[LLVMdev] data layout string questions.

[Reid Kleckner]

Looks like it's dead:

void DataLayout::parseSpecifier(StringRef Desc) {
  ...
    switch (Specifier) {
    case 's':
      // Ignored for backward compatibility.
      // FIXME: remove this on LLVM 4.0.
      break;
    case 'E':
      BigEndian = true;
      break;
    case 'e':
      BigEndian = false;
      break;

On Fri, Oct 31, 2014 at 3:42 PM, reed kotler <rkotler at mips.com> wrote:

>
> in the language reference manual there is no definition for the prefix "s"
> (only for large "S" which has to do with the stack).
>
> BTW for x86, I get
> s0:64:64
>
> Not sure what this means.
>
> The form:
> i<size>:<abi>:<pref>
>
> Seems to mean that llvm type i<size> and then abi == the number of bits.
>
> what is the acronym here for abi or does this mean that the ABI specified
> size?
>
> for "a", the definition is
>
> a:<abi>:<pref>
>
> But if I compile a program with x86 on a 64 bit machine I get:
>
> a0:0:64
>
> So this means that as long as the number of bits is greater than 0, the
> alignment is 64.
> 0 is like a wildcard in this case and the <abi> field is not relevant?
>
> In any case, the syntax is incorrect since there is no size field.
>
> BTW, it seems that S0 means the natural alignment is unspecified, which
> not in the document.
> Also, the absence of S means unspecified also.
>
> Tia.
>
> Reed
>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20141031/427a4b1b/attachment.html>