Support for HiPE-compatible code emission

Yiannis Tsiouris gtsiour at softlab.ntua.gr
Thu Feb 14 12:40:50 PST 2013


Hi Nadav,

On 02/14/2013 07:16 PM, Nadav Rotem wrote:
> Please put all of the "HiPE" logic in one place. Most people don't know what HiPE is, and would prefer to skip it when they read the code. Write something like:
>
> // Erlang stuff.
> if (CallingConvention == CallingConv::HiPE) {
>   ...
> }
>
>   if (Is64Bit)
>     return Primary ? X86::R11 : X86::R12;
>   ...
>    
That was how I had structured it in the beginning! :-) I changed it to 
match your suggestion.

> +/// GetScratchRegister - Get a temp register for performing work in the
> +/// segmented stack and/or HiPE stack prologue. Depending on platform and the
> +/// properties of the function either one or two registers will be needed. Set
> +/// primary to true for the first register, false for the second.
>
> Please remove " and/or HiPE" from the comment. This information does not help most people understand what this function does.
>    
I changed this to:

/// GetScratchRegister - Get a temp register for performing work in the
/// segmented stack and the Erlang/HiPE stack prologue. Depending on 
platform
/// and the properties of the function either one or two registers will be
/// needed. Set primary to true for the first register, false for the 
second.

If this is not clear enough, just tell me and I'll change it as you wish.

> In the test case "X86/hipe-prologue.ll"  Please add comment that describes what HIPE is.
>    
ACK.

Re-sending the whole patch for completeness.

Best,
yiannis

-- 
Yiannis Tsiouris
Ph.D. student,
Software Engineering Laboratory,
National Technical University of Athens
WWW: http://www.softlab.ntua.gr/~gtsiour

-------------- next part --------------
A non-text attachment was scrubbed...
Name: hipe-prologue-v4.patch
Type: text/x-patch
Size: 12015 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130214/8e388f66/attachment.bin>


More information about the llvm-commits mailing list