int amdgpu_gfx_kiq_init_ring(struct amdgpu_device *adev,
struct amdgpu_ring *ring,
struct amdgpu_irq_src *irq)
{
struct amdgpu_kiq *kiq = &adev->gfx.kiq;
int r = 0;
mutex_init(&kiq->ring_mutex);
r = amdgpu_wb_get(adev, &adev->virt.reg_val_offs);
if (r)
return r;
ring->adev = NULL;
ring->ring_obj = NULL;
ring->use_doorbell = true;
ring->doorbell_index = AMDGPU_DOORBELL_KIQ;
r = amdgpu_gfx_kiq_acquire(adev, ring);
if (r)
return r;
ring->eop_gpu_addr = kiq->eop_gpu_addr;
sprintf(ring->name, "kiq_%d.%d.%d", ring->me, ring->pipe, ring->queue);
r = amdgpu_ring_init(adev, ring, 1024,
irq, AMDGPU_CP_KIQ_IRQ_DRIVER0);
if (r)
dev_warn(adev->dev, "(%d) failed to init kiq ring\n", r);
return r;
}
/**
* si_dma_ring_test_ib - test an IB on the DMA engine
*
* @ring: amdgpu_ring structure holding ring information
*
* Test a simple IB in the DMA ring (VI).
* Returns 0 on success, error on failure.
*/
static int si_dma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct dma_fence *f = NULL;
unsigned index;
u32 tmp = 0;
u64 gpu_addr;
long r;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
return r;
}
gpu_addr = adev->wb.gpu_addr + (index * 4);
tmp = 0xCAFEDEAD;
adev->wb.wb[index] = cpu_to_le32(tmp);
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(adev, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
goto err0;
}
ib.ptr[0] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 0, 1);
ib.ptr[1] = lower_32_bits(gpu_addr);
ib.ptr[2] = upper_32_bits(gpu_addr) & 0xff;
ib.ptr[3] = 0xDEADBEEF;
ib.length_dw = 4;
r = amdgpu_ib_schedule(ring, 1, &ib, NULL, NULL, &f);
if (r)
goto err1;
r = dma_fence_wait_timeout(f, false, timeout);
if (r == 0) {
DRM_ERROR("amdgpu: IB test timed out\n");
r = -ETIMEDOUT;
goto err1;
} else if (r < 0) {
DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
goto err1;
}
tmp = le32_to_cpu(adev->wb.wb[index]);
if (tmp == 0xDEADBEEF) {
DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
r = 0;
} else {
DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
r = -EINVAL;
}
err1:
amdgpu_ib_free(adev, &ib, NULL);
dma_fence_put(f);
err0:
amdgpu_wb_free(adev, index);
return r;
}
/**
* amdgpu_ih_ring_init - initialize the IH state
*
* @adev: amdgpu_device pointer
*
* Initializes the IH state and allocates a buffer
* for the IH ring buffer.
* Returns 0 for success, errors for failure.
*/
int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
bool use_bus_addr)
{
u32 rb_bufsz;
int r;
/* Align ring size */
rb_bufsz = order_base_2(ring_size / 4);
ring_size = (1 << rb_bufsz) * 4;
adev->irq.ih.ring_size = ring_size;
adev->irq.ih.ptr_mask = adev->irq.ih.ring_size - 1;
adev->irq.ih.rptr = 0;
adev->irq.ih.use_bus_addr = use_bus_addr;
if (adev->irq.ih.use_bus_addr) {
if (!adev->irq.ih.ring) {
/* add 8 bytes for the rptr/wptr shadows and
* add them to the end of the ring allocation.
*/
adev->irq.ih.ring = pci_alloc_consistent(adev->pdev,
adev->irq.ih.ring_size + 8,
&adev->irq.ih.rb_dma_addr);
if (adev->irq.ih.ring == NULL)
return -ENOMEM;
memset((void *)adev->irq.ih.ring, 0, adev->irq.ih.ring_size + 8);
adev->irq.ih.wptr_offs = (adev->irq.ih.ring_size / 4) + 0;
adev->irq.ih.rptr_offs = (adev->irq.ih.ring_size / 4) + 1;
}
return 0;
} else {
r = amdgpu_wb_get(adev, &adev->irq.ih.wptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ih wptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &adev->irq.ih.rptr_offs);
if (r) {
amdgpu_wb_free(adev, adev->irq.ih.wptr_offs);
dev_err(adev->dev, "(%d) ih rptr_offs wb alloc failed\n", r);
return r;
}
return amdgpu_ih_ring_alloc(adev);
}
}
/**
* cik_sdma_ring_test_ring - simple async dma engine test
*
* @ring: amdgpu_ring structure holding ring information
*
* Test the DMA engine by writing using it to write an
* value to memory. (CIK).
* Returns 0 for success, error for failure.
*/
static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
unsigned i;
unsigned index;
int r;
u32 tmp;
u64 gpu_addr;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
return r;
}
gpu_addr = adev->wb.gpu_addr + (index * 4);
tmp = 0xCAFEDEAD;
adev->wb.wb[index] = cpu_to_le32(tmp);
r = amdgpu_ring_alloc(ring, 5);
if (r) {
DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
amdgpu_wb_free(adev, index);
return r;
}
amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0));
amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
amdgpu_ring_write(ring, 1); /* number of DWs to follow */
amdgpu_ring_write(ring, 0xDEADBEEF);
amdgpu_ring_commit(ring);
for (i = 0; i < adev->usec_timeout; i++) {
tmp = le32_to_cpu(adev->wb.wb[index]);
if (tmp == 0xDEADBEEF)
break;
DRM_UDELAY(1);
}
if (i < adev->usec_timeout) {
DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
} else {
DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
ring->idx, tmp);
r = -EINVAL;
}
amdgpu_wb_free(adev, index);
return r;
}
/**
* amdgpu_ring_init - init driver ring struct.
*
* @adev: amdgpu_device pointer
* @ring: amdgpu_ring structure holding ring information
* @max_ndw: maximum number of dw for ring alloc
* @nop: nop packet for this ring
*
* Initialize the driver information for the selected ring (all asics).
* Returns 0 on success, error on failure.
*/
int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
unsigned max_dw, u32 nop, u32 align_mask,
struct amdgpu_irq_src *irq_src, unsigned irq_type,
enum amdgpu_ring_type ring_type)
{
int r;
if (ring->adev == NULL) {
if (adev->num_rings >= AMDGPU_MAX_RINGS)
return -EINVAL;
ring->adev = adev;
ring->idx = adev->num_rings++;
adev->rings[ring->idx] = ring;
r = amdgpu_fence_driver_init_ring(ring,
amdgpu_sched_hw_submission);
if (r)
return r;
}
r = amdgpu_wb_get(adev, &ring->rptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &ring->wptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &ring->fence_offs);
if (r) {
dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &ring->next_rptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring next_rptr wb alloc failed\n", r);
return r;
}
ring->next_rptr_gpu_addr = adev->wb.gpu_addr + ring->next_rptr_offs * 4;
ring->next_rptr_cpu_addr = &adev->wb.wb[ring->next_rptr_offs];
r = amdgpu_wb_get(adev, &ring->cond_exe_offs);
if (r) {
dev_err(adev->dev, "(%d) ring cond_exec_polling wb alloc failed\n", r);
return r;
}
ring->cond_exe_gpu_addr = adev->wb.gpu_addr + (ring->cond_exe_offs * 4);
ring->cond_exe_cpu_addr = &adev->wb.wb[ring->cond_exe_offs];
spin_lock_init(&ring->fence_lock);
r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
if (r) {
dev_err(adev->dev, "failed initializing fences (%d).\n", r);
return r;
}
ring->ring_size = roundup_pow_of_two(max_dw * 4 *
amdgpu_sched_hw_submission);
ring->align_mask = align_mask;
ring->nop = nop;
ring->type = ring_type;
/* Allocate ring buffer */
if (ring->ring_obj == NULL) {
r = amdgpu_bo_create(adev, ring->ring_size, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_GTT, 0,
NULL, NULL, &ring->ring_obj);
if (r) {
dev_err(adev->dev, "(%d) ring create failed\n", r);
return r;
}
r = amdgpu_bo_reserve(ring->ring_obj, false);
if (unlikely(r != 0))
return r;
r = amdgpu_bo_pin(ring->ring_obj, AMDGPU_GEM_DOMAIN_GTT,
&ring->gpu_addr);
if (r) {
amdgpu_bo_unreserve(ring->ring_obj);
dev_err(adev->dev, "(%d) ring pin failed\n", r);
return r;
}
r = amdgpu_bo_kmap(ring->ring_obj,
(void **)&ring->ring);
amdgpu_bo_unreserve(ring->ring_obj);
if (r) {
dev_err(adev->dev, "(%d) ring map failed\n", r);
return r;
}
}
ring->ptr_mask = (ring->ring_size / 4) - 1;
ring->max_dw = max_dw;
if (amdgpu_debugfs_ring_init(adev, ring)) {
DRM_ERROR("Failed to register debugfs file for rings !\n");
}
return 0;
}
/**
* cik_sdma_ring_test_ib - test an IB on the DMA engine
*
* @ring: amdgpu_ring structure holding ring information
*
* Test a simple IB in the DMA ring (CIK).
* Returns 0 on success, error on failure.
*/
static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct fence *f = NULL;
unsigned i;
unsigned index;
int r;
u32 tmp = 0;
u64 gpu_addr;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
return r;
}
gpu_addr = adev->wb.gpu_addr + (index * 4);
tmp = 0xCAFEDEAD;
adev->wb.wb[index] = cpu_to_le32(tmp);
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(ring, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
goto err0;
}
ib.ptr[0] = SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
ib.ptr[1] = lower_32_bits(gpu_addr);
ib.ptr[2] = upper_32_bits(gpu_addr);
ib.ptr[3] = 1;
ib.ptr[4] = 0xDEADBEEF;
ib.length_dw = 5;
r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, &ib, 1, NULL,
AMDGPU_FENCE_OWNER_UNDEFINED,
&f);
if (r)
goto err1;
r = fence_wait(f, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
goto err1;
}
for (i = 0; i < adev->usec_timeout; i++) {
tmp = le32_to_cpu(adev->wb.wb[index]);
if (tmp == 0xDEADBEEF)
break;
DRM_UDELAY(1);
}
if (i < adev->usec_timeout) {
DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
ring->idx, i);
goto err1;
} else {
DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
r = -EINVAL;
}
err1:
fence_put(f);
amdgpu_ib_free(adev, &ib);
err0:
amdgpu_wb_free(adev, index);
return r;
}