Commit Graph

868 Commits (7766378d97323de375687285f2e125008b79045d)

Author SHA1 Message Date
Thomas Hellstrom 7766378d97 Initial support for fence object classes.
(Fence objects belonging to different command submission mechanisms).
2007-02-15 12:11:38 +01:00
Michel Dänzer 7803977aa9 Fix build against older kernels. 2007-02-15 09:26:16 +01:00
Thomas Hellstrom a78f70faad Merge branch 'ttm-vram-0-1-branch' 2007-02-14 15:33:40 +01:00
Thomas Hellstrom 35eb12d8bf Fix multiple spinlock unlocking 2007-02-14 15:32:08 +01:00
Thomas Hellstrom 1345076c8f Rename drm_ttm.h to drm_objects.h
Fix up some header incompatibilities in drm_fence.c caused by the previous
commit.
2007-02-14 14:10:10 +01:00
Thomas Hellstrom 8ffc1844b0 Move fence- and buffer-object related header stuff to drm_ttm.h 2007-02-14 14:05:40 +01:00
Thomas Hellstrom 5c9a7b0f94 Remove an intel-specific hack and replace it with a fence driver callback. 2007-02-14 13:31:35 +01:00
Thomas Hellstrom 04760563b8 Set the drm bus map type for each buffer object memory type. 2007-02-14 12:39:02 +01:00
Thomas Hellstrom 7bcb62b45d Rework buffer object vm code to use nopfn() for kernels >= 2.6.19. 2007-02-14 10:49:37 +01:00
Thomas Hellstrom 6b289db054 Remove debug printout. 2007-02-13 20:47:30 +01:00
Thomas Hellstrom e1460426b8 Bugzilla Bug #9457
Add refcounting of user waiters to the DRM hardware lock, so that we can use the
DRM_LOCK_CONT flag more conservatively.

Also add a kernel waiter refcount that if nonzero transfers the lock for the kernel context,
when it is released. This is useful when waiting for idle and can be used
for very simple fence object driver implementations for the new memory manager.

It also resolves the AIGLX startup deadlock for the sis and the via drivers.
i810, i830 still require that the hardware lock is really taken so the deadlock remains
for those two. I'm not sure about ffb. Anyone familiar with that code?
2007-02-13 20:47:30 +01:00
Thomas Hellstrom 9efdae317c More bugfixes.
Fixed memory, pinned buffers and unmappable memory now seems
fully functional.
2007-02-13 20:05:32 +01:00
Adam Jackson 5bd13c5e15 Fix some outdated URLs, remove others. 2007-02-12 15:45:51 -05:00
Thomas Hellstrom 398913dc0e Lindent. 2007-02-12 20:34:50 +01:00
Thomas Hellstrom b0c5339ed6 More bugfixes. 2007-02-12 20:32:03 +01:00
Thomas Hellstrom f02f83ee08 Cleanup and fix support for pinned buffers. 2007-02-12 17:47:57 +01:00
Thomas Hellstrom 85ee2a8d04 Various bugfixes. 2007-02-10 12:06:36 +01:00
Thomas Hellstrom 53aee3122a I915 accelerated blit copy functional.
Fixed - to System memory copies are implemented by
flipping in a cache-coherent TTM,
blitting to it, and then flipping it out.
2007-02-09 16:36:53 +01:00
Thomas Hellstrom 57df398072 Reinstate some LRU handling. 2007-02-09 12:43:18 +01:00
Thomas Hellstrom d32b21e016 Remove some code that should have gone in
commit 6a49d9a8ab
2007-02-09 00:11:53 +01:00
Thomas Hellstrom 99acdaee48 Fix copyright statements. 2007-02-09 00:07:29 +01:00
Thomas Hellstrom 6a49d9a8ab Fix evict_mutex locking range.
Implement unmappable buffers. (fault moves them to mappable when needed).
Various bugfixes.
2007-02-09 00:02:02 +01:00
Thomas Hellstrom b2bcbf874b Add an accelerated buffer copy cleanup helper.
Export helper functions and make some important buffer-object functions non-static.
Add an i915 accelerated blit buffer move for pci memory buffers.
2007-02-08 21:28:33 +01:00
Thomas Hellstrom a0ed808d05 Don't create a ttm just to copy from. 2007-02-08 19:06:39 +01:00
Thomas Hellstrom bf8f46d4c6 Fix mm_block leak.
Some other minor fixes.
2007-02-08 18:59:02 +01:00
Thomas Hellstrom e4b2da4406 A minor function interface change and some memcpy bugfixing.
Hooray!! it sort of works with a fixed AGP area as faked VRAM.
2007-02-08 16:21:38 +01:00
Thomas Hellstrom 1257907fa9 Simplify external ttm page allocation.
Implement a memcpy fallback for copying between buffers.
2007-02-08 13:29:08 +01:00
Thomas Hellstrom 09984ad77b Update memory compatibility tests.
Now only pinned buffers are broken.
2007-02-08 11:55:24 +01:00
Thomas Hellstrom c1fbd8a566 Checkpoint commit.
Flag handling and memory type selection cleanup.
glxgears won't start.
2007-02-07 17:25:13 +01:00
Thomas Hellstrom af24465b2e Fix a stray unlock_kernel() in drm_vm.c
Add a file for memory move helpers, drm_bo_move.c
Implement generic memory move.
Cached, no_move and unmapped memory temporarily broken.
2007-02-07 12:52:23 +01:00
Thomas Hellstrom 71b9e876f9 Simplify pci map vs no pci map choice. 2007-02-06 16:59:45 +01:00
Thomas Hellstrom 40ce53dfde Implement a drm_mem_reg_t substructure in the buffer object type. 2007-02-06 15:56:43 +01:00
Thomas Hellstrom 609e3b0375 Implement a policy for selecting memory types. 2007-02-06 14:20:33 +01:00
Thomas Hellstrom 2d962332de i915: Add copy-blit operation. 2007-02-05 16:13:32 +01:00
Stephane Marchesin ebf22aed9a nouveau: add missing nv04_graph.c symlink. 2007-02-03 06:02:12 +01:00
Stephane Marchesin d69902db3b nouveau: fix nv04 graph routines for new register names. 2007-02-03 05:25:36 +01:00
Thomas Hellstrom 63f2abd721 Make also later kernels work with buffer object vm
and clean up some function names.
2007-02-02 19:49:11 +01:00
Thomas Hellstrom c269d560e4 Make vm handle buffer objects instead of ttm objects.
Remove ttm objects.
Make vm aware of PCI memory type buffer objects.
(Only works for pre 2.6.16 kernels for now).
2007-02-02 14:47:44 +01:00
Thomas Hellstrom dd733dea38 Fix missing ttm_open_vma call from previous commit.
Honour the ttm backend cant-use-aperture flag.
2007-02-01 13:19:05 +01:00
Thomas Hellstrom 9677c5ecc6 Prepare for removal of the ttm_object type. 2007-02-01 10:53:55 +01:00
Thomas Hellstrom 333c6af47a Protect drm_mmap against disappearing maps.
The map lists and hash tables are protected using dev->struct_mutex,
but drm_mmap strangely never locked this mutex.
2007-02-01 00:38:57 +01:00
Thomas Hellstrom 3024f23c65 memory manager: Make device driver aware of different memory types.
Memory types are either fixed (on-card or pre-bound AGP) or not fixed
(dynamically bound) to an aperture. They also carry information about:

1) Whether they can be mapped cached.
2) Whether they are at all mappable.
3) Whether they need an ioremap to be accessible from kernel space.

In this way VRAM memory and, for example, pre-bound AGP appear
identical to the memory manager.

This also makes support for unmappable VRAM simple to implement.
2007-01-31 14:50:57 +01:00
Thomas Hellstrom 36d50687dd Fix an error-path oops. 2007-01-31 11:03:53 +01:00
Thomas Hellstrom d399fcf46f Add a buffer object transfer function.
Creates a placeholder for the old buffer contents
when it is transfered to / from static memory like VRAM.
2007-01-30 16:20:23 +01:00
Thomas Hellstrom 9bbdc0fb10 Clean up buffer object destruction somewhat. 2007-01-30 12:35:49 +01:00
Thomas Hellstrom 9a654e71bd Use pre-defined list_splice function. 2007-01-29 13:37:02 +01:00
Thomas Hellstrom 45418bb1b1 s/buf/bo/ for consistency. 2007-01-29 13:37:02 +01:00
Thomas Hellstrom 1e4c7d69f5 Some cleanup. A buffer object should only have one active memory type. 2007-01-29 13:37:02 +01:00
Thomas Hellstrom 582637641a Remove a scary error printed when we were leaking memory caches.
We don't use memory caches anymore...

Fix memory accounting initialization to only use low or DMA32 memory.
2007-01-25 14:27:29 +01:00
Matthieu Castet f04347f371 nouveau: nv20 graph ctx switch.
Untested...
2007-01-13 23:19:41 +01:00