[llvm-bugs] [Bug 36676] New: clang-cl asm output does not work with cc1as

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=36676

            Bug ID: 36676
           Summary: clang-cl asm output does not work with cc1as
           Product: clang
           Version: unspecified
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Frontend
          Assignee: unassignedclangbugs at nondot.org
          Reporter: nicolasweber at gmx.de
                CC: llvm-bugs at lists.llvm.org

thakis at thakis:~/src/chrome/src$ cat test.cc
int printf(const char*, ...);
void f() {
  char buffer[20];
  printf(buffer);
}
thakis at thakis:~/src/chrome/src$
third_party/llvm-build/Release+Asserts/bin/clang-cl /TP /c test.cc /FA -m32 
test.asm:29:7: error: unknown token in expression
        call    @__security_check_cookie at 4
                ^


/FA causes clang to not use the integrated assembler but instead to run a cc1as
pass on the generated .asm file -- and cc1as apparently doesn't like clang's
output.


I made /FA use intel assembly in r322652, which might be related.


clang-cl as far as I know doesn't have a flag like -S that means "only generate
assembly", it can only generate assembly as a side effect of compiling.


My motivation is to try and get whodis working with chrome/win translation
units in a cross build. whodis needs intel asm output. I'm trying to just tack
on '/FA /Z7 /Fafilename' to the compile command I get from the comp db, but hat
doesn't roundtrip. __security_check_cookie is one issue, another is `push
offset`:

/usr/local/google/home/thakis/src/chrome/src/out/gnwin/whodis.temp.asm:28933:14:
error: cannot use more than one symbol in memory operand
        push    offset "??_C at _02GFKOMOKH@?5?$CI?$AA@"
                       ^


I think these are the only two types of errors. So I suppose the fix is to
teach the assembly parser about these two constructs?

-- 
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/20180309/ee7a0672/attachment.html>