Commit bbec97aa authored by Christian König's avatar Christian König Committed by Alex Deucher

drm/amdgpu: add a fence timeout for the IB tests v2

10ms should be enough for now.

v2: fix some typos in CIK code
Signed-off-by: default avatarChristian König <christian.koenig@amd.com>
Reviewed-by: default avatarChunming Zhou <david1.zhou@amd.com>
Reviewed-by: default avatarEdward O'Callaghan <funfunctor@folklore1984.net>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
parent 8de190c9
......@@ -306,7 +306,7 @@ struct amdgpu_ring_funcs {
uint32_t oa_base, uint32_t oa_size);
/* testing functions */
int (*test_ring)(struct amdgpu_ring *ring);
int (*test_ib)(struct amdgpu_ring *ring);
int (*test_ib)(struct amdgpu_ring *ring, long timeout);
/* insert NOP packets */
void (*insert_nop)(struct amdgpu_ring *ring, uint32_t count);
/* pad the indirect buffer to the necessary number of dw */
......@@ -2245,7 +2245,7 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
#define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags)))
#define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib)))
#define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
#define amdgpu_ring_test_ib(r) (r)->funcs->test_ib((r))
#define amdgpu_ring_test_ib(r, t) (r)->funcs->test_ib((r), (t))
#define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r))
#define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r))
#define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r))
......
......@@ -33,6 +33,8 @@
#include "amdgpu.h"
#include "atom.h"
#define AMDGPU_IB_TEST_TIMEOUT msecs_to_jiffies(10)
/*
* IB
* IBs (Indirect Buffers) and areas of GPU accessible memory where
......@@ -286,7 +288,7 @@ int amdgpu_ib_ring_tests(struct amdgpu_device *adev)
if (!ring || !ring->ready)
continue;
r = amdgpu_ring_test_ib(ring);
r = amdgpu_ring_test_ib(ring, AMDGPU_IB_TEST_TIMEOUT);
if (r) {
ring->ready = false;
......
......@@ -1135,29 +1135,34 @@ void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring)
*
* Test if we can successfully execute an IB
*/
int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring)
int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct fence *fence = NULL;
int r;
struct fence *fence;
long r;
r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
if (r) {
DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
goto error;
}
r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence);
if (r) {
DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
goto error;
}
r = fence_wait(fence, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
goto error;
r = fence_wait_timeout(fence, false, timeout);
if (r == 0) {
DRM_ERROR("amdgpu: IB test timed out.\n");
r = -ETIMEDOUT;
} else if (r < 0) {
DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
} else {
DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
r = 0;
}
DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
error:
fence_put(fence);
return r;
......
......@@ -37,6 +37,6 @@ void amdgpu_uvd_free_handles(struct amdgpu_device *adev,
int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx);
void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring);
void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring);
int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring);
int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout);
#endif
......@@ -844,10 +844,10 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
* @ring: the engine to test on
*
*/
int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring)
int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct fence *fence = NULL;
int r;
long r;
/* skip vce ring1 ib test for now, since it's not reliable */
if (ring == &ring->adev->vce.ring[1])
......@@ -855,21 +855,25 @@ int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring)
r = amdgpu_vce_get_create_msg(ring, 1, NULL);
if (r) {
DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
goto error;
}
r = amdgpu_vce_get_destroy_msg(ring, 1, true, &fence);
if (r) {
DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
goto error;
}
r = fence_wait(fence, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
r = fence_wait_timeout(fence, false, timeout);
if (r == 0) {
DRM_ERROR("amdgpu: IB test timed out.\n");
r = -ETIMEDOUT;
} else if (r < 0) {
DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
} else {
DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
r = 0;
}
error:
fence_put(fence);
......
......@@ -39,7 +39,7 @@ void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib,
void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
unsigned flags);
int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring);
int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring);
int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout);
void amdgpu_vce_ring_begin_use(struct amdgpu_ring *ring);
void amdgpu_vce_ring_end_use(struct amdgpu_ring *ring);
......
......@@ -617,19 +617,19 @@ static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
* 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)
static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct fence *f = NULL;
unsigned index;
int r;
u32 tmp = 0;
u64 gpu_addr;
long r;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
return r;
}
......@@ -639,7 +639,7 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(adev, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
goto err0;
}
......@@ -654,14 +654,19 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
if (r)
goto err1;
r = fence_wait(f, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
r = 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;
......
......@@ -2105,25 +2105,25 @@ static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
* Provides a basic gfx ring test to verify that IBs are working.
* Returns 0 on success, error on failure.
*/
static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring)
static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct fence *f = NULL;
uint32_t scratch;
uint32_t tmp = 0;
int r;
long r;
r = amdgpu_gfx_scratch_get(adev, &scratch);
if (r) {
DRM_ERROR("amdgpu: failed to get scratch reg (%d).\n", r);
DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
return r;
}
WREG32(scratch, 0xCAFEDEAD);
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(adev, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
goto err1;
}
ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
......@@ -2135,14 +2135,19 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring)
if (r)
goto err2;
r = fence_wait(f, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
r = fence_wait_timeout(f, false, timeout);
if (r == 0) {
DRM_ERROR("amdgpu: IB test timed out\n");
r = -ETIMEDOUT;
goto err2;
} else if (r < 0) {
DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
goto err2;
}
tmp = RREG32(scratch);
if (tmp == 0xDEADBEEF) {
DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
r = 0;
} else {
DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
scratch, tmp);
......
......@@ -787,25 +787,25 @@ static int gfx_v8_0_ring_test_ring(struct amdgpu_ring *ring)
return r;
}
static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring)
static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct fence *f = NULL;
uint32_t scratch;
uint32_t tmp = 0;
int r;
long r;
r = amdgpu_gfx_scratch_get(adev, &scratch);
if (r) {
DRM_ERROR("amdgpu: failed to get scratch reg (%d).\n", r);
DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
return r;
}
WREG32(scratch, 0xCAFEDEAD);
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(adev, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
goto err1;
}
ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
......@@ -817,14 +817,19 @@ static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring)
if (r)
goto err2;
r = fence_wait(f, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
r = fence_wait_timeout(f, false, timeout);
if (r == 0) {
DRM_ERROR("amdgpu: IB test timed out.\n");
r = -ETIMEDOUT;
goto err2;
} else if (r < 0) {
DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
goto err2;
}
tmp = RREG32(scratch);
if (tmp == 0xDEADBEEF) {
DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
r = 0;
} else {
DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
scratch, tmp);
......
......@@ -668,19 +668,19 @@ static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring)
* Test a simple IB in the DMA ring (VI).
* Returns 0 on success, error on failure.
*/
static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring)
static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct fence *f = NULL;
unsigned index;
int r;
u32 tmp = 0;
u64 gpu_addr;
long r;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
return r;
}
......@@ -690,7 +690,7 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring)
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(adev, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
goto err0;
}
......@@ -709,14 +709,19 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring)
if (r)
goto err1;
r = fence_wait(f, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
r = fence_wait_timeout(f, false, timeout);
if (r == 0) {
DRM_ERROR("amdgpu: IB test timed out\n");
r = -ETIMEDOUT;
goto err1;
} else if (r) {
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;
......
......@@ -896,19 +896,19 @@ static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring)
* Test a simple IB in the DMA ring (VI).
* Returns 0 on success, error on failure.
*/
static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring)
static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_ib ib;
struct fence *f = NULL;
unsigned index;
int r;
u32 tmp = 0;
u64 gpu_addr;
long r;
r = amdgpu_wb_get(adev, &index);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
return r;
}
......@@ -918,7 +918,7 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring)
memset(&ib, 0, sizeof(ib));
r = amdgpu_ib_get(adev, NULL, 256, &ib);
if (r) {
DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
goto err0;
}
......@@ -937,14 +937,19 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring)
if (r)
goto err1;
r = fence_wait(f, false);
if (r) {
DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
r = 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;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment