[PATCH] D45653: [X86] Enable sibling-call optimization for functions returning structs

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added inline comments.


================
Comment at: test/CodeGen/X86/sibcall.ll:4
 ; RUN: llc < %s -mtriple=x86_64-linux -mcpu=core2 -mattr=+sse2 | FileCheck %s --check-prefix=X64
 ; RUN: llc < %s -mtriple=x86_64-linux-gnux32 -mcpu=core2 -mattr=+sse2  | FileCheck %s --check-prefix=X32ABI
 
----------------
This is very minor, but shouldn't the prefixes really be X86 (instead of X32), X64 and X32 (instead of X32ABI)?


================
Comment at: test/CodeGen/X86/sibcall.ll:615
-; 64-LABEL: t21_sret_to_sret:
-; 64: jmp {{_?}}t21_f_sret
 ; X32-LABEL: t21_sret_to_sret:
----------------
Old prefixes - this can just be removed as an NFC cleanup


================
Comment at: test/CodeGen/X86/sibcall.ll:654
-; 64: callq {{_?}}t21_f_non_sret
-; 64: retq
 ; X32-LABEL: t21_sret_to_non_sret:
----------------
Old prefixes - this can just be removed as an NFC cleanup


================
Comment at: test/CodeGen/X86/sibcall.ll:691
-; 64-LABEL: t21_non_sret_to_sret:
-; 64: jmp {{_?}}t21_f_sret
 ; X32-LABEL: t21_non_sret_to_sret:
----------------
Old prefixes - this can just be removed as an NFC cleanup


https://reviews.llvm.org/D45653