo-r10)
can't open /dev/v4l/video0: No data available
v4l-conf had some trouble, trying to continue anyway
v4l2: open /dev/v4l/video0: No data available
v4l: open /dev/v4l/video0: No data available
no video grabber device available

Thanks in advance if you help me :-)

-richard
ric...@co...

On Thu, 28 Nov 2002, Nikolai Zhubr wrote:

> Hi Vladimir,
> Wednesday, 27 November, 2002, 4:07:37, you wrote:
> > What did you have in mind ? I certainly don't mind km getting simpler or
> > making a small library out of it. In particular it might worth the effort
> > to investigate abstracting register accesses. Spending a few extra cycles
> > in cpu won't matter too much when we have to wait for PCI bus transfers to
> > finish.
> I think first of all it would be worth indeed to generalize registers,
> status bits and such, so that it would be possible then to get rid of
> separate irq handlers in mach64.c, r128.c and radeon.c. The point is
> that low-level stuff wouldn't need to know anything about actual
> processing sequence then, like whether we are going to transfer just
> all frames, or even-only, or whatever else one might invent. Lower-level
> code shouldn't care, just provide some generalized access.

I don't think separate irq handlers are wrong. However, I do plan some
abstraction for dma transfers and events like vsync/field capture/etc.

One problem here is that one cannot separate out dma start code from
interrupt handler by using wait queues as scheduler may not start the
woken up code on time. So we have to call that code direct one way or
another.

Also having all code in one function makes it a little faster.

I am still thinking about a good solution.

                          Vladimir Dergachev

>
> > Also there should/could be more additions to hardware-dependent part of
> > km, in particular to enable multiple DMA/AGP transfers and simple
> > rendering. (Aside from using overlay to display YUV data one could also
> > use texture units - at the cost of slight decrease in quality).
> Well, thats yet another point: I don't know much about this stuff so I
> wouldn't be able to handle it myself :( However, as soon as there is
> an appropriate low-level access layer available it should then be easy
> to put any abstract layer on top of it, whether v4l2 compatability or
> something completely different.
> --
> Best regards,
>  Nikolai Zhubr
>
> >                     best
>
> >                         Vladimir Dergachev
>
>

On Thu, 28 Nov 2002 je...@te... wrote:

> Hi all,
>
> First off, a kudos to the developers here on GATOS as I'm very impressed with
> what's available for the ATI cards.
>
> Now on to the question.  I'm trying to run my multimedia computer without a VGA
> monitor with all the graphics going to the TV.  Right now, I have a Rage 128
> Pro card (the one that has Video In, and Video/S-Video out).  I couldn't get
> the framebuffer console nor X to display on the TV.  I didn't have the CVS
> devel branch of ati.2, so I will be trying that.  I'm hopeful that I'll then
> have TV-Out for X.  But has anyone here had success getting TV-Out on a
> framebuffer console?  It works just fine on the regular console, but not with
> framebuffer support.

I do not believe this would work - r128 framebuffer does not know how to
setup TV-out. Try using VESA framebuffer.

                            Vladimir Dergachev

>
> I'll be trying CVS devel tonight, but can anyone tell me that I'll be able to
> have X on TV-Out for sure?  The reason I ask is that I have a AIW 7500 Radeon
> sitting in a shrink-wrapped box.  If I can't get X TV-Out with the Rage 128
> Pro, but I *CAN* get it with the 7500 Radeon, then I'd open up the box.
>
> Hopefully someone can help,
>
> Jean
>
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Get the new Palm Tungsten T
> handheld. Power & Color in a compact size!
> http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en
> _______________________________________________
> Gatos-devel mailing list
> Gat...@li...
> https://lists.sourceforge.net/lists/listinfo/gatos-devel
>