[llvm-bugs] [Bug 29171] New: Stack frame size can overflow signed 32-bit values on 64-bit targets.

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=29171

            Bug ID: 29171
           Summary: Stack frame size can overflow signed 32-bit values on
                    64-bit targets.
           Product: new-bugs
           Version: 3.9
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: new bugs
          Assignee: unassignedbugs at nondot.org
          Reporter: edy.burt at gmail.com
                CC: llvm-bugs at lists.llvm.org
    Classification: Unclassified

The following two functions generate differently signed stack offsets:

void good() {
  int buf[(1 << 29) - 1];
  buf[0] = 0;
}

void bad() {
  int buf[(1 << 29) + 1];
  buf[0] = 0;
}

good():                               # @good()
        push    rbp
        mov     rbp, rsp
        sub     rsp, 2147483520
        mov     dword ptr [rbp - 2147483648], 0
        add     rsp, 2147483520
        pop     rbp
        ret

bad():                                # @bad()
        push    rbp
        mov     rbp, rsp
        sub     rsp, 2147483536
        mov     dword ptr [rbp + 2147483632], 0
        add     rsp, 2147483536
        pop     rbp
        ret

As you can see, the mov ends up with a positive 64-bit offset from a negative
32-bit offset, instead of having it be sign-extended.
Also happens with stack pointer subtractions, on at least x64 and AArch64.

-- 
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/20160828/de9df252/attachment.html>