profile/ivi/libdrm.git
15 years agodrm_compat: remove kmap_atomic_prot_pfn()
Pekka Paalanen [Sat, 14 Feb 2009 20:22:39 +0000 (22:22 +0200)]
drm_compat: remove kmap_atomic_prot_pfn()

This function is unused, and yet creates build problems: the symbol
init_mm is not exported by the latest -rc kernels and I don't believe it
ever will be. Even CONFIG_UNUSED_SYMBOLS does not provide it anymore.

If this function is needed in the future, it needs to be reinvented in
any case. So remove it.

Signed-off-by: Pekka Paalanen <pq@iki.fi>
15 years agoRemove i915 driver sources from linux-core
Pekka Paalanen [Mon, 16 Feb 2009 18:43:02 +0000 (20:43 +0200)]
Remove i915 driver sources from linux-core

Intel developers have stated, that their DRM development continues
elsewhere in some Linux kernel trees. This makes the code in drm.git
just dead weight. This removal allows further cleanup of compatibility
code.

shared-core and bsd-core are left untouched this time.

Signed-off-by: Pekka Paalanen <pq@iki.fi>
Acked-by: Eric Anholt <eric@anholt.net>
15 years agointel: Clear ioctl parameter outvalues and pads that confuse valgrind.
Eric Anholt [Mon, 16 Feb 2009 02:35:03 +0000 (18:35 -0800)]
intel: Clear ioctl parameter outvalues and pads that confuse valgrind.

The minor CPU cost here is probably outweighed by bothering us with noise in
the tool.

15 years agonv40: fail completely if we don't have a ctxprog for the chipset
Ben Skeggs [Sun, 15 Feb 2009 12:06:18 +0000 (22:06 +1000)]
nv40: fail completely if we don't have a ctxprog for the chipset

15 years agonv50: context info for chipset 0xa0
Ben Skeggs [Sun, 15 Feb 2009 11:52:19 +0000 (21:52 +1000)]
nv50: context info for chipset 0xa0

15 years agolibdrm: make libdrm_nouveau conditional, disabled by default
Ben Skeggs [Wed, 11 Feb 2009 04:18:03 +0000 (14:18 +1000)]
libdrm: make libdrm_nouveau conditional, disabled by default

libdrm isn't supposed to ship APIs not present in a released kernel.

15 years agomode: Make xfdrmMode.[c|h] not depend on drm_mode.h
Jakob Bornecrantz [Wed, 11 Feb 2009 15:43:20 +0000 (16:43 +0100)]
mode: Make xfdrmMode.[c|h] not depend on drm_mode.h

15 years agonouveau: fix type warnings: void* arith, un/signed
Pekka Paalanen [Wed, 4 Feb 2009 19:44:50 +0000 (21:44 +0200)]
nouveau: fix type warnings: void* arith, un/signed

nouveau_notifier.c had two places where void* was used in arithmetic,
fixed by using char*.

nouveau_dma_wait(), nouveau_notifier_wait_status() and
nouveau_resource_alloc() had signed/unsigned comparison warnings, fixed
by changing the function parameter into an unsigned type.

Signed-off-by: Pekka Paalanen <pq@iki.fi>
15 years agodrm/nv50: fix nv9x chipsets
Ben Skeggs [Wed, 11 Feb 2009 00:48:36 +0000 (10:48 +1000)]
drm/nv50: fix nv9x chipsets

NVIDIA do this fun little sequence after updating the PRAMIN page tables.

On 9xxx chips, none of the PRAMIN BAR bindings (except the initial one)
worked, hence the majority of the setup needed to create a channel
ended up in the wrong place, causing all sorts of fun.

This is done by NVIDIA on nv8x chips also, so we'll do it for them too,
even though they appear to work without it.

15 years agodrm/nv50: add context info for nv98
Ben Skeggs [Wed, 11 Feb 2009 00:12:43 +0000 (10:12 +1000)]
drm/nv50: add context info for nv98

It won't work yet, just like the other 9xxx chips.  Real soon now :)

15 years agodrm/nv50: use a slightly different initial context for nv96
Ben Skeggs [Mon, 9 Feb 2009 23:05:09 +0000 (09:05 +1000)]
drm/nv50: use a slightly different initial context for nv96

I'm not 100% sure that the nv94 one we were using won't work.  The context
layouts are identical (well.. same ctxprog, so of course!), only a couple
of registers differ.  But, be safe until we actually get some 9xxx chips
working.

15 years agodrm/nv50: correct ramfc pointer in channel header
Ben Skeggs [Mon, 9 Feb 2009 23:02:17 +0000 (09:02 +1000)]
drm/nv50: correct ramfc pointer in channel header

Suprisingly the card still worked without this...

15 years agodrm/nv50: let the card handle the initial context switch
Ben Skeggs [Mon, 9 Feb 2009 22:57:21 +0000 (08:57 +1000)]
drm/nv50: let the card handle the initial context switch

Our PFIFO/PGRAPH context save/load functions don't really work well
(at all?) on nv5x yet.  Depending on what random state the card is
in before the drm loads, fbcon probably won't work correctly.

Luckily we've setup the GPU in such a way that it'll actually do a
hw context switch for the first context.  Not sure of how successful
this'd be currently on the older chips (actually, pretty sure it won't
work), so NV50 only for now.

15 years agonouveau: don't try to traverse non-existent lists
Stuart Bennett [Sat, 7 Feb 2009 21:20:17 +0000 (21:20 +0000)]
nouveau: don't try to traverse non-existent lists

Fixes nouveau_ioctl_mem_free Oops

15 years agonouveau: fix some issues where buffer objects never get freed
Ben Skeggs [Thu, 5 Feb 2009 13:03:29 +0000 (23:03 +1000)]
nouveau: fix some issues where buffer objects never get freed

15 years agomodetest: Add syntax to also specify a crtc when setting a mode.
Kristian Høgsberg [Wed, 4 Feb 2009 17:17:13 +0000 (12:17 -0500)]
modetest: Add syntax to also specify a crtc when setting a mode.

15 years agonouveau: install libdrm_nouveau with libdrm
Ben Skeggs [Fri, 30 Jan 2009 01:25:35 +0000 (11:25 +1000)]
nouveau: install libdrm_nouveau with libdrm

15 years agonouveau: bring in new mm api definitions, without the actual mm code
Ben Skeggs [Fri, 30 Jan 2009 01:18:54 +0000 (11:18 +1000)]
nouveau: bring in new mm api definitions, without the actual mm code

Use of the new bits is guarded with a mm_enabled=0 hardcode.

15 years agomodetest: Use cairo (if available) to paint a better pattern.
Kristian Høgsberg [Tue, 3 Feb 2009 20:03:41 +0000 (15:03 -0500)]
modetest: Use cairo (if available) to paint a better pattern.

15 years agomodetest: Don't sleep when just dumping state, wait for key for modeset.
Kristian Høgsberg [Tue, 3 Feb 2009 19:02:50 +0000 (14:02 -0500)]
modetest: Don't sleep when just dumping state, wait for key for modeset.

15 years agomodetest: Handle setting modes on multiple crtcs with one fb.
Kristian Høgsberg [Tue, 3 Feb 2009 19:00:00 +0000 (14:00 -0500)]
modetest: Handle setting modes on multiple crtcs with one fb.

15 years agomodetest: Use a more interesting test pattern.
Kristian Høgsberg [Tue, 3 Feb 2009 18:26:22 +0000 (13:26 -0500)]
modetest: Use a more interesting test pattern.

15 years agoRemove the "nv" driver.
Stephane Marchesin [Mon, 2 Feb 2009 22:47:11 +0000 (23:47 +0100)]
Remove the "nv" driver.

15 years agoNo need to pin buffer anymore in modetest.c
Kristian Høgsberg [Mon, 2 Feb 2009 20:47:09 +0000 (15:47 -0500)]
No need to pin buffer anymore in modetest.c

15 years agonouveau: don't save channel context if it has recently become invalid
Stuart Bennett [Thu, 29 Jan 2009 23:39:50 +0000 (23:39 +0000)]
nouveau: don't save channel context if it has recently become invalid

Bug exposed by DDX change d9da090c

15 years agonouveau: no suspend support for nv50+
Stuart Bennett [Sat, 25 Oct 2008 19:59:38 +0000 (20:59 +0100)]
nouveau: no suspend support for nv50+

15 years agoClean up README for the current state of kernel module affairs.
Eric Anholt [Wed, 28 Jan 2009 02:00:45 +0000 (18:00 -0800)]
Clean up README for the current state of kernel module affairs.

Bug #19608.

15 years agointel: don't count fences on 965 and later, as they don't use them.
Eric Anholt [Wed, 28 Jan 2009 01:16:11 +0000 (17:16 -0800)]
intel: don't count fences on 965 and later, as they don't use them.

15 years agointel: Fix under-counting of fences registers required in check_aperture.
Eric Anholt [Wed, 28 Jan 2009 00:54:11 +0000 (16:54 -0800)]
intel: Fix under-counting of fences registers required in check_aperture.

15 years agointel: libdrm support for fence management in execbuf
Jesse Barnes [Fri, 23 Jan 2009 22:13:45 +0000 (14:13 -0800)]
intel: libdrm support for fence management in execbuf

This patch tries to use the available fence count to figure out whether a
given batch can succeed or not (just like the aperture check).

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
15 years agodrm compat: fix euid for >=2.6.28
Pekka Paalanen [Tue, 27 Jan 2009 20:39:50 +0000 (22:39 +0200)]
drm compat: fix euid for >=2.6.28

drm_fops.c reads the current process' EUID directly from task_struct.
Apparently starting in 2.6.28-rc4 this fails to build.

In Linus' tree, commit b6dff3ec5e116e3af6f537d4caedcad6b9e5082a
"CRED: Separate task security context from task_struct"
moves the euid field from task_struct to another struct.

Earlier commit 9e2b2dc4133f65272a6d3c5dcb2ce63f8a87cae9
"CRED: Introduce credential access wrappers" implements the wrapper
macros to access e.g. euid. This is in 2.6.27-rc4, and this contains the
definition of current_euid() that will be used in the DRM compatibility header
for kernels before 2.6.27. That commit also creates <linux/cred.h>, which
contains the upstream definition of current_euid().

drm_fops.c is fixed to use current_euid(), and drm_compat.h will offer
the compatibility definition for kernels <2.6.27.

Signed-off-by: Pekka Paalanen <pq@iki.fi>
15 years agonv50: support chipset NV96
Ben Skeggs [Mon, 26 Jan 2009 22:36:33 +0000 (08:36 +1000)]
nv50: support chipset NV96

ctxprog seen in okias' trace identical to one we use on NV94, assuming
the initial context values for NV94 will work here too.

15 years agonv04-nv40: correct RAMHT size
Ben Skeggs [Thu, 13 Nov 2008 02:04:21 +0000 (13:04 +1100)]
nv04-nv40: correct RAMHT size

15 years agodrm: remove drmstat/dristat from linux-core build
Dave Airlie [Sun, 18 Jan 2009 20:46:23 +0000 (06:46 +1000)]
drm: remove drmstat/dristat from linux-core build

pointed out by pq

15 years agobump version to 2.4.4
Eric Anholt [Thu, 15 Jan 2009 00:11:32 +0000 (16:11 -0800)]
bump version to 2.4.4

15 years agointel: Retry pin ioctl on -EINTR.
Eric Anholt [Mon, 5 Jan 2009 01:37:18 +0000 (17:37 -0800)]
intel: Retry pin ioctl on -EINTR.

15 years agoDon't use DRM_BO_FLAG_NO_MOVE in bufmgr fake. It's a ttm flag.
Owain G. Ainsworth [Sat, 10 Jan 2009 18:48:46 +0000 (18:48 +0000)]
Don't use DRM_BO_FLAG_NO_MOVE in bufmgr fake. It's a ttm flag.

It's also unused, so worthless.

15 years agoRemove drmModeReplaceFb after it was removed from the kernel.
Owain G. Ainsworth [Sun, 11 Jan 2009 19:02:07 +0000 (19:02 +0000)]
Remove drmModeReplaceFb after it was removed from the kernel.

It is impossible to replace the original semantics of this call purely
in userland, since the fb_id would change.

after discussion with Dr_Jakob

Signed-Off-By: Owain Ainsworth <oga@openbsd.org>
Acked-By: Jakob Bornecrantz <jakob@vmware.com>
15 years agonv50: ack nsource to prevent continuous protection fault irqs
Ben Skeggs [Mon, 12 Jan 2009 00:16:07 +0000 (10:16 +1000)]
nv50: ack nsource to prevent continuous protection fault irqs

15 years agolibdrm: add autoconf check for clock_gettime
Jesse Barnes [Wed, 7 Jan 2009 19:47:52 +0000 (11:47 -0800)]
libdrm: add autoconf check for clock_gettime

Should be more portable this way.

15 years agolibdrm: only check for vblank timeout if we caught EINTR
Jesse Barnes [Wed, 7 Jan 2009 18:48:26 +0000 (10:48 -0800)]
libdrm: only check for vblank timeout if we caught EINTR

Michel caught a case where we might overwrite a success or other return
value with EBUSY, so check the return value before checking for the
timeout condition.

15 years agolibdrm: add timeout handling to drmWaitVBlank
Jesse Barnes [Wed, 7 Jan 2009 18:18:08 +0000 (10:18 -0800)]
libdrm: add timeout handling to drmWaitVBlank

In some cases, vblank interrupts may be disabled or otherwise broken.
The kernel has a 3s timeout builtin to handle these cases, but the X
server's SIGALM for cursor handling may interrupt vblank wait ioctls,
causing libdrm to restart the ioctl, making the kernel's timeout
useless.

This change tracks time across ioctl restarts and returns EBUSY to the
caller if the expected vblank sequence doesn't occur within 1s of the
first call.

Fixes fdo bz #18041, which is caused by a drmWaitVBlank hanging due to
the corresponding pipe getting disabled (thus preventing further events
from coming in).

15 years agoRemove executable from modeprint test dir
Jesse Barnes [Mon, 29 Dec 2008 17:21:58 +0000 (09:21 -0800)]
Remove executable from modeprint test dir

Accidentally committed it with the mode setting changes.

15 years ago[FreeBSD] Fix build on FreeBSD after modesetting import.
Robert Noland [Tue, 23 Dec 2008 18:56:23 +0000 (13:56 -0500)]
[FreeBSD] Fix build on FreeBSD after modesetting import.

15 years agoradeon: only write irq regs if irq is enabled
Dave Airlie [Tue, 23 Dec 2008 01:22:44 +0000 (11:22 +1000)]
radeon: only write irq regs if irq is enabled

15 years agointel: Rename plane[AB]* back to pipe[AB]*.
Eric Anholt [Wed, 17 Dec 2008 21:55:53 +0000 (13:55 -0800)]
intel: Rename plane[AB]* back to pipe[AB]*.

The values are really going to continue meaning pipe, not plane, and that's
what they're called in the kernel copy of the header.  Userland hasn't ever
made the switch to pipe!=plane, since userland checks are based on DRM
version, which is still stuck at 1.6.  However, Mesa did start using
plane[AB] names, so provide a compat define.

15 years agointel: Remove linux build of i915 DRM, as it's unmaintained and a user trap.
Eric Anholt [Wed, 17 Dec 2008 21:47:59 +0000 (13:47 -0800)]
intel: Remove linux build of i915 DRM, as it's unmaintained and a user trap.

The code's in the linux kernel.

15 years agointel: Sync GEM ioctl comments for easier diffing against the kernel.
Eric Anholt [Wed, 17 Dec 2008 21:45:01 +0000 (13:45 -0800)]
intel: Sync GEM ioctl comments for easier diffing against the kernel.

15 years agomodetest: Print more stuff. Prettier.
Kristian Høgsberg [Thu, 18 Dec 2008 05:02:43 +0000 (00:02 -0500)]
modetest: Print more stuff. Prettier.

15 years agolibdrm: Fix modetest/modeprint to use automake stuff.
Eric Anholt [Wed, 17 Dec 2008 18:41:21 +0000 (10:41 -0800)]
libdrm: Fix modetest/modeprint to use automake stuff.

15 years agointel: Cache tiling/swizzle state in user mode. Reset tiling on reuse.
Keith Packard [Mon, 15 Dec 2008 23:08:12 +0000 (15:08 -0800)]
intel: Cache tiling/swizzle state in user mode. Reset tiling on reuse.

Remember tiling mode values provided by appplications, and
record tiling mode when creating a buffer from another application. This
eliminates any need to ask the kernel for tiling values and also makes
reused buffers get the right tiling.

Signed-off-by: Keith Packard <keithp@keithp.com>
15 years agointel: Debug output %u vs uint64_t warning fix
Keith Packard [Mon, 15 Dec 2008 23:08:17 +0000 (15:08 -0800)]
intel: Debug output %u vs uint64_t warning fix

Signed-off-by: Keith Packard <keithp@keithp.com>
15 years agointel: return error status from drm_intel_gem_bo_map
Keith Packard [Sat, 13 Dec 2008 19:06:11 +0000 (11:06 -0800)]
intel: return error status from drm_intel_gem_bo_map

Applications may actually care if the mapping operation failed, so when
it happens, return an error indication. errno is probably trashed by
fprintf though.

Signed-off-by: Keith Packard <keithp@keithp.com>
15 years agointel: Dump out memory usage information when the kernel fails to pin
Keith Packard [Fri, 21 Nov 2008 09:49:39 +0000 (01:49 -0800)]
intel: Dump out memory usage information when the kernel fails to pin

The execbuffer ioctl returns ENOMEM when it fails to pin all of the buffers
in the GTT. This is usually caused by the DRM client attempting to use too
much memory in a single request. Dumping out the requested and available
memory values should help point out failures in the DRM code to catch over
commitments of this form.

Signed-off-by: Keith Packard <keithp@keithp.com>
15 years agolibdrm: add mode setting files
Jesse Barnes [Wed, 17 Dec 2008 18:09:49 +0000 (10:09 -0800)]
libdrm: add mode setting files

Add mode setting files to libdrm, including xf86drmMode.* and the new
drm_mode.h header.  Also add a couple of tests to sanity check the
kernel interfaces and update code to support them.

15 years agointel: don't skip set_domain on mapping of shared buffers.
Eric Anholt [Sun, 14 Dec 2008 22:39:24 +0000 (14:39 -0800)]
intel: don't skip set_domain on mapping of shared buffers.

15 years agointel: don't let named buffers into the BO cache.
Eric Anholt [Sun, 14 Dec 2008 22:35:48 +0000 (14:35 -0800)]
intel: don't let named buffers into the BO cache.

We wouldn't want some remaining 3D rendering to scribble on our batchbuffer.

15 years agointel: Remove the mapped flag, which is adequately covered by bo_gem->virtual.
Eric Anholt [Sun, 14 Dec 2008 22:32:09 +0000 (14:32 -0800)]
intel: Remove the mapped flag, which is adequately covered by bo_gem->virtual.

15 years agoRevert "Merge branch 'modesetting-gem'"
Jesse Barnes [Wed, 10 Dec 2008 23:47:28 +0000 (15:47 -0800)]
Revert "Merge branch 'modesetting-gem'"

This reverts commit 6656db10551bbb8770dd945b6d81d5138521f208.

We really just want the libdrm and ioctl bits, not all the driver
stuff.

15 years agoBump version to 2.4.3 to reflect addition of mode setting header
Jesse Barnes [Wed, 10 Dec 2008 23:37:45 +0000 (15:37 -0800)]
Bump version to 2.4.3 to reflect addition of mode setting header

15 years agoUpdate .gitignore with new modetest program
Jesse Barnes [Wed, 10 Dec 2008 23:37:29 +0000 (15:37 -0800)]
Update .gitignore with new modetest program

15 years agoMerge branch 'modesetting-gem'
Jesse Barnes [Tue, 9 Dec 2008 18:23:43 +0000 (10:23 -0800)]
Merge branch 'modesetting-gem'

15 years agolibdrm: bump to 2.4 for lib version
Dave Airlie [Mon, 8 Dec 2008 04:05:32 +0000 (14:05 +1000)]
libdrm: bump to 2.4 for lib version

we already have a 2.3.x version out there

15 years agoMerge branch 'master' into modesetting-gem
Jesse Barnes [Wed, 3 Dec 2008 19:54:07 +0000 (11:54 -0800)]
Merge branch 'master' into modesetting-gem

15 years agoMerge branch 'master' into modesetting-gem
Jesse Barnes [Wed, 3 Dec 2008 19:53:36 +0000 (11:53 -0800)]
Merge branch 'master' into modesetting-gem

15 years agoBump version to 2.4.2 to signal new libdrm_intel functions.
Eric Anholt [Mon, 1 Dec 2008 19:49:11 +0000 (11:49 -0800)]
Bump version to 2.4.2 to signal new libdrm_intel functions.

15 years agointel: Add a function for setting (GTT,GTT) domain, for use by UXA.
Eric Anholt [Thu, 13 Nov 2008 19:44:22 +0000 (11:44 -0800)]
intel: Add a function for setting (GTT,GTT) domain, for use by UXA.

This function can also serve the role that the bo_wait_rendering did, when
write_enable is unset.

15 years agonv50: support NV94 chipset
Ben Skeggs [Sun, 23 Nov 2008 07:49:09 +0000 (18:49 +1100)]
nv50: support NV94 chipset

15 years agonv50: update context-related tables for original 8800GTS
Ben Skeggs [Fri, 21 Nov 2008 01:43:44 +0000 (12:43 +1100)]
nv50: update context-related tables for original 8800GTS

I either messed up when I pulled these from a mmio-trace last time, or
the previous values didn't work on my card.  Hopefully it's the former!

In any case, at least one of the original NV50 chipsets work now.

15 years agoMerge branch 'master' into modesetting-gem
Jesse Barnes [Thu, 20 Nov 2008 19:01:24 +0000 (11:01 -0800)]
Merge branch 'master' into modesetting-gem

15 years agoDRM: make drm_map_type match upstream kernel
Jesse Barnes [Thu, 20 Nov 2008 19:00:29 +0000 (11:00 -0800)]
DRM: make drm_map_type match upstream kernel

Since the TTM type isn't upstream yet, we need to make sure libdrm uses
what the kernel uses, which is _DRM_GEM = 6.

15 years agoDRM: make drm_map_type match kernel
Jesse Barnes [Thu, 20 Nov 2008 18:57:33 +0000 (10:57 -0800)]
DRM: make drm_map_type match kernel

GEM is upstream, but TTM isn't, so _DRM_GEM needs to be 6, not 7.

15 years agoUnbreak drm build.
Stephane Marchesin [Wed, 19 Nov 2008 22:17:27 +0000 (23:17 +0100)]
Unbreak drm build.

15 years agoMerge branch 'modesetting-gem' of ssh://git.freedesktop.org/git/mesa/drm into modeset...
Jesse Barnes [Wed, 19 Nov 2008 20:42:49 +0000 (12:42 -0800)]
Merge branch 'modesetting-gem' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-gem

15 years agolibdrm_intel: fix merge error
Jesse Barnes [Wed, 19 Nov 2008 18:56:48 +0000 (10:56 -0800)]
libdrm_intel: fix merge error

don't take the lock twice

15 years agoUpdate modetest
Jesse Barnes [Wed, 19 Nov 2008 18:54:11 +0000 (10:54 -0800)]
Update modetest

15 years agoRevert "gem: protect idr_pre with the spinlock"
Jerome Glisse [Mon, 17 Nov 2008 12:35:51 +0000 (13:35 +0100)]
Revert "gem: protect idr_pre with the spinlock"

This reverts commit f46aba43329786c9b2776e82af31d90e1c67eae0.

15 years agogem: protect idr_pre with the spinlock
Jerome Glisse [Sun, 16 Nov 2008 17:13:16 +0000 (18:13 +0100)]
gem: protect idr_pre with the spinlock

There seems to be an issue if we don't protect idr_pre with
the spinlock we use for idr operations.

15 years agoradeon: protect cs ioctl atomic part with a mutex
Jerome Glisse [Sun, 16 Nov 2008 17:11:00 +0000 (18:11 +0100)]
radeon: protect cs ioctl atomic part with a mutex

A small subset of CS need to be atomic (relocation+IB commit to
ring) right now, because of the way relocation are handled, we
need to protect the whole ioctl.

15 years agolibdrm-radeon: add print callback to cs & small fixes
Jerome Glisse [Sun, 16 Nov 2008 17:04:43 +0000 (18:04 +0100)]
libdrm-radeon: add print callback to cs & small fixes

15 years agolibdrm-radeon: unref return current BO ptr to reflect BO destruction
Jerome Glisse [Sat, 15 Nov 2008 09:38:44 +0000 (10:38 +0100)]
libdrm-radeon: unref return current BO ptr to reflect BO destruction

15 years agodrm/bo: fix stupid lock imbalance
Dave Airlie [Sat, 15 Nov 2008 02:30:09 +0000 (12:30 +1000)]
drm/bo: fix stupid lock imbalance

15 years agolibdrm-radeon: unreference buffer once cs stream is submited or on cs clean
Jerome Glisse [Fri, 14 Nov 2008 11:13:53 +0000 (12:13 +0100)]
libdrm-radeon: unreference buffer once cs stream is submited or on cs clean

BO are referenced once by reloc to make sure that they not destroyed
before we get a chance to flush the cmd stream, so we need to unreference
them once in cs submit or cs erase if cs i never submitted so bo can
be destructed.

15 years agolibdrm-radeon: new tracker tools
Jerome Glisse [Fri, 14 Nov 2008 11:08:27 +0000 (12:08 +0100)]
libdrm-radeon: new tracker tools

To keep record of bo activities and print them when necessary,
should help in tracking unbalanced ref/unref calls.

15 years agoMerge branch 'master' into modesetting-gem
Jesse Barnes [Thu, 13 Nov 2008 23:30:06 +0000 (15:30 -0800)]
Merge branch 'master' into modesetting-gem

Conflicts:

libdrm/Makefile.am
libdrm/intel/intel_bufmgr.h
libdrm/intel/intel_bufmgr_fake.c
libdrm/intel/intel_bufmgr_gem.c
shared-core/drm.h
shared-core/i915_dma.c
shared-core/i915_irq.c
shared-core/radeon_cp.c
shared-core/radeon_drv.h

15 years agolibdrm_intel: fix warnings on 64 bit
Jesse Barnes [Thu, 13 Nov 2008 21:58:32 +0000 (13:58 -0800)]
libdrm_intel: fix warnings on 64 bit

Cast a couple of %llx args to unsigned long long.

15 years agolibdrm: add support for i915 GTT mapping ioctl
Jesse Barnes [Thu, 13 Nov 2008 21:52:04 +0000 (13:52 -0800)]
libdrm: add support for i915 GTT mapping ioctl

Add a drm_intel_gem_bo_map_gtt() function for mapping a buffer object
through the aperture rather than directly to its CPU cacheable memory.

15 years agomode: Minor reodering and renaming
Jakob Bornecrantz [Wed, 12 Nov 2008 18:17:18 +0000 (19:17 +0100)]
mode: Minor reodering and renaming

15 years agomode: Reorder the ioctls and numbering
Jakob Bornecrantz [Wed, 12 Nov 2008 18:10:50 +0000 (19:10 +0100)]
mode: Reorder the ioctls and numbering

This is to fill in the gaps left by the removal of
the hotplug ioctls. And they also look better :)

15 years agomode: Unify userspace API to uint_t
Jakob Bornecrantz [Wed, 12 Nov 2008 17:49:46 +0000 (18:49 +0100)]
mode: Unify userspace API to uint_t

15 years agomode: Remove hotplug support from ioctl interface
Jakob Bornecrantz [Wed, 12 Nov 2008 17:40:04 +0000 (18:40 +0100)]
mode: Remove hotplug support from ioctl interface

15 years agomode: Unify types for ids and strings
Jakob Bornecrantz [Wed, 12 Nov 2008 17:16:38 +0000 (18:16 +0100)]
mode: Unify types for ids and strings

15 years agoradeon+libdrm-radeon: change relocation informations
Jerome Glisse [Wed, 12 Nov 2008 14:56:40 +0000 (15:56 +0100)]
radeon+libdrm-radeon: change relocation informations

Relocation now consist of the following informations (in this order) :
handle          buffer object handle identifier
start_offset    start offset of first data of the buffer object used by the cs
end_offset      end offset of last data of the buffer object used by the cs
read_domain     read domain (either VRAM, or GTT as GPU is invalid for CS)
write_domain    write domain (either VRAM, or GTT as GPU is invalid for CS)
flags           flags used for further optimization (like discard previous
                buffer content or forget buffer content after cs which can
                help in avoiding moving content in or out)

15 years agoMerge branch 'modesetting-gem' of ssh://git.freedesktop.org/git/mesa/drm into modeset...
Jerome Glisse [Mon, 10 Nov 2008 21:21:32 +0000 (22:21 +0100)]
Merge branch 'modesetting-gem' of ssh://git.freedesktop.org/git/mesa/drm into modesetting-gem

15 years agolibdrm-radeon: be verbose on bo failure and cleanup cs a bit
Jerome Glisse [Mon, 10 Nov 2008 21:18:22 +0000 (22:18 +0100)]
libdrm-radeon: be verbose on bo failure and cleanup cs a bit

15 years agoradeon: add gart useable size to report to userspace
Dave Airlie [Mon, 10 Nov 2008 05:38:32 +0000 (15:38 +1000)]
radeon: add gart useable size to report to userspace

15 years agoradeon: upgrade atom headers
Dave Airlie [Mon, 10 Nov 2008 05:37:51 +0000 (15:37 +1000)]
radeon: upgrade atom headers

15 years agoradeon: fix fence race condition hopefully
Dave Airlie [Mon, 10 Nov 2008 05:37:36 +0000 (15:37 +1000)]
radeon: fix fence race condition hopefully

For some reason reading the SCRATCH reg from RAM causes some race to occur.
Hopefully fix this.

15 years agodrm/radeon: add dpms connector functions
Dave Airlie [Mon, 10 Nov 2008 05:37:16 +0000 (15:37 +1000)]
drm/radeon: add dpms connector functions

15 years agoradeon: avivo cursors are across the full surface.
Dave Airlie [Mon, 10 Nov 2008 05:35:34 +0000 (15:35 +1000)]
radeon: avivo cursors are across the full surface.

fixes cursor on second head