[release/9.0-staging] Fix return address hijacking with CET #109548
Merged
+29
−16
Conversation
This file contains 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
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There is a problematic case when return address is hijacked while in a managed method that tail calls a GC write barrier and when CET is enabled. The write barrier code can change while the handler for the hijacked address is executed from the vectored exception handler. When the vectored exception handler then returns to the write barrier to re-execute the `ret` instruction that has triggered the vectored exception handler due to the main stack containing a different address than the shadow stack (now with the main stack fixed), the instruction may no longer be `ret` due to the change of the write barrier change. This change fixes it by setting the context to return to from the vectored exception handler to point to the caller and setting the Rsp and SSP to match that. That way, the write barrier code no longer matters.
|
Tagging subscribers to this area: @mangod9 |
jkotas
approved these changes
Nov 5, 2024
jeffschwMSFT
approved these changes
Nov 5, 2024
Open
3 tasks
rbhanda
added
Servicing-approved
This was referenced Dec 4, 2024
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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.
Backport of #109074 to release/9.0-staging
/cc @janvorli
Customer Impact
There is a problematic case when return address is hijacked while in a managed method that tail calls a GC write barrier and when CET is enabled. The write barrier code can change while the handler for the hijacked address is executed from the vectored exception handler. When the vectored exception handler then returns to the write barrier to re-execute the ret instruction that has triggered the vectored exception handler due to the main stack containing a different address than the shadow stack (now with the main stack fixed), the instruction may no longer be ret due to the change of the write barrier change.
This was causing a 100% reproducible
System.AccessViolationExceptionin a customer's app.Regression
The issue was introduced by enabling CET by default in a .NET 9 preview
Testing
Customer provided test that was reliably crashing before the fix. The crash is gone with the fix. Also verified using coreclr tests with GC stress enabled.
Risk
Low, the fix just changes the location to return from the vectored exception handler from a ret instruction to where the instruction would return.