Add test that validates swt clone module function references. #1477
Conversation
Pavel-Durov
force-pushed
the
fix-function-reference-yk-module-clone-pass
branch
from
ce4bf45 to
f1d58ed
Compare
|
updated ykllvm submodule with the latest ykllvm version since ykjit/ykllvm#216 was merged. |
tests/c/swt_module_clone.c
Outdated
| #include <yk.h> | ||
| #include <yk_testing.h> | ||
|
|
||
| int add(int i, int j) { return i + j; } |
There was a problem hiding this comment.
These should probably be __attribute__((yk_outline)): at the moment I would expect that these are inlined, so the test will pass "by accident" (TM). Also, the stderr lang_test should check that the IR actually looks like we expect. Specifically, we should remove jit-pre-opt from YKD_LOG_IR but we should actually check that the functions, with the names we expect, are present in the AOT IR.
There was a problem hiding this comment.
OK, will update.
But there's a problem with adding IR checks here because we don't serialize the cloned/optimized functions at all.
There was a problem hiding this comment.
That's fine: if the AOT IR should only refer to the unoptimised functions, then that's our test!
There was a problem hiding this comment.
Should there be a call dec in the trace IR too?
There was a problem hiding this comment.
How do I get the trace IR?
There was a problem hiding this comment.
It depends which you want: aot, jit-pre-opt, or jit-post-opt.
There was a problem hiding this comment.
ok, got it.
I got confused cause originally we said to remove jit-pre-opt from YKD_LOG_IR
There was a problem hiding this comment.
Should there be a call dec in the trace IR too?
Added trace checks 👉 d4b95bf
There was a problem hiding this comment.
Can we check the call is to dec? I'm not totally sure off-hand if that is possible or not.
Add test to validate ykjit/ykllvm#216.