From a11024457d348672b26b3d4581ed19c793399b48 Mon Sep 17 00:00:00 2001 From: Felix Kuehling Date: Tue, 6 Feb 2018 20:32:46 -0500 Subject: uapi: Fix type used in ioctl parameter structures Use __u32 and __u64 instead of POSIX types that may not be defined in user mode builds. Signed-off-by: Felix Kuehling Reviewed-by: Oded Gabbay Signed-off-by: Oded Gabbay --- include/uapi/linux/kfd_ioctl.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'include/uapi') diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h index f4cab5b3ba9a..111d73ba2d96 100644 --- a/include/uapi/linux/kfd_ioctl.h +++ b/include/uapi/linux/kfd_ioctl.h @@ -263,10 +263,10 @@ struct kfd_ioctl_get_tile_config_args { }; struct kfd_ioctl_set_trap_handler_args { - uint64_t tba_addr; /* to KFD */ - uint64_t tma_addr; /* to KFD */ - uint32_t gpu_id; /* to KFD */ - uint32_t pad; + __u64 tba_addr; /* to KFD */ + __u64 tma_addr; /* to KFD */ + __u32 gpu_id; /* to KFD */ + __u32 pad; }; #define AMDKFD_IOCTL_BASE 'K' -- cgit v1.3-14-g43fede From 60bbade2a63de363a7acdb5e3b0f166a883fe3be Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Wed, 17 Jan 2018 13:18:47 +0800 Subject: drm/amdgpu: Expose more GPU sensor queries Add sub-queries for stable pstate shader/memory clock. Reviewed-by: Alex Deucher Signed-off-by: Rex Zhu Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 18 ++++++++++++++++++ include/uapi/drm/amdgpu_drm.h | 4 ++++ 2 files changed, 22 insertions(+) (limited to 'include/uapi') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index 9f189ab07cdd..b929986dd3d7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -758,6 +758,24 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file return -EINVAL; } break; + case AMDGPU_INFO_SENSOR_STABLE_PSTATE_GFX_SCLK: + /* get stable pstate sclk in Mhz */ + if (amdgpu_dpm_read_sensor(adev, + AMDGPU_PP_SENSOR_STABLE_PSTATE_SCLK, + (void *)&ui32, &ui32_size)) { + return -EINVAL; + } + ui32 /= 100; + break; + case AMDGPU_INFO_SENSOR_STABLE_PSTATE_GFX_MCLK: + /* get stable pstate mclk in Mhz */ + if (amdgpu_dpm_read_sensor(adev, + AMDGPU_PP_SENSOR_STABLE_PSTATE_MCLK, + (void *)&ui32, &ui32_size)) { + return -EINVAL; + } + ui32 /= 100; + break; default: DRM_DEBUG_KMS("Invalid request %d\n", info->sensor_info.type); diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h index 4d21191aaed0..1816bd8200d1 100644 --- a/include/uapi/drm/amdgpu_drm.h +++ b/include/uapi/drm/amdgpu_drm.h @@ -664,6 +664,10 @@ struct drm_amdgpu_cs_chunk_data { #define AMDGPU_INFO_SENSOR_VDDNB 0x6 /* Subquery id: Query graphics voltage */ #define AMDGPU_INFO_SENSOR_VDDGFX 0x7 + /* Subquery id: Query GPU stable pstate shader clock */ + #define AMDGPU_INFO_SENSOR_STABLE_PSTATE_GFX_SCLK 0x8 + /* Subquery id: Query GPU stable pstate memory clock */ + #define AMDGPU_INFO_SENSOR_STABLE_PSTATE_GFX_MCLK 0x9 /* Number of VRAM page faults on CPU access. */ #define AMDGPU_INFO_NUM_VRAM_CPU_PAGE_FAULTS 0x1E #define AMDGPU_INFO_VRAM_LOST_COUNTER 0x1F -- cgit v1.3-14-g43fede From 6a8bd08d0465b2b8d214007c58598e2c15312296 Mon Sep 17 00:00:00 2001 From: Rob Clark Date: Wed, 13 Dec 2017 15:12:57 -0500 Subject: drm/msm: add sudo flag to submit ioctl This flags cause cmdstream to be executed from the ringbuffer (RB) instead of IB1. Normally not something you'd ever want to do, but it is super useful for firmware debugging. Hidden behind CAP_SYS_RAWIO and a default=n kconfig option which depends on EXPERT (and has a suitably scary warning), to prevent it from being used on accident. Signed-off-by: Rob Clark --- drivers/gpu/drm/msm/Kconfig | 13 +++++++ drivers/gpu/drm/msm/adreno/a5xx_gpu.c | 65 +++++++++++++++++++++++++++++++++++ drivers/gpu/drm/msm/msm_gem.h | 1 + drivers/gpu/drm/msm/msm_gem_submit.c | 9 +++++ include/uapi/drm/msm_drm.h | 2 ++ 5 files changed, 90 insertions(+) (limited to 'include/uapi') diff --git a/drivers/gpu/drm/msm/Kconfig b/drivers/gpu/drm/msm/Kconfig index 99d39b2aefa6..3065cb290aa8 100644 --- a/drivers/gpu/drm/msm/Kconfig +++ b/drivers/gpu/drm/msm/Kconfig @@ -28,6 +28,19 @@ config DRM_MSM_REGISTER_LOGGING that can be parsed by envytools demsm tool. If enabled, register logging can be switched on via msm.reglog=y module param. +config DRM_MSM_GPU_SUDO + bool "Enable SUDO flag on submits" + depends on DRM_MSM && EXPERT + default n + help + Enable userspace that has CAP_SYS_RAWIO to submit GPU commands + that are run from RB instead of IB1. This essentially gives + userspace kernel level access, but is useful for firmware + debugging. + + Only use this if you are a driver developer. This should *not* + be enabled for production kernels. If unsure, say N. + config DRM_MSM_HDMI_HDCP bool "Enable HDMI HDCP support in MSM DRM driver" depends on DRM_MSM && QCOM_SCM diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c index 579c28c8c994..fa08b4897a56 100644 --- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c @@ -140,6 +140,65 @@ static void a5xx_flush(struct msm_gpu *gpu, struct msm_ringbuffer *ring) gpu_write(gpu, REG_A5XX_CP_RB_WPTR, wptr); } +static void a5xx_submit_in_rb(struct msm_gpu *gpu, struct msm_gem_submit *submit, + struct msm_file_private *ctx) +{ + struct msm_drm_private *priv = gpu->dev->dev_private; + struct msm_ringbuffer *ring = submit->ring; + struct msm_gem_object *obj; + uint32_t *ptr, dwords; + unsigned int i; + + for (i = 0; i < submit->nr_cmds; i++) { + switch (submit->cmd[i].type) { + case MSM_SUBMIT_CMD_IB_TARGET_BUF: + break; + case MSM_SUBMIT_CMD_CTX_RESTORE_BUF: + if (priv->lastctx == ctx) + break; + case MSM_SUBMIT_CMD_BUF: + /* copy commands into RB: */ + obj = submit->bos[submit->cmd[i].idx].obj; + dwords = submit->cmd[i].size; + + ptr = msm_gem_get_vaddr(&obj->base); + + /* _get_vaddr() shouldn't fail at this point, + * since we've already mapped it once in + * submit_reloc() + */ + if (WARN_ON(!ptr)) + return; + + for (i = 0; i < dwords; i++) { + /* normally the OUT_PKTn() would wait + * for space for the packet. But since + * we just OUT_RING() the whole thing, + * need to call adreno_wait_ring() + * ourself: + */ + adreno_wait_ring(ring, 1); + OUT_RING(ring, ptr[i]); + } + + msm_gem_put_vaddr(&obj->base); + + break; + } + } + + a5xx_flush(gpu, ring); + a5xx_preempt_trigger(gpu); + + /* we might not necessarily have a cmd from userspace to + * trigger an event to know that submit has completed, so + * do this manually: + */ + a5xx_idle(gpu, ring); + ring->memptrs->fence = submit->seqno; + msm_gpu_retire(gpu); +} + static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit, struct msm_file_private *ctx) { @@ -149,6 +208,12 @@ static void a5xx_submit(struct msm_gpu *gpu, struct msm_gem_submit *submit, struct msm_ringbuffer *ring = submit->ring; unsigned int i, ibs = 0; + if (IS_ENABLED(CONFIG_DRM_MSM_GPU_SUDO) && submit->in_rb) { + priv->lastctx = NULL; + a5xx_submit_in_rb(gpu, submit, ctx); + return; + } + OUT_PKT7(ring, CP_PREEMPT_ENABLE_GLOBAL, 1); OUT_RING(ring, 0x02); diff --git a/drivers/gpu/drm/msm/msm_gem.h b/drivers/gpu/drm/msm/msm_gem.h index 9320e184b48d..c5d9bd3e47a8 100644 --- a/drivers/gpu/drm/msm/msm_gem.h +++ b/drivers/gpu/drm/msm/msm_gem.h @@ -146,6 +146,7 @@ struct msm_gem_submit { struct msm_gpu_submitqueue *queue; struct pid *pid; /* submitting process */ bool valid; /* true if no cmdstream patching needed */ + bool in_rb; /* "sudo" mode, copy cmds into RB */ struct msm_ringbuffer *ring; unsigned int nr_cmds; unsigned int nr_bos; diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c index b8dc8f96caf2..7bd83e0afa97 100644 --- a/drivers/gpu/drm/msm/msm_gem_submit.c +++ b/drivers/gpu/drm/msm/msm_gem_submit.c @@ -430,6 +430,12 @@ int msm_ioctl_gem_submit(struct drm_device *dev, void *data, if (MSM_PIPE_FLAGS(args->flags) & ~MSM_SUBMIT_FLAGS) return -EINVAL; + if (args->flags & MSM_SUBMIT_SUDO) { + if (!IS_ENABLED(CONFIG_DRM_MSM_GPU_SUDO) || + !capable(CAP_SYS_RAWIO)) + return -EINVAL; + } + queue = msm_submitqueue_get(ctx, args->queueid); if (!queue) return -ENOENT; @@ -471,6 +477,9 @@ int msm_ioctl_gem_submit(struct drm_device *dev, void *data, goto out_unlock; } + if (args->flags & MSM_SUBMIT_SUDO) + submit->in_rb = true; + ret = submit_lookup_objects(submit, args, file); if (ret) goto out; diff --git a/include/uapi/drm/msm_drm.h b/include/uapi/drm/msm_drm.h index bbbaffad772d..c06d0a5bdd80 100644 --- a/include/uapi/drm/msm_drm.h +++ b/include/uapi/drm/msm_drm.h @@ -201,10 +201,12 @@ struct drm_msm_gem_submit_bo { #define MSM_SUBMIT_NO_IMPLICIT 0x80000000 /* disable implicit sync */ #define MSM_SUBMIT_FENCE_FD_IN 0x40000000 /* enable input fence_fd */ #define MSM_SUBMIT_FENCE_FD_OUT 0x20000000 /* enable output fence_fd */ +#define MSM_SUBMIT_SUDO 0x10000000 /* run submitted cmds from RB */ #define MSM_SUBMIT_FLAGS ( \ MSM_SUBMIT_NO_IMPLICIT | \ MSM_SUBMIT_FENCE_FD_IN | \ MSM_SUBMIT_FENCE_FD_OUT | \ + MSM_SUBMIT_SUDO | \ 0) /* Each cmdstream submit consists of a table of buffers involved, and -- cgit v1.3-14-g43fede From a446ae2c6e6555048301f2339cfd97b8eed6d0b7 Mon Sep 17 00:00:00 2001 From: Lionel Landwerlin Date: Tue, 6 Mar 2018 12:28:56 +0000 Subject: drm/i915: add query uAPI There are a number of information that are readable from hardware registers and that we would like to make accessible to userspace. One particular example is the topology of the execution units (how are execution units grouped in subslices and slices and also which ones have been fused off for die recovery). At the moment the GET_PARAM ioctl covers some basic needs, but generally is only able to return a single value for each defined parameter. This is a bit problematic with topology descriptions which are array/maps of available units. This change introduces a new ioctl that can deal with requests to fill structures of potentially variable lengths. The user is expected fill a query with length fields set at 0 on the first call, the kernel then sets the length fields to the their expected values. A second call to the kernel with length fields at their expected values will trigger a copy of the data to the pointed memory locations. The scope of this uAPI is only to provide information to userspace, not to allow configuration of the device. v2: Simplify dispatcher code iteration (Tvrtko) Tweak uapi drm_i915_query_item structure (Tvrtko) v3: Rename pad fields into flags (Chris) Return error on flags field != 0 (Chris) Only copy length back to userspace in drm_i915_query_item (Chris) v4: Use array of functions instead of switch (Chris) v5: More comments in uapi (Tvrtko) Return query item errors in length field (All) v6: Tweak uapi comments style to match the coding style (Lionel) v7: Add i915_query.h (Joonas) v8: (Lionel) Change the behavior of the item iterator to report invalid queries into the query item rather than stopping the iteration. This enables userspace applications to query newer items on older kernels and only have failure on the items that are not supported. v9: Edit copyright headers (Joonas) v10: Typos & comments in uapi (Joonas) Signed-off-by: Lionel Landwerlin Reviewed-by: Tvrtko Ursulin Acked-by: Chris Wilson Reviewed-by: Joonas Lahtinen Link: https://patchwork.freedesktop.org/patch/msgid/20180306122857.27317-6-lionel.g.landwerlin@intel.com --- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 2 ++ drivers/gpu/drm/i915/i915_query.c | 50 +++++++++++++++++++++++++++++++++++++++ drivers/gpu/drm/i915/i915_query.h | 15 ++++++++++++ include/uapi/drm/i915_drm.h | 46 ++++++++++++++++++++++++++++++++--- 5 files changed, 111 insertions(+), 3 deletions(-) create mode 100644 drivers/gpu/drm/i915/i915_query.c create mode 100644 drivers/gpu/drm/i915/i915_query.h (limited to 'include/uapi') diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile index 1bd9bc5b8c5c..4eee91a3a236 100644 --- a/drivers/gpu/drm/i915/Makefile +++ b/drivers/gpu/drm/i915/Makefile @@ -69,6 +69,7 @@ i915-y += i915_cmd_parser.o \ i915_gem_timeline.o \ i915_gem_userptr.o \ i915_gemfs.o \ + i915_query.o \ i915_request.o \ i915_trace_points.o \ i915_vma.o \ diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index c594ff5e57d0..d7c4de45644d 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -49,6 +49,7 @@ #include "i915_drv.h" #include "i915_trace.h" #include "i915_pmu.h" +#include "i915_query.h" #include "i915_vgpu.h" #include "intel_drv.h" #include "intel_uc.h" @@ -2832,6 +2833,7 @@ static const struct drm_ioctl_desc i915_ioctls[] = { DRM_IOCTL_DEF_DRV(I915_PERF_OPEN, i915_perf_open_ioctl, DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(I915_PERF_ADD_CONFIG, i915_perf_add_config_ioctl, DRM_UNLOCKED|DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(I915_PERF_REMOVE_CONFIG, i915_perf_remove_config_ioctl, DRM_UNLOCKED|DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(I915_QUERY, i915_query_ioctl, DRM_UNLOCKED|DRM_RENDER_ALLOW), }; static struct drm_driver driver = { diff --git a/drivers/gpu/drm/i915/i915_query.c b/drivers/gpu/drm/i915/i915_query.c new file mode 100644 index 000000000000..5582e6c3234a --- /dev/null +++ b/drivers/gpu/drm/i915/i915_query.c @@ -0,0 +1,50 @@ +/* + * SPDX-License-Identifier: MIT + * + * Copyright © 2018 Intel Corporation + */ + +#include "i915_drv.h" +#include "i915_query.h" +#include + +static int (* const i915_query_funcs[])(struct drm_i915_private *dev_priv, + struct drm_i915_query_item *query_item) = { +}; + +int i915_query_ioctl(struct drm_device *dev, void *data, struct drm_file *file) +{ + struct drm_i915_private *dev_priv = to_i915(dev); + struct drm_i915_query *args = data; + struct drm_i915_query_item __user *user_item_ptr = + u64_to_user_ptr(args->items_ptr); + u32 i; + + if (args->flags != 0) + return -EINVAL; + + for (i = 0; i < args->num_items; i++, user_item_ptr++) { + struct drm_i915_query_item item; + u64 func_idx; + int ret; + + if (copy_from_user(&item, user_item_ptr, sizeof(item))) + return -EFAULT; + + if (item.query_id == 0) + return -EINVAL; + + func_idx = item.query_id - 1; + + if (func_idx < ARRAY_SIZE(i915_query_funcs)) + ret = i915_query_funcs[func_idx](dev_priv, &item); + else + ret = -EINVAL; + + /* Only write the length back to userspace if they differ. */ + if (ret != item.length && put_user(ret, &user_item_ptr->length)) + return -EFAULT; + } + + return 0; +} diff --git a/drivers/gpu/drm/i915/i915_query.h b/drivers/gpu/drm/i915/i915_query.h new file mode 100644 index 000000000000..31dcef181f63 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_query.h @@ -0,0 +1,15 @@ +/* + * SPDX-License-Identifier: MIT + * + * Copyright © 2018 Intel Corporation + */ + +#ifndef _I915_QUERY_H_ +#define _I915_QUERY_H_ + +struct drm_device; +struct drm_file; + +int i915_query_ioctl(struct drm_device *dev, void *data, struct drm_file *file); + +#endif diff --git a/include/uapi/drm/i915_drm.h b/include/uapi/drm/i915_drm.h index 29fa48e4755d..eedd5a23a944 100644 --- a/include/uapi/drm/i915_drm.h +++ b/include/uapi/drm/i915_drm.h @@ -318,6 +318,7 @@ typedef struct _drm_i915_sarea { #define DRM_I915_PERF_OPEN 0x36 #define DRM_I915_PERF_ADD_CONFIG 0x37 #define DRM_I915_PERF_REMOVE_CONFIG 0x38 +#define DRM_I915_QUERY 0x39 #define DRM_IOCTL_I915_INIT DRM_IOW( DRM_COMMAND_BASE + DRM_I915_INIT, drm_i915_init_t) #define DRM_IOCTL_I915_FLUSH DRM_IO ( DRM_COMMAND_BASE + DRM_I915_FLUSH) @@ -375,6 +376,7 @@ typedef struct _drm_i915_sarea { #define DRM_IOCTL_I915_PERF_OPEN DRM_IOW(DRM_COMMAND_BASE + DRM_I915_PERF_OPEN, struct drm_i915_perf_open_param) #define DRM_IOCTL_I915_PERF_ADD_CONFIG DRM_IOW(DRM_COMMAND_BASE + DRM_I915_PERF_ADD_CONFIG, struct drm_i915_perf_oa_config) #define DRM_IOCTL_I915_PERF_REMOVE_CONFIG DRM_IOW(DRM_COMMAND_BASE + DRM_I915_PERF_REMOVE_CONFIG, __u64) +#define DRM_IOCTL_I915_QUERY DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_QUERY, struct drm_i915_query) /* Allow drivers to submit batchbuffers directly to hardware, relying * on the security mechanisms provided by hardware. @@ -1606,15 +1608,53 @@ struct drm_i915_perf_oa_config { __u32 n_flex_regs; /* - * These fields are pointers to tuples of u32 values (register - * address, value). For example the expected length of the buffer - * pointed by mux_regs_ptr is (2 * sizeof(u32) * n_mux_regs). + * These fields are pointers to tuples of u32 values (register address, + * value). For example the expected length of the buffer pointed by + * mux_regs_ptr is (2 * sizeof(u32) * n_mux_regs). */ __u64 mux_regs_ptr; __u64 boolean_regs_ptr; __u64 flex_regs_ptr; }; +struct drm_i915_query_item { + __u64 query_id; + + /* + * When set to zero by userspace, this is filled with the size of the + * data to be written at the data_ptr pointer. The kernel sets this + * value to a negative value to signal an error on a particular query + * item. + */ + __s32 length; + + /* + * Unused for now. Must be cleared to zero. + */ + __u32 flags; + + /* + * Data will be written at the location pointed by data_ptr when the + * value of length matches the length of the data to be written by the + * kernel. + */ + __u64 data_ptr; +}; + +struct drm_i915_query { + __u32 num_items; + + /* + * Unused for now. Must be cleared to zero. + */ + __u32 flags; + + /* + * This points to an array of num_items drm_i915_query_item structures. + */ + __u64 items_ptr; +}; + #if defined(__cplusplus) } #endif -- cgit v1.3-14-g43fede From c822e059185585f79b2007b1d2cafacf4264e610 Mon Sep 17 00:00:00 2001 From: Lionel Landwerlin Date: Tue, 6 Mar 2018 12:28:57 +0000 Subject: drm/i915: expose rcs topology through query uAPI With the introduction of asymmetric slices in CNL, we cannot rely on the previous SUBSLICE_MASK getparam to tell userspace what subslices are available. Here we introduce a more detailed way of querying the Gen's GPU topology that doesn't aggregate numbers. This is essential for monitoring parts of the GPU with the OA unit, because counters need to be normalized to the number of EUs/subslices/slices. The current aggregated numbers like EU_TOTAL do not gives us sufficient information. The Mesa series making use of this API is : https://patchwork.freedesktop.org/series/38795/ As a bonus we can draw representations of the GPU : https://imgur.com/a/vuqpa v2: Rename uapi struct s/_mask/_info/ (Tvrtko) Report max_slice/subslice/eus_per_subslice rather than strides (Tvrtko) Add uapi macros to read data from *_info structs (Tvrtko) v3: Use !!(v & DRM_I915_BIT()) for uapi macros instead of custom shifts (Tvrtko) v4: factorize query item writting (Tvrtko) tweak uapi struct/define names (Tvrtko) v5: Replace ALIGN() macro (Chris) v6: Updated uapi comments (Tvrtko) Moved flags != 0 checks into vfuncs (Tvrtko) v7: Use access_ok() before copying anything, to avoid overflows (Chris) Switch BUG_ON() to GEM_WARN_ON() (Tvrtko) v8: Tweak uapi comments style to match the coding style (Lionel) v9: Fix error in comment about computation of enabled subslice (Tvrtko) v10: Fix/update comments in uAPI (Sagar) v11: Drop drm_i915_query_(slice|subslice|eu)_info in favor of a single drm_i915_query_topology_info (Joonas) v12: Add subslice_stride/eu_stride in drm_i915_query_topology_info (Joonas) v13: Fix comment in uAPI (Joonas) Signed-off-by: Lionel Landwerlin Acked-by: Chris Wilson Reviewed-by: Joonas Lahtinen Link: https://patchwork.freedesktop.org/patch/msgid/20180306122857.27317-7-lionel.g.landwerlin@intel.com --- drivers/gpu/drm/i915/i915_query.c | 75 +++++++++++++++++++++++++++++++++++++++ include/uapi/drm/i915_drm.h | 62 ++++++++++++++++++++++++++++++++ 2 files changed, 137 insertions(+) (limited to 'include/uapi') diff --git a/drivers/gpu/drm/i915/i915_query.c b/drivers/gpu/drm/i915/i915_query.c index 5582e6c3234a..3ace929dd90f 100644 --- a/drivers/gpu/drm/i915/i915_query.c +++ b/drivers/gpu/drm/i915/i915_query.c @@ -8,8 +8,83 @@ #include "i915_query.h" #include +static int query_topology_info(struct drm_i915_private *dev_priv, + struct drm_i915_query_item *query_item) +{ + const struct sseu_dev_info *sseu = &INTEL_INFO(dev_priv)->sseu; + struct drm_i915_query_topology_info topo; + u32 slice_length, subslice_length, eu_length, total_length; + + if (query_item->flags != 0) + return -EINVAL; + + if (sseu->max_slices == 0) + return -ENODEV; + + BUILD_BUG_ON(sizeof(u8) != sizeof(sseu->slice_mask)); + + slice_length = sizeof(sseu->slice_mask); + subslice_length = sseu->max_slices * + DIV_ROUND_UP(sseu->max_subslices, + sizeof(sseu->subslice_mask[0]) * BITS_PER_BYTE); + eu_length = sseu->max_slices * sseu->max_subslices * + DIV_ROUND_UP(sseu->max_eus_per_subslice, BITS_PER_BYTE); + + total_length = sizeof(topo) + slice_length + subslice_length + eu_length; + + if (query_item->length == 0) + return total_length; + + if (query_item->length < total_length) + return -EINVAL; + + if (copy_from_user(&topo, u64_to_user_ptr(query_item->data_ptr), + sizeof(topo))) + return -EFAULT; + + if (topo.flags != 0) + return -EINVAL; + + if (!access_ok(VERIFY_WRITE, u64_to_user_ptr(query_item->data_ptr), + total_length)) + return -EFAULT; + + memset(&topo, 0, sizeof(topo)); + topo.max_slices = sseu->max_slices; + topo.max_subslices = sseu->max_subslices; + topo.max_eus_per_subslice = sseu->max_eus_per_subslice; + + topo.subslice_offset = slice_length; + topo.subslice_stride = DIV_ROUND_UP(sseu->max_subslices, BITS_PER_BYTE); + topo.eu_offset = slice_length + subslice_length; + topo.eu_stride = + DIV_ROUND_UP(sseu->max_eus_per_subslice, BITS_PER_BYTE); + + if (__copy_to_user(u64_to_user_ptr(query_item->data_ptr), + &topo, sizeof(topo))) + return -EFAULT; + + if (__copy_to_user(u64_to_user_ptr(query_item->data_ptr + sizeof(topo)), + &sseu->slice_mask, slice_length)) + return -EFAULT; + + if (__copy_to_user(u64_to_user_ptr(query_item->data_ptr + + sizeof(topo) + slice_length), + sseu->subslice_mask, subslice_length)) + return -EFAULT; + + if (__copy_to_user(u64_to_user_ptr(query_item->data_ptr + + sizeof(topo) + + slice_length + subslice_length), + sseu->eu_mask, eu_length)) + return -EFAULT; + + return total_length; +} + static int (* const i915_query_funcs[])(struct drm_i915_private *dev_priv, struct drm_i915_query_item *query_item) = { + query_topology_info, }; int i915_query_ioctl(struct drm_device *dev, void *data, struct drm_file *file) diff --git a/include/uapi/drm/i915_drm.h b/include/uapi/drm/i915_drm.h index eedd5a23a944..7f5634ce8e88 100644 --- a/include/uapi/drm/i915_drm.h +++ b/include/uapi/drm/i915_drm.h @@ -1619,6 +1619,7 @@ struct drm_i915_perf_oa_config { struct drm_i915_query_item { __u64 query_id; +#define DRM_I915_QUERY_TOPOLOGY_INFO 1 /* * When set to zero by userspace, this is filled with the size of the @@ -1655,6 +1656,67 @@ struct drm_i915_query { __u64 items_ptr; }; +/* + * Data written by the kernel with query DRM_I915_QUERY_TOPOLOGY_INFO : + * + * data: contains the 3 pieces of information : + * + * - the slice mask with one bit per slice telling whether a slice is + * available. The availability of slice X can be queried with the following + * formula : + * + * (data[X / 8] >> (X % 8)) & 1 + * + * - the subslice mask for each slice with one bit per subslice telling + * whether a subslice is available. The availability of subslice Y in slice + * X can be queried with the following formula : + * + * (data[subslice_offset + + * X * subslice_stride + + * Y / 8] >> (Y % 8)) & 1 + * + * - the EU mask for each subslice in each slice with one bit per EU telling + * whether an EU is available. The availability of EU Z in subslice Y in + * slice X can be queried with the following formula : + * + * (data[eu_offset + + * (X * max_subslices + Y) * eu_stride + + * Z / 8] >> (Z % 8)) & 1 + */ +struct drm_i915_query_topology_info { + /* + * Unused for now. Must be cleared to zero. + */ + __u16 flags; + + __u16 max_slices; + __u16 max_subslices; + __u16 max_eus_per_subslice; + + /* + * Offset in data[] at which the subslice masks are stored. + */ + __u16 subslice_offset; + + /* + * Stride at which each of the subslice masks for each slice are + * stored. + */ + __u16 subslice_stride; + + /* + * Offset in data[] at which the EU masks are stored. + */ + __u16 eu_offset; + + /* + * Stride at which each of the EU masks for each subslice are stored. + */ + __u16 eu_stride; + + __u8 data[]; +}; + #if defined(__cplusplus) } #endif -- cgit v1.3-14-g43fede From 1e09b05386f32efbebb798cf0341eca4b424c960 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Thu, 8 Mar 2018 18:01:24 +0800 Subject: drm/amdgpu: query vram type from atombios The vram type for dGPU is stored in umc_info while sys mem type for APU is stored in integratedsysteminfo Signed-off-by: Hawking Zhang Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c | 95 +++++++++++++++++++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h | 1 + drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 4 +- include/uapi/drm/amdgpu_drm.h | 1 + 4 files changed, 94 insertions(+), 7 deletions(-) (limited to 'include/uapi') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index ff8efd0f8fd5..a0f48cb9b8f0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c @@ -114,6 +114,9 @@ union igp_info { struct atom_integrated_system_info_v1_11 v11; }; +union umc_info { + struct atom_umc_info_v3_1 v31; +}; /* * Return vram width from integrated system info table, if available, * or 0 if not. @@ -143,6 +146,94 @@ int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev) return 0; } +static int convert_atom_mem_type_to_vram_type (struct amdgpu_device *adev, + int atom_mem_type) +{ + int vram_type; + + if (adev->flags & AMD_IS_APU) { + switch (atom_mem_type) { + case Ddr2MemType: + case LpDdr2MemType: + vram_type = AMDGPU_VRAM_TYPE_DDR2; + break; + case Ddr3MemType: + case LpDdr3MemType: + vram_type = AMDGPU_VRAM_TYPE_DDR3; + break; + case Ddr4MemType: + case LpDdr4MemType: + vram_type = AMDGPU_VRAM_TYPE_DDR4; + break; + default: + vram_type = AMDGPU_VRAM_TYPE_UNKNOWN; + break; + } + } else { + switch (atom_mem_type) { + case ATOM_DGPU_VRAM_TYPE_GDDR5: + vram_type = AMDGPU_VRAM_TYPE_GDDR5; + break; + case ATOM_DGPU_VRAM_TYPE_HBM: + vram_type = AMDGPU_VRAM_TYPE_HBM; + break; + default: + vram_type = AMDGPU_VRAM_TYPE_UNKNOWN; + break; + } + } + + return vram_type; +} +/* + * Return vram type from either integrated system info table + * or umc info table, if available, or 0 (TYPE_UNKNOWN) if not + */ +int amdgpu_atomfirmware_get_vram_type(struct amdgpu_device *adev) +{ + struct amdgpu_mode_info *mode_info = &adev->mode_info; + int index; + u16 data_offset, size; + union igp_info *igp_info; + union umc_info *umc_info; + u8 frev, crev; + u8 mem_type; + + if (adev->flags & AMD_IS_APU) + index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, + integratedsysteminfo); + else + index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, + umc_info); + if (amdgpu_atom_parse_data_header(mode_info->atom_context, + index, &size, + &frev, &crev, &data_offset)) { + if (adev->flags & AMD_IS_APU) { + igp_info = (union igp_info *) + (mode_info->atom_context->bios + data_offset); + switch (crev) { + case 11: + mem_type = igp_info->v11.memorytype; + return convert_atom_mem_type_to_vram_type(adev, mem_type); + default: + return 0; + } + } else { + umc_info = (union umc_info *) + (mode_info->atom_context->bios + data_offset); + switch (crev) { + case 1: + mem_type = umc_info->v31.vram_type; + return convert_atom_mem_type_to_vram_type(adev, mem_type); + default: + return 0; + } + } + } + + return 0; +} + union firmware_info { struct atom_firmware_info_v3_1 v31; }; @@ -151,10 +242,6 @@ union smu_info { struct atom_smu_info_v3_1 v31; }; -union umc_info { - struct atom_umc_info_v3_1 v31; -}; - int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev) { struct amdgpu_mode_info *mode_info = &adev->mode_info; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h index 288b97e54347..7689c961c4ef 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h @@ -28,6 +28,7 @@ bool amdgpu_atomfirmware_gpu_supports_virtualization(struct amdgpu_device *adev) void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev); int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev); int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev); +int amdgpu_atomfirmware_get_vram_type(struct amdgpu_device *adev); int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 67cd1fe17649..ceab14f16795 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -836,9 +836,9 @@ static int gmc_v9_0_sw_init(void *handle) spin_lock_init(&adev->gmc.invalidate_lock); + adev->gmc.vram_type = amdgpu_atomfirmware_get_vram_type(adev); switch (adev->asic_type) { case CHIP_RAVEN: - adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN; if (adev->rev_id == 0x0 || adev->rev_id == 0x1) { amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48); } else { @@ -849,8 +849,6 @@ static int gmc_v9_0_sw_init(void *handle) } break; case CHIP_VEGA10: - /* XXX Don't know how to get VRAM type yet. */ - adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM; /* * To fulfill 4-level page support, * vm size is 256TB (48bit), maximum size of Vega10, diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h index 1816bd8200d1..528f6d041e90 100644 --- a/include/uapi/drm/amdgpu_drm.h +++ b/include/uapi/drm/amdgpu_drm.h @@ -806,6 +806,7 @@ struct drm_amdgpu_info_firmware { #define AMDGPU_VRAM_TYPE_GDDR5 5 #define AMDGPU_VRAM_TYPE_HBM 6 #define AMDGPU_VRAM_TYPE_DDR3 7 +#define AMDGPU_VRAM_TYPE_DDR4 8 struct drm_amdgpu_info_device { /** PCI Device ID */ -- cgit v1.3-14-g43fede