[igt-dev] [PATCH i-g-t] tests/i915/gem_sync: Add support for local memory
Zbigniew Kempczyński
zbigniew.kempczynski at intel.com
Mon Jan 3 11:27:40 UTC 2022
On Sat, Dec 25, 2021 at 01:33:10AM +0530, sai.gowtham.ch at intel.com wrote:
> From: Ch Sai Gowtham <sai.gowtham.ch at intel.com>
>
> Add support for local memory region (Device memory)
>
> Signed-off-by: Ch Sai Gowtham <sai.gowtham.ch at intel.com>
> Cc: Zbigniew Kempczyński <zbigniew.kempczynski at intel.com>
> ---
> tests/i915/gem_sync.c | 193 ++++++++++++++++++++++++++++--------------
> 1 file changed, 131 insertions(+), 62 deletions(-)
>
> diff --git a/tests/i915/gem_sync.c b/tests/i915/gem_sync.c
> index 8c435845..8f8be2b1 100644
> --- a/tests/i915/gem_sync.c
> +++ b/tests/i915/gem_sync.c
> @@ -143,9 +143,20 @@ static void xchg_engine(void *array, unsigned i, unsigned j)
> igt_swap(E[i], E[j]);
> }
>
> +static uint32_t batch_create(int fd, uint32_t batch_size, uint32_t region)
> +{
> + const uint32_t bbe = MI_BATCH_BUFFER_END;
> + uint32_t handle;
> +
> + handle = gem_create_in_memory_regions(fd, 4096, region);
What batch_size is passed here for?
> + gem_write(fd, handle, 0, &bbe, sizeof(bbe));
> +
> + return handle;
> +}
> +
> static void
> sync_ring(int fd, const intel_ctx_t *ctx,
> - unsigned ring, int num_children, int timeout)
> + unsigned ring, int num_children, int timeout, uint32_t region)
> {
> struct intel_engine_data ied;
>
> @@ -155,15 +166,13 @@ sync_ring(int fd, const intel_ctx_t *ctx,
>
> intel_detect_and_clear_missed_interrupts(fd);
> igt_fork(child, num_children) {
> - const uint32_t bbe = MI_BATCH_BUFFER_END;
> struct drm_i915_gem_exec_object2 object;
> struct drm_i915_gem_execbuffer2 execbuf;
> double start, elapsed;
> unsigned long cycles;
>
> memset(&object, 0, sizeof(object));
> - object.handle = gem_create(fd, 4096);
> - gem_write(fd, object.handle, 0, &bbe, sizeof(bbe));
> + object.handle = batch_create(fd, 4096, region);
>
> memset(&execbuf, 0, sizeof(execbuf));
> execbuf.buffers_ptr = to_user_pointer(&object);
> @@ -193,9 +202,8 @@ sync_ring(int fd, const intel_ctx_t *ctx,
>
> static void
> idle_ring(int fd, const intel_ctx_t *ctx, unsigned int ring,
> - int num_children, int timeout)
> + int num_children, int timeout, uint32_t region)
> {
> - const uint32_t bbe = MI_BATCH_BUFFER_END;
> struct drm_i915_gem_exec_object2 object;
> struct drm_i915_gem_execbuffer2 execbuf;
> double start, elapsed;
> @@ -204,8 +212,7 @@ idle_ring(int fd, const intel_ctx_t *ctx, unsigned int ring,
> gem_require_ring(fd, ring);
>
> memset(&object, 0, sizeof(object));
> - object.handle = gem_create(fd, 4096);
> - gem_write(fd, object.handle, 0, &bbe, sizeof(bbe));
> + object.handle = batch_create(fd, 4096, region);
>
> memset(&execbuf, 0, sizeof(execbuf));
> execbuf.buffers_ptr = to_user_pointer(&object);
> @@ -234,7 +241,7 @@ idle_ring(int fd, const intel_ctx_t *ctx, unsigned int ring,
>
> static void
> wakeup_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> - int timeout, int wlen)
> + int timeout, int wlen, uint32_t region)
> {
> struct intel_engine_data ied;
> uint64_t ahnd = get_reloc_ahnd(fd, ctx->id);
> @@ -254,7 +261,7 @@ wakeup_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> ahnd = get_reloc_ahnd(fd, ctx->id);
>
> memset(&object, 0, sizeof(object));
> - object.handle = gem_create(fd, 4096);
> + object.handle = gem_create_in_memory_regions(fd, 4096, region);
You've created batch_create() for it, haven't you?
> object.offset = get_offset(ahnd, object.handle, 4096, 0);
> if (ahnd)
> object.flags = EXEC_OBJECT_PINNED;
> @@ -339,7 +346,7 @@ wakeup_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> }
>
> static void active_ring(int fd, const intel_ctx_t *ctx, unsigned int ring,
> - int num_children, int timeout)
> + int num_children, int timeout, uint32_t region)
> {
Here we have spinners which should be executed within region but we
got no such possibility right now. We should add .region to spinner
creation flags.
> struct intel_engine_data ied;
> uint64_t ahnd = get_reloc_ahnd(fd, ctx->id);
> @@ -398,7 +405,7 @@ static void active_ring(int fd, const intel_ctx_t *ctx, unsigned int ring,
>
> static void
> active_wakeup_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> - int timeout, int wlen)
> + int timeout, int wlen, uint32_t region)
> {
> struct intel_engine_data ied;
> uint64_t ahnd0 = get_reloc_ahnd(fd, 0);
> @@ -420,7 +427,7 @@ active_wakeup_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> ahnd = get_reloc_ahnd(fd, ctx->id);
>
> memset(&object, 0, sizeof(object));
> - object.handle = gem_create(fd, 4096);
> + object.handle = gem_create_in_memory_regions(fd, 4096, region);
Same with create_batch().
> object.offset = get_offset(ahnd, object.handle, 4096, 0);
> if (ahnd)
> object.offset = EXEC_OBJECT_PINNED;
> @@ -529,7 +536,7 @@ active_wakeup_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
>
> static void
> store_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> - int num_children, int timeout)
> + int num_children, int timeout, uint32_t region)
> {
> const unsigned int gen = intel_gen(intel_get_drm_devid(fd));
> struct intel_engine_data ied;
> @@ -541,7 +548,6 @@ store_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
>
> intel_detect_and_clear_missed_interrupts(fd);
> igt_fork(child, num_children) {
> - const uint32_t bbe = MI_BATCH_BUFFER_END;
> struct drm_i915_gem_exec_object2 object[2];
> struct drm_i915_gem_relocation_entry reloc[1024];
> struct drm_i915_gem_execbuffer2 execbuf;
> @@ -559,14 +565,13 @@ store_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> execbuf.rsvd1 = ctx->id;
>
> memset(object, 0, sizeof(object));
> - object[0].handle = gem_create(fd, 4096);
> - gem_write(fd, object[0].handle, 0, &bbe, sizeof(bbe));
> + object[0].handle = batch_create(fd, 4096, region);
> execbuf.buffer_count = 1;
> gem_execbuf(fd, &execbuf);
>
> object[0].flags |= EXEC_OBJECT_WRITE;
> object[0].flags |= has_relocs ? 0 : EXEC_OBJECT_PINNED;
> - object[1].handle = gem_create(fd, 20*1024);
> + object[1].handle = gem_create_in_memory_regions(fd, 20*1024, region);
>
> object[1].relocs_ptr = to_user_pointer(reloc);
> object[1].relocation_count = has_relocs ? 1024 : 0;
> @@ -629,7 +634,7 @@ store_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
>
> static void
> switch_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> - int num_children, int timeout)
> + int num_children, int timeout, uint32_t region)
> {
> const unsigned int gen = intel_gen(intel_get_drm_devid(fd));
> struct intel_engine_data ied;
> @@ -653,7 +658,6 @@ switch_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> unsigned long cycles;
>
> for (int i = 0; i < ARRAY_SIZE(contexts); i++) {
> - const uint32_t bbe = MI_BATCH_BUFFER_END;
> const uint32_t sz = 32 << 10;
> struct context *c = &contexts[i];
> uint32_t *batch, *b;
> @@ -670,13 +674,12 @@ switch_ring(int fd, const intel_ctx_t *ctx, unsigned ring,
> c->execbuf.rsvd1 = c->ctx->id;
>
> memset(c->object, 0, sizeof(c->object));
> - c->object[0].handle = gem_create(fd, 4096);
> - gem_write(fd, c->object[0].handle, 0, &bbe, sizeof(bbe));
> + c->object[0].handle = batch_create(fd, 4096, region);
> c->execbuf.buffer_count = 1;
> gem_execbuf(fd, &c->execbuf);
>
> c->object[0].flags |= EXEC_OBJECT_WRITE;
> - c->object[1].handle = gem_create(fd, sz);
> + c->object[1].handle = gem_create_in_memory_regions(fd, sz, region);
>
> c->object[1].relocs_ptr = to_user_pointer(c->reloc);
> c->object[1].relocation_count = has_relocs ? 1024 * i : 0;
> @@ -813,10 +816,9 @@ static void *waiter(void *arg)
>
> static void
> __store_many(int fd, const intel_ctx_t *ctx, unsigned ring,
> - int timeout, unsigned long *cycles)
> + int timeout, unsigned long *cycles, uint32_t region)
> {
> const unsigned int gen = intel_gen(intel_get_drm_devid(fd));
> - const uint32_t bbe = MI_BATCH_BUFFER_END;
> struct drm_i915_gem_exec_object2 object[2];
> struct drm_i915_gem_execbuffer2 execbuf;
> struct drm_i915_gem_relocation_entry reloc[1024];
> @@ -836,8 +838,7 @@ __store_many(int fd, const intel_ctx_t *ctx, unsigned ring,
> execbuf.rsvd1 = ctx->id;
>
> memset(object, 0, sizeof(object));
> - object[0].handle = gem_create(fd, 4096);
> - gem_write(fd, object[0].handle, 0, &bbe, sizeof(bbe));
> + object[0].handle = batch_create(fd, 4096, region);
> execbuf.buffer_count = 1;
> gem_execbuf(fd, &execbuf);
> object[0].flags |= EXEC_OBJECT_WRITE;
> @@ -945,7 +946,7 @@ __store_many(int fd, const intel_ctx_t *ctx, unsigned ring,
>
> static void
> store_many(int fd, const intel_ctx_t *ctx, unsigned int ring,
> - int num_children, int timeout)
> + int num_children, int timeout, uint32_t region)
> {
> struct intel_engine_data ied;
> unsigned long *shared;
> @@ -963,7 +964,8 @@ store_many(int fd, const intel_ctx_t *ctx, unsigned int ring,
> __store_many(fd, ctx,
> ied_flags(&ied, n),
> timeout,
> - &shared[n]);
> + &shared[n],
> + region);
> }
> igt_waitchildren();
>
> @@ -976,7 +978,7 @@ store_many(int fd, const intel_ctx_t *ctx, unsigned int ring,
> }
>
> static void
> -sync_all(int fd, const intel_ctx_t *ctx, int num_children, int timeout)
> +sync_all(int fd, const intel_ctx_t *ctx, int num_children, int timeout, uint32_t region)
> {
> struct intel_engine_data ied;
>
> @@ -985,15 +987,13 @@ sync_all(int fd, const intel_ctx_t *ctx, int num_children, int timeout)
>
> intel_detect_and_clear_missed_interrupts(fd);
> igt_fork(child, num_children) {
> - const uint32_t bbe = MI_BATCH_BUFFER_END;
> struct drm_i915_gem_exec_object2 object;
> struct drm_i915_gem_execbuffer2 execbuf;
> double start, elapsed;
> unsigned long cycles;
>
> memset(&object, 0, sizeof(object));
> - object.handle = gem_create(fd, 4096);
> - gem_write(fd, object.handle, 0, &bbe, sizeof(bbe));
> + object.handle = batch_create(fd, 4096, region);
>
> memset(&execbuf, 0, sizeof(execbuf));
> execbuf.buffers_ptr = to_user_pointer(&object);
> @@ -1023,7 +1023,7 @@ sync_all(int fd, const intel_ctx_t *ctx, int num_children, int timeout)
> }
>
> static void
> -store_all(int fd, const intel_ctx_t *ctx, int num_children, int timeout)
> +store_all(int fd, const intel_ctx_t *ctx, int num_children, int timeout, uint32_t region)
> {
> const unsigned int gen = intel_gen(intel_get_drm_devid(fd));
> struct intel_engine_data ied;
> @@ -1034,7 +1034,6 @@ store_all(int fd, const intel_ctx_t *ctx, int num_children, int timeout)
>
> intel_detect_and_clear_missed_interrupts(fd);
> igt_fork(child, num_children) {
> - const uint32_t bbe = MI_BATCH_BUFFER_END;
> struct drm_i915_gem_exec_object2 object[2];
> struct drm_i915_gem_relocation_entry reloc[1024];
> struct drm_i915_gem_execbuffer2 execbuf;
> @@ -1051,14 +1050,13 @@ store_all(int fd, const intel_ctx_t *ctx, int num_children, int timeout)
> execbuf.rsvd1 = ctx->id;
>
> memset(object, 0, sizeof(object));
> - object[0].handle = gem_create(fd, 4096);
> - gem_write(fd, object[0].handle, 0, &bbe, sizeof(bbe));
> + object[0].handle = batch_create(fd, 4096, region);
> execbuf.buffer_count = 1;
> gem_execbuf(fd, &execbuf);
>
> object[0].flags |= EXEC_OBJECT_WRITE;
> object[0].flags |= has_relocs ? 0 : EXEC_OBJECT_PINNED;
> - object[1].handle = gem_create(fd, 1024*16 + 4096);
> + object[1].handle = gem_create_in_memory_regions(fd, 1024*16 + 4096, region);
>
> object[1].relocs_ptr = to_user_pointer(reloc);
> object[1].relocation_count = has_relocs ? 1024 : 0;
Preempt wasn't touched. Is that your intention?
> @@ -1205,7 +1203,7 @@ igt_main
> const struct {
> const char *name;
> void (*func)(int fd, const intel_ctx_t *ctx, unsigned int engine,
> - int num_children, int timeout);
> + int num_children, int timeout, uint32_t memregion);
> int num_children;
> int timeout;
> } all[] = {
> @@ -1241,6 +1239,10 @@ igt_main
> const struct intel_execution_engine2 *e;
> const intel_ctx_t *ctx;
> int fd = -1;
> + struct drm_i915_query_memory_regions *query_info;
> + struct igt_collection *regions, *set;
> + char *sub_name;
> + uint32_t region;
>
> igt_fixture {
> fd = drm_open_driver(DRIVER_INTEL);
> @@ -1250,6 +1252,11 @@ igt_main
> ctx = intel_ctx_create_all_physical(fd);
>
> igt_fork_hang_detector(fd);
> + query_info = gem_get_query_memory_regions(fd);
> + igt_assert(query_info);
> + set = get_memory_region_set(query_info,
> + I915_SYSTEM_MEMORY,
> + I915_DEVICE_MEMORY);
> intel_allocator_multiprocess_start();
> }
>
> @@ -1257,41 +1264,101 @@ igt_main
> for_each_test(t, individual) {
> igt_subtest_with_dynamic_f("legacy-%s", t->name) {
> for (const struct intel_execution_ring *l = intel_execution_rings; l->name; l++) {
> - igt_dynamic_f("%s", l->name) {
> - t->func(fd, intel_ctx_0(fd), eb_ring(l),
> - t->num_children, t->timeout);
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + igt_dynamic_f("%s-%s", l->name, sub_name) {
> + t->func(fd, intel_ctx_0(fd), eb_ring(l),
> + t->num_children,
> + t->timeout,
> + region);
> + }
> + free(sub_name);
> }
> }
> }
> }
>
> - igt_subtest("basic-all")
> - sync_all(fd, ctx, 1, 2);
> - igt_subtest("basic-store-all")
> - store_all(fd, ctx, 1, 2);
> -
> - igt_subtest("all")
> - sync_all(fd, ctx, 1, 20);
> - igt_subtest("store-all")
> - store_all(fd, ctx, 1, 20);
> - igt_subtest("forked-all")
> - sync_all(fd, ctx, ncpus, 20);
> - igt_subtest("forked-store-all")
> - store_all(fd, ctx, ncpus, 20);
All below are asking for macro like in gem_exec_suspend.
--
Zbigniew
> + igt_subtest_with_dynamic("basic-all") {
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + igt_dynamic_f("%s", sub_name)
> + sync_all(fd, ctx, 1, 2, region);
> + free(sub_name);
> + }
> + }
> + igt_subtest_with_dynamic("basic-store-all") {
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + igt_dynamic_f("%s", sub_name)
> + store_all(fd, ctx, 1, 2, region);
> + free(sub_name);
> + }
> + }
> + igt_subtest_with_dynamic("all") {
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + igt_dynamic_f("%s", sub_name)
> + sync_all(fd, ctx, 1, 20, region);
> + free(sub_name);
> + }
> + }
> + igt_subtest_with_dynamic("store-all") {
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + igt_dynamic_f("%s", sub_name)
> + store_all(fd, ctx, 1, 20, region);
> + free(sub_name);
> + }
> + }
> + igt_subtest_with_dynamic("forked-all") {
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + igt_dynamic_f("%s", sub_name)
> + sync_all(fd, ctx, ncpus, 20, region);
> + free(sub_name);
> + }
> + }
> + igt_subtest_with_dynamic("forked-store-all") {
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + igt_dynamic_f("%s", sub_name)
> + store_all(fd, ctx, ncpus, 20, region);
> + free(sub_name);
> + }
> + }
>
> for_each_test(t, all) {
> - igt_subtest_f("%s", t->name)
> - t->func(fd, ctx, ALL_ENGINES, t->num_children, t->timeout);
> + igt_subtest_with_dynamic_f("%s", t->name) {
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + igt_dynamic_f("%s", sub_name)
> + t->func(fd, ctx, ALL_ENGINES, t->num_children,
> + t->timeout, region);
> + free(sub_name);
> + }
> + }
> }
>
> /* New way of selecting engines. */
> for_each_test(t, individual) {
> igt_subtest_with_dynamic_f("%s", t->name) {
> - for_each_ctx_engine(fd, ctx, e) {
> - igt_dynamic_f("%s", e->name) {
> - t->func(fd, ctx, e->flags,
> - t->num_children, t->timeout);
> - }
> + for_each_combination(regions, 1, set) {
> + sub_name = memregion_dynamic_subtest_name(regions);
> + region = igt_collection_get_value(regions, 0);
> + for_each_ctx_engine(fd, ctx, e)
> + igt_dynamic_f("%s-%s", e->name, sub_name)
> + t->func(fd, ctx, e->flags,
> + t->num_children,
> + t->timeout, region);
> + free(sub_name);
> }
> }
> }
> @@ -1314,6 +1381,8 @@ igt_main
> }
>
> igt_fixture {
> + free(query_info);
> + igt_collection_destroy(set);
> intel_allocator_multiprocess_stop();
> igt_stop_hang_detector();
> intel_ctx_destroy(fd, ctx);
> --
> 2.32.0
>
More information about the igt-dev
mailing list