[igt-dev] [PATCH v12 5/7] lib/i915: add gem_engine_topology library
Tvrtko Ursulin
tvrtko.ursulin at linux.intel.com
Tue Mar 19 10:18:49 UTC 2019
On 18/03/2019 23:28, Andi Shyti wrote:
> The gem_engine_topology library is a set of functions that
> interface with the query and getparam/setparam ioctls.
>
> The library's access point is the 'intel_init_engine_list()'
> function that, everytime is called, generates the list of active
> engines and returns them in a 'struct intel_engine_data'. The
> structure contains only the engines that are actively present in
> the GPU.
>
> The function can work in both the cases that the query and
> getparam ioctls are implemented or not by the running kernel. In
> case they are implemented, a query is made to the driver to fetch
> the list of active engines. In case they are not implemented, the
> list is taken from the 'intel_execution_engines2' array and
> stored only after checking their presence.
>
> The library provides three helper functions:
>
> 1. gem_topology_has_engine(): checks if the engine is present by
> querying the driver in a way that the caller function wouldn't
> know if the above ioctls are implemented.
>
> 2. gem_set_eb_flags(): sets the execution buffer flag depending
> whether it refers to context-engine mapping or not.
>
> 3. gem_context_execbuf(): is a wrapper that executes the buffer
> in a way that the calling function doesn't know about the
> ioctls in the driver.
>
> Signed-off-by: Andi Shyti <andi.shyti at intel.com>
> ---
> lib/Makefile.sources | 2 +
> lib/i915/gem_engine_topology.c | 210 +++++++++++++++++++++++++++++++++
> lib/i915/gem_engine_topology.h | 52 ++++++++
> lib/meson.build | 1 +
> 4 files changed, 265 insertions(+)
> create mode 100644 lib/i915/gem_engine_topology.c
> create mode 100644 lib/i915/gem_engine_topology.h
>
> diff --git a/lib/Makefile.sources b/lib/Makefile.sources
> index cf2720981707..757bd7a17ebe 100644
> --- a/lib/Makefile.sources
> +++ b/lib/Makefile.sources
> @@ -13,6 +13,8 @@ lib_source_list = \
> i915/gem_ring.c \
> i915/gem_mman.c \
> i915/gem_mman.h \
> + i915/gem_engine_topology.c \
> + i915/gem_engine_topology.h \
> i915_3d.h \
> i915_reg.h \
> i915_pciids.h \
> diff --git a/lib/i915/gem_engine_topology.c b/lib/i915/gem_engine_topology.c
> new file mode 100644
> index 000000000000..e836b3745e04
> --- /dev/null
> +++ b/lib/i915/gem_engine_topology.c
> @@ -0,0 +1,210 @@
> +/*
> + * Copyright © 2019 Intel Corporation
> + *
> + * Permission is hereby granted, free of charge, to any person obtaining a
> + * copy of this software and associated documentation files (the "Software"),
> + * to deal in the Software without restriction, including without limitation
> + * the rights to use, copy, modify, merge, publish, distribute, sublicense,
> + * and/or sell copies of the Software, and to permit persons to whom the
> + * Software is furnished to do so, subject to the following conditions:
> + *
> + * The above copyright notice and this permission notice (including the next
> + * paragraph) shall be included in all copies or substantial portions of the
> + * Software.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
> + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
> + * IN THE SOFTWARE.
> + */
> +
> +#include "drmtest.h"
> +#include "ioctl_wrappers.h"
> +
> +#include "i915/gem_engine_topology.h"
> +
> +#define SIZEOF_CTX_PARAM offsetof(struct i915_context_param_engines, \
> + class_instance[I915_EXEC_RING_MASK + 1])
> +#define SIZEOF_QUERY offsetof(struct drm_i915_query_engine_info, \
> + engines[I915_EXEC_RING_MASK + 1])
> +
> +static int __gem_query(int fd, struct drm_i915_query *q)
> +{
> + int err = 0;
> +
> + if (igt_ioctl(fd, DRM_IOCTL_I915_QUERY, q))
> + err = -errno;
> +
> + errno = 0;
> + return err;
> +}
> +
> +static void gem_query(int fd, struct drm_i915_query *q)
> +{
> + igt_assert_eq(__gem_query(fd, q), 0);
> +}
> +
> +static void query_engines(int fd,
> + struct drm_i915_query_engine_info *query_engines)
> +{
> + struct drm_i915_query_item item = { };
> + struct drm_i915_query query = { };
> +
> + item.query_id = DRM_I915_QUERY_ENGINE_INFO;
> + query.items_ptr = to_user_pointer(&item);
> + query.num_items = 1;
> + item.length = SIZEOF_QUERY;
> +
> + item.data_ptr = to_user_pointer(query_engines);
> +
> + gem_query(fd, &query);
> +}
> +
> +static void set_ctx_param_engines(struct intel_engine_data *ed)
> +{
> + struct i915_context_param_engines *ctx_engine;
> + struct drm_i915_gem_context_param ctx_param;
> + uint8_t buff[SIZEOF_CTX_PARAM] = { };
> + int i;
> +
> + ctx_engine = (struct i915_context_param_engines *) buff;
Tiny detail - you could move assignment to declaration of you reorder them.
> +
> + ctx_engine->extensions = 0;
No need since cleared when declared.
> + for (i = 0; i < ed->nengines; i++) {
> + ctx_engine->class_instance[i].engine_class =
> + ed->engines[i].class;
> + ctx_engine->class_instance[i].engine_instance =
> + ed->engines[i].instance;
Could make the assignments more readable by having a pointer p =
&ctx_engine->class_instance[0] and p++ in the for loop.
> + }
> +
> + ctx_param.size = offsetof(typeof(*ctx_engine), class_instance[i + 1]);
> + ctx_param.value = to_user_pointer(ctx_engine);
> + ctx_param.ctx_id = ed->ctx;
> + ctx_param.param = I915_CONTEXT_PARAM_ENGINES;
> +
> + gem_context_set_param(ed->fd, &ctx_param);
> +}
> +
> +static void init_engine_list(struct intel_engine_data *ed)
> +{
> + const char *class_names[] = { "rcs", "bcs", "vcs", "vecs" };
> + struct drm_i915_query_engine_info *query_engine;
> + unsigned char query_buffer[SIZEOF_QUERY] = { };
You used uint8_t above. I don't mind which one as long as it is consistent.
> + int i;
> +
> + query_engine = (struct drm_i915_query_engine_info *) query_buffer;
Could also assign in the declaration.
> + query_engines(ed->fd, query_engine);
> +
> + for (i = 0; i < query_engine->num_engines; i++) {
> + char *name;
> + __u16 class = query_engine->engines[i].engine_class;
> + __u16 instance = query_engine->engines[i].engine_instance;
> +
> + ed->engines[i].class = class;
> + ed->engines[i].instance = instance;
> +
> + /* if we don't recognise the class, then we mark it as "unk" */
> + if (class >= ARRAY_SIZE(class_names))
> + igt_assert(asprintf(&name, "unk-%d:%d",
> + class, instance) > 0);
> + else
> + igt_assert(asprintf(&name, "%s%d",
> + class_names[class], instance) > 0);
Names leak per iterator I think. Bummer, apart from using a big static
table of all possible names, and borrowing those pointers here, I don't
have any better ideas at the moment. Might actually be okay.
> +
> + ed->engines[i].name = name;
> + }
> +
> + ed->nengines = query_engine->num_engines;
> +}
> +
> +struct intel_engine_data intel_init_engine_list(int fd, uint32_t ctx_id)
> +{
> + struct drm_i915_gem_context_param ctx_param = {
> + .param = I915_CONTEXT_PARAM_ENGINES,
> + .ctx_id = ctx_id,
> + };
> + struct intel_engine_data engine_data = {
> + .n = 0,
No need.
> + .fd = fd,
> + .ctx = ctx_id,
> + };
> +
> + if (__gem_context_get_param(fd, &ctx_param)) {
> + const struct intel_execution_engine2 *e2;
> +
> + igt_debug("using pre-allocated engine list\n");
> +
> + __for_each_engine_class_instance(e2) {
> + if (!gem_has_engine(fd, e2->class, e2->instance))
> + continue;
> +
> + engine_data.engines[engine_data.nengines].name =
> + e2->name;
> + engine_data.engines[engine_data.nengines].instance =
> + e2->instance;
> + engine_data.engines[engine_data.nengines].class =
> + e2->class;
Also could keep a pointer to engine_data.engines[<current>] for more
readable assignments.
> + engine_data.nengines++;
> + }
> +
> + return engine_data;
> + }
> +
> + init_engine_list(&engine_data);
As far as I can see you missed the use case where we want to iterate
over engines already defined in the context engine map.
So I think if the above __gem_context_get_param succeeds with non-zero
size returned, you need to build engine_data based on those engines.
> +
> + if (ctx_param.size == sizeof(struct i915_context_param_engines))
> + set_ctx_param_engines(&engine_data);
> +
So this would change to something like:
if (ret == 0 && ctx_param.size == 0)
init_engine_list(&engine_data);
set_map = true;
else if (ret == 0)
init_engine_list_from_ctx_engine_map(&engine_data, ctx_param);
else
init_engine_list_from_static_tables(..);
set_map = true;
if (set_map)
set_ctx_param(...);
> + return engine_data;
> +}
> +
> +bool gem_topology_has_engine(int fd, struct intel_execution_engine2 *e,
> + uint8_t i, uint32_t ctx)
> +{
Why do we need this helper?
> + struct drm_i915_gem_context_param ctx_param = {
> + .param = I915_CONTEXT_PARAM_ENGINES,
> + };
> +
> + if (!e)
> + return false;
> +
> + if (__gem_context_get_param(fd, &ctx_param))
> + return gem_has_engine(fd, e->class, e->instance);
> +
> + return gem_context_has_engine(fd, i, ctx);
> +}
> +
> +void gem_set_eb_flags(int fd, struct drm_i915_gem_execbuffer2 *eb,
> + struct intel_execution_engine2 e2,
> + uint8_t engine, uint32_t ctx)
> +{
> + struct drm_i915_gem_context_param ctx_param = {
> + .param = I915_CONTEXT_PARAM_ENGINES,
> + };
> +
> + if (__gem_context_get_param(fd, &ctx_param)) {
> + eb->flags |= (I915_EXEC_RING_MASK & engine);
> + eb->rsvd1 = ctx;
> + } else {
> + eb->flags |= gem_class_instance_to_eb_flags(fd, e2.class,
> + e2.instance);
> + }
> +}
Store flags in struct intel_execution_engine2 while building the list
and then just eb->flags |= e2->eb_flags in tests?
> +
> +int __gem_context_execbuf(int fd, struct drm_i915_gem_execbuffer2 *eb,
> + struct intel_execution_engine2 e2,
> + uint8_t engine, uint32_t ctx)
> +{
> + gem_set_eb_flags(fd, eb, e2, engine, ctx);
> + return __gem_execbuf(fd, eb);
> +}
> +
> +void gem_context_execbuf(int fd, struct drm_i915_gem_execbuffer2 *eb,
> + struct intel_execution_engine2 e2,
> + uint8_t index_map, uint32_t ctx)
> +{
> + igt_assert_eq(__gem_context_execbuf(fd, eb, e2, index_map, ctx), 0);
> +}
Probably not want these two either.
> diff --git a/lib/i915/gem_engine_topology.h b/lib/i915/gem_engine_topology.h
> new file mode 100644
> index 000000000000..657b9397a181
> --- /dev/null
> +++ b/lib/i915/gem_engine_topology.h
> @@ -0,0 +1,52 @@
> +/*
> + * Copyright © 2019 Intel Corporation
> + *
> + * Permission is hereby granted, free of charge, to any person obtaining a
> + * copy of this software and associated documentation files (the "Software"),
> + * to deal in the Software without restriction, including without limitation
> + * the rights to use, copy, modify, merge, publish, distribute, sublicense,
> + * and/or sell copies of the Software, and to permit persons to whom the
> + * Software is furnished to do so, subject to the following conditions:
> + *
> + * The above copyright notice and this permission notice (including the next
> + * paragraph) shall be included in all copies or substantial portions of the
> + * Software.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
> + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
> + * IN THE SOFTWARE.
> + */
> +
> +#ifndef GEM_ENGINE_TOPOLOGY_H
> +#define GEM_ENGINE_TOPOLOGY_H
> +
> +#include "igt_gt.h"
> +
> +struct intel_engine_data {
> + int fd;
> + uint32_t ctx;
> +
> + uint32_t nengines;
> + uint32_t n;
This is the current engine index? Could it come handy to also have a
pointer to current engine in the iterator? Will see in later patches..
> + struct intel_execution_engine2 engines[I915_EXEC_RING_MASK + 1];
> +};
> +
> +struct intel_engine_data intel_init_engine_list(int fd, uint32_t ctx_id);
> +bool gem_topology_has_engine(int fd, struct intel_execution_engine2 *e,
> + uint8_t i, uint32_t ctx);
> +
> +void gem_set_eb_flags(int fd, struct drm_i915_gem_execbuffer2 *eb,
> + struct intel_execution_engine2 e2,
> + uint8_t engine, uint32_t ctx);
> +int __gem_context_execbuf(int fd, struct drm_i915_gem_execbuffer2 *eb,
> + struct intel_execution_engine2 e2,
> + uint8_t engine, uint32_t ctx);
> +void gem_context_execbuf(int fd, struct drm_i915_gem_execbuffer2 *eb,
> + struct intel_execution_engine2 e2,
> + uint8_t index_map, uint32_t ctx);
> +
> +#endif /* GEM_ENGINE_TOPOLOGY_H */
> diff --git a/lib/meson.build b/lib/meson.build
> index 0eb5585d72b9..3cc52f97c8bf 100644
> --- a/lib/meson.build
> +++ b/lib/meson.build
> @@ -5,6 +5,7 @@ lib_sources = [
> 'i915/gem_submission.c',
> 'i915/gem_ring.c',
> 'i915/gem_mman.c',
> + 'i915/gem_engine_topology.c',
> 'igt_color_encoding.c',
> 'igt_debugfs.c',
> 'igt_device.c',
>
Regards,
Tvrtko
More information about the igt-dev
mailing list