[ORC] Free ELF debug objects in dealloc action of corresponding code (wip) #168522
+245
−449
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
weliveindetail
commented
Nov 18, 2025
| Err = ES.getExecutorProcessControl().getBootstrapSymbols({ | ||
| {RegistrationAction, rt::RegisterJITLoaderGDBAllocActionName}, | ||
| {DeallocAction, rt::SimpleExecutorMemoryManagerReleaseWrapperName}, | ||
| {TargetMemMgr, rt::SimpleExecutorMemoryManagerInstanceName}, |
Member
Author
There was a problem hiding this comment.
InProcessMemoryManager does not expose bootstrap symbols yet. SimpleRemoteEPC does it in createDefaultMemoryManager() and we could do something similar in SelfExecutorProcessControl(). Or should we better enforce it in the JITLinkMemoryManager() base class ctor?
Contributor
There was a problem hiding this comment.
I think that that sounds reasonable, but we should wait until we've brought up the new ORC runtime: there might be an opportunity to define a widely implementable set of memory allocation primitives there.
| // FIXME: The lookup in the segment info here is a workaround. The below | ||
| // FA->release() is supposed to provide the base address in target memory, | ||
| // but InProcessMemoryManager returns the address of a FinalizedAllocInfo | ||
| // helper instead. |
Member
Author
There was a problem hiding this comment.
This is another inconsistency in InProcessMemoryManager. The workaround is ok, but not very reasonable.
Contributor
There was a problem hiding this comment.
There's a substantial further simplification that could be made here: If we're ok with the debug object living for as long as the corresponding JIT'd code then the debug object could be embedded in a special section in the graph, rather than having its own allocation. That would allow you to avoid managing the memory lifetime entirely, and you could just focus on populating the debug object bits and calling the registration and deregistration functions.
🐧 Linux x64 Test Results
Failed Tests(click on a test name to see its output) lldb-shelllldb-shell.Breakpoint/jit-loader_jitlink_elf.testlldb-shell.Breakpoint/jit-loader_rtdyld_elf.testLLVMLLVM.ExecutionEngine/OrcLazy/debug-descriptor.llLLVM.ExecutionEngine/OrcLazy/debug-objects.llLLVM-UnitLLVM-Unit.ExecutionEngine/Orc/_/OrcJITTests/OrcCAPITestBase/EnableDebugSupportIf these failures are unrelated to your changes (for example tests are broken or flaky at HEAD), please open an issue at https://github.com/llvm/llvm-project/issues and add the |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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 don't want to keep tracking allocations of debug objects in the controller after linking finished. So far, the plugin did that to free the target memory once the corresponding code is unlinked. This patch removes the tracking and instead attaches a dealloc action for it to the target memory of the code.