More conservative value analysis of pointer equality #516
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
A pointer "one past" a global "x" and a pointer to the beginning of a global "y" can be equal. This is undefined behavior in CompCert C, but defined behavior in ISO C. Thus, the value analysis should not infer that a pointer based at "x" and a pointer based at "y" cannot be equal. The only inference we keep is that a pointer based in the current stack block and a pointer based in a global variable cannot be equal.
xavierleroy
force-pushed
the
value-analysis-pointer-comparison
branch
from
b311744 to
918df5a
Compare
|
Follow-up commit: fd48dc8 |
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Outdated suggestions cannot be applied.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.
Consider two global variables
xandy. Currently, the value analysis considers that any pointer based onxcannot be equal to any pointer based ony. This reflects the CompCert C semantics for pointer equality: if the two pointers are within bounds, they must differ; and if one pointer is outside bounds (including if it is "one past"), comparison is undefined.However, a pointer "one past"
xand a pointer to the beginning ofycan be equal at the machine level, and comparing them for equality is not undefined behavior in ISO C, just an unpredictable result.This PR makes sure CompCert's value analysis says "I don't know" for equality tests between pointers based on different global variables. It still says "false or undefined" for equality between a pointer based in the current stack block and a pointer based elsewhere, as such pointers can never be equal even if "one past".