XVR-500 support in XVR-2500 driver?

XVR-500 support in XVR-2500 driver?

[Meelis Roos]

The newly merged XVR-500 and XVR-2500 drivers both seem to have
PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x7a2) (XVR-500) support in pci device 
table. Is this correct?

Just curiously reading new drivers,

-- 
Meelis Roos (mroos@linux.ee)

Re: XVR-500 support in XVR-2500 driver?

[David Miller]

From: Meelis Roos <mroos@linux.ee>
Date: Mon, 7 May 2007 23:21:44 +0300 (EEST)

> The newly merged XVR-500 and XVR-2500 drivers both seem to have
> PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x7a2) (XVR-500) support in pci device 
> table. Is this correct?
> 
> Just curiously reading new drivers,

Thanks for catching this.

Originally I handled both chips in one driver, then I split
them up, that's why it was there.  I'll fix that up :)

Re: XVR-500 support in XVR-2500 driver?

[David Miller]

From: David Miller <davem@davemloft.net>
Date: Mon, 07 May 2007 13:54:51 -0700 (PDT)

> From: Meelis Roos <mroos@linux.ee>
> Date: Mon, 7 May 2007 23:21:44 +0300 (EEST)
> 
> > The newly merged XVR-500 and XVR-2500 drivers both seem to have
> > PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x7a2) (XVR-500) support in pci device 
> > table. Is this correct?
> > 
> > Just curiously reading new drivers,
> 
> Thanks for catching this.
> 
> Originally I handled both chips in one driver, then I split
> them up, that's why it was there.  I'll fix that up :)

Here is that fix I will push to Linus, thanks again.

commit de9f0cf93fcbccc2de4e6bcb5ef90f997616be3b
Author: David S. Miller <davem@sunset.davemloft.net>
Date:   Mon May 7 14:02:51 2007 -0700

    [VIDEO] sunxvr2500: Fix PCI device ID table.
    
    Noticed by Meelis Roos.
    
    Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/drivers/video/sunxvr2500.c b/drivers/video/sunxvr2500.c
index 4010492..4316c7f 100644
--- a/drivers/video/sunxvr2500.c
+++ b/drivers/video/sunxvr2500.c
@@ -237,14 +237,14 @@ static void __devexit s3d_pci_unregister(struct pci_dev *pdev)
 }
 
 static struct pci_device_id s3d_pci_table[] = {
-	{	/* XVR-2500 */
-		PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0032),
-		.driver_data = 1,
-	},
-	{	/* XVR-500 */
-		PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x07a2),
-		.driver_data = 0,
-	},
+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002c),	},
+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002d),	},
+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002e),	},
+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002f),	},
+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0030),	},
+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0031),	},
+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0032),	},
+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0033),	},
 	{ 0, }
 };
 

Re: XVR-500 support in XVR-2500 driver?

[Tom "spot" Callaway]

On Mon, 2007-05-07 at 14:03 -0700, David Miller wrote:
> From: David Miller <davem@davemloft.net>
> Date: Mon, 07 May 2007 13:54:51 -0700 (PDT)
> 
> > From: Meelis Roos <mroos@linux.ee>
> > Date: Mon, 7 May 2007 23:21:44 +0300 (EEST)
> > 
> > > The newly merged XVR-500 and XVR-2500 drivers both seem to have
> > > PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x7a2) (XVR-500) support in pci device 
> > > table. Is this correct?
> > > 
> > > Just curiously reading new drivers,
> > 
> > Thanks for catching this.
> > 
> > Originally I handled both chips in one driver, then I split
> > them up, that's why it was there.  I'll fix that up :)
> 
> Here is that fix I will push to Linus, thanks again.

Dave,

Thanks for these drivers, fills a much requested void!

Do you know if there is a matching xorg driver to go along with the
kernel implementation?

~spot

Re: XVR-500 support in XVR-2500 driver?

[David Miller]

From: "Tom \"spot\" Callaway" <tcallawa@redhat.com>
Date: Mon, 07 May 2007 16:23:41 -0500

> Thanks for these drivers, fills a much requested void!
> 
> Do you know if there is a matching xorg driver to go along with the
> kernel implementation?

I reverse engineered the hardware by poking at these chips from the
firmware prompt for nearly a week, so what's there is the best I could
come up with.

Using the 'fbdev' driver, Xorg should just work with the XVR2500 but
there will be an ugly white boarder around the graphics screen just as
the current kernel console will have.  The board comes up in 32bpp
mode so that's what you'll need to use in xorg.conf

I cannot say the same for the XVR500 because of the funny mode the
framebuffer comes up in from the OBP.

So until I make some more progress reverse engineering these cards,
support is still currently half-assed.

At least there is some native kernel text console for this chips now.

Ideally I should be able to get it to the point where dumb framebuffer
Xorg support should work fully for both chips.  However, I doubt I'll
be able to add hw acceleration any time soon as these chips are very
complicated (they use DMA commands to do hardware rendering etc.), are
completely undocumented, and there's zero chance 3dlabs is going to
ever release the specs for these board to help the effort along.

Re: XVR-500 support in XVR-2500 driver?

[Jan Engelhardt]

On May 7 2007 14:03, David Miller wrote:
>
>diff --git a/drivers/video/sunxvr2500.c b/drivers/video/sunxvr2500.c
>index 4010492..4316c7f 100644
>--- a/drivers/video/sunxvr2500.c
>+++ b/drivers/video/sunxvr2500.c
>@@ -237,14 +237,14 @@ static void __devexit s3d_pci_unregister(struct pci_dev *pdev)
> }
> 
> static struct pci_device_id s3d_pci_table[] = {
>-	{	/* XVR-2500 */
>-		PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0032),
>-		.driver_data = 1,
>-	},
>-	{	/* XVR-500 */
>-		PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x07a2),
>-		.driver_data = 0,
>-	},
>+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002c),	},
>+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002d),	},
>+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002e),	},
>+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002f),	},
>+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0030),	},
>+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0031),	},
>+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0032),	},
>+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0033),	},

The extra comma inside the { } is not needed, is it?



Jan
-- 

Re: XVR-500 support in XVR-2500 driver?

[David Miller]

From: Jan Engelhardt <jengelh@linux01.gwdg.de>
Date: Tue, 8 May 2007 10:50:56 +0200 (MEST)

> 
> On May 7 2007 14:03, David Miller wrote:
> >
> >diff --git a/drivers/video/sunxvr2500.c b/drivers/video/sunxvr2500.c
> >index 4010492..4316c7f 100644
> >--- a/drivers/video/sunxvr2500.c
> >+++ b/drivers/video/sunxvr2500.c
> >@@ -237,14 +237,14 @@ static void __devexit s3d_pci_unregister(struct pci_dev *pdev)
> > }
> > 
> > static struct pci_device_id s3d_pci_table[] = {
> >-	{	/* XVR-2500 */
> >-		PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0032),
> >-		.driver_data = 1,
> >-	},
> >-	{	/* XVR-500 */
> >-		PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x07a2),
> >-		.driver_data = 0,
> >-	},
> >+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002c),	},
> >+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002d),	},
> >+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002e),	},
> >+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x002f),	},
> >+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0030),	},
> >+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0031),	},
> >+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0032),	},
> >+	{	PCI_DEVICE(PCI_VENDOR_ID_3DLABS, 0x0033),	},
> 
> The extra comma inside the { } is not needed, is it?

No, but if later I want to fill in more fields per entry I will indeed
need that comma there.