[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

libdrm: Changes to 'refs/tags/libdrm-2.4.5-1'

Tag 'libdrm-2.4.5-1' created by Julien Cristau <jcristau@debian.org> at 2009-03-24 21:25 +0000

Tagging upload of libdrm 2.4.5-1 to unstable.

Changes since libdrm-2.3.1-2:
Alan Hourihane (96):
      Merge remote branch 'origin/modesetting-101' into modesetting-101
      Remove extraneous drm_crtc.h include
      Use drm_framebuffer instead of drm_crtc for fb & fbt
      Correct PCI ID for i845
      Fix SDVO outputs
      Fix a register read that was swapped SDVOB/SDVOC
      Move SDVO PENDING check to read_response so all SDVO commands
      Check status after SDVO command for sdvo_set_control_bus_switch
      Support the RGB outputs of SDVO cards, which are essentially
      Merge remote branch 'origin/modesetting-101' into modesetting-101
      When no valid EDID modes are available, we bail.
      free the duplicated mode.
      Fix build problem
      Fix return status
      Shouldn't free the mode here. It's done later on.
      Just some minor cleanups.
      Fix build problem
      Grab the default mode if the preferred mode isn't available.
      Move destruction of crtc as intelfb_remove uses the crtc to locate the fb.
      bring in change from drm_fb.c
      Set crtcinfo on temporary mode
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      Fix merge problem.
      Move fbo creation to the specified fb driver which gives
      Change some printk's to DRM_DEBUG's
      Fix type/flags usage problem to check for preferred modes.
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      merge fixes
      Bring back code from merge that was accidentally removed.
      Move out the code from i915_dma_cleanup to unload to match
      don't copy back if an error was returned.
      no need to copy to/from user as the unlocked ioctl does
      Add brackets
      Add some more checks to modelist walking for matching
      Create memory pool for TT memory
      Set the fb_base, so userspace applications can actually work
      Fix some buffer teardown problems.
      should be unlocked
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      Fix a crash on X startup
      Need fb attached
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      pass pointer for drmModeRmFB
      Pass pointer to drmModeRmMode.
      Use _size
      add missing lock
      remove duplicate and obsolete ioctl statements
      remove unused define
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      Remove duplicate
      Add missing round_jiffies_relative() for older kernels
      Fix function declaration
      build fix for older kernels
      fix some warnings
      build fix for older kernels
      DRM_CONTROL_ALLOW logic was reversed
      make modefb/modedemo match each others test output.
      tweak it
      update app to cycle through 4 fbdev's
      Revert "DRM_CONTROL_ALLOW logic was reversed"
      After the previous revert fix libdrm to start at minor 1
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      Fix up conflicts for DRI2 (untested)
      compatibility code for  pci_get_bus_and_slot()
      minor test fixes
      fix SAREA
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      silence warning
      Implement short circuit for base change only
      define PRETHAW
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      Add FENCE registers to MMIO list
      Fix connector description table
      propogate failed fixups back up
      Use ARRAY_SIZE
      Check mode before adding to EDID
      Add property info.
      build fix
      Add support for monitor hotplug signals/waits
      global hotplug events happen in the pipe A stat register,
      Add sample code to test hotplug events
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      Fix green offset
      Older kernels don't have kobject_uevent_env(), so punt
      Fix test applications for recent DRM changes
      Fix build problems
      Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      Fix warnings
      more checks for NULL encoder so we don't segfault.
      Synchronize the DDC EDID read to it's fb_ddc.c counterpart
      set to 2.4.0

Alex Deucher (46):
      RADEON: fix typo in last commit
      RADEON: switch IGP gart to use radeon_write_agp_base()
      RADEON: Add untested support for RS400 chips
      RADEON: 0x1002 0x5657 is actually an RV410
      NV50: s/FALSE/false/
      on_each_cpu() compat fixup from krh
      Brute force port of legacy crtc/encoder code
      unify connector, i2c handling for atom and legacy
      Add legacy dac detect stubs
      various cleanups
      LUT updates
      atom: implement crtc lock
      Restructure cursor handling and add support for legacy cursors
      Convert COM BIOS to table offset lookup function
      set base in legacy crtc mode set
      Get legacy working finally
      Fix warnings
      Add additional quirks from ddx
      Fill in and make use of more com bios tables on legacy chips
      Add com bios asic init bits
      radeon: get primary dac adj info from bios tables
      radeon: remove unused legacy state
      radeon: first pass at bios scratch regs
      radeon: first pass at legacy dac detect
      radeon: legacy lvds updates
      radeon: fix warning from radeon_legacy_state removal
      Merge branch 'modesetting-gem' of git+ssh://agd5f@git.freedesktop.org/git/mesa/drm into modesetting-gem
      radeon: fix legacy LVDS
      Export drm_put_minor
      Radeon: restructure PLL data
      radeon: add function to configure PCIE lanes
      radeon: Add functions to set mem/eng clocks
      radeon: pll and interlace updates from the ddx
      radeon: rmx_fixup() fixes for legacy chips
      radeon: first pass at using atombios on r4xx hw
      Merge branch 'modesetting-gem' of git+ssh://agd5f@git.freedesktop.org/git/mesa/drm into modesetting-gem
      radeon: make atom on r4xx a module option
      radeon: use atom for ext tmds on r4xx
      radeon: Add support for HD2100 IGP (RS740)
      radeon: fix bus master enabled bits on newer asics
      radeon: fix duplicate define in my last commit
      radeon: add comment to clarify bus mastering on PCIE chips
      radeon: pull in recent fixes from ddx
      radeon: fix some fallout from the busmaster disable cleanup
      radeon: fix error in busmaster enable logic

Arjan van de Ven (1):
      Apply a few stylistic cleanups to match kernel code.

Ben Skeggs (34):
      nouveau: allocate drm-use vram buffers from end of vram.
      nv50: use same dma object for fb/tt access
      nv50: oops, keep VRAM allocations aligned at 64KiB - that's our page size..
      nv50: when destroying a channel make sure it's not still current on PFIFO
      nouveau: allocate drm-use vram buffers from end of vram.
      nv50: when destroying a channel make sure it's not still current on PFIFO
      nouveau: interface changes for nv5x 3d
      nouveau: interface changes for nv5x 3d
      nv50: use same dma object for fb/tt access
      nv50: add support for chipset 0x92
      nv50: add ctxprog for chipset 0xaa
      nv50: add ctxprog for chipset 0x50
      nv50: add initial context to match ctxprog for chipset 0x50
      nv50: add initial context for chipset 0xaa
      nv50: move context-related tables a separate header file
      nv50: symlink nv50_grctx.h to linux-core...
      nv50: update context-related tables for original 8800GTS
      nv50: support NV94 chipset
      nv50: ack nsource to prevent continuous protection fault irqs
      nv04-nv40: correct RAMHT size
      nv50: support chipset NV96
      nouveau: bring in new mm api definitions, without the actual mm code
      nouveau: install libdrm_nouveau with libdrm
      nouveau: fix some issues where buffer objects never get freed
      drm/nv50: let the card handle the initial context switch
      drm/nv50: correct ramfc pointer in channel header
      drm/nv50: use a slightly different initial context for nv96
      drm/nv50: add context info for nv98
      drm/nv50: fix nv9x chipsets
      libdrm: make libdrm_nouveau conditional, disabled by default
      nv50: context info for chipset 0xa0
      nv40: fail completely if we don't have a ctxprog for the chipset
      libdrm/nouveau: fix dma debugging
      libdrm/nouveau: free drmVersion after we're done with it

Branden Robinson (4):
      Set svn:keywords property to "Id" (keyword already present).
      Note probable out-of-dateness of maplink() function body.
      Add high-level documentation to XSFBS shell library, intended to reduce
      Test for existence of debian/patches directory before creating a symlink to it.

Brice Goglin (5):
      Fix displaying of patches applied by quilt.
      Minor fixes in the patching system.
      Fix "display the output of quilt push/pop".
      Do not call laptop-detect, let the only user call it directly
      Update xsfbs/quilt patching rules to be compatible with the

Coleman Kane (1):
      Change prototype of drmIoctl to unsigned long request.

Dave Airlie (408):
      copy over some files and reorg radeon to add ttm fencing not working yet
      radeon: de-static irq function, fixup fence/buffer
      radeon: enable buffer manager
      add initial PCIE GART framework for TTM
      radeon: add setup for PCIE GART ttm
      borrow edid.h from fb directory
      Initial import of modesetting for intel driver in DRM
      initial userspace interface to get modes
      checkpoint commit: added getresources, crtc and output
      checkpoint commit: implement SetCrtc so modes can in theory be set from user
      add back compat for bool
      make crtc_config be named mode_config
      use the baseaddr at least
      only bo finish at driver unload
      only init at driver load
      comment out unworkable code
      add initial drm_fb framebuffer
      add copyright statement
      add getfb ioctl
      allow framebuffer changes on the crtc setup
      fix unbalanced lock and make sure mode list has modes so lvds code doesn't crash
      set bracing style like Linux
      only initialise modes when fbcon or fbset asks for it
      cleanup framebuffers on drm unload
      use FB everywhere
      another large overhaul of interactions with userspace...
      add some missing export symbols
      I don't think this dec is necessary and my stuff all works without ..
      backout alanh's broken commit
      don't crash if no desired mode
      Merge branch 'origin' into modesetting-101
      make sure TMDS gets a crtc
      backout SDVO control bus check and comment why this is a bad idea
      fixup vrefresh reporting, it should now be *1000 in userspace
      drm: make mode numbers no change by comparing probed modes
      i915: fix vblank pipe setup
      Merge branch 'origin' into modesetting-101
      fix unusued variable
      Add support for user defined modes
      cleanup usermodes on drm mode setting shutdown
      Merge branch 'origin' into radeon-ttm
      cleanup pcigart ttm for new backend layout
      WIP cleanup
      complete PCIE backend for ttm
      take the lock earlier in ttmtest
      invalidate gart tlb on PCIE after table change
      add wbinvd calls
      set start to gart_vm_start at least
      radeon: PCIGART memory is Can't map aperture as well there is one
      Merge branch 'radeon-ttm' of git://people.freedesktop.org/~airlied/drm into radeon-ttm
      radeon: add VRAM support for radeon ttm
      drm/radeon/ttm: more VRAM fixes
      radeon: make vram fixed type
      i915: cleanup pageflip derefs sarea even if no sarea exists
      Merge branch 'master' into modesetting-101
      tests: update for new drm interface
      intel: fixup || vs |
      drm: don't remove mappings added by the driver
      modes: pass type to userspace for preferred showing
      drm: send correct fb id to userspace
      libdrm: add crtc/output ids to userspace interface
      libdrm: return crtc id to userspace
      Merge branch 'origin' into modesetting-101
      drm: make fb modes use usermode add/remove interface
      add _ioctl to the end of two more ioctls
      fb: make fb interface use user mode attach/detach for adding modes
      drm/modesetting: add initial gettable properites code.
      modesetting API change for removing mode ids and making modes per output.
      finish of mode add/remove, just have attach/detach modes
      mode: copy back the mode if is valid correctly
      add flags to mode debug print
      hooks up sync and flags and also clocks to get mode matching better
      arrgggh.. make all ioctl structs 32/64-bit compatible hopefully.
      more WIP on blobs..
      add property blobs and edid reporting support
      oops initialise variable to false
      check previous mode first
      retab intelfb code
      modesetting: add dpms property and initial settable property ioctl
      modesetting: fixup property setting and add connector property
      Merge branch 'master' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      remove output names
      we should not be unlocking this here
      lockdep warned about a possible locking dependency
      HERE BEZ HACKZ.. magic variable to make shit work
      Merge branch 'modesetting-airlied' into modesetting-101
      drm: move drm_head to drm_minor and fix up users
      crtc: fixup allocation size
      drm: add initial support for a drm control device node
      drm: only call suspend/resume on control node
      add control ioctls
      allow control getversion
      add internals for opening a control node
      add control node open
      fixup crtcinfo on modes from userspace
      Merge remote branch 'origin/master' into modesetting-101
      don't reinit ring if already initialised
      if irq already enabled well just keep trucking
      drm: re-write minor number allocation to use an idr.
      Revert "After the previous revert fix libdrm to start at minor 1"
      major port of multi-master ideas into modesetting
      start moving over to proper hierarchy wrt master accesses
      missing bits
      remove drm_minors_limit
      switch naming to new proposed scheme
      various fixes from trying to get userspace started
      fb: fixup the offset by getting it from the right place
      ttm: make sure userspace can't destroy kernel create memory managers
      drm/fb: get rid of offset from structure use bo offset
      remove more offset
      hopefully shit works now without this...
      add ioctl to get back memory managed area sized - used for kernel inited areas
      modesetting: fix memory leak and misallocation
      drm: fix pick crtcs mode selection code
      i915_mmio: add overlay regs
      i915: fix typos
      drm: add a check for if modesetting is supported.
      drm: add modesetting as a driver feature.
      drm: change fb api to take a bo handle not the bo pointer.
      remove unused functions + include header file
      Merge remote branch 'origin/master' into modesetting-101
      libdrm: fix warnings in mode code
      drm: fixup compat with old x.org drivers
      fixup previous merge
      Merge branch 'master' of ../../drm into modesetting-101
      Merge branch 'master' of ../../drm into modesetting-101
      woah somehow got these upstream
      drm: fixup for new sysfs API
      I really screwed up that merge somehow
      Merge branch 'master' of ../../drm into modesetting-101
      drm/modesetting: fixup irq removal on exit
      worst merge effort ever
      actually turn the irq off
      make startup of Xorg smoother if the mode doesn't change.
      drm: we already worked out the pitch. multiplying by 4 is just madness..
      drm: hopefully fix cursors on 965
      drm: fix oops on unload.
      drm: fix fd closing ordering.
      drm: select the correct master to attempt to remove the lock from.
      drm: pick correct master for cleaning up
      i915: safety check the sarea map still exists
      drm: add master set/drop protocol
      drm: remove authentication on master exit.
      Merge remote branch 'origin/master' into modesetting-101
      cursor: pass handle not BO.
      drm: fix typo from previous merge
      drm_mode: initial replacefb implemenation
      drm: fix replacefb to change fb properties
      intel: set correct limits on screen width/height from DDX
      drm: set crtc->fb to NULL
      drm: check for NULL fb here, shouldn't happen but avoid oops for now
      i915: fix vbl swap for multi-master
      Revert "i915: fix vbl swap for multi-master"
      drm: fix oops on reading proc file with no master
      drm: masters are always authenticated
      drm: remove root only from a lot of drm ioctls to get stuff running as non-root
      modeset: disable radeon ms by default
      modesetting: reorganise code into core and helper functions.
      modesetting: reorganise out crtc/outputs are allocated.
      drm/modesetting: add initial encoder structures and setup functions
      drm: remove unused init func from outputs
      drm: add encoder / get encoder to the modesetting resources interface
      libdrm: add encoder retrival
      drm: add more encoder interfaces
      drm: add encoder ids to the output handling
      drm: add red hat copyright.
      drm: init the encoder list/count
      drm: add encoder free function
      tests: add basic encoder reading to test
      drm: add encoder attach/detach
      drm: fix a couple of bugs in the encoder return to userspace
      drm: attach an encoder.
      modesetting: the great renaming.
      modesetting: drop crtcs/clones from the connectors
      drm: switch possible crtc/clones over to encoders
      r500: attempt to make AGP work by programming agp base in the MC correctly
      drm/modesetting: another re-org of some internals.
      drm: fixup some interfaces so test code works again
      drm/modesetting: move some connector functions to helper.
      drm/modesetting: add best encoder finding for modesetting
      drm: fixup encoder picking in set_config stage
      Merge branch 'modesetting-101-encoders' into modesetting-101
      drm: add functions to get/set gamma ramps
      drm: only report framebuffers available on this fd.
      drm/modesetting: redo object handles around a core object.
      drm: initial mode object groups.
      drm/ati_pcigart: use proper page mapping function
      drm: sg alloc should write back the handle to userspace
      drm: make mode comparison more betterer.
      drm/modesetting: overhaul the fb create/delete.
      drm/modesetting: pass object handle to driver !bo
      drm: remove sysfs in driver for now.. should probably be in helper
      drm/modesetting: bo not used anymore
      intel: use kzalloc
      drm/sysfs: don't try an unregister if not registered
      drm: introduce generation counter to interface.
      libdrm: shouldn't rely on this
      intel: report a known connector
      drm: fix hotplug oops
      modesetting: initial attempt at debonging fb
      modesetting: add surface width/heights
      modesetting: use surface width height for buffer allocs
      modesetting: fix fb clearing up
      drm: modesetting unify the hotplug init paths a lot.
      drm/modesetting: attempt to make fb code more sane
      drm/modesetting: more fb interface cleanups
      drm/modeset: add more debugging and fixup some fb enable/disabe bits
      intel: if no spare crtc exists don't just take one.
      drm/intel: make hotplug just be an event
      intelfb: move mode sets into the intel crtcs
      intelfb: admit fbdev is crap and punt on trying to resize to a larger fbdev.
      intelfb: remove duplicate define
      intelfb: add multi fb paths
      Merge remote branch 'origin/modesetting-101' into modesetting-101-fb
      intel: don't set the mode on the framebuffer if isn't set to scanout our framebuffer
      drm: fix up fb resize again
      r300/r500: add hier-z regs
      ati_pcigart: split out the page insert function
      drm: fix the ioctl to not believe userspace.
      r300: fix warning
      drm: only use kernel ioctl cmd when doing a core ioctl.
      agp: use true/false instead of TRUE/FALSE
      drm: fix encoders get permissions
      modesetting: lookup blob using correct identifier.
      modesetting: rip out all of the generation code.
      Merge remote branch 'origin/modesetting-101' into modesetting-gem
      modesetting/helper: fix array overrun - count should be reset here
      drm/pcigart: fix warning
      drm: add fix for PAT on radeon with 2.6.26
      drm: fix missing symbol export
      drm/radeon: fixup 0 vs NULL
      radeon: remove microcode version
      radeon: fix type DST vs Z cache flush
      radeon: fix typo with a better typo
      drm: use correct mode destructor
      drm: don't set the signal blocker on the master process.
      Merge remote branch 'origin/modesetting-101' into modesetting-gem
      modesetting: pass file_priv into cursor set functions
      nv50: remove TRUE/FALSE
      modesetting: pass file priv to cursor
      drm: don't set the signal blocker on the master process.
      Merge remote branch 'origin/modesetting-101' into modesetting-gem
      radeon: add initial atombios modesetting and GEM -> TTM translation layer.
      tests: add some basic radeon gem tests
      radeon command submission start
      r300: initial command stream parser for packet 0.
      r300: add some packet 3 decodes
      radeon_cs: extract some offsets from packet data
      radeon_cs: add relocate hook for mm and non-mm relocations
      radeon: move code around putting emit into cs
      i915: add version checks for opregion on old kernels
      i915: more version checks
      radeon: add some handles to the sarea + kernel mm available check
      radeon: hack gem to get an offset back for Mesa.
      TTM: remove API and userspace objects.
      drm: add fault handler support so as to be more like possible upstream
      drm: add fault handler support so as to be more like possible upstream
      drm: userspace rip out TTM API
      drm: remove object hash
      Revert "i915: Move all of the irq install/uninstall to load time."
      drm: remove unneeded debugging
      modesetting: pick_crtcs can't be static
      i915/radeon: use rmmap locked
      drm: fixup master code to use krefs
      drm: remove unused types
      radeon: don't use ring if cp not going
      radeon: add userspace mm enable switch
      drm: finish bo after lastclose
      radeon: fix blit due to registers wrong
      radeon: fix setting new memmap in right place
      pcigart: fixup memset + remove wbinvd
      drm: don't teardown things in modeset paths
      radeon: just evict to TT not cached
      radeon: fixup PCI GART table with GEM enabled
      radeon: add initial support for legacy crtc/encoders.
      drm: TRUE/true
      drm: add OS_HAS_GEM option.
      drm_vm: fix for build on 2.6.22
      radeon: FEDORA: add old DMA buffers on top of GEM
      radeon: add userspace call for mm support check
      radeon: make buffer swap for older drivers work again on GEM
      radeon: use mm_enabled variable to denote memory manager running
      radeon: remove debugging
      radeon: FEDORA: patch to make 3D driver work
      radeon: fix kernel_mm properly
      Merge branch 'radeon-gem-cs' into modesetting-gem
      i915: fixup from last merge hopefully
      radeon: add copy/solid regs for rn50
      radeon: add support for memory map init
      radeon: fix LVDS modes problem
      radeon: reserve 64k of VRAM for now for text mode so we don't trample it
      radeon: oops set correct scratch
      drm: add gpu create script
      libdrm: add udev support.
      drm: add memory clean flag.
      radeon: fixup domains and use them properly
      drm/ttm: export drm_bo_add_ttm
      drm: add move zero function to memset unclean buffers
      radeon/ttm: add support for zeroing the contents of VRAM buffers
      radeon: avoid oops on encoders with no crtc set
      radeon: braino pointed out on mailing list
      drm: fix some whitespace
      radeon: fixup checks for crtc in dpms paths
      radeon: limit LVDS to first CRTC for now
      drm: port X crtc picking algorithm.
      radeon: get lvds atombios info
      radeon: fixup LVDS mode getting
      radeon: make writeback work after suspend/resume.
      drm: fix sysfs error path.
      radeon: fixup a number of avivo checks for rs690
      radeon: rs690 GART tables need to be in uncached memory.
      radeon: change interface from headers add pin_domain into padding
      modesetting: Add helper to force restore modes on crtcs at resume time
      radeon: add initial suspend/resume support
      make text reserve 256k
      radeon: don't do full edid for detection purposes
      radeon: remove unneeded debugging
      radeon: sort out atom vs combios tables for r400 cards
      radeon: fixup reference counting properly
      radeon: fix return value
      radeon: do proper memory controller init and setup
      radeon: fail properly if we can't create the ring.
      radeon: disable blit moves
      radeon: fixup GEM domain setting - allows more userspace paths
      radeon: fixup buffer and cs bits
      fixup radeon stuff - need to checkout irqs
      radeon: tmds bracket failure found by krh.
      radeon: fix voverplus calculation
      drm: fix brace placement
      radeon: fix minor cursor issues
      radeon: add parsing for r6xx object tables
      radeon: add r600 modesetting registers writes
      Merge remote branch 'origin/master' into modesetting-gem
      Merge remote branch 'origin/modesetting-gem' into modesetting-gem
      drm/radeon: fixup clean flag handling
      radeon: fix pin ioctl interface to mesa can find offset for pinned buffers
      radeon: fix alignment so Xv works again
      drm: add create gpu tree script
      radeon: PCIE cards don't appear to have explicit bus master
      libdrm: don't depend or link to libdrm_intel
      drm/radeon: initial suspend/resume fix.
      drm: add discardable flag.
      radeon: use discardable flags on no backing store objects
      drm: remove stray debug code
      radeon: move memcpy until after CP is stopped
      radeon: fix buffer copying for VRAM->TT
      radeon: re-enable hw blits for copying from VRAM
      radeon: fixup suspend/resume bus master enable
      radeon: fixup interrupt suspend/resume
      radeon: add CS support for r100/r200 in 2D driver
      radeon: add initial agp support.
      libdrm: don't depend or link to libdrm_intel
      link libdrm_intel properly
      drm: make handles 32-bits again not sure why they changed
      radeon: add some more r100 support to test AGP
      radeon: fix unused agp functionality
      radeon: export radeon_modeset
      radeon: don't enable dynclks on rs48x
      radeon: workaround failure to parse some rs48x edid
      radeon: fix small typo in agp code
      radeon: fix some warnings
      radeon: make writeback work again
      radeon: fixup scratch register interactions properly
      radeon: update proper chip family
      radeon: fix accessible VRAM sizing
      radeon: pull bus master enable into its own function
      radeon: add r423 bits to modesetting
      radeon: release agp on module unload
      radeon/drm: fixup ref counting in on fb objs
      radeon: make new CS2 command submission interface port older interface to this
      radeon: don't copy to user the cs ids
      radeon: CS2 make it all work with new relocs style
      radeon: fix free after refcount
      radeon: fix some warnings
      radeon: remove unused gem indirect ioctl
      radeon: rs480 fixes for bus mastering
      radeon: add wait rendering API
      radeon: only enable dynclks if asked for
      radeon: add proc debugging for interrupts/ring
      radeon: fix race in sysfs
      radeon: overhaul ring interactions
      radeon: add more debugging
      radeon: setup isync cntl properly
      radeon: enable DVI-D + HDMI connectors.
      radeon: set dma bufs bo type to a kernel type
      radeon: add more HDMI bits
      radeon: commit ring after emitting the buffer discards
      radeon: disable debugging message
      radeon: disable AGP for certain chips if not specified until we figure it out
      radeon: add mtrr support for VRAM aperture.
      radeon: fix ROP values for the paint ROP
      radeon: make build again
      radeon: fixup vram visible calculation to take a/c pinned objects for now
      radeon: force all ring writes to 16-dword alignment.
      radeon: disable HDP read cache for now
      radeon: fix ring tail overflow issue since alignment
      drm/radeon: add uncached allocator to drm ttm code.
      modesetting: set the crtc x,y after the mode base change
      radeon: avivo cursors are across the full surface.
      drm/radeon: add dpms connector functions
      radeon: fix fence race condition hopefully
      radeon: upgrade atom headers
      radeon: add gart useable size to report to userspace
      drm/bo: fix stupid lock imbalance
      libdrm: bump to 2.4 for lib version
      radeon: only write irq regs if irq is enabled
      drm: remove drmstat/dristat from linux-core build

David Airlie (20):
      mode: fixup problems with framebuffer add function
      fixup numerous issues with adding framebuffer support
      set the base address of the CRTC correctly
      export output name to userspace
      re-tab and fixup the i915GM get core clock function to actually work
      fixup true/false in intel_sdvo.c
      fixup sarea writes for set pipe base and dpms
      add sdvo debugging output
      fixup calculation to make sdvo work
      use fb pitch and fix up some whitespace
      add an fb count + id get to the get resources code path
      add support for setting a framebuffer depth
      fixup framebuffer depth
      oops for 32 pitch.. hey I can see stuff on fbcon now.. it looks like text.. just a bit garbled
      line_length calculation was incorrect.. I now can get fbcon to run
      revert LVDS destroy - this oops on sysfs on sdvo init of i2c bus
      export drm_bo_driver_finish symbol
      i915/drm: clean up a lot of the i915/drm startup/teardown sequences
      clean up ring buffer and TTM in i915_driver_unload
      WIP more code for radeon

David Nusinow (13):
      Move xsfbs back to its rightful place
      Make us error out if we call patch and don't have quilt installed
      Fix the last commit and fix spelling error. Thanks Brice Goglin for the spelling catch.
      Fix for bug #371152. Thanks Joey Hess.
      Pull patches from ubuntu for ARCHITECTURE, LAPTOP, RECONFIGURE, FIRSTINST, and UPGRADE variables
      Move the files so they'll be put in debian/xsfbs automatically when pulled
      Update serverabi generation to also account for driver Provides
      Fix name of serverminver
      Add support for X server input ABI versioning
      Fix typo that prevented the input abi versioning from working.
      Fix one missed bit from the serverabiver -> serverminver rename
      Remove make-orig-tar-gz target
      Kill the manifest code.

Dennis Kasprzyk (1):
      radeon: Restore software interrupt on resume.

Eric Anholt (137):
      Add required permission notices for code copied from X.Org source.
      Clarify that drm_agp_ttm_backend is associated with a drm_ttm.
      Initial add of mmfs module.
      Move mmfs.h userland interface to shared-core.
      Extend the mmfs basic test to do a couple of ioctls.
      Add pread/pwrite ioctls to mmfs.
      Add mmap ioctl to mmfs.
      Rename drm_mm.c and its fuctions to drm_memrange.
      Move mmfs ioctls into the DRM. Untested.
      Move mmfs tests over to be drm tests.
      Remove the remainder of the mmfs device.
      Hacking towards hooking up execbuffer.
      checkpoint: rename to GEM and a few more i915 bits.
      checkpoint: gtt binding written.
      checkpoint: relocations support.
      Add pin/unpin object ioctls for gem.
      Remove _args from gem ioctl argument structure tags.
      Update mm tests for GEM rename.
      Make GEM object handles be nonzero.
      Fix missing member settings in obj/obj_priv, and some error paths.
      Fix offset passed to AGP to be pages instead of bytes.
      Unbind objects when freeing, fix some error paths, and warn in others.
      Don't include the tail guard memrange in foreach callbacking.
      Don't forget to set the memrange private, and reset ring on kernel entry.
      GEM: Replace drm_memrange_for_each with just evicting what we brought in.
      GEM: Save the last ioremapped page for relocations in case we need it again.
      GEM: Skip relocation if presumed offset matches.
      GEM: Use irq-based fencing rather than syncing and evicting every exec.
      GEM: Extend cache domain stuff for 965.
      GEM: Wait for existing rendering to complete before writing relocation data.
      GEM: fix testcases for new ioctl args.
      GEM: Fix oops on NULL dereference when we try clflushing when we don't need to.
      GEM: Avoid leaking refs on target objects on presumed offset success.
      GEM: Clear obj_priv->agp_mem when we free it.
      GEM: Separate the LRU into execution list and LRU list.
      GEM: Fix arguments to drm_memrange_init so we don't exceed our allocation.
      [GEM] Typo (and thinking) fixes in drm-gem.txt and doxygen.
      [GEM] Update testcases for new API.
      [gem] typo fix in comment.
      [gem] Rename the GTT LRU lists to active (executing) and inactive (idle).
      [gem] Hold dev->struct_mutex to protect structure data.
      [gem] Clean up active/inactive list handling using helper functions.
      [gem] Rename sequence numbers from "cookie" to "seqno"
      [gem] Use a separate sequence number field from classic/ttm
      [gem] Hold a reference on the object in i915_gem_wait_space.
      [gem] Fix bad test for list_for_each completion.
      [gem] Replace ring throttling hack with actual time measurement.
      [gem] Release GEM buffers from work task scheduled from IRQ.
      [intel-gem] Replace idlelock usage with real lock acquisition.
      [intel-gem] Write the presumed_offset back out after updating it.
      [intel-gem] Add an option to check GTT versus CPU coherency at execbuf time.
      [intel-gem] Only update obj->write_domain if we're actually changing it.
      Merge commit 'origin/master' into drm-gem
      Import bufmgr code to libdrm.  Not yet hooked up to the build.
      Fix and hook up bufmgr code to the build.
      Fix libdrm to actually include the new code instead of just building it.
      Add a function to bufmgr_fake to evict all buffers in the GTT.
      [gem] Don't forget to munmap in the non-bo-reuse object-freeing case.
      [gem] Manage the ringbuffer from the kernel in the GEM case.
      [intel] Fix BUG_ON trigger in irq masking if you did on/off with irqs disabled.
      [gem] Remove carefully-sprinkled i915_kernel_lost_context().
      Remove override of drm module list in preparation for merge.
      [gem] Move potentially device-specific ioctls to the intel driver.
      [gem] Another round of cleanups from checkpatch.pl
      Move the renaming of mm.c symbols to #defines in the header.
      [gem] Remove the drm_client_lock_take in set_domain.
      Don't clear USER_INTERRUPT if we're not handling it.
      Bump libdrm version to 2.4.0.
      [gem] Don't require the lock in execbuf now that it's not needed for the ring.
      [gem] Remove the interrupt handler for retiring requests.
      [gem] Catch -EINTR from blocking ioctls and restart them.
      Fix i915_wait_irq in the presence of interrupt masking.
      [linux] Use the device's irq for handler setup instead of stale dev->irq.
      [intel] Quirk away MSI support on 945G/GM.
      [intel-gem] typo fix in DRM_ERROR
      intel-gem: Add two new ioctls for managing tiling on objects.
      intel-gem: Disable tiling if we get junk from the MCHBAR read.
      intel-gem: Leave 8xx tiling on until we find any issues.
      intel-gem: Set up HWS when it needs a vaddr during GEM init.
      intel-gem: Remove recently added pci_read_base prototype.
      intel-gem: Move /proc debugging to a separate file.
      intel-gem: Don't do the GTT-pwrite shortcut on tiled buffers.
      intel-gem: Add a quick hack to reduce clflushing on pread.
      intel-gem: Fix pread math and logic errors.
      intel-gem: Move debug-only functions to a separate file.
      intel-gem: Speed up tiled readpixels by tracking which pages have been flushed.
      intel-gem: Fix regression tests.
      intel-gem: Another checkpatch.pl pass.
      Merge commit 'origin/master' into drm-gem
      intel-gem: checkpatch.pl on drm_proc.c new contents.
      intel: Fix typo in unused register definition name.
      intel-gem: Check return value of pci_read_base.
      intel-gem: actually add i915_gem_debug.c.
      Revert "Rename drm_mm.c and its fuctions to drm_memrange."
      intel-gem: Remove dead field from reverted changes.
      intel-gem: Replace version bump signalling GEM with I915_PARAM_HAS_GEM.
      Replace the check_aperture API with one we can make thread-safe.
      Merge branch 'drm-gem'
      Fix compile warning from check_aperture change.
      Drop TTM interfaces from the userland library.
      drm: Add tests for GEM_FLINK ioctl.
      Move intel libdrm stuff to libdrm_intel.so
      intel: Move IRQ emit/wait from callbacks into the bufmgr.
      intel: move drm calls to exec buffers to libdrm_intel.
      intel: don't forget to include config.h in bufmgr code.
      Add missing \ to libdrm_la_SOURCES
      intel: Do strerror on errno, not on the -1 return value from ioctl.
      intel: Replace wraparound test logic in bufmgr_fake.  Again.
      intel: Allow up to 15 seconds chewing on one buffer before acknowledging -EBUSY.
      intel: Protect bufmgr objects with a pthread mutex.
      intel: Avoid pthread mutex recursion in bufmgr_fake.
      intel: Add interface for getting tiling mode of a bo.
      intel: Fix compile warning.
      intel: Protect bufmgr objects with a pthread mutex.
      intel: Add dri_bufmgr_check_aperture support for bufmgr_gem.
      intel: Rename dri_ and intel_ symbols to drm_intel_.
      libdrm 2.4.1.
      intel: Restart on interrupt of bo_wait_rendering instead of complaining.
      intel: Add a function for setting (GTT,GTT) domain, for use by UXA.
      Bump version to 2.4.2 to signal new libdrm_intel functions.
      intel: Remove the mapped flag, which is adequately covered by bo_gem->virtual.
      intel: don't let named buffers into the BO cache.
      intel: don't skip set_domain on mapping of shared buffers.
      libdrm: Fix modetest/modeprint to use automake stuff.
      intel: Sync GEM ioctl comments for easier diffing against the kernel.
      intel: Remove linux build of i915 DRM, as it's unmaintained and a user trap.
      intel: Rename plane[AB]* back to pipe[AB]*.
      intel: Retry pin ioctl on -EINTR.
      bump version to 2.4.4
      intel: Fix under-counting of fences registers required in check_aperture.
      intel: don't count fences on 965 and later, as they don't use them.
      Clean up README for the current state of kernel module affairs.
      intel: Clear ioctl parameter outvalues and pads that confuse valgrind.
      intel: Add a new bufmgr alloc function to get BOs ready for rendering to.
      Bump version to 2.4.5 for new API additions.
      Fix distcheck for optional nouveau stuff.
      Add libdrm_intel.pc by popular demand.

Hong Liu (12):
      Porting DVO stuff
      fix removing output_attrs
      clear interrupt status before install irq
      add sysfs entry for DVO output
      fixup i915 workqueue handling when modeset=1
      fix G33 hardware status page in modeset
      modeset init code cleanup
      free dummy read page if fail to init mm
      fix kernel oops when removing fb
      i915: init bo mm at driver init only when modeset=1
      i915: check dummy page before freeing
      This is a modified version of Hong's patch from last month, with a few

Ian Romanick (4):
      xgi: Fix 64-bit kernel / 32-bit user issue.
      xgixp: Remove dependency on TTM fences
      xgi: Bump kernel version
      drm-gem: Fix build

Ivan Kokshaysky (1):
      Fix DRM_CAS() on Alpha. (#16549)

Jakob Bornecrantz (45):
      add addfb/rmfb ioctls
      Added debug messages so we know which output we are dealing with
      Fix offset should from pci device address
      Lvds now power up backlight on commit
      Cleaned up userspace interface for modesetting.
      Fix unlocking of spinlock when we should not
      Changed mode config spinlock to mutex
      Added small modesetting test
      Modesetting Hotplug
      Fix and cleanup of Hotplug
      Fixed build
      Updated test mode and added modedemo
      Added fixed misc framebuffer problems
      Updated the modedemo test
      Panning now works without modeset
      Fix for X axis panning problem
      Made radeon_ms not always compile
      Added cursor support
      Misc panning fixes for intel_fb
      ModeFB demo now display cursor
      Fix for cursor off
      Modedemo now uses two crtc and output pairs
      Small update to modedemo
      Added hotplug ioctl
      Added kernel part of hotplug ioctl
      Added userspace part of hotplug ioctl and demo
      Wrong open call
      Added you can now clone displays in modedemo
      drm: Made set_config use drm_mode_set as a argument
      i915: Fixed indent in intel_fb.c
      i915: Changed intel_fb to use the new drm_crtc_set_config interface
      i915: execbuf now works without i915_dma_init being called
      tests: Improved and renamed the mode app to modeprint
      tests: Updated modeprint
      tests: modeprint s/fb/fd/
      tests: Fix faulty error messages in modeprint
      Seperate modesetting userspace bits into drm_mode.h
      mode: Indent defines
      mode: Try to settle on a standard for struct fields
      mode: Unify types for ids and strings
      mode: Remove hotplug support from ioctl interface
      mode: Unify userspace API to uint_t
      mode: Reorder the ioctls and numbering
      mode: Minor reodering and renaming
      mode: Make xfdrmMode.[c|h] not depend on drm_mode.h

Jerome Glisse (60):
      Merge branch 'radeon-ttm' of git://people.freedesktop.org/~airlied/drm into modesetting-101
      drm: split edid handling in get_edid & add_edid_mode
      drm: check edid data, so we deal well with broken driver.
      drm: fix dead lock in drm_buffer_object_transfer
      drm: don't reset to 0 irq_enabled when client open file descriptor
      radeon_ms: radeon modesetting first commit.
      radeon_ms: avoid to unintialize things which haven't been initialized
      Merge commit 'origin/modesetting-101' into modesetting-radeon
      radeon_ms: update to lastest fb change
      radeon_ms: fix fbcon by fixing palette
      radeon_ms: fix pll computation to follow hw constraint
      radeon_ms: add sarea & install header
      radeon_ms: update to follow lastest modesetting change
      radeon_ms: add rom parsing & adapt code
      radeon_ms: cope with lastest drm modesetting change
      radeon_ms: use radeon connector type insted of drm
      radeon_ms: bring radeon_ms up to date with lastest changes
      radeon: remove TTM from an earlier merge
      rradeon_ms: rework fence code and bring radeon ms up to date
      Merge branch 'modesetting-101' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      radeon_ms: fix fence
      Merge branch 'modesetting-101' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      radeon_ms: this is a modesetting driver, bring things up to date
      radeon_ms: add hang debuging helper functions
      radeon_ms: initial pass at command buffer validation
      radeon_ms: small fix & cleanup to command checking
      radeon_ms: fixes fb handling
      radeon_ms: add crtc set base callback & fix palette
      radeon_ms: check for NULL fb
      radeon_ms: fix framebuffer code
      radeon_ms: another fb fix reset mode if fb changed
      radeon_ms: command buffer validation use array of function pointer
      radeon_ms: rework command submission ioctl & cleanup
      Merge branch 'modesetting-101' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      radeon: r345xx fixe hard lockup
      radeon: bump driver date to know if lockup fix is in
      radeon: fix screen corruption introduced by last patch
      radeon: actualy try to fix the corruption
      radeon: *really* fix screen corruption thanks to Lukasz Krotowski
      radeon: radeon util library
      Merge branch 'modesetting-gem' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-gem
      radeon: reloc are backend dependant
      radeon: libdrm_radeon updates bo & cs interfaces
      Merge branch 'modesetting-gem' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-gem
      radeon: debug bo
      Merge branch 'modesetting-gem' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-gem
      radeon: libdrm_radeon add handle to debug string
      radeon: lib radeon add bo & cs gem backend
      libdrm-radeon: update libdrm-radeon to match current CS relocation structures
      radeon: add more packet3 relocations handling
      libdrm-radeon: be verbose on bo failure and cleanup cs a bit
      Merge branch 'modesetting-gem' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-gem
      radeon+libdrm-radeon: change relocation informations
      libdrm-radeon: new tracker tools
      libdrm-radeon: unreference buffer once cs stream is submited or on cs clean
      libdrm-radeon: unref return current BO ptr to reflect BO destruction
      libdrm-radeon: add print callback to cs & small fixes
      radeon: protect cs ioctl atomic part with a mutex
      gem: protect idr_pre with the spinlock
      Revert "gem: protect idr_pre with the spinlock"

Jesse Barnes (154):
      Add copyrights before I forget
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into modesetting-101
      make drmP.h include drm_crtc.h for CRTC related stuff.
      various cleanups to EDID code:
      document drm_mode_duplicate and fix vrefresh calculation (off by 1000 error)
      just codingstyle cleanups
      remove a printk to make things less verbose
      Fix refresh calculation (mistakenly removed 1000 factor needed for integer calulations, fixed mode printout debugging routine instead).
      Slam in most of X.Org's i830_lvds (not quite done yet so removed from Makefile.kernel too).
      Finish bringing in LVDS code, re-add to Makefile.  Needed other changes too:
      Remove some delays from Intel i2c code, we'll need a more comprehensive fix
      Remove some debug #if 0 codes and add a reminder to check locking around
      Add save/restore state for LVDS code, along with a few other LVDS related
      Move i915 init code to new file, i915_init.c, and create a new high level
      Export drm_setup for use by new driver init code.
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into origin/modesetting-101
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into origin/modesetting-101
      Various changes for in-kernel modesetting:
      Fixup DDC probing.  We only have one DDC bus so we have to use it only
      export vblank routine for use by intel_display.c and intel_sdvo.c.
      fix modeset cleanup for LVDS and reenable it in i915.
      Fix EDID pixel clock calculation.
      Whitespace cleanups.
      Fix i2c unregistration, cleanup panel_fixed_mode assignment.
      Add aperture size and preallocation probing (from intelfb), cleanup load code to be more general.
      Add new buffer object type for kernel allocations that don't initially have a user mapping.
      Use new kernel buffer object type and cleanup agp probing.
      Remove debug statement about buffer objects
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into origin/modesetting-101
      Revert "Export drm_setup for use by new driver init code.", we don't really want to use this function
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into origin/modesetting-101
      Whitespace cleanup
      Move i2c init back to where it belongs and add i2c unregistration in *_destroy.
      Don't use drm_setup, do SAREA allocation and mapping directly instead.
      Initialize the hw lock waitqueue so we don't hang in drm_lastclose.
      Don't need a NULL check prior to calling kfree.
      Fix 945+ hotplug detection, remove some unused variables.
      Add new function for getting a CRTC pointer given a pipe number.
      Use crtc_from_pipe call in intel_lvds.c and add get_mode panel mode detection.  Also fix up error case for when LVDS mode can't be determined.  Leave placeholder code in place for BIOS mode probing and platform quirks.
      Oops, forgot to push the msleep() in the hotplug test.  Wouldn't want to
      Move driver load call to after AGP init, in case the load routine needs AGP stuff.
      Fix PRIV0 memory initialization (mm_init takes pages, not bytes), align fb
      Use drm_mem_reg_ioremap to map buffer object.
      Add prototypes for drm_bo_init_mm and drm_buffer_object_create for use by
      Cleanup whitespace, rename macro argument.
      Move initial framebuffer allocation and configuration to drm_initial_config,
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into origin/modesetting-101
      Add framebuffer bo freeing to drm_mode_config_cleanup (seems like the best
      Just use drm_output_destroy to cleanup LVDS failures.  It'll call our cleanup
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into origin/modesetting-101
      Document main drm_crtc.c functions, and rename drm_crtc_mode_create to
      Add a mode name generation wrapper to make name format changes easier.
      Document drm_modes.c functions.
      Add a monitor information structure separate from the EDID data for tracking
      Oops, edid data should just be a char *, as edid.h is private to edid.c
      Fix FB pitch value (we had it wrong and were working around it in a few
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into origin/modesetting-101
      Remove unused edid.h header from X.Org.
      Add locking.  The main lock is dev->mode_config.config_lock.  It should be
      Merge branch 'modesetting-101' of git+ssh://git.freedesktop.org/git/mesa/drm into origin/modesetting-101
      Call preallocated space VRAM instead of PRIV0 to be more consistent with
      Suspend/resume support (incomplete).
      Suspend/resume shouldn't call drm_initial_config (seems to work ok for me now),
      Merge branch 'master' into modesetting-101 - TTM & typedef removal
      Add 965GM macro bits
      Cleanup SDVO debug output
      Remove buffer object user list check in drm_bo_destroy_unlocked
      Move map hash destruction to after driver unload hook is called
      Hack out i915_mem_takedown
      Merge branch 'master' into modesetting-101
      Use user copy routines for writing modes & ids back to userspace
      Fix typo in copy_to_user calls

Reply to: