If the requested number of DWs on the ring is larger than
the size of the ring itself, return an error.

In testing with large VM updates, we've seen crashes when we
try and allocate more space on the ring than the total size
of the ring without checking.

This prevents the crash but for large VM updates or bo moves
of very large buffers, we will need to break the transaction
down into multiple batches.  I have patches to use IBs for
the next kernel.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Print 32dword before last know rptr as problem most likely comes
from previous command. Also small cosmetic change to the printing.
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Signed-off-by: NPaul Bolle <pebolle@tiscali.nl>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Pass the vm and ring index rather than an IB.  This allows
us to use the vm_flush interface for non-IB cases in the
future.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>

Convert #include "..." to #include <path/...> in drivers/gpu/.
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Acked-by: NDave Airlie <airlied@redhat.com>
Acked-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NThomas Gleixner <tglx@linutronix.de>
Acked-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: NDave Jones <davej@redhat.com>

Even GPUs can have a null pointer dereference, so move
the IB pool to another offset to catch those.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>

Let's allow GCC to optimize better.

This exposed some five unused functions, but this patch doesn't remove them.
Signed-off-by: NLauri Kasanen <cand@gmx.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Move flushing the VMs as function into the rings.
First step to make VM operations async.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>

Store a reference to the VM into the IB structure, that
makes calculating the IBs address a bit less complicated.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>

Reset the lockup timeout on ring (re-)initialisation.

Otherwise we get error messages like this on gpu resets:
[ 1559.949177] radeon 0000:01:00.0: GPU lockup CP stall for more than 1482270msec
Signed-off-by: NChristian König <deathsimple@vodafone.de>
cc: stable@vger.kernel.org
Reviewed-by: NMichel Dänzer <michel.daenzer@amd.com>

Static checkers complain if this we don't check for allocation failure.
Also we can use the new kmalloc_array() function here as a cleanup.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Adds documentation to most of the functions in
radeon_ring.c

v2: adjust per Christian's suggestions
v3: adjust per Christian's latest patches
v4: adjust per my latest changes
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>

Add support for using memory buffers rather than
scratch registers.  Some rings may not be able to
write to scratch registers.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>

Just store the index in the ring structure.
Idea taken from one of Jerome's wip rptr patches.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NChristian König <christian.koenig@amd.com>

Const IBs are executed on the CE not the CP, so we can't
fence them in the normal way.

So submit them directly before the IB instead, just as
the documentation says.

v2: keep the extra documentation
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>

Try to save whatever is on the rings when
we encounter an lockup.

v2: Fix spelling error. Free saved ring data if reset fails.
    Add documentation for the new functions.
v3: Some more spelling fixes
v4: It doesn't make sense to save anything if all fences
    are signaled
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NMichel Dänzer <michel.daenzer@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>

Before emitting any indirect buffer, emit the offset of the next
valid ring content if any. This allow code that want to resume
ring to resume ring right after ib that caused GPU lockup.

v2: use scratch registers instead of storing it into memory
v3: skip over the surface sync for ni and si as well
v4: use SET_CONFIG_REG instead of PACKET0
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>

The IB pool is in gart memory, so it is completely
superfluous to unpin / repin it on suspend / resume.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>

We don't need to pad anything if the number of dwords
written to the ring already matches the requirements.

Fixes some "writting more dword to ring than expected"
warnings.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>
Reviewed-by: NMichel Dänzer <michel.daenzer@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>

Move inter ring syncing with semaphores into the
existing ring allocations, with that we need to
lock the ring mutex only once.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>

It is completely unnecessary to create fences
before they are emitted, so remove it and a bunch
of checks if fences are emitted or not.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>

Just move its only caller into the same file as it and make it static.
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This adds prime->fd and fd->prime support to radeon.
It passes the sg object to ttm and then populates
the gart entries using it.

Compile tested only.

v2: stub kmap + use new helpers + add reimporting
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

No need to malloc it any more.
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

It never really belonged there in the first place.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

It isn't necessary any more and the suballocator seems to perform
even better.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

A startover with a new idea for a multiple ring allocator.
Should perform as well as a normal ring allocator as long
as only one ring does somthing, but falls back to a more
complex algorithm if more complex things start to happen.

We store the last allocated bo in last, we always try to allocate
after the last allocated bo. Principle is that in a linear GPU ring
progression was is after last is the oldest bo we allocated and thus
the first one that should no longer be in use by the GPU.

If it's not the case we skip over the bo after last to the closest
done bo if such one exist. If none exist and we are not asked to
block we report failure to allocate.

If we are asked to block we wait on all the oldest fence of all
rings. We just wait for any of those fence to complete.

v2: We need to be able to let hole point to the list_head, otherwise
    try free will never free the first allocation of the list. Also
    stop calling radeon_fence_signalled more than necessary.

v3: Don't free allocations without considering them as a hole,
    otherwise we might lose holes. Also return ENOMEM instead of ENOENT
    when running out of fences to wait for. Limit the number of holes
    we try for each ring to 3.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Define the interface without modifying the allocation
algorithm in any way.

v2: rebase on top of fence new uint64 patch
v3: add ring to debugfs output
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Allocating and freeing it seperately.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Dumping the current allocations.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Instead of hacking the calculation multiple times.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Some callers illegal called fence_wait_next/empty
while holding the ring emission mutex. So don't
relock the mutex in that cases, and move the actual
locking into the fence code.

v2: Don't try to unlock the mutex if it isn't locked.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This convert fence to use uint64_t sequence number intention is
to use the fact that uin64_t is big enough that we don't need to
care about wrap around.

Tested with and without writeback using 0xFFFFF000 as initial
fence sequence and thus allowing to test the wrap around from
32bits to 64bits.

v2: Add comment about possible race btw CPU & GPU, add comment
    stressing that we need 2 dword aligned for R600_WB_EVENT_OFFSET
    Read fence sequenc in reverse order of GPU write them so we
    mitigate the race btw CPU and GPU.

v3: Drop the need for ring to emit the 64bits fence, and just have
    each ring emit the lower 32bits of the fence sequence. We
    handle the wrap over 32bits in fence_process.

v4: Just a small optimization: Don't reread the last_seq value
    if loop restarts, since we already know its value anyway.
    Also start at zero not one for seq value and use pre instead
    of post increment in emmit, otherwise wait_empty will deadlock.
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

A single global mutex for ring submissions seems sufficient.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Nothing chipset or ring specific with it,
so also move it to radon_ring.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Fixing just another deadlock problem with gpu reset tests.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Don't hard code the 10 seconds timeout. Compute jobs
can run much longer.
Signed-off-by: NChristian König <deathsimple@vodafone.de>
Reviewed-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>