Re: [PATCH v6 3/3] drm/doc/rfc: VM_BIND uapi definition

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 30/06/2022 06:11, Jason Ekstrand wrote:
On Sat, Jun 25, 2022 at 8:49 PM Niranjana Vishwanathapura <niranjana.vishwanathapura@xxxxxxxxx <mailto:niranjana.vishwanathapura@xxxxxxxxx>> wrote:

    VM_BIND and related uapi definitions

    v2: Reduce the scope to simple Mesa use case.
    v3: Expand VM_UNBIND documentation and add
         I915_GEM_VM_BIND/UNBIND_FENCE_VALID
         and I915_GEM_VM_BIND_TLB_FLUSH flags.
    v4: Remove I915_GEM_VM_BIND_TLB_FLUSH flag and add additional
         documentation for vm_bind/unbind.
    v5: Remove TLB flush requirement on VM_UNBIND.
         Add version support to stage implementation.
    v6: Define and use drm_i915_gem_timeline_fence structure for
         all timeline fences.
    v7: Rename I915_PARAM_HAS_VM_BIND to I915_PARAM_VM_BIND_VERSION.
         Update documentation on async vm_bind/unbind and versioning.
         Remove redundant vm_bind/unbind FENCE_VALID flag, execbuf3
         batch_count field and I915_EXEC3_SECURE flag.

    Signed-off-by: Niranjana Vishwanathapura
    <niranjana.vishwanathapura@xxxxxxxxx
    <mailto:niranjana.vishwanathapura@xxxxxxxxx>>
    Reviewed-by: Daniel Vetter <daniel.vetter@xxxxxxxx
    <mailto:daniel.vetter@xxxxxxxx>>
    ---
      Documentation/gpu/rfc/i915_vm_bind.h | 280 +++++++++++++++++++++++++++
      1 file changed, 280 insertions(+)
      create mode 100644 Documentation/gpu/rfc/i915_vm_bind.h

    diff --git a/Documentation/gpu/rfc/i915_vm_bind.h
    b/Documentation/gpu/rfc/i915_vm_bind.h
    new file mode 100644
    index 000000000000..a93e08bceee6
    --- /dev/null
    +++ b/Documentation/gpu/rfc/i915_vm_bind.h
    @@ -0,0 +1,280 @@
    +/* SPDX-License-Identifier: MIT */
    +/*
    + * Copyright © 2022 Intel Corporation
    + */
    +
    +/**
    + * DOC: I915_PARAM_VM_BIND_VERSION
    + *
    + * VM_BIND feature version supported.
    + * See typedef drm_i915_getparam_t param.
    + *
    + * Specifies the VM_BIND feature version supported.
    + * The following versions of VM_BIND have been defined:
    + *
    + * 0: No VM_BIND support.
    + *
    + * 1: In VM_UNBIND calls, the UMD must specify the exact mappings
    created
    + *    previously with VM_BIND, the ioctl will not support unbinding
    multiple
    + *    mappings or splitting them. Similarly, VM_BIND calls will not
    replace
    + *    any existing mappings.
    + *
    + * 2: The restrictions on unbinding partial or multiple mappings is
    + *    lifted, Similarly, binding will replace any mappings in the
    given range.
    + *
    + * See struct drm_i915_gem_vm_bind and struct drm_i915_gem_vm_unbind.
    + */
    +#define I915_PARAM_VM_BIND_VERSION     57
    +
    +/**
    + * DOC: I915_VM_CREATE_FLAGS_USE_VM_BIND
    + *
    + * Flag to opt-in for VM_BIND mode of binding during VM creation.
    + * See struct drm_i915_gem_vm_control flags.
    + *
    + * The older execbuf2 ioctl will not support VM_BIND mode of operation.
    + * For VM_BIND mode, we have new execbuf3 ioctl which will not
    accept any
    + * execlist (See struct drm_i915_gem_execbuffer3 for more details).
    + */
    +#define I915_VM_CREATE_FLAGS_USE_VM_BIND       (1 << 0)
    +
    +/* VM_BIND related ioctls */
    +#define DRM_I915_GEM_VM_BIND           0x3d
    +#define DRM_I915_GEM_VM_UNBIND         0x3e
    +#define DRM_I915_GEM_EXECBUFFER3       0x3f
    +
+#define DRM_IOCTL_I915_GEM_VM_BIND  DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GEM_VM_BIND, struct
    drm_i915_gem_vm_bind)
+#define DRM_IOCTL_I915_GEM_VM_UNBIND  DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GEM_VM_UNBIND, struct
    drm_i915_gem_vm_bind)
+#define DRM_IOCTL_I915_GEM_EXECBUFFER3  DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GEM_EXECBUFFER3, struct
    drm_i915_gem_execbuffer3)
    +
    +/**
    + * struct drm_i915_gem_timeline_fence - An input or output timeline
    fence.
    + *
    + * The operation will wait for input fence to signal.
    + *
    + * The returned output fence will be signaled after the completion
    of the
    + * operation.
    + */
    +struct drm_i915_gem_timeline_fence {
    +       /** @handle: User's handle for a drm_syncobj to wait on or
    signal. */
    +       __u32 handle;
    +
    +       /**
    +        * @flags: Supported flags are:
    +        *
    +        * I915_TIMELINE_FENCE_WAIT:
    +        * Wait for the input fence before the operation.
    +        *
    +        * I915_TIMELINE_FENCE_SIGNAL:
    +        * Return operation completion fence as output.
    +        */
    +       __u32 flags;
    +#define I915_TIMELINE_FENCE_WAIT            (1 << 0)
    +#define I915_TIMELINE_FENCE_SIGNAL          (1 << 1)
    +#define __I915_TIMELINE_FENCE_UNKNOWN_FLAGS
    (-(I915_TIMELINE_FENCE_SIGNAL << 1))
    +
    +       /**
    +        * @value: A point in the timeline.
    +        * Value must be 0 for a binary drm_syncobj. A Value of 0 for a
    +        * timeline drm_syncobj is invalid as it turns a drm_syncobj
    into a
    +        * binary one.
    +        */
    +       __u64 value;
    +};
    +
    +/**
    + * struct drm_i915_gem_vm_bind - VA to object mapping to bind.
    + *
    + * This structure is passed to VM_BIND ioctl and specifies the
    mapping of GPU
    + * virtual address (VA) range to the section of an object that
    should be bound
    + * in the device page table of the specified address space (VM).
    + * The VA range specified must be unique (ie., not currently bound)
    and can
    + * be mapped to whole object or a section of the object (partial
    binding).
    + * Multiple VA mappings can be created to the same section of the
    object
    + * (aliasing).
    + *
    + * The @start, @offset and @length must be 4K page aligned. However
    the DG2
    + * and XEHPSDV has 64K page size for device local-memory and has
    compact page
    + * table. On those platforms, for binding device local-memory
    objects, the
    + * @start must be 2M aligned, @offset and @length must be 64K aligned.


This is not acceptable.  We need 64K granularity.  This includes the starting address, the BO offset, and the length.  Why?  The tl;dr is that it's a requirement for about 50% of D3D12 apps if we want them to run on Linux via D3D12.  A longer explanation follows.  I don't necessarily expect kernel folks to get all the details but hopefully I'll have left enough of a map that some of the Intel Mesa folks can help fill in details.

Many modern D3D12 apps have a hard requirement on Tier2 tiled resources.  This is a feature that Intel has supported in the D3D12 driver since Skylake.  In order to implement this feature, VKD3D requires the various sparseResidencyImage* and sparseResidency*Sampled Vulkan features.  If we want those apps to work (there's getting to be quite a few of them), we need to implement the Vulkan sparse residency features.
|
|
What is sparse residency?  I'm glad you asked!  The sparse residency features allow a client to separately bind each miplevel or array slice of an image to a chunk of device memory independently, without affecting any other areas of the image.  Once you get to a high enough miplevel that everything fits inside a single sparse image block (that's a technical Vulkan term you can search for in the spec), you can enter a "miptail" which contains all the remaining miplevels in a single sparse image block.

The term "sparse image block" is what the Vulkan spec uses.  On Intel hardware and in the docs, it's what we call a "tile".  Specifically, the image needs to use Yf or Ys tiling on SKL-TGL or a Tile64 on DG2+.  This is because Tile4 and legacy X and Y-tiling don't provide any guarantees about page alignment for slices.  Yf, Ys, and Tile64, on the other hand, align all slices of the image to a tile boundary, allowing us to map memory to different slices independently, assuming we have 64K (or 4K for Yf) VM_BIND granularity.  (4K isn't actually a requirement for SKL-TGL; we can use Ys all the time which has 64K tiles but there's no reason to not support 4K alignments on integrated.)

Someone may be tempted to ask, "Can't we wiggle the strides around or something to make it work?"  I thought about that and no, you can't. The problem here is LOD2+.  Sure, you can have a stride such that the image is a multiple of 2M worth of tiles across.  That'll work fine for LOD0 and LOD1; both will be 2M aligned.  However, LOD2 won't be and there's no way to control that.  The hardware will place it to the right of LOD1 by ROUND_UP(width, tile_width) pixels and there's nothing you can do about that.  If that position doesn't happen to hit a 2M boundary, you're out of luck.

I hope that explanation provides enough detail.  Sadly, this is one of those things which has a lot of moving pieces all over different bits of the hardware and various APIs and they all have to work together just right for it to all come out in the end.  But, yeah, we really need 64K aligned binding if we want VKD3D to work.

Just to confirm, the new model would be to enforce 64K GTT alignment for lmem pages, and then for smem pages we would only require 4K alignment, but with the added restriction that userspace will never try to mix the two (lmem vs smem) within the same 2M va range (page-table). The kernel will verify this and throw an error if needed. This model should work with the above?


--Jason

    + * Also, for such mappings, i915 will reserve the whole 2M range
    for it so as
    + * to not allow multiple mappings in that 2M range (Compact page
    tables do not
    + * allow 64K page and 4K page bindings in the same 2M range).
    + *
    + * Error code -EINVAL will be returned if @start, @offset and
    @length are not
    + * properly aligned. In version 1 (See I915_PARAM_VM_BIND_VERSION),
    error code
    + * -ENOSPC will be returned if the VA range specified can't be
    reserved.
    + *
    + * VM_BIND/UNBIND ioctl calls executed on different CPU threads
    concurrently
    + * are not ordered. Furthermore, parts of the VM_BIND operation can
    be done
    + * asynchronously, if valid @fence is specified.
    + */
    +struct drm_i915_gem_vm_bind {
    +       /** @vm_id: VM (address space) id to bind */
    +       __u32 vm_id;
    +
    +       /** @handle: Object handle */
    +       __u32 handle;
    +
    +       /** @start: Virtual Address start to bind */
    +       __u64 start;
    +
    +       /** @offset: Offset in object to bind */
    +       __u64 offset;
    +
    +       /** @length: Length of mapping to bind */
    +       __u64 length;
    +
    +       /**
    +        * @flags: Supported flags are:
    +        *
    +        * I915_GEM_VM_BIND_READONLY:
    +        * Mapping is read-only.
    +        *
    +        * I915_GEM_VM_BIND_CAPTURE:
    +        * Capture this mapping in the dump upon GPU error.
    +        */
    +       __u64 flags;
    +#define I915_GEM_VM_BIND_READONLY      (1 << 1)
    +#define I915_GEM_VM_BIND_CAPTURE       (1 << 2)
    +
    +       /**
    +        * @fence: Timeline fence for bind completion signaling.
    +        *
    +        * It is an out fence, hence using I915_TIMELINE_FENCE_WAIT flag
    +        * is invalid, and an error will be returned.
    +        */
    +       struct drm_i915_gem_timeline_fence fence;
    +
    +       /**
    +        * @extensions: Zero-terminated chain of extensions.
    +        *
    +        * For future extensions. See struct i915_user_extension.
    +        */
    +       __u64 extensions;
    +};
    +
    +/**
    + * struct drm_i915_gem_vm_unbind - VA to object mapping to unbind.
    + *
    + * This structure is passed to VM_UNBIND ioctl and specifies the
    GPU virtual
    + * address (VA) range that should be unbound from the device page
    table of the
    + * specified address space (VM). VM_UNBIND will force unbind the
    specified
    + * range from device page table without waiting for any GPU job to
    complete.
    + * It is UMDs responsibility to ensure the mapping is no longer in
    use before
    + * calling VM_UNBIND.
    + *
    + * If the specified mapping is not found, the ioctl will simply
    return without
    + * any error.
    + *
    + * VM_BIND/UNBIND ioctl calls executed on different CPU threads
    concurrently
    + * are not ordered. Furthermore, parts of the VM_UNBIND operation
    can be done
    + * asynchronously, if valid @fence is specified.
    + */
    +struct drm_i915_gem_vm_unbind {
    +       /** @vm_id: VM (address space) id to bind */
    +       __u32 vm_id;
    +
    +       /** @rsvd: Reserved, MBZ */
    +       __u32 rsvd;
    +
    +       /** @start: Virtual Address start to unbind */
    +       __u64 start;
    +
    +       /** @length: Length of mapping to unbind */
    +       __u64 length;
    +
    +       /** @flags: Currently reserved, MBZ */
    +       __u64 flags;
    +
    +       /**
    +        * @fence: Timeline fence for unbind completion signaling.
    +        *
    +        * It is an out fence, hence using I915_TIMELINE_FENCE_WAIT flag
    +        * is invalid, and an error will be returned.
    +        */
    +       struct drm_i915_gem_timeline_fence fence;
    +
    +       /**
    +        * @extensions: Zero-terminated chain of extensions.
    +        *
    +        * For future extensions. See struct i915_user_extension.
    +        */
    +       __u64 extensions;
    +};
    +
    +/**
    + * struct drm_i915_gem_execbuffer3 - Structure for
    DRM_I915_GEM_EXECBUFFER3
    + * ioctl.
    + *
    + * DRM_I915_GEM_EXECBUFFER3 ioctl only works in VM_BIND mode and
    VM_BIND mode
    + * only works with this ioctl for submission.
    + * See I915_VM_CREATE_FLAGS_USE_VM_BIND.
    + */
    +struct drm_i915_gem_execbuffer3 {
    +       /**
    +        * @ctx_id: Context id
    +        *
    +        * Only contexts with user engine map are allowed.
    +        */
    +       __u32 ctx_id;
    +
    +       /**
    +        * @engine_idx: Engine index
    +        *
    +        * An index in the user engine map of the context specified
    by @ctx_id.
    +        */
    +       __u32 engine_idx;
    +
    +       /**
    +        * @batch_address: Batch gpu virtual address/es.
    +        *
    +        * For normal submission, it is the gpu virtual address of
    the batch
    +        * buffer. For parallel submission, it is a pointer to an
    array of
    +        * batch buffer gpu virtual addresses with array size equal
    to the
    +        * number of (parallel) engines involved in that submission (See
    +        * struct i915_context_engines_parallel_submit).
    +        */
    +       __u64 batch_address;
    +
    +       /** @flags: Currently reserved, MBZ */
    +       __u64 flags;
    +
    +       /** @rsvd1: Reserved, MBZ */
    +       __u32 rsvd1;
    +
    +       /** @fence_count: Number of fences in @timeline_fences array. */
    +       __u32 fence_count;
    +
    +       /**
    +        * @timeline_fences: Pointer to an array of timeline fences.
    +        *
    +        * Timeline fences are of format struct
    drm_i915_gem_timeline_fence.
    +        */
    +       __u64 timeline_fences;
    +
    +       /** @rsvd2: Reserved, MBZ */
    +       __u64 rsvd2;
    +
    +       /**
    +        * @extensions: Zero-terminated chain of extensions.
    +        *
    +        * For future extensions. See struct i915_user_extension.
    +        */
    +       __u64 extensions;
    +};
    +
    +/**
    + * struct drm_i915_gem_create_ext_vm_private - Extension to make
    the object
    + * private to the specified VM.
    + *
    + * See struct drm_i915_gem_create_ext.
    + */
    +struct drm_i915_gem_create_ext_vm_private {
    +#define I915_GEM_CREATE_EXT_VM_PRIVATE         2
    +       /** @base: Extension link. See struct i915_user_extension. */
    +       struct i915_user_extension base;
    +
    +       /** @vm_id: Id of the VM to which the object is private */
    +       __u32 vm_id;
    +};
-- 2.21.0.rc0.32.g243a4c7e27




[Index of Archives]     [AMD Graphics]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux