[llvm-bugs] [Bug 39574] New: Clang generates incorrect code when compiling chromium's sql_transaction.cc

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=39574

            Bug ID: 39574
           Summary: Clang generates incorrect code when compiling
                    chromium's sql_transaction.cc
           Product: clang
           Version: trunk
          Hardware: Other
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: C++
          Assignee: unassignedclangbugs at nondot.org
          Reporter: ptworek at vewd.com
                CC: dgregor at apple.com, llvm-bugs at lists.llvm.org,
                    richard-llvm at metafoo.co.uk

I've recently run into strange crashes in a chromium based product compiled
with the recent HEAD versions of clang. After some investigation I've
discovered that blink::SQLTransaction::StateFunctionFor seems to return bad
pointers. This is rather strange since the function itself is pretty trivial
https://cs.chromium.org/chromium/src/third_party/blink/renderer/modules/webdatabase/sql_transaction.cc?q=sql_transaction.cc&sq=package:chromium&dr&l=136.
What is even more interesting, the problem occurs only if this one specific
function is compiled with optimizations enabled. Adding "pragma clang optimize
on/off" around the function fixes the problem. Another way to solve it is to
simply comment out the second DCHECK in this function. The DCHECK failure is
not a problem however, only its presence affects the generated code somehow.

The issue happens only on armv7, I cannot reproduce it on x86_64.

So far I've been unable to prepare a minimized TC that reproduces this problem
outside of chromium codebase.

-- 
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/20181107/1d9e367b/attachment.html>