[llvm-bugs] [Bug 41490] New: _LIBCPP_INLINE_VISIBILITY doesn't always hide the symbols (Ex. __emplace_back_slow_path)

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=41490

            Bug ID: 41490
           Summary: _LIBCPP_INLINE_VISIBILITY doesn't always hide the
                    symbols (Ex. __emplace_back_slow_path)
           Product: libc++
           Version: unspecified
          Hardware: PC
                OS: All
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: All Bugs
          Assignee: unassignedclangbugs at nondot.org
          Reporter: eric at efcs.ca
                CC: llvm-bugs at lists.llvm.org, mclow.lists at gmail.com

The `inline` keyword placed on the declarations on `__push_back_slow_path` and
`__emplace_back_slow_path` is needed to prevent them from appearing in
libc++.so's dynamic symbol table (for instantiations used inside the library). 

However, they are marked also given hidden visibility, which should prevent
this in the first place.

Adding `inline` is problematic for two reasons:

1) It's not a real fix.
2) It can needlessly increase the binary size of users (because clang actually
uses `inline` as an inlining hint).

The second problem cause chrome to have a 1% binary size increase.

We should investigate why Clang is not hiding this symbol. We should also
seriously consider moving to explicit export lists.

@Louis, any thoughts?

-- 
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/20190413/e3dd3e22/attachment.html>