int main(int argc, char **argv)
{
uint32_t batch[2] = {MI_BATCH_BUFFER_END};
uint32_t handle;
uint32_t devid;
int fd;
drmtest_subtest_init(argc, argv);
fd = drm_open_any();
devid = intel_get_drm_devid(fd);
handle = gem_create(fd, 4096);
gem_write(fd, handle, 0, batch, sizeof(batch));
if (drmtest_run_subtest("render"))
loop(fd, handle, I915_EXEC_RENDER, "render");
if (drmtest_run_subtest("bsd"))
if (HAS_BSD_RING(devid))
loop(fd, handle, I915_EXEC_BSD, "bsd");
if (drmtest_run_subtest("blt"))
if (HAS_BLT_RING(devid))
loop(fd, handle, I915_EXEC_BLT, "blt");
gem_close(fd, handle);
close(fd);
return skipped_all ? 77 : 0;
}
void
intel_batchbuffer_flush(struct intel_batchbuffer *batch)
{
int ring = 0;
if (HAS_BLT_RING(batch->devid))
ring = I915_EXEC_BLT;
intel_batchbuffer_flush_on_ring(batch, ring);
}
int main(int argc, char **argv)
{
int fd;
int devid;
igt_skip_on_simulation();
if (argc != 1) {
fprintf(stderr, "usage: %s\n", argv[0]);
igt_fail(-1);
}
fd = drm_open_any();
devid = intel_get_drm_devid(fd);
if (!HAS_BLT_RING(devid)) {
fprintf(stderr, "not (yet) implemented for pre-snb\n");
return 77;
}
bufmgr = drm_intel_bufmgr_gem_init(fd, 4096);
if (!bufmgr) {
fprintf(stderr, "failed to init libdrm\n");
igt_fail(-1);
}
drm_intel_bufmgr_gem_enable_reuse(bufmgr);
batch = intel_batchbuffer_alloc(bufmgr, devid);
if (!batch) {
fprintf(stderr, "failed to create batch buffer\n");
igt_fail(-1);
}
target_buffer = drm_intel_bo_alloc(bufmgr, "target bo", 4096, 4096);
if (!target_buffer) {
fprintf(stderr, "failed to alloc target buffer\n");
igt_fail(-1);
}
blt_bo = drm_intel_bo_alloc(bufmgr, "target bo", 4*4096*4096, 4096);
if (!blt_bo) {
fprintf(stderr, "failed to alloc blt buffer\n");
igt_fail(-1);
}
dummy_reloc_loop();
drm_intel_bo_unreference(target_buffer);
intel_batchbuffer_free(batch);
drm_intel_bufmgr_destroy(bufmgr);
close(fd);
return 0;
}
static void make_busy(int fd, uint32_t handle)
{
struct drm_i915_gem_execbuffer2 execbuf;
struct drm_i915_gem_exec_object2 obj[2];
struct drm_i915_gem_relocation_entry reloc[2];
uint32_t batch[20];
uint32_t tmp;
int count;
tmp = gem_create(fd, 1024*1024);
obj[0].handle = tmp;
obj[0].relocation_count = 0;
obj[0].relocs_ptr = 0;
obj[0].alignment = 0;
obj[0].offset = 0;
obj[0].flags = 0;
obj[0].rsvd1 = 0;
obj[0].rsvd2 = 0;
obj[1].handle = handle;
obj[1].relocation_count = 2;
obj[1].relocs_ptr = (uintptr_t) reloc;
obj[1].alignment = 0;
obj[1].offset = 0;
obj[1].flags = 0;
obj[1].rsvd1 = 0;
obj[1].rsvd2 = 0;
execbuf.buffers_ptr = (uintptr_t)obj;
execbuf.buffer_count = 2;
execbuf.batch_start_offset = 0;
execbuf.batch_len = gem_linear_blt(fd, batch, tmp, tmp, 1024*1024,reloc);
execbuf.cliprects_ptr = 0;
execbuf.num_cliprects = 0;
execbuf.DR1 = 0;
execbuf.DR4 = 0;
execbuf.flags = 0;
if (HAS_BLT_RING(intel_get_drm_devid(fd)))
execbuf.flags |= I915_EXEC_BLT;
i915_execbuffer2_set_context_id(execbuf, 0);
execbuf.rsvd2 = 0;
gem_write(fd, handle, 0, batch, execbuf.batch_len);
for (count = 0; count < 10; count++)
do_ioctl(fd, DRM_IOCTL_I915_GEM_EXECBUFFER2, &execbuf);
gem_close(fd, tmp);
}
int main(int argc, char **argv)
{
int fd;
int devid;
if (argc != 1) {
fprintf(stderr, "usage: %s\n", argv[0]);
exit(-1);
}
fd = drm_open_any();
devid = intel_get_drm_devid(fd);
if (!HAS_BLT_RING(devid)) {
fprintf(stderr, "inter ring check needs gen6+\n");
return 77;
}
bufmgr = drm_intel_bufmgr_gem_init(fd, 4096);
if (!bufmgr) {
fprintf(stderr, "failed to init libdrm\n");
exit(-1);
}
drm_intel_bufmgr_gem_enable_reuse(bufmgr);
batch = intel_batchbuffer_alloc(bufmgr, devid);
if (!batch) {
fprintf(stderr, "failed to create batch buffer\n");
exit(-1);
}
target_buffer = drm_intel_bo_alloc(bufmgr, "target bo", 4096, 4096);
if (!target_buffer) {
fprintf(stderr, "failed to alloc target buffer\n");
exit(-1);
}
store_dword_loop(I915_EXEC_RENDER);
drm_intel_bo_unreference(target_buffer);
intel_batchbuffer_free(batch);
drm_intel_bufmgr_destroy(bufmgr);
close(fd);
return 0;
}
int main(int argc, char **argv)
{
int fd;
int devid;
if (argc != 1) {
fprintf(stderr, "usage: %s\n", argv[0]);
igt_fail(-1);
}
fd = drm_open_any();
devid = intel_get_drm_devid(fd);
if (HAS_BSD_RING(devid))
num_rings++;
if (HAS_BLT_RING(devid))
num_rings++;
printf("num rings detected: %i\n", num_rings);
bufmgr = drm_intel_bufmgr_gem_init(fd, 4096);
if (!bufmgr) {
fprintf(stderr, "failed to init libdrm\n");
igt_fail(-1);
}
drm_intel_bufmgr_gem_enable_reuse(bufmgr);
batch = intel_batchbuffer_alloc(bufmgr, devid);
if (!batch) {
fprintf(stderr, "failed to create batch buffer\n");
igt_fail(-1);
}
mi_lri_loop();
gem_quiescent_gpu(fd);
intel_batchbuffer_free(batch);
drm_intel_bufmgr_destroy(bufmgr);
close(fd);
return 0;
}
static int
copy(int fd, uint32_t dst, uint32_t src, uint32_t *all_bo, int n_bo)
{
uint32_t batch[12];
struct drm_i915_gem_relocation_entry reloc[2];
struct drm_i915_gem_exec_object2 *obj;
struct drm_i915_gem_execbuffer2 exec;
uint32_t handle;
int n, ret, i=0;
batch[i++] = (XY_SRC_COPY_BLT_CMD |
XY_SRC_COPY_BLT_WRITE_ALPHA |
XY_SRC_COPY_BLT_WRITE_RGB | 6);
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i - 1] += 2;
batch[i++] = (3 << 24) | /* 32 bits */
(0xcc << 16) | /* copy ROP */
WIDTH*4;
batch[i++] = 0; /* dst x1,y1 */
batch[i++] = (HEIGHT << 16) | WIDTH; /* dst x2,y2 */
batch[i++] = 0; /* dst reloc */
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i++] = 0; /* FIXME */
batch[i++] = 0; /* src x1,y1 */
batch[i++] = WIDTH*4;
batch[i++] = 0; /* src reloc */
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i++] = 0; /* FIXME */
batch[i++] = MI_BATCH_BUFFER_END;
batch[i++] = MI_NOOP;
handle = gem_create(fd, 4096);
gem_write(fd, handle, 0, batch, sizeof(batch));
reloc[0].target_handle = dst;
reloc[0].delta = 0;
reloc[0].offset = 4 * sizeof(batch[0]);
reloc[0].presumed_offset = 0;
reloc[0].read_domains = I915_GEM_DOMAIN_RENDER;
reloc[0].write_domain = I915_GEM_DOMAIN_RENDER;
reloc[1].target_handle = src;
reloc[1].delta = 0;
reloc[1].offset = 7 * sizeof(batch[0]);
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
reloc[1].offset += sizeof(batch[0]);
reloc[1].presumed_offset = 0;
reloc[1].read_domains = I915_GEM_DOMAIN_RENDER;
reloc[1].write_domain = 0;
obj = calloc(n_bo + 1, sizeof(*obj));
for (n = 0; n < n_bo; n++)
obj[n].handle = all_bo[n];
obj[n].handle = handle;
obj[n].relocation_count = 2;
obj[n].relocs_ptr = (uintptr_t)reloc;
exec.buffers_ptr = (uintptr_t)obj;
exec.buffer_count = n_bo + 1;
exec.batch_start_offset = 0;
exec.batch_len = i * 4;
exec.DR1 = exec.DR4 = 0;
exec.num_cliprects = 0;
exec.cliprects_ptr = 0;
exec.flags = HAS_BLT_RING(intel_get_drm_devid(fd)) ? I915_EXEC_BLT : 0;
i915_execbuffer2_set_context_id(exec, 0);
exec.rsvd2 = 0;
ret = drmIoctl(fd, DRM_IOCTL_I915_GEM_EXECBUFFER2, &exec);
if (ret)
ret = errno;
gem_close(fd, handle);
free(obj);
return ret;
}
static void
copy(int fd, uint32_t dst, uint32_t src, unsigned int error)
{
uint32_t batch[12];
struct drm_i915_gem_relocation_entry reloc[2];
struct drm_i915_gem_exec_object2 obj[3];
struct drm_i915_gem_execbuffer2 exec;
uint32_t handle;
int ret, i=0;
batch[i++] = XY_SRC_COPY_BLT_CMD |
XY_SRC_COPY_BLT_WRITE_ALPHA |
XY_SRC_COPY_BLT_WRITE_RGB;
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i - 1] |= 8;
else
batch[i - 1] |= 6;
batch[i++] = (3 << 24) | /* 32 bits */
(0xcc << 16) | /* copy ROP */
WIDTH*4;
batch[i++] = 0; /* dst x1,y1 */
batch[i++] = (HEIGHT << 16) | WIDTH; /* dst x2,y2 */
batch[i++] = 0; /* dst reloc */
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i++] = 0;
batch[i++] = 0; /* src x1,y1 */
batch[i++] = WIDTH*4;
batch[i++] = 0; /* src reloc */
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i++] = 0;
batch[i++] = MI_BATCH_BUFFER_END;
batch[i++] = MI_NOOP;
handle = gem_create(fd, 4096);
gem_write(fd, handle, 0, batch, sizeof(batch));
reloc[0].target_handle = dst;
reloc[0].delta = 0;
reloc[0].offset = 4 * sizeof(batch[0]);
reloc[0].presumed_offset = 0;
reloc[0].read_domains = I915_GEM_DOMAIN_RENDER;
reloc[0].write_domain = I915_GEM_DOMAIN_RENDER;
reloc[1].target_handle = src;
reloc[1].delta = 0;
reloc[1].offset = 7 * sizeof(batch[0]);
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
reloc[1].offset += sizeof(batch[0]);
reloc[1].presumed_offset = 0;
reloc[1].read_domains = I915_GEM_DOMAIN_RENDER;
reloc[1].write_domain = 0;
memset(obj, 0, sizeof(obj));
exec.buffer_count = 0;
obj[exec.buffer_count++].handle = dst;
if (src != dst)
obj[exec.buffer_count++].handle = src;
obj[exec.buffer_count].handle = handle;
obj[exec.buffer_count].relocation_count = 2;
obj[exec.buffer_count].relocs_ptr = (uintptr_t)reloc;
exec.buffer_count++;
exec.buffers_ptr = (uintptr_t)obj;
exec.batch_start_offset = 0;
exec.batch_len = i * 4;
exec.DR1 = exec.DR4 = 0;
exec.num_cliprects = 0;
exec.cliprects_ptr = 0;
exec.flags = HAS_BLT_RING(intel_get_drm_devid(fd)) ? I915_EXEC_BLT : 0;
i915_execbuffer2_set_context_id(exec, 0);
exec.rsvd2 = 0;
ret = drmIoctl(fd, DRM_IOCTL_I915_GEM_EXECBUFFER2, &exec);
if (ret)
ret = errno;
if (error == ~0)
igt_assert_neq(ret, 0);
else
igt_assert(ret == error);
gem_close(fd, handle);
}
static void busy(data_t *data, uint32_t handle, int size, int loops)
{
struct drm_i915_gem_relocation_entry reloc[20];
struct drm_i915_gem_exec_object2 gem_exec[2];
struct drm_i915_gem_execbuffer2 execbuf;
struct drm_i915_gem_pwrite gem_pwrite;
struct drm_i915_gem_create create;
uint32_t buf[170], *b;
int i;
memset(reloc, 0, sizeof(reloc));
memset(gem_exec, 0, sizeof(gem_exec));
memset(&execbuf, 0, sizeof(execbuf));
b = buf;
for (i = 0; i < 20; i++) {
*b++ = XY_COLOR_BLT_CMD_NOLEN |
((data->intel_gen >= 8) ? 5 : 4) |
COLOR_BLT_WRITE_ALPHA | XY_COLOR_BLT_WRITE_RGB;
*b++ = 0xf0 << 16 | 1 << 25 | 1 << 24 | 4096;
*b++ = 0;
*b++ = size >> 12 << 16 | 1024;
reloc[i].offset = (b - buf) * sizeof(uint32_t);
reloc[i].target_handle = handle;
reloc[i].read_domains = I915_GEM_DOMAIN_RENDER;
reloc[i].write_domain = I915_GEM_DOMAIN_RENDER;
*b++ = 0;
if (data->intel_gen >= 8)
*b++ = 0;
*b++ = canary;
}
*b++ = MI_BATCH_BUFFER_END;
if ((b - buf) & 1)
*b++ = 0;
gem_exec[0].handle = handle;
gem_exec[0].flags = EXEC_OBJECT_NEEDS_FENCE;
create.handle = 0;
create.size = 4096;
drmIoctl(data->fd, DRM_IOCTL_I915_GEM_CREATE, &create);
gem_exec[1].handle = create.handle;
gem_exec[1].relocation_count = 20;
gem_exec[1].relocs_ptr = (uintptr_t)reloc;
execbuf.buffers_ptr = (uintptr_t)gem_exec;
execbuf.buffer_count = 2;
execbuf.batch_len = (b - buf) * sizeof(buf[0]);
execbuf.flags = 1 << 11;
if (HAS_BLT_RING(data->devid))
execbuf.flags |= I915_EXEC_BLT;
gem_pwrite.handle = gem_exec[1].handle;
gem_pwrite.offset = 0;
gem_pwrite.size = execbuf.batch_len;
gem_pwrite.data_ptr = (uintptr_t)buf;
if (drmIoctl(data->fd, DRM_IOCTL_I915_GEM_PWRITE, &gem_pwrite) == 0) {
while (loops--)
drmIoctl(data->fd, DRM_IOCTL_I915_GEM_EXECBUFFER2, &execbuf);
}
drmIoctl(data->fd, DRM_IOCTL_GEM_CLOSE, &create.handle);
}
static int
blit(int fd, uint32_t dst, uint32_t src, uint32_t *all_bo, int n_bo)
{
uint32_t batch[12];
struct drm_i915_gem_relocation_entry reloc[2];
struct drm_i915_gem_exec_object2 *obj;
struct drm_i915_gem_execbuffer2 exec;
uint32_t handle;
int n, ret, i=0;
batch[i++] = XY_SRC_COPY_BLT_CMD |
XY_SRC_COPY_BLT_WRITE_ALPHA |
XY_SRC_COPY_BLT_WRITE_RGB;
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i - 1] |= 8;
else
batch[i - 1] |= 6;
batch[i++] = (3 << 24) | /* 32 bits */
(0xcc << 16) | /* copy ROP */
WIDTH*4;
batch[i++] = 0; /* dst x1,y1 */
batch[i++] = (HEIGHT << 16) | WIDTH; /* dst x2,y2 */
batch[i++] = 0; /* dst reloc */
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i++] = 0;
batch[i++] = 0; /* src x1,y1 */
batch[i++] = WIDTH*4;
batch[i++] = 0; /* src reloc */
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
batch[i++] = 0;
batch[i++] = MI_BATCH_BUFFER_END;
batch[i++] = MI_NOOP;
handle = gem_create(fd, 4096);
gem_write(fd, handle, 0, batch, sizeof(batch));
reloc[0].target_handle = dst;
reloc[0].delta = 0;
reloc[0].offset = 4 * sizeof(batch[0]);
reloc[0].presumed_offset = 0;
reloc[0].read_domains = I915_GEM_DOMAIN_RENDER;
reloc[0].write_domain = I915_GEM_DOMAIN_RENDER;
reloc[1].target_handle = src;
reloc[1].delta = 0;
reloc[1].offset = 7 * sizeof(batch[0]);
if (intel_gen(intel_get_drm_devid(fd)) >= 8)
reloc[1].offset += sizeof(batch[0]);
reloc[1].presumed_offset = 0;
reloc[1].read_domains = I915_GEM_DOMAIN_RENDER;
reloc[1].write_domain = 0;
memset(&exec, 0, sizeof(exec));
obj = calloc(n_bo + 1, sizeof(*obj));
for (n = 0; n < n_bo; n++)
obj[n].handle = all_bo[n];
obj[n].handle = handle;
obj[n].relocation_count = 2;
obj[n].relocs_ptr = (uintptr_t)reloc;
exec.buffers_ptr = (uintptr_t)obj;
exec.buffer_count = n_bo + 1;
exec.flags = HAS_BLT_RING(intel_get_drm_devid(fd)) ? I915_EXEC_BLT : 0;
ret = __gem_execbuf(fd, &exec);
gem_close(fd, handle);
free(obj);
return ret;
}
static void run(int object_size)
{
struct drm_i915_gem_execbuffer2 execbuf;
struct drm_i915_gem_exec_object2 exec[3];
struct drm_i915_gem_relocation_entry reloc[4];
uint32_t buf[20];
uint32_t handle, src, dst;
int fd, len, count;
int ring;
fd = drm_open_any();
handle = gem_create(fd, 4096);
src = gem_create(fd, object_size);
dst = gem_create(fd, object_size);
len = gem_linear_blt(buf, src, dst, object_size, reloc);
gem_write(fd, handle, 0, buf, len);
exec[0].handle = src;
exec[0].relocation_count = 0;
exec[0].relocs_ptr = 0;
exec[0].alignment = 0;
exec[0].offset = 0;
exec[0].flags = 0;
exec[0].rsvd1 = 0;
exec[0].rsvd2 = 0;
exec[1].handle = dst;
exec[1].relocation_count = 0;
exec[1].relocs_ptr = 0;
exec[1].alignment = 0;
exec[1].offset = 0;
exec[1].flags = 0;
exec[1].rsvd1 = 0;
exec[1].rsvd2 = 0;
exec[2].handle = handle;
exec[2].relocation_count = len > 40 ? 4 : 2;
exec[2].relocs_ptr = (uintptr_t)reloc;
exec[2].alignment = 0;
exec[2].offset = 0;
exec[2].flags = 0;
exec[2].rsvd1 = 0;
exec[2].rsvd2 = 0;
ring = 0;
if (HAS_BLT_RING(intel_get_drm_devid(fd)))
ring = I915_EXEC_BLT;
execbuf.buffers_ptr = (uintptr_t)exec;
execbuf.buffer_count = 3;
execbuf.batch_start_offset = 0;
execbuf.batch_len = len;
execbuf.cliprects_ptr = 0;
execbuf.num_cliprects = 0;
execbuf.DR1 = 0;
execbuf.DR4 = 0;
execbuf.flags = ring;
i915_execbuffer2_set_context_id(execbuf, 0);
execbuf.rsvd2 = 0;
for (count = 1; count <= 1<<17; count <<= 1) {
struct timeval start, end;
gettimeofday(&start, NULL);
if (gem_exec(fd, &execbuf, count))
exit(1);
gem_sync(fd, handle);
gettimeofday(&end, NULL);
printf("Time to blt %d bytes x %6d: %7.3fµs, %s\n",
object_size, count,
elapsed(&start, &end, count),
bytes_per_sec((char *)buf, object_size/elapsed(&start, &end, count)*1e6));
fflush(stdout);
}
gem_close(fd, handle);
close(fd);
}
