summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2009-05-04ttm: Fix up printouts.modesetting-newttmThomas Hellstrom11-66/+64
Signed-off-by: Thomas Hellstrom <thomas-at-shipmail-dot-org>
2009-05-04ttm: Replace the ttm_buffer_object::mutex with a spinlock.Thomas Hellstrom7-159/+128
Signed-off-by: Thomas Hellstrom <thomas-at-shipmail-dot-org>
2009-05-04ttm: s/flags/placement/Thomas Hellstrom8-61/+59
Signed-off-by: Thomas Hellstrom <thomas-at-shipmail-dot-org>
2009-05-04ttm: Remove the struct ttm_mem_reg::proposed_flags member.Thomas Hellstrom7-57/+38
Signed-off-by: Thomas Hellstrom <thomas-at-shipmail-dot-org>
2009-05-04ttm: Update buffer object members and protection documentation.Thomas Hellstrom5-60/+85
Use linux bitmaps for bo private flags. Signed-off-by: Thomas Hellstrom <thomas-at-shipmail-dot-org>
2009-04-30openchrome: Use cached system memory when evicting from VRAM.Thomas Hellstrom1-1/+5
2009-04-28ttm: Don't set the ghost object to NULL.Thomas Hellstrom1-1/+1
Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
2009-04-28ttm: Fix an illegal state transition.Thomas Hellstrom1-1/+1
Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
2009-04-28ttm: Fix a return value.Thomas Hellstrom1-1/+1
Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
2009-04-28ttm: Fix uninitialized buffer object sizes.Thomas Hellstrom2-1/+9
Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
2009-04-05openchrome: Reinstate the old register used for the pause address.Thomas Hellstrom1-1/+1
0x40c isn't implemented on old unichromes, and seems to cause occasional lockups on newer ones.
2009-03-23ttm: Work around a locking order reversal in fault().Thomas Hellstrom1-4/+22
Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
2009-03-23ttm: Clear the bo->vm_node pointerThomas Hellstrom2-1/+8
a) When the vm_node is freed in ttm_bo.c b) When we transfer a buffer object in ttm_bo_util.c Reported by Jerome Glisse.
2009-03-22ttm: Fix a bo vm error path.Thomas Hellstrom1-2/+1
Reported by Jerome Glisse.
2009-03-22ttm: Optimize delayed buffer destruction.Thomas Hellstrom1-8/+14
Don't walk through the delayed delete list completely on each try, but stop after the first buffer with an unsignaled fence. Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
2009-03-19ttm: Fix buffer object transfer.Thomas Hellstrom1-7/+24
Fix destruction and initalization of "ghost" buffer objects that are created on pipelined buffer moves. Reported and suggestions by Jerome Glisse. Signed-off-by: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
2009-03-11openchrome: Fix a performance regression.Thomas Hellstrom1-3/+34
Unichrome's don't like small command submissions, so pad the command stream with dummy data. Also use 0x40c as pause address register instead of 0x418, as reading the latter seems to sometimes stall the PCI bus until the AGP command reader has completed a cycle.
2009-03-05openchrome: Adjust DMA tracker interval.Thomas Hellstrom1-1/+1
2009-03-05openchrome: Further deferred fencing tweaks.Thomas Hellstrom1-10/+18
2009-03-05openchrome: Add deferred fence emission.Thomas Hellstrom6-26/+80
2009-02-23ttm: Some commenting fixups.Thomas Hellstrom2-31/+2
2009-02-23ttm: Don't take a buffer object off lru lists if reservation fails.Thomas Hellstrom1-2/+3
2009-02-20Update to latest libdrm.Thomas Hellstrom31-27/+11628
2009-02-17ttm: Adjust swapping target so that current allocation will succeed.Thomas Hellstrom1-3/+7
2009-02-17ttm: Fix a spinlock bug.Thomas Hellstrom1-4/+5
2009-02-13Fix refcounting bugs in swap code.Thomas Hellstrom1-2/+3
2009-02-13More bool conversion.Thomas Hellstrom7-15/+15
2009-02-13Another checkpoint: Use bool for boolean.Thomas Hellstrom7-23/+23
2009-02-13Checkpoint: Use bool for boolean.Thomas Hellstrom19-189/+189
2009-02-11ttm: indentation fixesJerome Glisse1-145/+184
2009-02-10Fix openchrome backwards compat.Thomas Hellstrom1-0/+27
2009-02-10Fix TTM backwards compat to 2.6.25Thomas Hellstrom1-8/+20
2009-02-05openchrome: Use ttm_lock_set_kill functionality to kill processes attempting ↵Thomas Hellstrom1-0/+6
to use the TTM functionality when it is gone.
2009-02-05openchrome: Use the BKL for old drm ioctls as some of them don't appear ↵Thomas Hellstrom1-1/+28
thread-safe, but keep unlocked_ioctl for opencrhome- and TTM code.
2009-02-05drm_core: Avoid dri client deadlocks when master disappears by 1) Notify ↵Thomas Hellstrom3-9/+15
lock waiters when we close the master file descriptor. Not when the master structure is removed. 2) Send a SIGTERM to all clients waiting for the lock.
2009-02-05Work around some X86 PAT oddities by using VM_MIXEDMAP insteadThomas Hellstrom1-6/+6
of VM_PFNMAP.
2009-02-05drm core: Wake up _all_ lock waiters when the master is gone.Thomas Hellstrom2-2/+2
2009-02-04Don't return -ERESTARTSYS to user-space.Thomas Hellstrom1-1/+1
2009-02-04ttm: Add a kill capability to the ttm lock.Thomas Hellstrom2-3/+47
2009-02-03Silence a couple of compilation warnings.Thomas Hellstrom3-3/+4
2009-02-03openchrome: s/1/true/Thomas Hellstrom1-1/+1
2009-02-03radeon: add space accounting to cs codeDave Airlie3-1/+140
2009-02-03Fix for 2.6.29Thomas Hellstrom2-1/+5
2009-02-03Fix a segfault due to the ttm file object disappearing at release while we ↵Thomas Hellstrom1-5/+34
still hold a reference to identify object creator.
2009-02-03openchrome: Fix for 2.6.29Thomas Hellstrom1-1/+0
2009-01-26Check for null pointer when determining VRAM start.Thomas Hellstrom1-9/+13
(Reported by Erik-Jan on openchrome-devel)
2009-01-26Silence error message.Thomas Hellstrom1-2/+0
2009-01-26bump version to 2.4.4Eric Anholt1-1/+1
2009-01-26intel: Retry pin ioctl on -EINTR.Eric Anholt1-1/+4
2009-01-26libdrm: add autoconf check for clock_gettimeJesse Barnes2-1/+8
Should be more portable this way.
generated by cgit v1.2.3 (git 2.39.1) at 2024-05-12 22:29:38 +0000