Handle stepping through language thunks that call other functions than the target #10145
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…read plans (llvm#129301) Jonas recently added a trampoline handling strategy for simple language thunks that does: "step through language thunks stepping in one level deep and stopping if you hit user code". That was actually pulled over from the swift implementation. However, this strategy and the strategy we have to "step out past language thunks" when stepping out come into conflict if the thunk you are stepping through calls some other function before dispatching to the intended method. When you step out of the called function back into the thunk, should you keep stepping out past the thunk or not? In most cases, you want to step out past the thunk, but in this particular case you don't. This patch adds a way to inform the thread plan (or really it's ShouldStopHere behavior) of which behavior it should have, and passes the don't step through thunks to the step through plan it uses to step through thunks. I didn't add a test for this because I couldn't find a C++ thunk that calls another function before getting to the target function. I asked the clang folks here if they could think of a case where clang would do this, and they couldn't. If anyone can think of such a construct, it will be easy to write the step through test for it... This does happen in swift, however, so when I cherry-pick this to the swift fork I'll test it there. (cherry picked from commit ddbce2f)
jimingham
requested review from
felipepiovezan and
adrian-prantl
and removed request for
a team
felipepiovezan
approved these changes
Mar 1, 2025
| self.assertEqual(var.GetSummary(), '"foo"') | ||
|
|
||
| # Now make sure that stepping out steps past the thunk: | ||
| thread.StepOut() |
There was a problem hiding this comment.
is there any benefit in also branching on use_api here?
There was a problem hiding this comment.
Sure, I changed the test to use use_api consistently for all the steps.
felipepiovezan
approved these changes
Mar 3, 2025
JDevlieghere
approved these changes
Mar 3, 2025
|
@swift-ci please test |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
We have a general feature of "stepping out past language thunks". But we also have a strategy for resolving the target of a thunk which involves stepping through it, stepping in one level deep, and if we get somewhere interesting we stop. If we try to do the latter maneuver, we need to turn off the stepping out past thunks or the first time we step out of something "not interesting" we'll also step past the thunk, defeating the strategy.
The fixes that by adding a way to say "this plan should or should not do the step out past thunk, then setting that for the step through plan.
Also added tests for this.