Skip to content

Fix getParent / getNodeBaseParent to return pointer to const (#94224) #96053

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from

Conversation

dklimkin
Copy link
Member

Fix getParent / getNodeBaseParent to return pointer to const as opposed to const pointer.

This is particularly problematic as the returned object iterators get const bit unset.

@llvmbot
Copy link
Member

llvmbot commented Jun 19, 2024

@llvm/pr-subscribers-llvm-adt

Author: Danial Klimkin (dklimkin)

Changes

Fix getParent / getNodeBaseParent to return pointer to const as opposed to const pointer.

This is particularly problematic as the returned object iterators get const bit unset.


Full diff: https://github.com/llvm/llvm-project/pull/96053.diff

2 Files Affected:

  • (modified) llvm/include/llvm/ADT/ilist_node.h (+4-1)
  • (modified) llvm/include/llvm/ADT/ilist_node_base.h (+4-1)
diff --git a/llvm/include/llvm/ADT/ilist_node.h b/llvm/include/llvm/ADT/ilist_node.h
index 209fd209167e8..db0319d487433 100644
--- a/llvm/include/llvm/ADT/ilist_node.h
+++ b/llvm/include/llvm/ADT/ilist_node.h
@@ -17,6 +17,7 @@
 
 #include "llvm/ADT/ilist_node_base.h"
 #include "llvm/ADT/ilist_node_options.h"
+#include <type_traits>
 
 namespace llvm {
 
@@ -29,7 +30,9 @@ struct NodeAccess;
 /// has been set in the list options.
 template <class NodeTy, class ParentPtrTy> class node_parent_access {
 public:
-  inline const ParentPtrTy getParent() const {
+  using ConstParentPtrTy =
+      std::add_pointer_t<std::add_const_t<std::remove_pointer_t<ParentPtrTy>>>;
+  inline ConstParentPtrTy getParent() const {
     return static_cast<const NodeTy *>(this)->getNodeBaseParent();
   }
   inline ParentPtrTy getParent() {
diff --git a/llvm/include/llvm/ADT/ilist_node_base.h b/llvm/include/llvm/ADT/ilist_node_base.h
index caad87cdd4d71..5a8244336188f 100644
--- a/llvm/include/llvm/ADT/ilist_node_base.h
+++ b/llvm/include/llvm/ADT/ilist_node_base.h
@@ -10,6 +10,7 @@
 #define LLVM_ADT_ILIST_NODE_BASE_H
 
 #include "llvm/ADT/PointerIntPair.h"
+#include <type_traits>
 
 namespace llvm {
 
@@ -48,10 +49,12 @@ template <class NodeBase> class node_base_prevnext<NodeBase, true> {
 
 template <class ParentPtrTy> class node_base_parent {
   ParentPtrTy Parent = nullptr;
+  using ConstParentPtrTy =
+      std::add_pointer_t<std::add_const_t<std::remove_pointer_t<ParentPtrTy>>>;
 
 public:
   void setNodeBaseParent(ParentPtrTy Parent) { this->Parent = Parent; }
-  inline const ParentPtrTy getNodeBaseParent() const { return Parent; }
+  inline ConstParentPtrTy getNodeBaseParent() const { return Parent; }
   inline ParentPtrTy getNodeBaseParent() { return Parent; }
 };
 template <> class node_base_parent<void> {};

@dklimkin dklimkin requested a review from SLTozer June 19, 2024 10:55
Copy link

⚠️ We detected that you are using a GitHub private e-mail address to contribute to the repo.
Please turn off Keep my email addresses private setting in your account.
See LLVM Discourse for more information.

@dklimkin
Copy link
Member Author

Ideally, I'd prefer ParentPtrTy to be ParentTy and used as ParentTy* throughout but it's a much larger change.

Copy link
Contributor

@SLTozer SLTozer left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ideally, I'd prefer ParentPtrTy to be ParentTy and used as ParentTy* throughout but it's a much larger change.

I just commented exactly that on the original PR - I originally passed around ParentPtrTy to preserve the possibility of a void* parent pointer since it seemed like a 0-cost decision that could possibly be useful in future, but since it now results in messier template logic I'm fine to drop it.

This patch LGTM, so if you need to merge it to unblock something downstream then go ahead; if it's not so urgent, then I'll upload a patch that fixes it as described above instead.

@dklimkin
Copy link
Member Author

I've put an internal fix for this now so we are good to wait for the patch and reduce churn. Thank you.

@dklimkin dklimkin closed this Jun 19, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants