[functorch] fix unexpected successes
diff --git a/functorch/test/test_ops.py b/functorch/test/test_ops.py
index bc11dc2..fa94b8c 100644
--- a/functorch/test/test_ops.py
+++ b/functorch/test/test_ops.py
@@ -269,9 +269,6 @@
 vjp_fail = {
     skip('nn.functional.dropout'),  # randomness testing artifact
     skip('nn.functional.rrelu'),  # randomness testing artifact
-    xfail('linalg.cholesky'),
-    xfail('linalg.inv'),
-    xfail('linalg.matrix_power'),
     xfail('tensor_split'),
     xfail('to_sparse'),
     xfail('nn.functional.ctc_loss'),
@@ -356,10 +353,7 @@
 
         # Composite ops that do bad things. Need to be fixed in PyTorch core.
         # RuntimeError: Cannot access data pointer of Tensor that doesn't have storage
-        xfail('linalg.inv'),
         xfail('linalg.eigvals'),
-        xfail('linalg.matrix_power'),
-        xfail('linalg.cholesky'),
         xfail('tensor_split'),
 
         # Causing a CUDA assert, needs investigation
diff --git a/functorch/test/test_pythonkey.py b/functorch/test/test_pythonkey.py
index 2f3fb77..6dfd116 100644
--- a/functorch/test/test_pythonkey.py
+++ b/functorch/test/test_pythonkey.py
@@ -182,9 +182,6 @@
     xfail('to_sparse'),
     xfail('allclose'),
     xfail('rsub', 'rsub_scalar'),
-    xfail('linalg.matrix_power'),
-    xfail('linalg.inv'),
-    xfail('linalg.cholesky'),
     xfail('nn.functional.dropout'),
     xfail('linalg.eigvals'),
     xfail('nn.functional.ctc_loss'),
@@ -326,7 +323,6 @@
     @skipOps('TestEagerFusionOpInfo', 'test_aot_autograd_exhaustive', {
         xfail('__rmatmul__'),
         xfail('linalg.cholesky'),
-        xfail('linalg.inv'),
         xfail('matmul'),
         xfail('msort'),
         xfail('nn.functional.linear'),