Am 2023-02-08 um 14:57 schrieb Xiaogang.Chen:
From: Xiaogang Chen <xiaogang.c...@amd.com>
When xnack is on user space can use svm page restore to set a vm range without
setup it first, then use regular api to register. Currently kfd api and svm are
not interoperable. We already have check on that, but for user buffer the
mapping
address is not same as buffer cpu virtual address. Add checking on that to
avoid error propagate to hmm.
---
drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index f79b8e964140..6d9cf860d2da 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -1065,6 +1065,21 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file
*filep,
mutex_unlock(&p->svms.lock);
return -EADDRINUSE;
}
+
+ /* When register user buffer check if it has been registered by svm by
+ * buffer cpu virtual address.
+ */
+ if ((flags & KFD_IOC_ALLOC_MEM_FLAGS_USERPTR) &&
+ interval_tree_iter_first(&p->svms.objects,
+ args->mmap_offset >> PAGE_SHIFT,
+ (args->mmap_offset + args->size - 1) >> PAGE_SHIFT)) {
The indentation would be more readable if it was properly aligned with
the opening (, both for the if statement and the
interval_tree_iter_first call. Our coding style is not completely
consistent in this, but the existing call of interval_tree_iter_first in
this function uses that style for reference.
Regards,
Felix
+
+ pr_err("User Buffer Address: 0x%llx already allocated by SVM\n",
+ args->mmap_offset);
+ mutex_unlock(&p->svms.lock);
+ return -EADDRINUSE;
+ }
+
mutex_unlock(&p->svms.lock);
#endif
mutex_lock(&p->mutex);
