[llvm-bugs] [Bug 33081] New: [ubsan] Consider not instrumenting volatile loads and stores with null checks

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=33081

            Bug ID: 33081
           Summary: [ubsan] Consider not instrumenting volatile loads and
                    stores with null checks
           Product: compiler-rt
           Version: unspecified
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: compiler-rt
          Assignee: unassignedbugs at nondot.org
          Reporter: rnk at google.com
                CC: llvm-bugs at lists.llvm.org

The idea behind volatile is that accessing the memory in question has some
implementation-defined semantics and side effects. Using volatile is
essentially a request to get the behavior of the current implementation,
whatever it may be, such as trapping to the signal handler or loading from
memory that's actually mapped at zero.

This came up with I added LLVM's CrashRecoveryContext tests, which
intentionally dereference null. It's actually interesting to test SIGSEGV
separately from testing SIGILL, which is what __builtin_trap would normally
raise.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20170517/db8175cb/attachment.html>