Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Oak Zeng <oak.zeng@intel.com>
To: intel-xe@lists.freedesktop.org
Subject: [CI 09/43] drm/svm: Mark drm_gpuvm to participate SVM
Date: Tue, 11 Jun 2024 22:25:31 -0400	[thread overview]
Message-ID: <20240612022605.385062-9-oak.zeng@intel.com> (raw)
In-Reply-To: <20240612022605.385062-1-oak.zeng@intel.com>

A mm_struct field is added to drm_gpuvm. Also add a parameter to
drm_gpuvm_init to say whether this gpuvm participate svm (shared
virtual memory with CPU process). Under SVM, CPU program and GPU
program share one process virtual address space.

Cc: Dave Airlie <airlied@redhat.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Felix Kuehling <felix.kuehling@amd.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Brian Welty <brian.welty@intel.com>
Cc: <dri-devel@lists.freedesktop.org>
Suggested-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Oak Zeng <oak.zeng@intel.com>
---
 drivers/gpu/drm/drm_gpuvm.c            | 7 ++++++-
 drivers/gpu/drm/nouveau/nouveau_uvmm.c | 2 +-
 drivers/gpu/drm/xe/xe_vm.c             | 2 +-
 include/drm/drm_gpuvm.h                | 8 +++++++-
 4 files changed, 15 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/drm_gpuvm.c b/drivers/gpu/drm/drm_gpuvm.c
index 7402ed6f1d33..5f246ca472a8 100644
--- a/drivers/gpu/drm/drm_gpuvm.c
+++ b/drivers/gpu/drm/drm_gpuvm.c
@@ -984,6 +984,8 @@ EXPORT_SYMBOL_GPL(drm_gpuvm_resv_object_alloc);
  * @reserve_offset: the start of the kernel reserved GPU VA area
  * @reserve_range: the size of the kernel reserved GPU VA area
  * @ops: &drm_gpuvm_ops called on &drm_gpuvm_sm_map / &drm_gpuvm_sm_unmap
+ * @participate_svm: whether this gpuvm participat shared virtual memory
+ * with CPU mm
  *
  * The &drm_gpuvm must be initialized with this function before use.
  *
@@ -997,7 +999,7 @@ drm_gpuvm_init(struct drm_gpuvm *gpuvm, const char *name,
 	       struct drm_gem_object *r_obj,
 	       u64 start_offset, u64 range,
 	       u64 reserve_offset, u64 reserve_range,
-	       const struct drm_gpuvm_ops *ops)
+	       const struct drm_gpuvm_ops *ops, bool participate_svm)
 {
 	gpuvm->rb.tree = RB_ROOT_CACHED;
 	INIT_LIST_HEAD(&gpuvm->rb.list);
@@ -1016,6 +1018,9 @@ drm_gpuvm_init(struct drm_gpuvm *gpuvm, const char *name,
 	gpuvm->drm = drm;
 	gpuvm->r_obj = r_obj;
 
+	if (participate_svm)
+		gpuvm->mm = current->mm;
+
 	drm_gem_object_get(r_obj);
 
 	drm_gpuvm_warn_check_overflow(gpuvm, start_offset, range);
diff --git a/drivers/gpu/drm/nouveau/nouveau_uvmm.c b/drivers/gpu/drm/nouveau/nouveau_uvmm.c
index ee02cd833c5e..0d11f1733e29 100644
--- a/drivers/gpu/drm/nouveau/nouveau_uvmm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_uvmm.c
@@ -1861,7 +1861,7 @@ nouveau_uvmm_ioctl_vm_init(struct drm_device *dev,
 		       NOUVEAU_VA_SPACE_END,
 		       init->kernel_managed_addr,
 		       init->kernel_managed_size,
-		       &gpuvm_ops);
+		       &gpuvm_ops, false);
 	/* GPUVM takes care from here on. */
 	drm_gem_object_put(r_obj);
 
diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
index 99bf7412475c..14069fc9d820 100644
--- a/drivers/gpu/drm/xe/xe_vm.c
+++ b/drivers/gpu/drm/xe/xe_vm.c
@@ -1363,7 +1363,7 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags)
 	}
 
 	drm_gpuvm_init(&vm->gpuvm, "Xe VM", DRM_GPUVM_RESV_PROTECTED, &xe->drm,
-		       vm_resv_obj, 0, vm->size, 0, 0, &gpuvm_ops);
+		       vm_resv_obj, 0, vm->size, 0, 0, &gpuvm_ops, false);
 
 	drm_gem_object_put(vm_resv_obj);
 
diff --git a/include/drm/drm_gpuvm.h b/include/drm/drm_gpuvm.h
index 429dc0d82eba..838dd7137f07 100644
--- a/include/drm/drm_gpuvm.h
+++ b/include/drm/drm_gpuvm.h
@@ -242,6 +242,12 @@ struct drm_gpuvm {
 	 * @drm: the &drm_device this VM lives in
 	 */
 	struct drm_device *drm;
+	/**
+	 * @mm: the process &mm_struct which create this gpuvm.
+	 * This is only used for shared virtual memory where virtual
+	 * address space is shared b/t CPU and GPU program.
+	 */
+	struct mm_struct *mm;
 
 	/**
 	 * @mm_start: start of the VA space
@@ -342,7 +348,7 @@ void drm_gpuvm_init(struct drm_gpuvm *gpuvm, const char *name,
 		    struct drm_gem_object *r_obj,
 		    u64 start_offset, u64 range,
 		    u64 reserve_offset, u64 reserve_range,
-		    const struct drm_gpuvm_ops *ops);
+		    const struct drm_gpuvm_ops *ops, bool participate_svm);
 
 /**
  * drm_gpuvm_get() - acquire a struct drm_gpuvm reference
-- 
2.26.3


  parent reply	other threads:[~2024-06-12  2:15 UTC|newest]

Thread overview: 46+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-06-12  2:25 [CI 01/43] mm/hmm: let users to tag specific PFNs Oak Zeng
2024-06-12  2:25 ` [CI 02/43] dma-mapping: provide an interface to allocate IOVA Oak Zeng
2024-06-12  2:25 ` [CI 03/43] dma-mapping: provide callbacks to link/unlink pages to specific IOVA Oak Zeng
2024-06-12  2:25 ` [CI 04/43] iommu/dma: Provide an interface to allow preallocate IOVA Oak Zeng
2024-06-12  2:25 ` [CI 05/43] iommu/dma: Prepare map/unmap page functions to receive IOVA Oak Zeng
2024-06-12  2:25 ` [CI 06/43] iommu/dma: Implement link/unlink page callbacks Oak Zeng
2024-06-12  2:25 ` [CI 07/43] drm: move xe_sg_segment_size to drm layer Oak Zeng
2024-06-12  2:25 ` [CI 08/43] drm: Move GPUVA_START/LAST to drm_gpuvm.h Oak Zeng
2024-06-12  2:25 ` Oak Zeng [this message]
2024-06-12  2:25 ` [CI 10/43] drm/svm: introduce drm_mem_region concept Oak Zeng
2024-06-12  2:25 ` [CI 11/43] drm/svm: introduce hmmptr and helper functions Oak Zeng
2024-06-12  2:25 ` [CI 12/43] drm/svm: Introduce helper to remap drm memory region Oak Zeng
2024-06-12  2:25 ` [CI 13/43] drm/svm: handle CPU page fault Oak Zeng
2024-06-12  2:25 ` [CI 14/43] drm/svm: Migrate a range of hmmptr to vram Oak Zeng
2024-06-12  2:25 ` [CI 15/43] drm/svm: Add DRM SVM documentation Oak Zeng
2024-06-12  2:25 ` [CI 16/43] drm/xe: s/xe_tile_migrate_engine/xe_tile_migrate_exec_queue Oak Zeng
2024-06-12  2:25 ` [CI 17/43] drm/xe: Add xe_vm_pgtable_update_op to xe_vma_ops Oak Zeng
2024-06-12  2:25 ` [CI 18/43] drm/xe: Convert multiple bind ops into single job Oak Zeng
2024-06-12  2:25 ` [CI 19/43] drm/xe: Update VM trace events Oak Zeng
2024-06-12  2:25 ` [CI 20/43] drm/xe: Update PT layer with better error handling Oak Zeng
2024-06-12  2:25 ` [CI 21/43] drm/xe: Retry BO allocation Oak Zeng
2024-06-12  2:25 ` [CI 22/43] drm/xe/uapi: Add DRM_XE_VM_BIND_FLAG_SYSTEM_ALLOCATOR flag Oak Zeng
2024-06-12  2:25 ` [CI 23/43] drm/xe: Add a helper to calculate userptr end address Oak Zeng
2024-06-12  2:25 ` [CI 24/43] drm/xe: Add dma_addr res cursor Oak Zeng
2024-06-12  2:25 ` [CI 25/43] drm/xe: Use drm_mem_region for xe Oak Zeng
2024-06-12  2:25 ` [CI 26/43] drm/xe: use drm_hmmptr in xe Oak Zeng
2024-06-12  2:25 ` [CI 27/43] drm/xe: Moving to range based vma invalidation Oak Zeng
2024-06-12  2:25 ` [CI 28/43] drm/xe: Support range based page table update Oak Zeng
2024-06-12  2:25 ` [CI 29/43] drm/xe/uapi: Add DRM_XE_VM_CREATE_FLAG_PARTICIPATE_SVM flag Oak Zeng
2024-06-12  2:25 ` [CI 30/43] drm/xe/svm: Create userptr if page fault occurs on system_allocator VMA Oak Zeng
2024-06-12  2:25 ` [CI 31/43] drm/xe/svm: Add faulted userptr VMA garbage collector Oak Zeng
2024-06-12  2:25 ` [CI 32/43] drm/xe: Introduce helper to get tile from memory region Oak Zeng
2024-06-12  2:25 ` [CI 33/43] drm/xe/svm: implement functions to allocate and free device memory Oak Zeng
2024-06-12  2:25 ` [CI 34/43] drm/xe/svm: Get drm device from drm memory region Oak Zeng
2024-06-12  2:25 ` [CI 35/43] drm/xe/svm: Get page map owner of a " Oak Zeng
2024-06-12  2:25 ` [CI 36/43] drm/xe/svm: Add migrate layer functions for SVM support Oak Zeng
2024-06-12  2:25 ` [CI 37/43] drm/xe/svm: introduce svm migration function Oak Zeng
2024-06-12  2:26 ` [CI 38/43] drm/xe/svm: Register xe memory region to drm layer Oak Zeng
2024-06-12  2:26 ` [CI 39/43] drm/xe/svm: Introduce DRM_XE_SVM kernel config Oak Zeng
2024-06-12  2:26 ` [CI 40/43] drm/xe/svm: Migration from sram to vram for system allocator Oak Zeng
2024-06-12  2:26 ` [CI 41/43] drm/xe/svm: Determine a vma is backed by device memory Oak Zeng
2024-06-12  2:26 ` [CI 42/43] drm/xe/svm: Introduce hmm_pfn array based resource cursor Oak Zeng
2024-06-12  2:26 ` [CI 43/43] drm/xe: Enable system allocator uAPI Oak Zeng
2024-06-12  3:14 ` ✓ CI.Patch_applied: success for series starting with [CI,01/43] mm/hmm: let users to tag specific PFNs Patchwork
2024-06-12  3:15 ` ✗ CI.checkpatch: warning " Patchwork
2024-06-12  3:16 ` ✗ CI.KUnit: failure " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20240612022605.385062-9-oak.zeng@intel.com \
    --to=oak.zeng@intel.com \
    --cc=intel-xe@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox