libdrm: Changes to 'refs/tags/libdrm-2.3.1-1'
Tag 'libdrm-2.3.1-1' created by Julien Cristau <jcristau@debian.org> at 2008-07-13 17:10 +0000
Tagging upload of libdrm 2.3.1-1 to unstable.
Changes since libdrm-2.3.0-4:
Aapo Tahkola (1):
Sync r300_reg.h from mesa driver. #10210
Adam Jackson (1):
Fix some outdated URLs, remove others.
Adrian Bunk (1):
drm_rmmap_ioctl(): remove dead code
Alan Hourihane (9):
Remove old i830 kernel driver.
remove i830 reference
don't copy back if an error was returned.
Add brackets
catch an out of memory condition
fix SAREA
Add error message
Fix for debug memory routines
fix build problems
Alex Deucher (16):
r300: Document more of the RADEON_RBBM_STATUS register.
Fix chip family for RV550
RADEON: production microcode for all radeons, r1xx-r6xx
RADEON: switch over to new production microcode
Radeon IGP: clean up registers and magic numbers
Radeon IGP: wrap MCIND access
Radeon IGP: merge RS4xx/RS6xx gart setup
R300+: fixup PURGE/FLUSH macros
RADEON: write AGP_BASE_2 on chips that support it
RS4xx: fix MCIND index mask
R300+: fixup pixcache flush
RADEON: cleanup radeon_do_engine_reset()
R3/4/5: init pipe setup in drm
RADEON: fix copy/pasto in last commit
RS4xx: separate out RS400 and RS480 IGP chips
RADEON: add get_param for number of GB pipes
Arthur Huillet (12):
Made drm_sg_alloc accessible from inside the DRM - drm_sg_alloc_ioctl is the ioctl wrapper
Added support for PCIGART for PCI(E) cards. Bumped DRM interface patchlevel.
fixed bug that prevented PCIE cards from actually using PCIGART - NV50 will probably still have a problem
NV50 will not attempt to use PCIGART now
fixed object creation code to not Oops on 64bits, worked around memalloc not working on 64bit for PCIGART
now attempting to create PCI object only when there is a pci_heap
now attempting to create PCI object only when there is a pci_heap
Merge commit 'public/master'
applied patch from Ian Romanick fixing PCI DMA object creation code
nouveau: creating notifier in PCI memory for PCIGART
nouveau: added support for software methods, and implemented those necessary for NV04 (TNT1) to start X
nouveau: mandatory "oops I forgot half of the files" commit
B. Rathmann (1):
nouveau: fix memory initialization with multiple cards.
Ben Skeggs (171):
Add stub {get,set}param ioctls.
Use DMA_IN_MEMORY for DMA objects. This is needed for a DDX change that will
Allow cmdbuf location(AGP,VRAM) and size to be configured.
Fix second start of X server without module reload beforehand, and a couple of other fixes.
Add copyright notices while I still remember..
Add pciid for GeForce Go 6150 (0x0244).
Fix some randomness in activating a second channel on NV40 (odd GET/PUT vals). Ch 1 GET now advances, but no ctx_switch.
Some info on NV40's RAMFC
Setup NV40 RAMFC (in wrong location.. but anyway), rearrange the RAMFC setup code a bit.
NV40: FIFO context switching now WorksForMe(tm)
NV40: *Now* fifo ctx switching works for me..
typo
Useful output on a FIFO error interrupt.
Oops, we have more than 4 subchannels..
Remove hack which delays activation of a additional channel. The previously active channel's state is saved to RAMFC before PFIFO gets clobbered.
Leave the bottom 64kb of RAMIN untouched.
Hack around yet another "X restart borkage without nouveau.ko reload" problem.
Merge branch 'nouveau-1' of git+ssh://git.freedesktop.org/git/mesa/drm into nouveau-1
Restructure initialisation a bit.
Completely untested NV10/20/30 FIFO context switching changes.
Dump some useful info when a PGRAPH error occurs.
Check some return vals, fixes a couple of oopses.
Only return FIFO number if the FIFO is marked as in use..
Don't spam dmesg if PMC_INTSTAT is 0
Wrap access to objects in RAMIN.
Use nouveau_mem.c to allocate RAMIN.
Port remaining NV4 RAMIN access from the ddx into the drm.
nouveau: save/restore endianness flag on FIFO switch
nouveau: Alloc cmdbuf for each channel individually
nouveau: return the *actual* type of memory alloc'd to userspace
nouveau: Only clobber PFIFO if no channels are already alloc'd
nouveau: Add nv40-specific PGRAPH code, not hooked up yet.
nouveau: Hook up grctx code for NV4x.
nouveau: Hookup nv40_graph_init.
nouveau: oops, forgot to free RAMIN..
nouveau: Use PMC_BOOT_0 to determine which ctx_voodoo to load.
nouveau: get c51 doing glxgears without the binary driver's help.
nouveau: map pci resource 2 on >=nv40
nouveau: avoid allocating vram that's used as instance memory.
nouveau: nv43 context stuff
nouveau: oops
nouveau: re-add 6150 Go pciid (0x0244)
nouveau: fix getparam from 32-bit client on 64-bit kernel
nouveau: always print nsource/nstatus regs on PGRAPH errors
nouveau: nv4c default context
nouveau: determine chipset type at startup, instead of every time we use it.
nouveau: ack PFIFO interrupts at PFIFO, not PMC.
nouveau: remove a hack that's not needed since the last interface change.
nouveau: intrusive drm interface changes
nouveau: s/fifo/channel/
nouveau: associate all created objects with a channel + cleanups
nouveau: make sure cmdbuf object gets destroyed
nouveau: support multiple channels per client (breaks drm interface)
nouveau: remove unused cruft
nouveau: rework nouveau_fifo_alloc() so the drm can create internal FIFOs
nouveau: move card initialisation into the drm
nouveau: Power up all card units by default on startup.
nouveau: make a note about a bit that breaks some cards
nouveau: NV46 support
nouveau: rename engtab functions
nouveau: prototype PFIFO/PGRAPH engtab API
nouveau: (mostly) hook up put_base again
nouveau: split PFIFO/PGRAPH context creation
nouveau: NV4X PFIFO engtab functions
nouveau: NV4X PGRAPH engtab functions
nouveau: NV04 PFIFO engtab functions
nouveau: NV1X/2X/3X PFIFO engtab functions
nouveau: NV3X PGRAPH engtab functions
nouveau: NV04/NV10/NV20 PGRAPH engtab functions
nouveau: kill some dead code
nouveau: NV49/NV4B PGRAPH setup from jb17bsome and stephan_2303
nouveau/nv04: Set NV_PFIFO_CACHE1_PUSH1 correctly + small tweaks
nouveau: Nuke DMA_OBJECT_INIT ioctl (bumps interface to 0.0.7)
nouveau/nv50: skeletal backend
nouveau: name some regs
nouveau: simplify PRAMIN access
nouveau: never touch PRAMIN with NV_WRITE, cleanup RAMHT code a bit
nouveau/nv10: Fix earlier NV1x chips
nouveau: Hack around possible Xv blit adaptor breakage
nouveau: small RAMFC cleanups
nouveau: rewrite gpu object code
nouveau: enable reporting for all PFIFO/PGRAPH irqs
nouveau/nv50: Initial channel/object support
nouveau: Avoid oops
nouveau: Don't be so strict on <NV50
nouveau: Allocate mappable VRAM for notifiers..
nouveau: Some checks on userspace object handles.
nouveau/nv50: G80 fixes.
nouveau: separate region_offset into map_handle and offset.
nouveau: nuke left over debug message
nouveau: mem_alloc() returns offsets, not absolute addresses now.
nouveau: unbreak AGP
nouveau: nuke internal typedefs, and drm_device_t use.
nouveau: G8x PCIEGART
nouveau: Destroy PGRAPH context table on PGRAPH takedown
nouveau/nv50: get non-default push buffer sizes working.
nouveau: Pass channel struct around instead of channel id.
nouveau: Various internal and external API changes
nouveau: Determine trapped channel id from active grctx on >=NV40
nouveau: Give DRM its own gpu channel
nouveau: Remove PGRAPH_SURFACE hack, it wont work now anyway.
Export some useful ttm functions to drivers.
nouveau: ouch, add nouveau_dma.[ch] files..
nouveau: Init global gpuobj list early, unbreaks sgdma code.
nouveau: enable/disable engine-specific interrupts in _init()/_takedown()
nouveau/nv50: hack up initial channel context from current state
nouveau: return channel id
nouveau: Always allocate drm's push buffer in VRAM
nouveau/nv40: add some missing pciids.
nouveau: silence irq handler a bit
nouveau: Allow creation of gpuobjs before any other init has taken place.
nouveau/nv50: demagic instmem setup.
nouveau/nv40: Fix channel scheduling.
nouveau: Catch all NV4x chips instead of just NV_40.
nouveau: Workaround mysterious PRAMIN clobbering by the card.
nouveau: Allow GART notifiers when using sgdma code.
nouveau: Turn some messages into DRM_DEBUGs..
nouveau: Use count parameter in nouveau_notifier_alloc().
nouveau: Detect memory on NFORCE/NFORCE2 correctly.
nouveau: Poke 0x2230 on NV47 also.
nouveau: Add NV44 ctx ucode. Patch from stillunknown.
nouveau/nv50: Correct thinko for 8800 chips + cleanup a bit.
nouveau: NV4c ctx ucode.
nouveau/nv40: Dump extra info on ucode state if ctx switch fails.
nouveau/nv40: Preserve other bits in 0x400304/0x400310 like NVIDIA do.
nouveau: Use nv41 ctxprog/vals on nv42.
nouveau: Handle multiple PFIFO exceptions per irq, cleanup output.
nouveau: Survive PFIFO_CACHE_ERROR.
nouveau: Cleanup PGRAPH handler, attempt to survive PGRAPH exceptions.
nouveau: revert unintended change.
nouveau: crappy ttm mm init, disabled for now.
nouveau: hook up an inital fence irq handler
nouveau: vram is bar1
nouveau: only pass annoying messages if irq isn't handled fully.
nouveau: cleanups
Fill fence sequence after emit ioctl.
nouveau: disable m2mf buffer move for the moment
nouveau: Use a sw method instead of notify interrupt to signal fence completion.
nouveau: stub superioctl
nouveau: funcs to determine active channel on PFIFO.
nouveau: store user control reg offsets in channel struct
nouveau: Use "new" NV40 USER control regs.
nouveau: Attempt to wait for channel idle before we destroy it.
Merge branch 'fifo-cleanup' into upstream-master
Revert "nouveau: stub superioctl"
nouveau: Also wait until CACHE1 gets emptied.
nv50: some cleanups + small changes
nv50: use dummy page in gart tables
nv50: some needed ctx vals
nv50: abort on chips without ctx ucode
nv50: hook up timer funcs...
nv50: oops, lost some state saving along the way somewhere.
nv50: more small changes
nv05: enable ctx/op methods, and ignore patch valid failures.
nv40: some more nv67 changes
nv40: actually init all tile regs.
nouveau: don't touch NV_USER regs on channel destroy.
nouveau: redo channel idle detection
nv50: force channel vram access through vm
nv40: allocate massive amount of PRAMIN for grctx on all chipsets.
nv40: voodoo - not quite.
nouveau: silence warning
nouveau: enable m2mf for tt<->vram moves, fix fence_poll
nouveau: enable accelerated move to sysmem
nv50: I cave... Add nv84 initial context values.
nv50: PGRAPH exception handling completely different from earlier chips
nouveau: gather nsource in trap_info()
nouveau: restore original NV_PFIFO_CACHES_REASSIGN value in fifo handler
nouveau: disable all card interrupts when unknown PFIFO IRQ occurs.
nouveau: guard against channels potentially not having a context, fix nv50
nv50: enable 0x400500 bit 0 after PGRAPH exception also
Benjamin Herrenschmidt (1):
drm: Fix for non-coherent DMA PowerPC
Brian (6):
More detailed instructions, tips.
s/OJBECTS/OBJECTS/
Clean-ups and reformatting.
reformatting, clean-ups
Added idr_replace() function which was apparently added in Linux 2.6.18
Added bool typedef added in kernel 2.6.19
Brice Goglin (2):
Misc fixes to debian/control and debian/copyright
Install the upstream ChangeLog
Christoph Hellwig (1):
drm: remove drm_follow_page, and drm_ioremap and ioremapfree
Dave Airlie (243):
drm: i810_dma.c: fix pointer arithmetic for 64-bit target
fixup some of the comments in drm_context.c
remove some DRM_ARRAY_SIZE from linux core code
remove local copies of pci domain/bus/slot/num
fix const pointer warnings with file_operations
add static function, and remove bad attributions
drm: lots of small cleanups and whitespace issues fixed up
initial import of nouveau code from nouveau CVS
add pci ids for nouveau
Merge branch 'master' into nouveau-1
remove config.h
add powerpc mmio swapper to NV_READ/WRITE macros
nouveau: add compat ioc32 support
Merge branch 'master' into nouveau-1
use a uint64_t for this not a pointer
fixup fifo size so it is page aligned
drm: fixup page alignment on SAREA map on ppc64
drm: ioremap balanced with iounmap for drivers/char/drm
Revert "drm: ioremap balanced with iounmap for drivers/char/drm"
fix missing DRM_ERR from kernel
use spin_lock_init in via dmablit
make a savage function static from kernel
remove inline from large function
make sizeof match the copy struct
fix some sizes in sis_drv.h
fix comment in r128
use unifdef to clean up some code
fix irq args compatiblity with pre 2.6.19
mm: incorrect VM_FAULT_OOM returns from drivers
[SPARC]: Respect vm_page_prot in io_remap_page_range().
drm: remove all 2.4 support for drm development tree.
remove legacy taskqueue code
remove drm pci from 2.5 days
fixup inclusion of agp.h
fixup i915 defines in create script
fixup i915 return values from kernel
remove do munmap 4 args
add kcalloc compat for before 2.6.10
remove unused via/sis files from lk build
add nouveau symlinks via git
fixup symlinks via Makefile
Merge branch 'nouveau-1'
fixup permission along line of kernel
make build against 2.6.20 hopefully
fixup i810/i830 to use drm_core_ioremap instead of drm_ioremap
i830: complete fix for i830 maps
add export symbol for memory manager
novueau: try resource 3 if resource 2 is 0 length
ttm: make ttm alloc/free into alloc_pages/free_pages
Revert "nouveau: Fill in context_init for nv10-nv3x."
nouveau: add missing symlink
drm: remove last usage of VM_OFFSET
ati: make pcigart code able to handle variable size PCI GART aperture
radeon: make PCI GART aperture size variable, but making table size variable
remove drm_lookup_map unused now
deinline agp_remap along lines of kernel
clean up more of inline functions agp_remap/drm_lookup_map
whitespace cleanup pending a kernel merge
more tab/space conversion
use ARRAY_SIZE
make drm fops const from kernel
oops missing else
cleanup ioctl expansion code
more whitespace issues
fixup return values in drm ioctl
more return values fixup
rename badly named define
drm: remove second spinlock init for tasklet lock
cleanup more whitespace from ttm merge
vm: cleanup drm_vm.c along lines of cleanups queued for kernel
drm/ttm: make sure dev_mapping is set-up for the first opener of the drm
drm/i915: set the bo up at firstopen time not after DMA init
drm/bo: avoid oops if the memory manager for this type isn't initialised
add an install target to the drm modules makefile
i915: use breadcrumb macro everywhere
fixup install target, not sure what I was smokin...
radeon: add support for reverse engineered xpress200m
radeon: bump version for IGPGART support
drm: nopfn warning on 64-bit
remove some whitespace violation
whitespace cleanup
update create_lk_drm.sh script
remove DRM_GETSAREA and replace with drm_getsarea function
remove register usage
drm: add new drm_wait_on function to replace macro
move i915 to new drm_wait_on function
drm: remove unused define DRM_HAS_TTM
drm/ttm: cleanup mm_ioctl ioctls to be separate ioctls.
drm/ttm: cleanup most of fence ioctl split out
libdrm: fix typo on ttm fence interface
drm/ttm: ioctl cleanup for buffer object - user side only
drm/ttm: remove old fence ioctl
ttm: complete drm buffer object ioctl split
drm: cleanup use of Linux list handling macros
drm/ttm: cleanup mm_ioctl ioctls to be separate ioctls.
drm/ttm: cleanup most of fence ioctl split out
ati_pcigart: cut to 80 chars
whitespace fixups from kernel
Revert "drm/ttm: cleanup most of fence ioctl split out"
Revert "drm/ttm: cleanup mm_ioctl ioctls to be separate ioctls."
radeon: add other IGP chipsets
drm: convert drawable handling to use Linux idr
drm: move context handling code to use linux idr
drm: fixup initialisation of list heads and idr
radeon: add support for vblank on crtc2
radeon: refine irq acking for vbl on crtc 2
remove include of linux ioctl32.h from drm drivers
drm: fix radeon setparam alignment issues on 32/64-bit
oops must fix this properly at some point
use krh's idr mods to remove lists from idr code
Merge branch 'master' into cleanup
Merge branch 'drm-ttm-cleanup-branch'
drm: remove typedefs in drm.h to their own section
drm: fix typedef in drm_os_linux.h
drm: detypedef drm.h and fixup all problems
drm: remove ttm userspace typedefs
drm: remove drmP.h internal typedefs
drm: remove hashtab/sman and object typedefs
drm: detypedef ttm/bo/fence code
drm: remove drm_buf_t
drm: remove drm_ref_t
drm: remove internal sman typedef
drm: fixup compat wrappers
drm: fixup old kernel compat code
drm_context: fix braino
drm: remove drm_u64_t, replace with uint64_t everwhere
drm: idr stuff is upstream for 2.6.23
fix some missing whitespace/tab
fix drm no-compile due to BSD :-)
fix missing brace placement for IOC_IN
drm: fix size argument for copy to/from user
drm/bo: set the req pointer for each buffer to validate
drm: fix fencing refcount error
drm/fence: shut up lockdep
drm: add unlocked ioctl code path - not used yet
i915: i965 non-secure batchbuffer bit has moved.
drm: remove XFREE86_VERSION macros
drm_sysfs: update sysfs code from kernel
drm/ttm: fixup fence class naming and interfaces
drm: use fence_class as name instead of class
ttm: returning into dummy causes a buffer object leak
drm: fix page count calculation
Merge branch 'pre-superioctl-branch'
i915: check mask instead of flags for buffer fence types
i915: add superioctl support to i915
i915: fix vbl_swap allocation
drm: rename drmBOUnReference to drmBOUnreference for consistency
drm: drop drm bo list handling code
i915: lock struct mutex about buffer object lookups
i915: split reloc execution into separate function
i915: require mfence before submitting batchbuffer
i915: use a drm memory barrier define
i915: remove relocatee kernel mapping sooner stops mutex taking during sleep
i915: relocate buffers before validation add memory barrier between two
drm/i915: add driver cache flush entry point
drm: call driver load after initing agp subsystem
i915: add backwards compat chipset flushing code
drm: add chipset flushing via agp support
i915: fix compat code on 965/g33
drm/ttm: add support for cached un-snooped mappings.
radeon: set the address to access the aperture on the CPU side correctly
drm: fix oops since user objects..
drm: move some of the OS stuff into the OS header
drm: remove lots of spurious whitespace.
drm: update kernel generator script using v4l script
drm: fix the the typo
drm/ttm: kernel coding style for bo.c and objects.h
drm/ttm: apply linux kernel coding style to bo_lock/move/object/ttm.c
drm: more kernel coding style cleanups
i915: cleanup most of the whitespace
drm/agp: kernel style fixes
i915: compat code doesn't work in i8xx hw.
i915: disable TTM on 8xx chips for now until flushing is solved
i915: oops disable TTM is backwards
intel: add flushing for i8xx chipsets.
i915: remove excess debug output
radeon: refactor out the fb/agp location read/write.
radeon: add initial r5xx support
r500: suggestion from glisse to not add cliprect offset on r5xx
r500: add pci id for X1650
i915: add context handle to superioctl struct
drm: cleanup drm_regman.c coding style
drm: major whitespace/coding style realignment with kernel
r500: add a bunch of all r5xx pci ids..
i965: oops force mi batchbuffer start
drm: more cleanups
drm: oops not a cleanup..
drm: enable udev node creation
patch from -mm kernel to use upper_32_bits
take down stuff after asking driver to unload
drm: move agp include outside CONFIG_AGP as it isn't dependant on agp in kernel
Merge branch 'master' into r500-support
remove duplicate pciids
Merge branch 'r500-support'
mach64: cleanup some of the macro formatting
mach64: some more minor cleanups
fixup i915 compat resource allocation
Revert "Fix pipe<->plane mapping vs. vblank handling (again)"
drm/i915: add support for E7221
i915/flush: get the ret the right way around
i915 make relocs use copy from user
i915: fix missing header when copying data from userspace
drm: add support for passing state into the suspend hooks.
agp: export the correct symbol
drm: reorganise minor number handling using code from modesetting branch
drm/ttm: add ioctl to get back memory managed area sized
ttm: make sure userspace can't destroy kernel create memory managers
drm/bo: allow non-suser priv to add kernel BOs.
flush_agp_mappings commit
drm/radeon: check sarea_priv exists
remove unneeded load call
drm: this u32 should be a dma_addr_t
drm: set rs690 gart base completly.
drm/rs690: set AGP_BASE_2 to 0
ati: fix rs690 igp gart by allocating the page table in 32-bit memory
drm/pcigart: fix the pci gart to use the drm_pci wrapper.
initial r500 RS and FP register and upload code
drm: add new rs690 pci id
move some more r300 regs into not allowed on r500
drm: fixup r500fp submission
r500: fragment program upload is also used to upload constants.
Merge branch 'r500-fp'
i915: fix oops on agp=off
drm/r300: fix wait interface mixup
nouveau: fix return from function..
i915: gfx hw and i945gme fixes from upstream
r500: add allowed range for us config/pixsize
drm: nopage compat fixup for drm_vm
ati_pcigart: fill out 40-bit gart table support properly
ati_pcigart: fixup properly this version might even work
Revert "ati_pcigart: fixup properly this version might even work"
ati_pcigart: stop working in the evenings you mess up too often
ati_pcigart: oops wrong way around not that it actually mattered
r500: add more register ranges for Mesa driver
rs690/r500: vblank support.
drm: fix nouveau warning
r500: add two more register ranges for mesa driver to setup
radeon: bump release date/version for r500 3D support
i915: fix BSD bh, DRI2 not uses anywhere else
radeon: split microcode out into a separate header file.
libdrm: make a branch for libdrm which drops all the TTM apis.
remove include
drm: sg alloc should write back the handle to userspace
Dave Jones (1):
add missing quadro id
David Woodhouse (1):
fix radeon setparam on 32/64 systems, harder.
Eric Anholt (56):
Add chip family flags to i915 driver, and fix a missing '"' in mach64 ID list.
Define __iomem for systems without it.
Warning fix: correct type of i915_mmio argument.
Catch up to new interrupt API, and retire FreeBSD 4.x support here.
Merge branch 'origin'
Merge branch 'crestline-qa', adding support for the 965GM chipset.
Fix FreeBSD build.
Add dry-coded DRM drawable private information storage for FreeBSD.
Add current BSD stuff to .gitignore.
Add some trivial regression tests, one of which fails.
Fix the getclient test (Need this feature for future tests).
Add a test for drawable add, remove, and update.
FreeBSD: Fix the recently added drawable add/remove/update code.
Merge branch 'origin'
FreeBSD warnings cleanup.
Fix linux spinlock macros after the last commit.
Remove DRM_ERR OS macro.
BSD: Replace symlink building with symlinks in git.
Replace filp in ioctl arguments with drm_file *file_priv.
Replace DRM_IOCTL_ARGS with (dev, data, file_priv) and remove DRM_DEVICE.
linux: Make DRM_IOCTL_GET_CLIENT return EINVAL when it can't find client #idx.
Fix copy'n'paste-o in FreeBSD drawable code.
Improve the drawable test to use multiple drawables.
Replace NO_MOVE/NO_EVICT flags to buffer objects with an ioctl to set pinning.
Copy the important parts of object_validate into object_create().
debug print ioctl return value as -integer rather than fffffwhatever.
Add a couple of doxygen comments from reading the code.
Remove the pinned buffer from the LRU when pinning.
Initialize the AGP structure's base address at init rather than enable.
Add doxygen and fix whitespace for drm_auth.c
Add a regression test for authentication.
Bug #11895: Only add the AGP base to map offset if the caller didn't.
BSD: Replace brief description in each file's first line with doxygen later on.
Add simple regression test for getstats (does it not crash the kernel?).
Add a regression test for the setversion interface.
BSD: Fix regression in setversion ioctl (current version not returned).
Require master in setversion test, since it requires auth.
Fix a bad error message in auth.c regression test.
Add a set of tests for DRM locking, exposing issues on BSD.
BSD: Return EINVAL if drm_unlock is called on an unheld or other-owner lock.
Fix dev->agp->base initialization on BSD, and fix addmap range check on Linux.
Merge branch 'master' into bo-set-pin
Merge branch 'origin' into bo-set-pin
Fix mapCount refcounting on unmap, even though the value is unused.
Add some more verbosity to drm_bo_set_pin_req comments.
Merge branch 'bo-set-pin'
FreeBSD: Fill in domain field when supported.
Fix missing \n on some DRM_ERROR in i915_dma.c
Fix capitalization of __linux__ define.
Add additional explanation of DRM_BO_FLAG_CACHED_MAPPED before I forget again.
Merge commit 'airlied/i915-ttm-cfu'
Don't shortcut the info syscall for drmBOBusy on nonshareable objects.
Clarify through the names what drm_ttm_alloc_pages() and friend actually did.
Remove unused DRM_FENCE_FLAG_WAIT_IGNORE_SIGNALS.
Clarify when WAIT_LAZY is relevant to users.
[FreeBSD] Convert from drm_device_t to struct drm_device for consistency.
George Sapountzis (5):
libdrm: remove HAVE_XORG_CONFIG_H and XFree86LOADER ifdef's.
freebsd: remove stray apperance of IN_MODULE.
linux: minor indent.
Revert "bug 7092 : add pci ids for mach64 in Dell poweredge 4200"
mach64: fix after vblank-rework
Haihao Xiang (1):
Delay for a usec while spinning waiting for ring buffer space.
Hasso Tepper (1):
Add DragonFly BSD support for device creation
Ian Romanick (117):
Add support SiS based XGI chips to SiS DRM.
Initial XP10 code drop from XGI.
dos2unix and Lindent
Gut support for pre-2.6 kernels.
linux/config.h is deprecated or gone.
Clean up compile-time kernel feature detection.
Add XGI driver to Makefiles.
Revert over-zealous change from previous commit.
Clean up mixed declarations and code.
Clean up warnings about unused variables and functions.
Convert comment header of xgi_find_pcie_virt to kernel doc format.
Clean up debug log messages in xgi_find_pcie_block.
Minor clean up of variable declarations in xgi_find_pcie_virt.
Remove unused type 'struct xgi_pcie_list_s' / xgi_pcie_list_t.
Fix return type of xgi_find_pcie_block.
Eliminate structure typedefs
Eliminate unused integer and float typedefs.
Replace BOOL with bool.
Eliminate unnecessary defines of TRUE and FALSE.
Replace U(8|16) with u(8|16).
Delete unused arrays s_emptyBegin and s_flush2D.
Convert some PCI-e GART related variable to generic types.
Stop-gap fix in xgi_submit_cmdlist
Clean up xgi_pcie_heap_check
Convert open coded list iterators to either list_for_each_entry or list_for_each_entry_safe
Convert xgi_mem_location enum values to less generic names.
Convert a few more U32 variables to more appropriate, generic types.
Convert weird rtdsc usage to get_cycles.
Major clean up of xgi_ge_irq_handler
Remove XGI_IOCTL_CPUID and associated cruft.
Merge xgi_mem_req and xgi_mem_alloc into a single type.
Adjust the types of the fields of xgi_aperture.
Correct types that are shared with user mode.
Move types shared with user mode to xgi_drm.h.
Eliminiate fields in xgi_info that are duplicates of fields in pci_dev.
Fix ioctl types.
Convert occurances of U32 to other types.
Merge branch 'master' into xgi-0-0-2
Correct errors in the usage of pci_map_page.
Massive log message clean up in xgi_submit_cmdlist.
Eliminate several useless ioctls and associated cruft.
Eliminate unnecessary structures and defines.
Clean ups (primarilly log messages) in xgi_test_rwinkernel.
Log message clean up in WriteRegDWord. Remove unused inline functions.
Make drm_sg_free callable in-kernel.
Add XP5 and XP10 PCI IDs.
Initial pass at converting driver to DRM infrastructure.
Move MMIO drm_addmap (and code that depends on it) to xgi_bootstrap.
Rework xgi_(pcie|fb)_free_all to prevent deadlock.
Debug message and comment clean up in xgi_submit_cmdlist.
Fix error handing related to xgi_cmdlist_initialize.
Delete unused variable in xgi_driver_load.
Change handling of begin types slightly.
Clean up generation of begin commands in xgi_submit_cmdlist
Clean up flush command generation in addFlush2D.
Clean up xgi_cmd_info and associated code.
Make s_cmdring a field in the xgi_info structure instead of a global.
Rename and document fields of xgi_cmdring_info.
xgi_mem_alloc::offset is a hardware offset, so it should be u32, not long.
Remove some extraneous debug messages.
Bump version.
Eliminate XGI_CHECK_PCI_CONFIG.
Minor log message clean up.
Minor log message clean up.
Fix bug preventing X server from restarting.
Fix flags for serveral ioctls.
Fix license formatting.
Pass correct offset to xgi_find_pcie_virt.
Fix typo on previous commit. Sigh...
Merge branch 'master' of ssh+git://git.freedesktop.org/git/mesa/drm into xgi-0-0-2
Eliminate unnecessary (and now wrong) call gto drm_sg_free.
Eliminate use of DRM_ERR.
Convert to new ioctl interface between core DRM and device-specific module.
Use OUT3C5B macro instead of assuming little-endian byte order.
Use DRM_READ/DRM_WRITE macros instead of directly accessing MMIO space.
Fix GE shut-down sequence.
Move additional GE initialization into the kernel.
Refactor register dumping code.
Eliminate allocation "owner" usage.
Eliminate special-case handling of framebuffer (fake) allocation.
Finish removing allocation "owner" infrastructure.
Eliminate unnecessary function xgi_pcie_free_locked.
Unify infrastructure for allocating (not yet freeing) on-card / GART memory.
Replace per-heap semaphores with drm_device::struct_mutex.
Refactor xgi_(fb|pcie)_free_all into xgi_free_all.
Release client memory in reclaim_buffers_idlelocked instead of preclose.
Unify infrastructure for freeing on-card / GART memory.
Associate master file pointer with command list buffer.
Unify alloc and free ioctls.
Use sman memory manager instead of internal version.
Minor clean up of IRQ code. Much, much more to come.
Mask off correct bits in M2REG_AUTO_LINK_STATUS for interrupt handling.
Use DRM_MEMORYBARRIER() macro instead of mb().
After calling drm_sman_cleanup, mark both heaps as uninitialized.
Strobe magic 0xB03F register to flush PCI-e GART table.
Clean up xgi_(en|dis)able_(mmio|ge) and move to xgi_misc.c.
Remove unused interrupt related functions.
Move dwWriteReg to xgi_cmdlist.c, the only file where it is used.
Clean up remaining C++ style comments.
Eliminate unused / useless ioctls.
Implement fence support.
Forgot to add this file on the last commit.
Remove unnecessary include.
Use DRM_SPINLOCK / DRM_UNSPINLOCK macros.
Fix late night dumb-dumb mistake.
Use ati_pcigart for PCI-e GART table handling.
Acutally emit the IRQ (duh) when setting the fence post.
Merge branch 'master' of ssh+git://git.freedesktop.org/git/mesa/drm into xgi-0-0-2
Bump version to 1.0.0.
Merge branch 'xgi-0-0-2'
Fix ioc32 compat layer
Add ioc32 compat layer for XGI DRM.
First round of byte-ordering fixes for PowerPC.
Use 'ifdef __BIG_ENDIAN' instead of 'if __BIG_ENDIAN'
Fix command list submission on big-endian.
Eliminate trailing whitespace from last commit.
Initial pass at porting MGA to vblank-rework
Jakob Bornecrantz (2):
drm: remove unnecessary NULL checks, and fix some indents..
Fixed unlock check on EAGAIN
Jay Estabrook (1):
Fix Alpha domain/bus issue
Jeremy Kolb (17):
nouveau: get nv30 context switching to work.
nouveau: Try to get nv35 pgraph switching working. Doesn't quite yet.
nouveau: Remove write to CTX_SIZE. This gives us proper nv3x PGRAPH switching.
nouveau: cleanup nv30_graph.c
nouveau: fix warning.
nouveau: fix warning.
nouveau: ttm stubs
Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm
nouveau: add missing file.
nouveau: more filling in for ttm. Change copyright since it's based off of radeon code.
nouveau: incorporate darktama's buffer object work.
nouveau: Add darktama's fencing code. Restructure some stuff.
nouveau: put it all together.
nouveau: add darktama's email address to authors section.
nouveau: flip buffer into gart.
nouveau: rename
nouveau: Add ctx_voodoo for NV86
Jeremy Kolb jkolb@brandeis.edu (1):
nouveau: Fill in context_init for nv10-nv3x.
Jerome Glisse (2):
drm: fix dead lock in drm_buffer_object_transfer
drm: don't reset to 0 irq_enabled when client open file descriptor
Jesper Juhl (1):
While reading some code I stumbled across the use of 'err' in
Jesse Barnes (41):
Add new buffer object type for kernel allocations that don't initially have a user mapping.
drm/ttm: allow drm_buffer_object_create to be used in-kernel
Initial checkin of vblank rework. Code attempts to reduce the number
Update vblank code:
Remove unnecessary (and uncommented!) read barrier from the interrupt
Comment new vblank routines and fixup several issues:
First cut at radeon support for the vblank rework.
Remove broken crtc enable checks, radeon does it slightly differently
Remove broken CRTC enable checks and incorrect user irq enable in set_pipe
RADEON: fix race in vblank interrupt handling
Merge branch 'vblank-rework' into vblank
more vblank rework
Merge branch 'vblank-rework' into vblank
Disambiguate planes & pipes for swap operations
Merge branch 'master' of ssh://git.freedesktop.org/git/mesa/drm
Remove plane->pipe mapping from SAREA private after all
Revert drm_i915_flip_t braindamage
update DRM sysfs support
i915: suspend/resume support
Remove unused memory save areas
Merge branch 'master' into vblank-rework, fixup remaining drivers
Cleanup vblank_init and fix drm_irq_install
Use unsigned long instead of u64 in drm_modeset_ctl_t
Make sure PLLs are enabled before writing pipe configuration regs
Fix pipe<->plane mapping vs. vblank handling (again)
Merge branch 'master' into vblank-rework, including mach64 support
i915 irq fixes
Correct vblank count value
Fix IS_I915G macro
Fix thinko in get_vblank_counter
Remove broken 'in vblank' accounting
Fixup modeset ioctl number & typedef usage
Fix hibernate save/restore of VGA attribute regs
Fix vblank enable/disable callbacks
i915: save/restore interrupt state
Fix saveGR array size
Restore pipeconf regs unconditionally
i915: put ARX back into index mode before doing restore
Enum-ectomy of vblank modesetting ioctl
Use fixed sized types in new ioctls
i915: fix off by one in VGA save/restore of AR & CR regs
Jie Luo (1):
[i915] Fix typo in (unused) START_ADDR definition.
Jiri Slaby (1):
Beside the emitted warning, the added cast (u64 -> unsigned) strips out
José Fonseca (3):
mach64: use utf-8
mach64: make buffer emission macros normal functions
mach64: comment bus master / ring buffer behavior and security
Julien Cristau (9):
Merge branch 'libdrm-2_3-branch' of git://anongit.freedesktop.org/git/mesa/drm into debian-experimental
autoreconf
Update changelogs
add missing files
Merge tag 'libdrm-2.3.1' into debian-experimental
update changelogs and refresh patch
Add a note about breaking i915tex_dri.so
debian/rules fixups
Prepare changelog for upload
Jung-uk Kim (2):
Update a bunch of FreeBSD port code.
Bug #11870: FreeBSD hardware lock cleanup fix with multiple opens by a process.
Keith Packard (26):
Revert i915 drm driver name to i915; miniglx doesn't work otherwise
Merge branch 'master' into crestline
Align whitespace with master
Allow vblank interrupts to remain disabled across VT switch.
Add register defines for hw binning
Allow parallel module compile
Allow relocation to be skipped when buffers don't move.
Bump driver minor for relocation optimzations
i915: wait for buffer idle before writing relocations
Make relocation validate client computed values when debugging
Make ttm create/destroy APIs consistent. Pass page_flags in create.
Document drm_bo_do_validate. Remove spurious 'do_wait' parameter.
Document drm_bo_handle_validate. Match drm_bo_do_validate parameter order.
Document fence_class mess in drm_bo_setstatus_ioctl
Document drm_buffer_object_validate function.
Document drm_ttm_set_user.
Clean up and document drm_ttm.c APIs. drm_bind_ttm -> drm_ttm_bind.
Move dummy_read_page from drm_ttm_set_user to drm_ttm_create.
Use dummy_read_page for unpopulated kernel-allocated ttm pages.
Rename inappropriately named 'mask' fields to 'proposed_flags' instead.
Change drm_bo_type_dc to drm_bo_type_device and comment usage of this value.
[915]: more registers for S3 (DSPCLK_GATE_D, CACHE_MODE_0, MI_ARB_STATE)
Switch from PIPE_VBLANK to PIPE_EVENT interrupts.
Save and restore dsparb and d_state regs
On I965, use correct 3DSTATE_DRAWING_RECTANGLE command in vblank
[I915] Handle tiled buffers in vblank tasklet
Kristian Høgsberg (17):
Initialize rwlock using rwlock_init to appease lockdep validator.
Drop drm_drawable_list and add drm_drawable_info directly to the idr.
Fix must-check warnings and implement a few error paths.
Don't take dev->struct_mutex twice in drm_setsareactx.
Use idr_replace trick to eliminate struct drm_ctx_sarea_list.
Take bo type argument out of the ioctl interface.
Drop destroy ioctls for fences and buffer objects.
Move struct drm_drawable_info out of public header file.
Add new shared header file drm_internal.h.
Remove references to the sarea_priv perf_boxes field.
i915: Make sarea_priv setup optional.
i915: Add I915_PARAM_CHIPSET_ID param to get chipset ID.
i915: Only look up dev_priv->mmio_map if it's not already set up
i915: Add a dri2 init path that gets the lock from the dri2 sarea.
Don't free irq resources until after we've unregistered the handler.
i915: Remove leading underscore from struct tags.
Fix one last occurance of struct _drm_i915_batchbuffer.
Kyle McMartin (1):
i915: fix invalid opcode exception on cpus without clflush
Li Zefan (2):
drm: fix memset size error
drm: don't cast a pointer to pointer of list_head
Maarten Maathuis (20):
replace instance of SA_SHIRQ with IRQF_SHARED
Add context init voodoo and context switch code for NV41.
nouveau: Add Quadro NVS 140 pciid
nouveau: Fix dereferencing a NULL pointer when erroring out during initialization.
linux-drm: Obey device class requirements when detecting devices.
nouveau: Switch over to using PMC_BOOT_0 for card detection.
nouveau: NV47 context switching voodoo + warning
nouveau: Remove excess device classes.
nouveau: A char is signed, so it may overflow for >NV50.
nouveau: PMC_BOOT_1 was not mapped.
nouveau: Properly identify NV40 and NV44 generation.
nouveau: Fix warning in nouveau_mem.c
nouveau: NV40 can/should now be able to run after the blob.
nouveau: A single define of dma skips is more than enough.
nouveau: Remove some random (french) comment.
nouveau: Add ctx values for nv86.
nouveau: forgot to add a break
nv50: primitive display interrupt handler.
nv50: primitive i2c interrupt handler
nouveau: NV9x cards exist as well.
Maciej Cencora (1):
drm: add initial rs690 support for drm.
Matthieu Castet (35):
For nv10, bit 16 of RAMFC need to be set for 64 bytes fifo context.
Add basic pgraph context for nv10.
Merge branch 'master' of git+ssh://matc@git.freedesktop.org/git/mesa/drm/
nouveau : remove useless init : we clear RAMIN before
nouveau: add and indent pgraph regs
nouveau: first step to make graph ctx works
nouveau: nv20 graph ctx switch.
nouveau: opps restored the wrong channel
nouveau: opps nv20 ctx ramin size was wrong
make works ctx switch on nv10.
nouveau: nv10 ctx switch, some regs are nv17+ only
nouveau: nv ctx switch opps the size of array was wrong
nouveau : set the correct PGRAPH_CTX_CONTROL register
nouveau : nv10 ctx switch fix
nouveau: fix usage of PGRAPH_CTX_CONTROL on nv20+
nouveau: nv10 per channel init from ddx
nouveau : fix fifo context size for nv10
nouveau : nv10 graph clipping values were forgoten in ddx to drm commit
nouveau : nv10 graph move clipping value to per channel init
nouveau : fix enable irq (in the previous code all irq were masked by engine
nouveau : nv10, nv20, nv30 : don't save all channel in the same RAMFC entry
nouveau : fix some potential crashes with objects causing hash collision
nouveau : nv1x graph reworks
nouveau : add NV04_PGRAPH_TRAPPED_ADDR definition
nouveau : nv10 pipe ctx switch load/save.
nouveau : nv10 fix NV10_PGRAPH_CTX_USER save/load
nouveau : clean chan->pgraph_ctx stuff. We now do a static init of the array.
nouveau : nv1x fix strange corruption
nouveau : stop the fifo of the channel we are deleting
nouveau : nv04 don't use chan->pgraph_ctx array
nouveau : pgraph_ctx dynamic alloc for nv04, nv10
nouveau : nv20_graph replace nouveau_graph_wait_idle by nouveau_wait_for_idle
nouveau : nv30 remove harcoded NV20_PGRAPH_CHANNEL_CTX_TABLE
nouveau : print correct value in nouveau_graph_dump_trap_info for nv04
nouveau : nv10 and nv04 PGRAPH_NSTATUS are different
Maurice van der Pot (1):
nouveau: fix RAMHT wrapping
Michael Buesch (1):
drm/linux-core: drmP.h compilation fix
Michel Dänzer (49):
i915_vblank_tasklet: Try harder to avoid tearing.
Track linux-core symlinks in git.
Unshare drm_drawable.c again for now.
Core build fix for BSD.
Unify radeon offset checking.
linux-core: Make git ignore generated module symbol version files.
Make git ignore Emacs style backup files and cscope files.
i915: Fix a DRM_ERROR that should be DRM_DEBUG.
radeon: Fix u32 overflows when determining AGP base address in card space.
Make git ignore generated config.h.in.
i915: Unify breadcrumb emission.
i915: Page flipping enhancements.
i915: Improved page flipping support, including triple buffering.
Add DRM_VBLANK_FLIP.
i915: Add support for scheduled buffer swaps to be done as flips.
i915: Fix test for synchronous flip affecting both pipes.
i915: Don't emit waits for pending flips before emitting synchronous flips.
i915: Only clean up page flipping when the last client goes away, not any one.
i915: Eliminate dev_priv->current_page.
i915: Do not wait for pending flips on both pipes at the same time.
i915: Only wait for pending flips before asynchronous flips again.
Merge branch 'i915-pageflip'
Bump version patchlevel so it can be tested for new functionality.
Fix build against older kernels.
radeon: Don't mess up page flipping when a file descriptor is closed.
Fix userspace ABI breakage from 3c384a9ad5f964709a237cfe035ea5d6df2da5fa.
drm: make sure the drawable code doesn't call malloc(0).
Fix memory leaks in vblank error paths.
Return current sequence number to userspace after blocking wait for vblank.
i915: Fix tests for vblank interrupts being enabled on CRTC by X server.
Wake up vblank waitqueue in drm_handle_vblank().
Make vblank waitqueue per CRTC.
Remove DRIVER_IRQ_VBL(2).
i915: Fix handling of breadcrumb counter wraparounds.
radeon: VBlank rework fixups.
Remove mask parameter from radeon_acknowledge_irqs().
drm_modeset_ctl_t fixes.
Use drm_calloc instead of assigning 0.
Fix vblank wait condition.
radeon: Acknowledge all interrupts we're interested in.
Restore pre-idr semantics for drawable information.
Simplification for previous commit.
One more spinlock initializer cleanup.
radeon: Improve vblank counter.
i915: Fix scheduled buffer swaps.
i915: Reinstate check that drawable has valid information in i915_vblank_swap.
Don't build without any optimization on Linux.
Revert part of earlier commit that caused an unresolved symbol for i915.
linux: Clean up vblank related resources from drm_irq_uninstall().
Mike Isely (1):
drm: Fix race that can lockup the kernel
Mirko (1):
Add new RV380 pci id
Márton Németh (1):
drm: cleanup DRM_DEBUG() parameters
Nian Wu (17):
Merge branch 'master' into crestline
Merge branch 'master' into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Merge git://proxy01.pd.intel.com:9419/git/mesa/drm into crestline
Oliver McFadden (19):
Guess another unknown register used for R300 pacification.
Reply to: