[llvm-bugs] [Bug 35489] New: +g and +X inline asm constraints generate worse code than +r

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=35489

            Bug ID: 35489
           Summary: +g and +X inline asm constraints generate worse code
                    than +r
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Common Code Generator Code
          Assignee: unassignedbugs at nondot.org
          Reporter: richard-llvm at metafoo.co.uk
                CC: llvm-bugs at lists.llvm.org

Testcase:

int f(int n) { asm volatile("":"+g"(n)); return n; }
int k = f(4) + 1;


Targeting x86_64, we get this mess:

        movl    $4, -4(%rsp)
        movl    $4, %eax
        #APP
        #NO_APP
        movl    %eax, -4(%rsp)
        addl    $1, %eax
        movl    %eax, k(%rip)

A "+r" constraint, which should be *more* constraining (and thus produce worse
code in general) gives this:

        movl    $4, %eax
        #APP
        #NO_APP
        addl    $1, %eax
        movl    %eax, k(%rip)

GCC produces the better code for both constraints.

-- 
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/20171201/e3a73b97/attachment.html>