[PATCH 15/21] drm/amdgpu: untie user ring ids from kernel ring ids v3
Christian König
deathsimple at vodafone.de
Tue Mar 7 10:31:17 UTC 2017
Am 07.03.2017 um 02:10 schrieb Andres Rodriguez:
> Add amdgpu_queue_mgr, a mechanism that allows disjointing usermode's
> ring ids from the kernel's ring ids.
>
> The queue manager maintains a per-file descriptor map of user ring ids
> to amdgpu_ring pointers. Once a map is created it is permanent (this is
> required to maintain FIFO execution guarantees for a context's ring).
>
> Different queue map policies can be configured for each HW IP.
> Currently all HW IPs use the identity mapper, i.e. kernel ring id is
> equal to the user ring id.
>
> The purpose of this mechanism is to distribute the load across multiple
> queues more effectively for HW IPs that support multiple rings.
> Userspace clients are unable to check whether a specific resource is in
> use by a different client. Therefore, it is up to the kernel driver to
> make the optimal choice.
>
> v2: remove amdgpu_queue_mapper_funcs
> v3: made amdgpu_queue_mgr per context instead of per-fd
>
> Signed-off-by: Andres Rodriguez <andresx7 at gmail.com>
One nit pick and three problems in the error code path below.
> ---
> drivers/gpu/drm/amd/amdgpu/Makefile | 2 +-
> drivers/gpu/drm/amd/amdgpu/amdgpu.h | 27 ++-
> drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 95 +++--------
> drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 6 +
> drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c | 230 ++++++++++++++++++++++++++
> drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c | 45 +++++
> drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 2 +
> 7 files changed, 328 insertions(+), 79 deletions(-)
> create mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
> index 2814aad..0081d0c 100644
> --- a/drivers/gpu/drm/amd/amdgpu/Makefile
> +++ b/drivers/gpu/drm/amd/amdgpu/Makefile
> @@ -17,21 +17,21 @@ amdgpu-y := amdgpu_drv.o
> amdgpu-y += amdgpu_device.o amdgpu_kms.o \
> amdgpu_atombios.o atombios_crtc.o amdgpu_connectors.o \
> atom.o amdgpu_fence.o amdgpu_ttm.o amdgpu_object.o amdgpu_gart.o \
> amdgpu_encoders.o amdgpu_display.o amdgpu_i2c.o \
> amdgpu_fb.o amdgpu_gem.o amdgpu_ring.o \
> amdgpu_cs.o amdgpu_bios.o amdgpu_benchmark.o amdgpu_test.o \
> amdgpu_pm.o atombios_dp.o amdgpu_afmt.o amdgpu_trace_points.o \
> atombios_encoders.o amdgpu_sa.o atombios_i2c.o \
> amdgpu_prime.o amdgpu_vm.o amdgpu_ib.o amdgpu_pll.o \
> amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \
> - amdgpu_gtt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o
> + amdgpu_gtt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o amdgpu_queue_mgr.o
>
> # add asic specific block
> amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \
> ci_smc.o ci_dpm.o dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o \
> amdgpu_amdkfd_gfx_v7.o
>
> amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o si_dpm.o si_smc.o
>
> amdgpu-y += \
> vi.o mxgpu_vi.o
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> index 377f58a..734d941 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> @@ -648,32 +648,56 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
> int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
> struct amdgpu_job **job);
>
> void amdgpu_job_free_resources(struct amdgpu_job *job);
> void amdgpu_job_free(struct amdgpu_job *job);
> int amdgpu_job_submit(struct amdgpu_job *job, struct amdgpu_ring *ring,
> struct amd_sched_entity *entity, void *owner,
> struct dma_fence **f);
>
> /*
> + * Queue manager
> + */
> +struct amdgpu_queue_mapper {
> + int hw_ip;
> + struct mutex lock;
> + /* protected by lock */
> + struct amdgpu_ring *queue_map[AMDGPU_MAX_RINGS];
> +};
> +
> +struct amdgpu_queue_mgr {
> + struct amdgpu_queue_mapper mapper[AMDGPU_MAX_IP_NUM];
> +};
> +
> +int amdgpu_queue_mgr_init(struct amdgpu_device *adev,
> + struct amdgpu_queue_mgr *mgr);
> +int amdgpu_queue_mgr_fini(struct amdgpu_device *adev,
> + struct amdgpu_queue_mgr *mgr);
> +int amdgpu_queue_mgr_map(struct amdgpu_device *adev,
> + struct amdgpu_queue_mgr *mgr,
> + int hw_ip, int instance, int ring,
> + struct amdgpu_ring **out_ring);
> +
> +/*
Not completely mandatory, but would be nice to amdgpu_queue_mgr.h
instead of amdgpu.h.
> * context related structures
> */
>
> struct amdgpu_ctx_ring {
> uint64_t sequence;
> struct dma_fence **fences;
> struct amd_sched_entity entity;
> };
>
> struct amdgpu_ctx {
> struct kref refcount;
> struct amdgpu_device *adev;
> + struct amdgpu_queue_mgr queue_mgr;
> unsigned reset_counter;
> spinlock_t ring_lock;
> struct dma_fence **fences;
> struct amdgpu_ctx_ring rings[AMDGPU_MAX_RINGS];
> bool preamble_presented;
> };
>
> struct amdgpu_ctx_mgr {
> struct amdgpu_device *adev;
> struct mutex lock;
> @@ -1723,23 +1747,20 @@ static inline bool amdgpu_is_mec_queue_enabled(struct amdgpu_device *adev,
> #define amdgpu_gds_switch(adev, r, v, d, w, a) (adev)->gds.funcs->patch_gds_switch((r), (v), (d), (w), (a))
>
> /* Common functions */
> int amdgpu_gpu_reset(struct amdgpu_device *adev);
> bool amdgpu_need_backup(struct amdgpu_device *adev);
> void amdgpu_pci_config_reset(struct amdgpu_device *adev);
> bool amdgpu_need_post(struct amdgpu_device *adev);
> void amdgpu_update_display_priority(struct amdgpu_device *adev);
>
> int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data);
> -int amdgpu_cs_get_ring(struct amdgpu_device *adev, u32 ip_type,
> - u32 ip_instance, u32 ring,
> - struct amdgpu_ring **out_ring);
> void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes);
> void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *abo, u32 domain);
> bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo);
> int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages);
> int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
> uint32_t flags);
> bool amdgpu_ttm_tt_has_userptr(struct ttm_tt *ttm);
> struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm);
> bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
> unsigned long end);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> index 57301f5..67de115 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> @@ -23,76 +23,20 @@
> *
> * Authors:
> * Jerome Glisse <glisse at freedesktop.org>
> */
> #include <linux/pagemap.h>
> #include <drm/drmP.h>
> #include <drm/amdgpu_drm.h>
> #include "amdgpu.h"
> #include "amdgpu_trace.h"
>
> -int amdgpu_cs_get_ring(struct amdgpu_device *adev, u32 ip_type,
> - u32 ip_instance, u32 ring,
> - struct amdgpu_ring **out_ring)
> -{
> - /* Right now all IPs have only one instance - multiple rings. */
> - if (ip_instance != 0) {
> - DRM_ERROR("invalid ip instance: %d\n", ip_instance);
> - return -EINVAL;
> - }
> -
> - switch (ip_type) {
> - default:
> - DRM_ERROR("unknown ip type: %d\n", ip_type);
> - return -EINVAL;
> - case AMDGPU_HW_IP_GFX:
> - if (ring < adev->gfx.num_gfx_rings) {
> - *out_ring = &adev->gfx.gfx_ring[ring];
> - } else {
> - DRM_ERROR("only %d gfx rings are supported now\n",
> - adev->gfx.num_gfx_rings);
> - return -EINVAL;
> - }
> - break;
> - case AMDGPU_HW_IP_COMPUTE:
> - if (ring < adev->gfx.num_compute_rings) {
> - *out_ring = &adev->gfx.compute_ring[ring];
> - } else {
> - DRM_ERROR("only %d compute rings are supported now\n",
> - adev->gfx.num_compute_rings);
> - return -EINVAL;
> - }
> - break;
> - case AMDGPU_HW_IP_DMA:
> - if (ring < adev->sdma.num_instances) {
> - *out_ring = &adev->sdma.instance[ring].ring;
> - } else {
> - DRM_ERROR("only %d SDMA rings are supported\n",
> - adev->sdma.num_instances);
> - return -EINVAL;
> - }
> - break;
> - case AMDGPU_HW_IP_UVD:
> - *out_ring = &adev->uvd.ring;
> - break;
> - case AMDGPU_HW_IP_VCE:
> - if (ring < adev->vce.num_rings){
> - *out_ring = &adev->vce.ring[ring];
> - } else {
> - DRM_ERROR("only %d VCE rings are supported\n", adev->vce.num_rings);
> - return -EINVAL;
> - }
> - break;
> - }
> - return 0;
> -}
> -
> static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
> struct drm_amdgpu_cs_chunk_fence *data,
> uint32_t *offset)
> {
> struct drm_gem_object *gobj;
> unsigned long size;
>
> gobj = drm_gem_object_lookup(p->filp, data->handle);
> if (gobj == NULL)
> return -EINVAL;
> @@ -868,23 +812,22 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
> struct drm_amdgpu_cs_chunk_ib *chunk_ib;
> struct amdgpu_ring *ring;
>
> chunk = &parser->chunks[i];
> ib = &parser->job->ibs[j];
> chunk_ib = (struct drm_amdgpu_cs_chunk_ib *)chunk->kdata;
>
> if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
> continue;
>
> - r = amdgpu_cs_get_ring(adev, chunk_ib->ip_type,
> - chunk_ib->ip_instance, chunk_ib->ring,
> - &ring);
> + r = amdgpu_queue_mgr_map(adev, &parser->ctx->queue_mgr, chunk_ib->ip_type,
> + chunk_ib->ip_instance, chunk_ib->ring, &ring);
> if (r)
> return r;
>
> if (ib->flags & AMDGPU_IB_FLAG_PREAMBLE) {
> parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT;
> if (!parser->ctx->preamble_presented) {
> parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
> parser->ctx->preamble_presented = true;
> }
> }
> @@ -972,30 +915,31 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
>
> deps = (struct drm_amdgpu_cs_chunk_dep *)chunk->kdata;
> num_deps = chunk->length_dw * 4 /
> sizeof(struct drm_amdgpu_cs_chunk_dep);
>
> for (j = 0; j < num_deps; ++j) {
> struct amdgpu_ring *ring;
> struct amdgpu_ctx *ctx;
> struct dma_fence *fence;
>
> - r = amdgpu_cs_get_ring(adev, deps[j].ip_type,
> - deps[j].ip_instance,
> - deps[j].ring, &ring);
> - if (r)
> - return r;
> -
> ctx = amdgpu_ctx_get(fpriv, deps[j].ctx_id);
> if (ctx == NULL)
> return -EINVAL;
>
> + r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr,
> + deps[j].ip_type,
> + deps[j].ip_instance,
> + deps[j].ring, &ring);
> + if (r)
> + return r;
> +
You need to call amdgpu_ctx_put() in the error path here. Otherwise we
won't unlock the ctx any more if that fails.
> fence = amdgpu_ctx_get_fence(ctx, ring,
> deps[j].handle);
> if (IS_ERR(fence)) {
> r = PTR_ERR(fence);
> amdgpu_ctx_put(ctx);
> return r;
>
> } else if (fence) {
> r = amdgpu_sync_fence(adev, &p->job->sync,
> fence);
> @@ -1107,29 +1051,30 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
> struct drm_file *filp)
> {
> union drm_amdgpu_wait_cs *wait = data;
> struct amdgpu_device *adev = dev->dev_private;
> unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout);
> struct amdgpu_ring *ring = NULL;
> struct amdgpu_ctx *ctx;
> struct dma_fence *fence;
> long r;
>
> - r = amdgpu_cs_get_ring(adev, wait->in.ip_type, wait->in.ip_instance,
> - wait->in.ring, &ring);
> - if (r)
> - return r;
> -
> ctx = amdgpu_ctx_get(filp->driver_priv, wait->in.ctx_id);
> if (ctx == NULL)
> return -EINVAL;
>
> + r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr,
> + wait->in.ip_type, wait->in.ip_instance,
> + wait->in.ring, &ring);
> + if (r)
> + return r;
> +
Dito.
> fence = amdgpu_ctx_get_fence(ctx, ring, wait->in.handle);
> if (IS_ERR(fence))
> r = PTR_ERR(fence);
> else if (fence) {
> r = dma_fence_wait_timeout(fence, true, timeout);
> dma_fence_put(fence);
> } else
> r = 1;
>
> amdgpu_ctx_put(ctx);
> @@ -1151,29 +1096,29 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
> */
> static struct dma_fence *amdgpu_cs_get_fence(struct amdgpu_device *adev,
> struct drm_file *filp,
> struct drm_amdgpu_fence *user)
> {
> struct amdgpu_ring *ring;
> struct amdgpu_ctx *ctx;
> struct dma_fence *fence;
> int r;
>
> - r = amdgpu_cs_get_ring(adev, user->ip_type, user->ip_instance,
> - user->ring, &ring);
> - if (r)
> - return ERR_PTR(r);
> -
> ctx = amdgpu_ctx_get(filp->driver_priv, user->ctx_id);
> if (ctx == NULL)
> return ERR_PTR(-EINVAL);
>
> + r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr, user->ip_type,
> + user->ip_instance, user->ring, &ring);
> + if (r)
> + return ERR_PTR(r);
> +
And once more here.
With those fixed the patch is Reviewed-by: Christian König
<christian.koenig at amd.com>.
Regards,
Christian.
> fence = amdgpu_ctx_get_fence(ctx, ring, user->seq_no);
> amdgpu_ctx_put(ctx);
>
> return fence;
> }
>
> /**
> * amdgpu_cs_wait_all_fence - wait on all fences to signal
> *
> * @adev: amdgpu device
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
> index 400c66b..3ddc8db 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
> @@ -51,20 +51,24 @@ static int amdgpu_ctx_init(struct amdgpu_device *adev, struct amdgpu_ctx *ctx)
> struct amdgpu_ring *ring = adev->rings[i];
> struct amd_sched_rq *rq;
>
> rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_NORMAL];
> r = amd_sched_entity_init(&ring->sched, &ctx->rings[i].entity,
> rq, amdgpu_sched_jobs);
> if (r)
> goto failed;
> }
>
> + r = amdgpu_queue_mgr_init(adev, &ctx->queue_mgr);
> + if (r)
> + goto failed;
> +
> return 0;
>
> failed:
> for (j = 0; j < i; j++)
> amd_sched_entity_fini(&adev->rings[j]->sched,
> &ctx->rings[j].entity);
> kfree(ctx->fences);
> ctx->fences = NULL;
> return r;
> }
> @@ -79,20 +83,22 @@ static void amdgpu_ctx_fini(struct amdgpu_ctx *ctx)
>
> for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
> for (j = 0; j < amdgpu_sched_jobs; ++j)
> dma_fence_put(ctx->rings[i].fences[j]);
> kfree(ctx->fences);
> ctx->fences = NULL;
>
> for (i = 0; i < adev->num_rings; i++)
> amd_sched_entity_fini(&adev->rings[i]->sched,
> &ctx->rings[i].entity);
> +
> + amdgpu_queue_mgr_fini(adev, &ctx->queue_mgr);
> }
>
> static int amdgpu_ctx_alloc(struct amdgpu_device *adev,
> struct amdgpu_fpriv *fpriv,
> uint32_t *id)
> {
> struct amdgpu_ctx_mgr *mgr = &fpriv->ctx_mgr;
> struct amdgpu_ctx *ctx;
> int r;
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c
> new file mode 100644
> index 0000000..3e9ac80
> --- /dev/null
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_queue_mgr.c
> @@ -0,0 +1,230 @@
> +/*
> + * Copyright 2017 Valve 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 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 COPYRIGHT HOLDER(S) OR AUTHOR(S) 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.
> + *
> + * Authors: Andres Rodriguez
> + */
> +
> +#include "amdgpu.h"
> +#include "amdgpu_ring.h"
> +
> +static int amdgpu_queue_mapper_init(struct amdgpu_queue_mapper *mapper,
> + int hw_ip)
> +{
> + if (!mapper)
> + return -EINVAL;
> +
> + if (hw_ip > AMDGPU_MAX_IP_NUM)
> + return -EINVAL;
> +
> + mapper->hw_ip = hw_ip;
> + mutex_init(&mapper->lock);
> +
> + memset(mapper->queue_map, 0, sizeof(mapper->queue_map));
> +
> + return 0;
> +}
> +
> +static struct amdgpu_ring *amdgpu_get_cached_map(struct amdgpu_queue_mapper *mapper,
> + int ring)
> +{
> + return mapper->queue_map[ring];
> +}
> +
> +static int amdgpu_update_cached_map(struct amdgpu_queue_mapper *mapper,
> + int ring, struct amdgpu_ring *pring)
> +{
> + if (WARN_ON(mapper->queue_map[ring])) {
> + DRM_ERROR("Un-expected ring re-map\n");
> + return -EINVAL;
> + }
> +
> + mapper->queue_map[ring] = pring;
> +
> + return 0;
> +}
> +
> +static int amdgpu_identity_map(struct amdgpu_device *adev,
> + struct amdgpu_queue_mapper *mapper,
> + int ring,
> + struct amdgpu_ring **out_ring)
> +{
> + switch (mapper->hw_ip) {
> + case AMDGPU_HW_IP_GFX:
> + *out_ring = &adev->gfx.gfx_ring[ring];
> + break;
> + case AMDGPU_HW_IP_COMPUTE:
> + *out_ring = &adev->gfx.compute_ring[ring];
> + break;
> + case AMDGPU_HW_IP_DMA:
> + *out_ring = &adev->sdma.instance[ring].ring;
> + break;
> + case AMDGPU_HW_IP_UVD:
> + *out_ring = &adev->uvd.ring;
> + break;
> + case AMDGPU_HW_IP_VCE:
> + *out_ring = &adev->vce.ring[ring];
> + break;
> + default:
> + *out_ring = NULL;
> + DRM_ERROR("unknown HW IP type: %d\n", mapper->hw_ip);
> + return -EINVAL;
> + }
> +
> + return amdgpu_update_cached_map(mapper, ring, *out_ring);
> +}
> +
> +/**
> + * amdgpu_queue_mgr_init - init an amdgpu_queue_mgr struct
> + *
> + * @adev: amdgpu_device pointer
> + * @mgr: amdgpu_queue_mgr structure holding queue information
> + *
> + * Initialize the the selected @mgr (all asics).
> + *
> + * Returns 0 on success, error on failure.
> + */
> +int amdgpu_queue_mgr_init(struct amdgpu_device *adev,
> + struct amdgpu_queue_mgr *mgr)
> +{
> + int i, r;
> +
> + if (!adev || !mgr)
> + return -EINVAL;
> +
> + memset(mgr, 0, sizeof(*mgr));
> +
> + for (i = 0; i < AMDGPU_MAX_IP_NUM; ++i) {
> + r = amdgpu_queue_mapper_init(&mgr->mapper[i], i);
> + if (r)
> + return r;
> + }
> +
> + return 0;
> +}
> +
> +/**
> + * amdgpu_queue_mgr_fini - de-initialize an amdgpu_queue_mgr struct
> + *
> + * @adev: amdgpu_device pointer
> + * @mgr: amdgpu_queue_mgr structure holding queue information
> + *
> + * De-initialize the the selected @mgr (all asics).
> + *
> + * Returns 0 on success, error on failure.
> + */
> +int amdgpu_queue_mgr_fini(struct amdgpu_device *adev,
> + struct amdgpu_queue_mgr *mgr)
> +{
> + return 0;
> +}
> +
> +/**
> + * amdgpu_queue_mgr_map - Map a userspace ring id to an amdgpu_ring
> + *
> + * @adev: amdgpu_device pointer
> + * @mgr: amdgpu_queue_mgr structure holding queue information
> + * @hw_ip: HW IP enum
> + * @instance: HW instance
> + * @ring: user ring id
> + * @our_ring: pointer to mapped amdgpu_ring
> + *
> + * Map a userspace ring id to an appropriate kernel ring. Different
> + * policies are configurable at a HW IP level.
> + *
> + * Returns 0 on success, error on failure.
> + */
> +int amdgpu_queue_mgr_map(struct amdgpu_device *adev,
> + struct amdgpu_queue_mgr *mgr,
> + int hw_ip, int instance, int ring,
> + struct amdgpu_ring **out_ring)
> +{
> + int r, ip_num_rings;
> + struct amdgpu_queue_mapper *mapper = &mgr->mapper[hw_ip];
> +
> + if (!adev || !mgr || !out_ring)
> + return -EINVAL;
> +
> + if (hw_ip >= AMDGPU_MAX_IP_NUM)
> + return -EINVAL;
> +
> + if (ring >= AMDGPU_MAX_RINGS)
> + return -EINVAL;
> +
> + /* Right now all IPs have only one instance - multiple rings. */
> + if (instance != 0) {
> + DRM_ERROR("invalid ip instance: %d\n", instance);
> + return -EINVAL;
> + }
> +
> + switch (hw_ip) {
> + case AMDGPU_HW_IP_GFX:
> + ip_num_rings = adev->gfx.num_gfx_rings;
> + break;
> + case AMDGPU_HW_IP_COMPUTE:
> + ip_num_rings = adev->gfx.num_compute_rings;
> + break;
> + case AMDGPU_HW_IP_DMA:
> + ip_num_rings = adev->sdma.num_instances;
> + break;
> + case AMDGPU_HW_IP_UVD:
> + ip_num_rings = 1;
> + break;
> + case AMDGPU_HW_IP_VCE:
> + ip_num_rings = adev->vce.num_rings;
> + break;
> + default:
> + DRM_ERROR("unknown ip type: %d\n", hw_ip);
> + return -EINVAL;
> + }
> +
> + if (ring >= ip_num_rings) {
> + DRM_ERROR("Ring index:%d exceeds maximum:%d for ip:%d\n",
> + ring, ip_num_rings, hw_ip);
> + return -EINVAL;
> + }
> +
> + mutex_lock(&mapper->lock);
> +
> + *out_ring = amdgpu_get_cached_map(mapper, ring);
> + if (*out_ring) {
> + /* cache hit */
> + r = 0;
> + goto out_unlock;
> + }
> +
> + switch (mapper->hw_ip) {
> + case AMDGPU_HW_IP_GFX:
> + case AMDGPU_HW_IP_COMPUTE:
> + case AMDGPU_HW_IP_DMA:
> + case AMDGPU_HW_IP_UVD:
> + case AMDGPU_HW_IP_VCE:
> + r = amdgpu_identity_map(adev, mapper, ring, out_ring);
> + break;
> + default:
> + *out_ring = NULL;
> + r = -EINVAL;
> + DRM_ERROR("unknown HW IP type: %d\n", mapper->hw_ip);
> + }
> +
> +out_unlock:
> + mutex_unlock(&mapper->lock);
> + return r;
> +}
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> index 7c842b7..43cd539 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> @@ -45,20 +45,65 @@
> * pointers are equal, the ring is idle. When the host
> * writes commands to the ring buffer, it increments the
> * wptr. The GPU then starts fetching commands and executes
> * them until the pointers are equal again.
> */
> static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
> struct amdgpu_ring *ring);
> static void amdgpu_debugfs_ring_fini(struct amdgpu_ring *ring);
>
> /**
> + * amdgpu_ring_is_valid_index - check if a ring idex is valid for a HW IP
> + *
> + * @adev: amdgpu_device pointer
> + * @ip_type: The HW IP to check against
> + * @ring: the ring index
> + *
> + * Check if @ring is a valid index for @ip_type (all asics).
> + * Returns 0 on success, error on failure.
> + */
> +int amdgpu_ring_is_valid_index(struct amdgpu_device *adev,
> + int ip_type, int ring)
> +{
> + int ip_num_rings;
> +
> + switch (ip_type) {
> + case AMDGPU_HW_IP_GFX:
> + ip_num_rings = adev->gfx.num_gfx_rings;
> + break;
> + case AMDGPU_HW_IP_COMPUTE:
> + ip_num_rings = adev->gfx.num_compute_rings;
> + break;
> + case AMDGPU_HW_IP_DMA:
> + ip_num_rings = adev->sdma.num_instances;
> + break;
> + case AMDGPU_HW_IP_UVD:
> + ip_num_rings = 1;
> + break;
> + case AMDGPU_HW_IP_VCE:
> + ip_num_rings = adev->vce.num_rings;
> + break;
> + default:
> + DRM_ERROR("unknown ip type: %d\n", ip_type);
> + return -EINVAL;
> + }
> +
> + if (ring >= ip_num_rings) {
> + DRM_ERROR("Ring index:%d exceeds maximum:%d for ip:%d\n",
> + ring, ip_num_rings, ip_type);
> + return -EINVAL;
> + }
> +
> + return 0;
> +}
> +
> +/**
> * amdgpu_ring_alloc - allocate space on the ring buffer
> *
> * @adev: amdgpu_device pointer
> * @ring: amdgpu_ring structure holding ring information
> * @ndw: number of dwords to allocate in the ring buffer
> *
> * Allocate @ndw dwords in the ring buffer (all asics).
> * Returns 0 on success, error on failure.
> */
> int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw)
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> index 2345b398..35da5c5 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> @@ -168,20 +168,22 @@ struct amdgpu_ring {
> uint64_t current_ctx;
> char name[16];
> unsigned cond_exe_offs;
> u64 cond_exe_gpu_addr;
> volatile u32 *cond_exe_cpu_addr;
> #if defined(CONFIG_DEBUG_FS)
> struct dentry *ent;
> #endif
> };
>
> +int amdgpu_ring_is_valid_index(struct amdgpu_device *adev,
> + int hw_ip, int ring);
> int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw);
> void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
> void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
> void amdgpu_ring_commit(struct amdgpu_ring *ring);
> void amdgpu_ring_undo(struct amdgpu_ring *ring);
> int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
> unsigned ring_size, struct amdgpu_irq_src *irq_src,
> unsigned irq_type);
> void amdgpu_ring_fini(struct amdgpu_ring *ring);
>
More information about the amd-gfx
mailing list