[LLVMbugs] [Bug 22748] New: Incorrect handling of /openmp- compiler option in --drive-mode=cl mode.

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Mon Mar 2 01:11:49 PST 2015


http://llvm.org/bugs/show_bug.cgi?id=22748

            Bug ID: 22748
           Summary: Incorrect handling of /openmp- compiler option in
                    --drive-mode=cl mode.
           Product: clang
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: normal
          Priority: P
         Component: C++
          Assignee: unassignedclangbugs at nondot.org
          Reporter: eremeev at viva64.com
                CC: dgregor at apple.com, llvmbugs at cs.uiuc.edu
    Classification: Unclassified

We were using a 3.5 clang for preprocessing of C/C++ source files intended for
Visual C++ cl.exe compiler. For this, we were utilizing a '--drive-mode=cl',
which should automatically interchange cl compiler options to their gcc/clang
analogs.

Now, several flags, such as '/openmp', for example, are unused by clang, so the
warning is generated. But it seems that '--drive-mode=cl' is unaware of
'/openmp-' option. In 3.5, it is still treated as an unknow option '/o penmp-'.

But, in the current trunk (3.7), the preprocessor starts treating '/openmp-' as
the output (-o) option of gcc/clang. This has some unforeseen consequences, as
we use clang for preprocessing and we are expecting it to behave like cl when
'--drive-mode=cl' is enabled - that is, starting 

clang --driver-mode=cl /openmp- /P hello.cpp

We expect the preprocessed file (hello.i) to be generated in the current
directory. But, as the current trunk treat the '/openmp-' as /o penmp-, instead
we get the 'penmp-.i' file as the output of the preprocessor, which is
unexpected and incorrect from the cl.exe point of view.

-- 
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/20150302/07c37b1c/attachment.html>


More information about the llvm-bugs mailing list