D3DM err

Hi,

I have Sandgate II-G (PXA270+2700G) and CE5.0. Now I try to add
Direct3DMobile in the CE image. When I execute the D3DM sample code
from Microsoft, the call CreateDevice fails with return value
D3DMERR_INVALIDDEVICE. That reads the hook can't find according to MS
doc.
Anyone here experiences similar situation with D3DM?


Regards,
Mark

Don
Sat Jan 08 21:30:38 CST 2005

I would imagine that you're seeing that error because there isn't a D3DM
driver on your device.


"Mark" <harvanaisland@yahoo.com> wrote in message
news:1105200537.327607.214990@c13g2000cwb.googlegroups.com...
> Hi,
>
> I have Sandgate II-G (PXA270+2700G) and CE5.0. Now I try to add
> Direct3DMobile in the CE image. When I execute the D3DM sample code
> from Microsoft, the call CreateDevice fails with return value
> D3DMERR_INVALIDDEVICE. That reads the hook can't find according to MS
> doc.
> Anyone here experiences similar situation with D3DM?
>
>
> Regards,
> Mark
>

Mark
Sun Jan 09 09:57:03 CST 2005

I try to use D3DM reference driver. Same result.



Don Crouch wrote:
> I would imagine that you're seeing that error because there isn't a
D3DM
> driver on your device.
>
>
> "Mark" <harvanaisland@yahoo.com> wrote in message
> news:1105200537.327607.214990@c13g2000cwb.googlegroups.com...
> > Hi,
> >
> > I have Sandgate II-G (PXA270+2700G) and CE5.0. Now I try to add
> > Direct3DMobile in the CE image. When I execute the D3DM sample code
> > from Microsoft, the call CreateDevice fails with return value
> > D3DMERR_INVALIDDEVICE. That reads the hook can't find according to
MS
> > doc.
> > Anyone here experiences similar situation with D3DM?
> >
> >
> > Regards,
> > Mark
> >

doncr
Mon Jan 10 21:22:56 CST 2005

If you're invoking the SDK sample with the -ref command line switch and
CreateDevice is still failing (and you've verified that d3dmref.dll is
in your release directory), then it's probably not a driver issue.

You can get the D3DM runtime to output more detailed debug messages
when something fails by enabling the CE Debug Zones for d3dm.dll.
Unfortunately you can't use the easy way of doing this from PB's menu
since d3dm.dll isn't already loaded by the OS. If you add the
following DWORD registry value on your host PC (not the target
device!), PB will enable all of the possible debug zones for the D3DM
runtime:

HKCU\Pegasus\Zones\Direct3D Mobile

Set this DWORD value to 0x0000003F.

This will produce quite a bit of spew to the debugger's output pane,
but will probably help you pinpoint the reason CreateDevice is failing.

The bit values for the above DWORD value corresponde to the following
types of messages (debug zones):

bit 0 = Errors
bit 1 = Params
bit 2 = Memory
bit 3 = Sync
bit 4 = Msg
bit 5 = RefCount

The default value is to report Errors (bit 0 set = 0x00000001). If you
keep the registry value around, you'll probably want to set it back to
0x1 once you've determined why CreateDevice is failing, otherwise
you'll be flooded with debugging messages anytime you run a D3DM
application under the debugger.

Mark
Tue Jan 11 15:06:36 CST 2005

Thanks.

I haven't get the debug out. I figure that I have to make a debug
build.

Got another question. I have graphics accelator hardware (intel 2700G).
Say I want to take advantage of the hardware, should I use RemoteHook
to the driver, is that right?

According to MS document, only remote hooking will allow driver to
access hw. If the software simulation like d3dmref.dll is used, then
should use LocalHook. Is that correct?


Mark


doncr@hotmail.com wrote:
> If you're invoking the SDK sample with the -ref command line switch
and
> CreateDevice is still failing (and you've verified that d3dmref.dll
is
> in your release directory), then it's probably not a driver issue.
>
> You can get the D3DM runtime to output more detailed debug messages
> when something fails by enabling the CE Debug Zones for d3dm.dll.
> Unfortunately you can't use the easy way of doing this from PB's menu
> since d3dm.dll isn't already loaded by the OS. If you add the
> following DWORD registry value on your host PC (not the target
> device!), PB will enable all of the possible debug zones for the D3DM
> runtime:
>
> HKCU\Pegasus\Zones\Direct3D Mobile
>
> Set this DWORD value to 0x0000003F.
>
> This will produce quite a bit of spew to the debugger's output pane,
> but will probably help you pinpoint the reason CreateDevice is
failing.
>
> The bit values for the above DWORD value corresponde to the following
> types of messages (debug zones):
>
> bit 0 = Errors
> bit 1 = Params
> bit 2 = Memory
> bit 3 = Sync
> bit 4 = Msg
> bit 5 = RefCount
>
> The default value is to report Errors (bit 0 set = 0x00000001). If
you
> keep the registry value around, you'll probably want to set it back
to
> 0x1 once you've determined why CreateDevice is failing, otherwise
> you'll be flooded with debugging messages anytime you run a D3DM
> application under the debugger.

Don
Wed Jan 12 10:39:09 CST 2005

Yes, you'll need a debug build in order for the debug zone messages to work.

Your assumptions regarding local vs. remote hooking are pretty much correct.
I suppose it is possible that a locally-hooked D3DM driver could talk to the
hardware using something like the ExtEscape mechanism, but that would tend
to be a rarity.

Just as the display driver executing in the gwes.exe process, it's more
likely that a D3DM hardware driver would do the same (they may actually be
the same driver DLL), and that would require the remote hook in order for
the D3DM runtime to be able to communicate with it.


"Mark" <harvanaisland@yahoo.com> wrote in message
news:1105477596.433493.29640@f14g2000cwb.googlegroups.com...
> Thanks.
>
> I haven't get the debug out. I figure that I have to make a debug
> build.
>
> Got another question. I have graphics accelator hardware (intel 2700G).
> Say I want to take advantage of the hardware, should I use RemoteHook
> to the driver, is that right?
>
> According to MS document, only remote hooking will allow driver to
> access hw. If the software simulation like d3dmref.dll is used, then
> should use LocalHook. Is that correct?
>
>
> Mark