kparzysz-quic commented on a change in pull request #8955:
URL: https://github.com/apache/tvm/pull/8955#discussion_r703863612
##########
File path: tests/python/unittest/test_target_codegen_hexagon.py
##########
@@ -17,30 +17,21 @@
import numpy as np
import os
+import pytest
import re
+import sys
import tvm
import tvm.relay
+import tvm.testing
import tvm.contrib.hexagon as hexagon
-def check_prereq_and_setup():
- if tvm.target.codegen.llvm_version_major() <= 7:
- print("Skipping test: need LLVM 7 or later for codegen")
- return False
- if os.name != "posix":
- print("Skipping test on non-POSIX platforms")
- return False
- if not tvm.runtime.enabled("hexagon"):
- print("Hexagon runtime not enabled")
- return False
- # Register a phony linker, so that we can test codegen without a Hexagon
toolchain.
- hexagon.register_linker(lambda: "/bin/true")
- return True
+# Register a phony linker, so that we can test codegen without a Hexagon
toolchain.
+hexagon.register_linker(lambda: "/bin/true")
Review comment:
One problem is that if all of these tests run from a common
"super"-module, and are executed in parallel, then the linker may get
unregistered by one test before the other tests are done using it. Right now
this is the only tests that does this, but if there are more added later, then
it may become an issue.
Another thing may be that we may want to do lazy linking in Hexagon module,
so that we're not looking for the linker each time when `tvm.build` is
executed...
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
