2013-07-25 11:11:15 +02:00
|
|
|
; Test LOCR and LOCGR.
|
|
|
|
|
;
|
|
|
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z196 | FileCheck %s
|
|
|
|
|
|
|
|
|
|
; Test LOCR.
|
|
|
|
|
define i32 @f1(i32 %a, i32 %b, i32 %limit) {
|
|
|
|
|
; CHECK-LABEL: f1:
|
|
|
|
|
; CHECK: clfi %r4, 42
|
[SystemZ] Be more careful about inverting CC masks
System z branches have a mask to select which of the 4 CC values should
cause the branch to be taken. We can invert a branch by inverting the mask.
However, not all instructions can produce all 4 CC values, so inverting
the branch like this can lead to some oddities. For example, integer
comparisons only produce a CC of 0 (equal), 1 (less) or 2 (greater).
If an integer EQ is reversed to NE before instruction selection,
the branch will test for 1 or 2. If instead the branch is reversed
after instruction selection (by inverting the mask), it will test for
1, 2 or 3. Both are correct, but the second isn't really canonical.
This patch therefore keeps track of which CC values are possible
and uses this when inverting a mask.
Although this is mostly cosmestic, it fixes undefined behavior
for the CIJNLH in branch-08.ll. Another fix would have been
to mask out bit 0 when generating the fused compare and branch,
but the point of this patch is that we shouldn't need to do that
in the first place.
The patch also makes it easier to reuse CC results from other instructions.
llvm-svn: 187495
2013-07-31 14:30:20 +02:00
|
|
|
; CHECK: locrhe %r2, %r3
|
2013-07-25 11:11:15 +02:00
|
|
|
; CHECK: br %r14
|
|
|
|
|
%cond = icmp ult i32 %limit, 42
|
|
|
|
|
%res = select i1 %cond, i32 %a, i32 %b
|
|
|
|
|
ret i32 %res
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Test LOCGR.
|
|
|
|
|
define i64 @f2(i64 %a, i64 %b, i64 %limit) {
|
|
|
|
|
; CHECK-LABEL: f2:
|
|
|
|
|
; CHECK: clgfi %r4, 42
|
[SystemZ] Be more careful about inverting CC masks
System z branches have a mask to select which of the 4 CC values should
cause the branch to be taken. We can invert a branch by inverting the mask.
However, not all instructions can produce all 4 CC values, so inverting
the branch like this can lead to some oddities. For example, integer
comparisons only produce a CC of 0 (equal), 1 (less) or 2 (greater).
If an integer EQ is reversed to NE before instruction selection,
the branch will test for 1 or 2. If instead the branch is reversed
after instruction selection (by inverting the mask), it will test for
1, 2 or 3. Both are correct, but the second isn't really canonical.
This patch therefore keeps track of which CC values are possible
and uses this when inverting a mask.
Although this is mostly cosmestic, it fixes undefined behavior
for the CIJNLH in branch-08.ll. Another fix would have been
to mask out bit 0 when generating the fused compare and branch,
but the point of this patch is that we shouldn't need to do that
in the first place.
The patch also makes it easier to reuse CC results from other instructions.
llvm-svn: 187495
2013-07-31 14:30:20 +02:00
|
|
|
; CHECK: locgrhe %r2, %r3
|
2013-07-25 11:11:15 +02:00
|
|
|
; CHECK: br %r14
|
|
|
|
|
%cond = icmp ult i64 %limit, 42
|
|
|
|
|
%res = select i1 %cond, i64 %a, i64 %b
|
|
|
|
|
ret i64 %res
|
|
|
|
|
}
|
[SystemZ] Be more careful about inverting CC masks
System z branches have a mask to select which of the 4 CC values should
cause the branch to be taken. We can invert a branch by inverting the mask.
However, not all instructions can produce all 4 CC values, so inverting
the branch like this can lead to some oddities. For example, integer
comparisons only produce a CC of 0 (equal), 1 (less) or 2 (greater).
If an integer EQ is reversed to NE before instruction selection,
the branch will test for 1 or 2. If instead the branch is reversed
after instruction selection (by inverting the mask), it will test for
1, 2 or 3. Both are correct, but the second isn't really canonical.
This patch therefore keeps track of which CC values are possible
and uses this when inverting a mask.
Although this is mostly cosmestic, it fixes undefined behavior
for the CIJNLH in branch-08.ll. Another fix would have been
to mask out bit 0 when generating the fused compare and branch,
but the point of this patch is that we shouldn't need to do that
in the first place.
The patch also makes it easier to reuse CC results from other instructions.
llvm-svn: 187495
2013-07-31 14:30:20 +02:00
|
|
|
|
|
|
|
|
; Test LOCR in a case that could use COMPARE AND BRANCH. We prefer using
|
|
|
|
|
; LOCR if possible.
|
|
|
|
|
define i32 @f3(i32 %a, i32 %b, i32 %limit) {
|
|
|
|
|
; CHECK-LABEL: f3:
|
|
|
|
|
; CHECK: chi %r4, 42
|
|
|
|
|
; CHECK: locrlh %r2, %r3
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%cond = icmp eq i32 %limit, 42
|
|
|
|
|
%res = select i1 %cond, i32 %a, i32 %b
|
|
|
|
|
ret i32 %res
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; ...and again for LOCGR.
|
|
|
|
|
define i64 @f4(i64 %a, i64 %b, i64 %limit) {
|
|
|
|
|
; CHECK-LABEL: f4:
|
|
|
|
|
; CHECK: cghi %r4, 42
|
|
|
|
|
; CHECK: locgrlh %r2, %r3
|
|
|
|
|
; CHECK: br %r14
|
|
|
|
|
%cond = icmp eq i64 %limit, 42
|
|
|
|
|
%res = select i1 %cond, i64 %a, i64 %b
|
|
|
|
|
ret i64 %res
|
|
|
|
|
}
|