It makes no difference to kmalloc if the structure
is 48 or 64 bytes in size.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/396950/

We can still allocate 16TiB with that.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/396946/

Neither page allocation backend nor the driver should mess with that.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Reviewed-by: NMadhav Chauhan <madhav.chauhan@amd.com>
Link: https://patchwork.freedesktop.org/patch/396948/

The move notify callback is only used in one place, this should
be removed in the future, but for now just rename it to the use
case which is to notify the driver that the GPU memory is to be
deleted.

Drivers can be cleaned up after this separately.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201021044031.1752624-2-airlied@gmail.com

This moves the call to tt binding into the driver move,
and drops the driver callback.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201020010319.1692445-8-airlied@gmail.com

The drivers now do this in the move callback.

move_notify is still needed in the destroy path.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201020010319.1692445-7-airlied@gmail.com

The drivers now control this, so drop unbinding.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201020010319.1692445-5-airlied@gmail.com

This moves the to system move into the drivers, and moves all
the unbinds in the move path under driver control

Note: radeon/nouveau already wait so don't duplicate it.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201020010319.1692445-4-airlied@gmail.com

This just gives the driver control over some of the bind paths.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201020010319.1692445-2-airlied@gmail.com

Uninline ttm_bo_move_ttm. Eventually want to unhook the unbind out.
Reviewed-by: NBen Skeggs <bskeggs@redhat.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201019071314.1671485-5-airlied@gmail.com

In all 3 drivers there is a case where the driver knows the
bo is in SYSTEM so don't call the api that checks that.
Reviewed-by: NBen Skeggs <bskeggs@redhat.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201019071314.1671485-4-airlied@gmail.com

Changing the caching on the fly never really worked
flawlessly.

So stop this completely and just let drivers specific the
desired caching in the tt or bus object.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMichael J. Ruhl <michael.j.ruhl@intel.com>
Link: https://patchwork.freedesktop.org/patch/394256/

And implement setting it up correctly in the drivers.

This allows getting rid of the placement flags for this.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMichael J. Ruhl <michael.j.ruhl@intel.com>
Link: https://patchwork.freedesktop.org/patch/394254/

All drivers can determine the tt caching state at creation time,
no need to do this on the fly during every validation.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NMichael J. Ruhl <michael.j.ruhl@intel.com>
Link: https://patchwork.freedesktop.org/patch/394253/

We already implemented the fault handler ourself,
just open code what is necessary here.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/392322/

The core move code currently handles use_tt moves, for radeon
this was being handled also in the driver, but not using the same
paths.

If moving between TT/SYSTEM (all the use_tt paths on radeon) use
the core move function.

Eventually the core will be flipped over to calling the driver.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200924051845.397177-5-airlied@gmail.com

Stop using TTM_PL_FLAG_NO_EVICT.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Reviewed-by: NHuang Rui <ray.huang@amd.com>
Link: https://patchwork.freedesktop.org/patch/391610/?series=81973&rev=1

I'm thinking of pushing the wait into the drivers.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200923030454.362731-6-airlied@gmail.com

Just pass it around move, and remove unused pieces
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200923030454.362731-4-airlied@gmail.com

The two accel cleanup paths were mostly the same once refactored.

Just pass a bool to say if the evictions are to be pipelined.
Signed-off-by: NDave Airlie <airlied@redhat.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200917064132.148521-2-airlied@gmail.com

Now the bind functions have all the protection explicitly the
drivers can just call them directly, and the api can be unexported
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200917043040.146575-5-airlied@gmail.com

This moves unbind into the driver side on destroy paths.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200917043040.146575-4-airlied@gmail.com

Call the driver first and have it call the common code cleanup.

This is useful later to fix unbind.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200917043040.146575-3-airlied@gmail.com

This moves the generic tracking into the drivers and protects
against reentrancy in the drivers. It fixes up radeon and agp
to be able to query the bound status as that is required.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200917043040.146575-2-airlied@gmail.com

Move these up to the bo level, moving ttm_tt to just being
backing store. Next step is to move the bound flag out.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200915024007.67163-6-airlied@gmail.com

Drivers have to call populate themselves now before binding.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200915024007.67163-5-airlied@gmail.com

This adds 2 getters and 4 setters, however unbound and populated
are currently the same thing, this will change, it also drops
a BUG_ON that seems not that useful.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200915024007.67163-2-airlied@gmail.com

Instead of letting TTM make an educated guess based on
some mask all drivers should just specify what caching
they want for their CPU mappings.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/390207/

As far as I can tell this was never used either and we just
always fallback to the order cached > wc > uncached anyway.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/390142/

It's not supported to specify more than one of those flags.
So it never made sense to make this a flag in the first place.

Nuke the flags and specify directly which memory type to use.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/389826/?series=81551&rev=1

Since the agp bind/unbind/destroy are now getting called from drivers
rather than via the func table, drop the bdev parameter.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NBen Skeggs <bskeggs@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200907204630.1406528-13-airlied@gmail.com

Reviewed-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NBen Skeggs <bskeggs@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200907204630.1406528-12-airlied@gmail.com

Do agp decision in the driver, instead of special binding funcs

v2: use container_of, drop some {}.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200907204630.1406528-6-airlied@gmail.com

This pattern is cut-n-pasted across 4 drivers, switch it to
a WARN_ON instead, as BUG_ON is considered a bad idea usually.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NBen Skeggs <bskeggs@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200907204630.1406528-2-airlied@gmail.com

This is used by TTM to communicate the physical address
which should be used with ioremap(), ioremap_wc(). We don't
need to separate the base and offset in any way here.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/389457/

I want to split this structure up and use it differently,
step one remove bdev pointer from it and pass it explicitly.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200826014428.828392-4-airlied@gmail.com

This can be gotten back from bdev.
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200826014428.828392-3-airlied@gmail.com

This is always calculated the same, and only used in a couple of places.
Signed-off-by: NDave Airlie <airlied@redhat.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200811074658.58309-2-airlied@gmail.com

The drivers all do the same thing here.

Reviewed-by: Christian König <christian.koenig@amd.com> for both.
Signed-off-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200811074658.58309-1-airlied@gmail.com

The object flags created in radeon_ttm_placement_from_domain take care that
we use the correct caching for AGP, this is just superflous.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Reviewed-by: NDave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/384336/?series=80346&rev=1