maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/X86/anyext.ll

19 lines
426 B
LLVM
Raw Normal View History

As Dan pointed out, movzbl, movsbl, and friends are nicer than their alias (movzx/movsx) because they give more information. Revert that part of the patch. llvm-svn: 129498
2011-04-14 09:46:37 +08:00
; RUN: llc < %s -march=x86-64 | grep movzbl | count 2
Don't use INSERT_SUBREG to model anyext operations on x86-64, as it leads to partial-register definitions. To help avoid redundant zero-extensions, also teach the h-register matching patterns that use movzbl to match anyext as well as zext. llvm-svn: 80099
2009-08-26 22:59:13 +08:00
As Dan pointed out, movzbl, movsbl, and friends are nicer than their alias (movzx/movsx) because they give more information. Revert that part of the patch. llvm-svn: 129498
2011-04-14 09:46:37 +08:00
; Use movzbl to avoid partial-register updates.
Don't use INSERT_SUBREG to model anyext operations on x86-64, as it leads to partial-register definitions. To help avoid redundant zero-extensions, also teach the h-register matching patterns that use movzbl to match anyext as well as zext. llvm-svn: 80099
2009-08-26 22:59:13 +08:00
define i32 @foo(i32 %p, i8 zeroext %x) nounwind {
%q = trunc i32 %p to i8
%r = udiv i8 %q, %x
%s = zext i8 %r to i32
%t = and i32 %s, 1
ret i32 %t
}
define i32 @bar(i32 %p, i16 zeroext %x) nounwind {
%q = trunc i32 %p to i16
%r = udiv i16 %q, %x
%s = zext i16 %r to i32
%t = and i32 %s, 1
ret i32 %t
}