[llvm-bugs] [Bug 50917] New: [Miscompilation] Inlining Causes Jump to Wrong Address, Maybe Because of setjmp()

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=50917

            Bug ID: 50917
           Summary: [Miscompilation] Inlining Causes Jump to Wrong
                    Address, Maybe Because of setjmp()
           Product: clang
           Version: 12.0
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: C
          Assignee: unassignedclangbugs at nondot.org
          Reporter: gavin.d.howard at gmail.com
                CC: blitzrakete at gmail.com, dgregor at apple.com,
                    erik.pilkington at gmail.com, llvm-bugs at lists.llvm.org,
                    richard-llvm at metafoo.co.uk

Clang is inlining a function, and when returning from the function, it is
jumping to the wrong location.

The code in question is at
https://git.yzena.com/gavin/bc/src/commit/c2d2977f2c6c9fe3cd8dc9b07fc8913008c81895/src/vm.c#L780-L842
.

As near as I can tell, when the call to `bc_vm_readLine()` returns `false`, the
code then does not jump to the end of the loop. Instead, it jumps to line 815,
which previously did a `setjmp()`. That `setjmp()` then thinks that a
`longjmp()` happens, which causes it to jump past the end of the loop and to
the  `err:` label (the code inside `BC_SETJMP_LOCKED(err)` essentially does a
`if (setjmp(buf)) goto err;`).

I have confirmed the bug exists in Clang 12.0.1-rc3, 11.1.0, 10.0.1, and 9.0.1.
I could not get 8.0.1 to compile.

-- 
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/20210628/d5734143/attachment.html>