Hey Matthew, all,
This sounds like a good thing to have.
There are a number of DG2 machines where we have a small BAR and this is
causing more apps to fail.
Anv currently reports 3 memory heaps to the app :
- local device only (not host visible) -> mapped to lmem
- device/cpu -> mapped to smem
- local device but also host visible -> mapped to lmem
So we could use this straight away, by just not putting the
I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS flag on the allocation of the
first heap.
One thing I don't see in this proposal is how can we get the size of the
2 lmem heap : cpu visible, cpu not visible
We could use that to report the appropriate size to the app.
We probably want to report a new drm_i915_memory_region_info and either :
- put one of the reserve field to use to indicate : cpu visible
- or define a new enum value in drm_i915_gem_memory_class
Cheers,
-Lionel
On 18/02/2022 13:22, Matthew Auld wrote:
Add an entry for the new uapi needed for small BAR on DG2+.
Signed-off-by: Matthew Auld <matthew.auld@xxxxxxxxx>
Cc: Thomas Hellström <thomas.hellstrom@xxxxxxxxxxxxxxx>
Cc: Jon Bloomfield <jon.bloomfield@xxxxxxxxx>
Cc: Daniel Vetter <daniel.vetter@xxxxxxxx>
Cc: Jordan Justen <jordan.l.justen@xxxxxxxxx>
Cc: Kenneth Graunke <kenneth@xxxxxxxxxxxxx>
Cc: mesa-dev@xxxxxxxxxxxxxxxxxxxxx
---
Documentation/gpu/rfc/i915_small_bar.h | 153 +++++++++++++++++++++++
Documentation/gpu/rfc/i915_small_bar.rst | 40 ++++++
Documentation/gpu/rfc/index.rst | 4 +
3 files changed, 197 insertions(+)
create mode 100644 Documentation/gpu/rfc/i915_small_bar.h
create mode 100644 Documentation/gpu/rfc/i915_small_bar.rst
diff --git a/Documentation/gpu/rfc/i915_small_bar.h b/Documentation/gpu/rfc/i915_small_bar.h
new file mode 100644
index 000000000000..fa65835fd608
--- /dev/null
+++ b/Documentation/gpu/rfc/i915_small_bar.h
@@ -0,0 +1,153 @@
+/**
+ * struct __drm_i915_gem_create_ext - Existing gem_create behaviour, with added
+ * extension support using struct i915_user_extension.
+ *
+ * Note that in the future we want to have our buffer flags here, at least for
+ * the stuff that is immutable. Previously we would have two ioctls, one to
+ * create the object with gem_create, and another to apply various parameters,
+ * however this creates some ambiguity for the params which are considered
+ * immutable. Also in general we're phasing out the various SET/GET ioctls.
+ */
+struct __drm_i915_gem_create_ext {
+ /**
+ * @size: Requested size for the object.
+ *
+ * The (page-aligned) allocated size for the object will be returned.
+ *
+ * Note that for some devices we have might have further minimum
+ * page-size restrictions(larger than 4K), like for device local-memory.
+ * However in general the final size here should always reflect any
+ * rounding up, if for example using the I915_GEM_CREATE_EXT_MEMORY_REGIONS
+ * extension to place the object in device local-memory.
+ */
+ __u64 size;
+ /**
+ * @handle: Returned handle for the object.
+ *
+ * Object handles are nonzero.
+ */
+ __u32 handle;
+ /**
+ * @flags: Optional flags.
+ *
+ * Supported values:
+ *
+ * I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS - Signal to the kernel that
+ * the object will need to be accessed via the CPU.
+ *
+ * Only valid when placing objects in I915_MEMORY_CLASS_DEVICE, and
+ * only strictly required on platforms where only some of the device
+ * memory is directly visible or mappable through the CPU, like on DG2+.
+ *
+ * One of the placements MUST also be I915_MEMORY_CLASS_SYSTEM, to
+ * ensure we can always spill the allocation to system memory, if we
+ * can't place the object in the mappable part of
+ * I915_MEMORY_CLASS_DEVICE.
+ *
+ * Note that buffers that need to be captured with EXEC_OBJECT_CAPTURE,
+ * will need to enable this hint, if the object can also be placed in
+ * I915_MEMORY_CLASS_DEVICE, starting from DG2+. The execbuf call will
+ * throw an error otherwise. This also means that such objects will need
+ * I915_MEMORY_CLASS_SYSTEM set as a possible placement.
+ *
+ * Without this hint, the kernel will assume that non-mappable
+ * I915_MEMORY_CLASS_DEVICE is preferred for this object. Note that the
+ * kernel can still migrate the object to the mappable part, as a last
+ * resort, if userspace ever CPU faults this object, but this might be
+ * expensive, and so ideally should be avoided.
+ */
+#define I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS (1 << 0)
+ __u32 flags;
+ /**
+ * @extensions: The chain of extensions to apply to this object.
+ *
+ * This will be useful in the future when we need to support several
+ * different extensions, and we need to apply more than one when
+ * creating the object. See struct i915_user_extension.
+ *
+ * If we don't supply any extensions then we get the same old gem_create
+ * behaviour.
+ *
+ * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see
+ * struct drm_i915_gem_create_ext_memory_regions.
+ *
+ * For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see
+ * struct drm_i915_gem_create_ext_protected_content.
+ */
+#define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0
+#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1
+ __u64 extensions;
+};
+
+#define DRM_I915_QUERY_VMA_INFO 5
+
+/**
+ * struct __drm_i915_query_vma_info
+ *
+ * Given a vm and GTT address, lookup the corresponding vma, returning its set
+ * of attributes.
+ *
+ * .. code-block:: C
+ *
+ * struct drm_i915_query_vma_info info = {};
+ * struct drm_i915_query_item item = {
+ * .data_ptr = (uintptr_t)&info,
+ * .query_id = DRM_I915_QUERY_VMA_INFO,
+ * };
+ * struct drm_i915_query query = {
+ * .num_items = 1,
+ * .items_ptr = (uintptr_t)&item,
+ * };
+ * int err;
+ *
+ * // Unlike some other types of queries, there is no need to first query
+ * // the size of the data_ptr blob here, since we already know ahead of
+ * // time how big this needs to be.
+ * item.length = sizeof(info);
+ *
+ * // Next we fill in the vm_id and ppGTT address of the vma we wish
+ * // to query, before then firing off the query.
+ * info.vm_id = vm_id;
+ * info.offset = gtt_address;
+ * err = ioctl(fd, DRM_IOCTL_I915_QUERY, &query);
+ * if (err || item.length < 0) ...
+ *
+ * // If all went well we can now inspect the returned attributes.
+ * if (info.attributes & DRM_I915_QUERY_VMA_INFO_CPU_VISIBLE) ...
+ */
+struct __drm_i915_query_vma_info {
+ /**
+ * @vm_id: The given vm id that contains the vma. The id is the value
+ * returned by the DRM_I915_GEM_VM_CREATE. See struct
+ * drm_i915_gem_vm_control.vm_id.
+ */
+ __u32 vm_id;
+ /** @pad: MBZ. */
+ __u32 pad;
+ /**
+ * @offset: The corresponding ppGTT address of the vma which the kernel
+ * will use to perform the lookup.
+ */
+ __u64 offset;
+ /**
+ * @attributes: The returned attributes for the given vma.
+ *
+ * Possible values:
+ *
+ * DRM_I915_QUERY_VMA_INFO_CPU_VISIBLE - Set if the pages backing the
+ * vma are currently CPU accessible. If this is not set then the vma is
+ * currently backed by I915_MEMORY_CLASS_DEVICE memory, which the CPU
+ * cannot directly access(this is only possible on discrete devices with
+ * a small BAR). Attempting to MMAP and fault such an object will
+ * require the kernel first synchronising any GPU work tied to the
+ * object, before then migrating the pages, either to the CPU accessible
+ * part of I915_MEMORY_CLASS_DEVICE, or I915_MEMORY_CLASS_SYSTEM, if the
+ * placements permit it. See I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS.
+ *
+ * Note that this is inherently racy.
+ */
+#define DRM_I915_QUERY_VMA_INFO_CPU_VISIBLE (1<<0)
+ __u64 attributes;
+ /** @rsvd: MBZ */
+ __u32 rsvd[4];
+};
diff --git a/Documentation/gpu/rfc/i915_small_bar.rst b/Documentation/gpu/rfc/i915_small_bar.rst
new file mode 100644
index 000000000000..fea92d3d69ab
--- /dev/null
+++ b/Documentation/gpu/rfc/i915_small_bar.rst
@@ -0,0 +1,40 @@
+==========================
+I915 Small BAR RFC Section
+==========================
+Starting from DG2 we will have resizable BAR support for device local-memory,
+but in some cases the final BAR size might still be smaller than the total
+local-memory size. In such cases only part of local-memory will be CPU
+accessible, while the remainder is only accessible via the GPU.
+
+I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS flag
+----------------------------------------------
+New gem_create_ext flag to tell the kernel that a BO will require CPU access.
+The becomes important when placing an object in LMEM, where underneath the
+device has a small BAR, meaning only part of it is CPU accessible. Without this
+flag the kernel will assume that CPU access is not required, and prioritize
+using the non-CPU visible portion of LMEM(if present on the device).
+
+Related to this, we now also reject any objects marked with
+EXEC_OBJECT_CAPTURE, which are also not tagged with NEEDS_CPU_ACCESS. This only
+impacts DG2+.
+
+XXX: One open here is whether we should extend the memory region query to return
+the CPU visible size of the region. For now the IGTs just use debugfs to query
+the size. However, if userspace sees a real need for this then extending the
+region query would be a lot nicer.
+
+.. kernel-doc:: Documentation/gpu/rfc/i915_small_bar.h
+ :functions: __drm_i915_gem_create_ext
+
+DRM_I915_QUERY_VMA_INFO query
+-----------------------------
+Query the attributes of some vma. Given a vm and GTT offset, find the
+respective vma, and return its set of attrubutes. For now we only support
+DRM_I915_QUERY_VMA_INFO_CPU_VISIBLE, which is set if the object/vma is
+currently placed in memory that is accessible by the CPU. This should always be
+set on devices where the CPU visible size of LMEM matches the probed size. If
+this is not set then CPU faulting the object will first require migrating the
+pages.
+
+.. kernel-doc:: Documentation/gpu/rfc/i915_small_bar.h
+ :functions: __drm_i915_query_vma_info
diff --git a/Documentation/gpu/rfc/index.rst b/Documentation/gpu/rfc/index.rst
index 018a8bf317a6..5b8495bdc1fd 100644
--- a/Documentation/gpu/rfc/index.rst
+++ b/Documentation/gpu/rfc/index.rst
@@ -19,3 +19,7 @@ host such documentation:
.. toctree::
i915_scheduler.rst
+
+.. toctree::
+
+ i915_small_bar.rst
