[LLVMbugs] [Bug 5990] [x86, x86-64] MMX vectors trample on FP stack, breaking 80-bit floats

bugzilla-daemon at cs.uiuc.edu bugzilla-daemon at cs.uiuc.edu
Mon Jan 11 10:59:28 PST 2010


Dale Johannesen <dalej at apple.com> changed:

           What    |Removed                     |Added
                 CC|                            |dalej at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID

--- Comment #4 from Dale Johannesen <dalej at apple.com>  2010-01-11 12:59:25 ---
You don't provide source.  If the compiler is producing the MM vector operation
from source that doesn't use MM vectors, that's wrong.  However, if the vector
operation is in the source, it's the user's responsibility to insert EMMS at
the right places.  Both gcc (4.2) and llvm-gcc work this way.  This is
obviously not ideal, but fixing it is a big job and there is not much interest
in doing so.

Long doubles are passed to printf normally; there wouldn't be much point in the
%L modifier if you couldn't.

Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

More information about the llvm-bugs mailing list