[PATCH v4 7/9] gpu/drm/amdgpu: optimize out the case when a range is updated to read only

jglisse at redhat.com jglisse at redhat.com
Wed Jan 23 22:23:13 UTC 2019


From: Jérôme Glisse <jglisse at redhat.com>

When range of virtual address is updated read only and corresponding
user ptr object are already read only it is pointless to do anything.
Optimize this case out.

Signed-off-by: Jérôme Glisse <jglisse at redhat.com>
Cc: Christian König <christian.koenig at amd.com>
Cc: Jan Kara <jack at suse.cz>
Cc: Felix Kuehling <Felix.Kuehling at amd.com>
Cc: Jason Gunthorpe <jgg at mellanox.com>
Cc: Andrew Morton <akpm at linux-foundation.org>
Cc: Matthew Wilcox <mawilcox at microsoft.com>
Cc: Ross Zwisler <zwisler at kernel.org>
Cc: Dan Williams <dan.j.williams at intel.com>
Cc: Paolo Bonzini <pbonzini at redhat.com>
Cc: Radim Krčmář <rkrcmar at redhat.com>
Cc: Michal Hocko <mhocko at kernel.org>
Cc: Ralph Campbell <rcampbell at nvidia.com>
Cc: John Hubbard <jhubbard at nvidia.com>
Cc: kvm at vger.kernel.org
Cc: dri-devel at lists.freedesktop.org
Cc: linux-rdma at vger.kernel.org
Cc: linux-fsdevel at vger.kernel.org
Cc: Arnd Bergmann <arnd at arndb.de>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
index 3e6823fdd939..7880eda064cd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
@@ -294,6 +294,7 @@ static int amdgpu_mn_invalidate_range_start_hsa(struct mmu_notifier *mn,
 {
 	struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn);
 	struct interval_tree_node *it;
+	bool update_to_read_only;
 	unsigned long end;
 
 	/* notification is exclusive, but interval is inclusive */
@@ -302,6 +303,8 @@ static int amdgpu_mn_invalidate_range_start_hsa(struct mmu_notifier *mn,
 	if (amdgpu_mn_read_lock(amn, range->blockable))
 		return -EAGAIN;
 
+	update_to_read_only = mmu_notifier_range_update_to_read_only(range);
+
 	it = interval_tree_iter_first(&amn->objects, range->start, end);
 	while (it) {
 		struct amdgpu_mn_node *node;
@@ -317,6 +320,16 @@ static int amdgpu_mn_invalidate_range_start_hsa(struct mmu_notifier *mn,
 
 		list_for_each_entry(bo, &node->bos, mn_list) {
 			struct kgd_mem *mem = bo->kfd_bo;
+			bool read_only;
+
+			/*
+			 * If it is already read only and we are updating to
+			 * read only then we do not need to change anything.
+			 * So save time and skip this one.
+			 */
+			read_only = amdgpu_ttm_tt_is_readonly(bo->tbo.ttm);
+			if (update_to_read_only && read_only)
+				continue;
 
 			if (amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm,
 							 range->start,
-- 
2.17.2



More information about the dri-devel mailing list