[llvm] r213892 - [X86] Optimize stackmap shadows on X86.
echristo at gmail.com
Thu Jul 24 16:26:39 PDT 2014
On Thu, Jul 24, 2014 at 4:06 PM, Lang Hames <lhames at gmail.com> wrote:
> They're essentially unrelated. I moved the other Lower* methods
> (LowerSTACKMAP, LowerPATCHPOINT) in to X86AsmPrinter, and did the same for
> LowerTlsAddr for consistency. I forgot to back this cosmetic change to
> LowerTlsAddr out before I committed. My apologies for that.
> There is no functional change due to the signature change of LowerTlsAddr,
> so it can be safely reverted if you would prefer. The changes to the body of
> LowerTlsAddr should remain if possible though - they are valuable (though
> not strictly necessary) for the optimization.
FWIW I like the changes here, just would have preferred to see them
separately. I really dislike the "static function that passes half of
the class member variables as arguments" "pattern".
> - Lang.
> On Thu, Jul 24, 2014 at 2:36 PM, Philip Reames <listmail at philipreames.com>
>> Was there a review thread for this I missed? I would have liked to see
>> this before it went in.
>> On 07/24/2014 01:40 PM, Lang Hames wrote:
>>> case X86::TLS_base_addr64:
>>> - return LowerTlsAddr(OutStreamer, MCInstLowering, *MI,
>>> + return LowerTlsAddr(MCInstLowering, *MI);
>> This looks like it might be an accidental unrelated change?
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
More information about the llvm-commits