[PATCH v2] [X86] Support 'V' register operand modifier

[David Woodhouse via llvm-commits]

On Thu, 2018-02-08 at 11:27 +0000, Chandler Carruth wrote:
> This looks pretty good to me, but Craig should probably have a look
> before it lands.
> 
> If you don't have commit access, you may need to send it as an
> attachment -- my email client at least destroys the patch file when
> in-line so that I can't apply it and commit it easily for you.

I have had commit access in the past. I'll see if git-svn or I can
remember my password ;)

The other 32-bit build failure looks slightly more complex. It looks
like there's a switch(sizeof(arg)) which is obviously (to me but not
LLVM) constant, and the inline asm for the case which can never
actually happen, would have bogus constraints because of the size.
GCC optimises it away before ever caring, and LLVM doesn't.

That was very late last night though; I'll take another look today.

I'll also look at the per-function attribute to disable retpoline.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 5213 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180208/9eadfa7d/attachment.bin>