Re: pbuffers, extensions to DRM

--- Keith Whitwell <keith-CdwZJljFklH+2FeEXyspIVaTQe2KTcn/@public.gmane.org> wrote:

> > A work-around would be to draw a filled polygon to implement the clear, 
> > but I'm wondering where else there might be similar assumptions to trip 
> > us up.
> 
> Why not just fix the ioctl?

Yes fix the ioctl. I think there are four cases of clear that need to be fixed.
It is pretty obvious looking at the code. From the last time I looked at this I
believe that clear is the only command that needs to be fixed.

> > The other thing we need is a mechanism to allocate the memory for 
> > pbuffers.  My first guess is that this would have to be allocated out of 
> > the space used for textures.  I think the DRI texture memory allocator 
> > could be adapted to do this.
> 
> Correct.  There's a memory manager there for AGP space, but it's not 
> integrated with the texture manager.  In the i830 driver, particularly the one
> 
> on the agp-i865-0-1-branch, I have a kernel memory manager which integrates 
> fairly smoothly with the client-driver's texture memory manager.  This is 
> probably a decent starting point.
> 
We need a unified memory manager strategy . Putting it the driver is a good
idea. The code could be a feature common to all DRM modules. 

I would also use this manager in my initialization code when I init the first
display buffer and allocate memory for the hardware cursor. After thinking about
this for a second I shouldn't allocate any framebuffers at init time, instead I
should wait for a higher layer to tell me to make one.


DRM also needs a couple of minor extensions. I've asked Eric to add these but I
haven't heard back yet.

--------------------------------------------------------------

Could you provide me with an offical API for getting these items? It's stupid to
overlay the stats function to conserve IOCTLs.

DRM_STAT_PCI_VENDOR].value = dev->pdev->vendor;
DRM_STAT_PCI_DEVICE].value = dev->pdev->device;
DRM_STAT_PCI_SUB_VENDOR].value = dev->pdev->subsystem_vendor;
DRM_STAT_PCI_SUB_DEVICE].value = dev->pdev->subsystem_device;
DRM_STAT_IRQ_NUMBER].value = dev->pdev->irq;
DRM_STAT_PCI_MMIO_START].value = 0;
DRM_STAT_PCI_MMIO_LENGTH].value = 0;
DRM_STAT_PCI_FB_START].value = 0;
DRM_STAT_PCI_FB_LENGTH].value = 0;

The driver also needs to do pci_enable_deivce and pci_disable_device. The
version in the kernel tree does this but the one in DRI doesn't. Linus must have
added it.

Last thing I need is a function for returning a copy of the ROM. This could be
done generically. The radeon has a bug where the RADEON_MPP_TB_CONFIG register
needs to be written. This only has to be done once when the driver is loaded. 

It's also not clear to me the X needs to know the IRQ number. Why doesn't the
DRM module just send a signal on retrace?

=====
Jon Smirl
jonsmirl-/[email protected]

__________________________________
Do you Yahoo!?
New Yahoo! Photos - easier uploading and sharing.
http://photos.yahoo.com/