| Age | Commit message (Collapse) | Author | 
|---|
|  | server falls back to PCI GART or disables the DRI gracefully | 
|  | kernel tree. Added comments to that effect. | 
|  | Dawes to avoid backword incompatibilities... | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | from linux/mm.h now and wrapper.h has been removed in 2.5.69 and later. | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | newdrm-0-0-1-branch. | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | structures if AGP support is enabled in the kernel (__REALLY_HAVE_AGP).
    This fixes the compile errors on kernels without AGP support. | 
|  |  | 
|  |  | 
|  |  | 
|  | and must be done if in future a driver tries to talk to the AGP
    directly from kernelspace instead of userspace, and the AGP is not
    present. | 
|  | initializing the DRM device - this was causing a kernel oops when the
    AGPGART module wasn't loaded. | 
|  |  | 
|  |  | 
|  |  | 
|  | respectively. Splited the work out of the ioctls and renamed (with the
    _ioctl prefix). Added some more documentation. Did the same for
    drm_sgpsupport.h. | 
|  |  | 
|  |  | 
|  | with the documentation merge. | 
|  |  | 
|  |  |