Files
clang-p2996/llvm/test/CodeGen/SystemZ/cond-move-04.mir
Jonas Paulsson 16b7cc69ef [SystemZ] Eliminate call sequence instructions early. (#77812)
On SystemZ, the outgoing argument area which is big enough for all calls
in the function is created once during the prolog, as opposed to
adjusting the stack around each call. The call-sequence instructions are
therefore not really useful any more than to compute the maximum call
frame size, which has so far been done by PEI, but can just as well be
done at an earlier point.

This patch removes the mapping of the CallFrameSetupOpcode and
CallFrameDestroyOpcode and instead computes the MaxCallFrameSize
directly after instruction selection and then removes the ADJCALLSTACK
pseudos. This removes the confusing pseudos and also avoids the problem
of having to keep the call frame size accurate when creating new MBBs.

This fixes #76618 which exposed the need to maintain the call frame size
when splitting blocks (which was not done).
2024-03-28 18:26:38 +01:00

75 lines
1.9 KiB
YAML

# RUN: llc -mtriple=s390x-linux-gnu -mcpu=z13 -start-before=greedy %s -o - \
# RUN: | FileCheck %s
#
# Test that regalloc manages (via regalloc hints) to avoid a LOCRMux jump
# sequence expansion.
--- |
declare ptr @foo(ptr, i32 signext, i32 signext) local_unnamed_addr
define ptr @fun(ptr returned) {
br label %2
; <label>:2: ; preds = %6, %1
%3 = zext i16 undef to i32
switch i32 %3, label %4 [
i32 15, label %6
i32 125, label %5
]
; <label>:4: ; preds = %2
br label %6
; <label>:5: ; preds = %2
br label %6
; <label>:6: ; preds = %5, %4, %2
%7 = phi i32 [ 4, %2 ], [ undef, %4 ], [ 10, %5 ]
%8 = call ptr @foo(ptr undef, i32 signext undef, i32 signext %7)
br label %2
}
...
# CHECK: locr
# CHECK-NOT: risblg
---
name: fun
alignment: 4
tracksRegLiveness: true
registers:
- { id: 0, class: gr32bit }
- { id: 1, class: gr64bit }
- { id: 2, class: grx32bit }
- { id: 3, class: grx32bit }
- { id: 4, class: grx32bit }
- { id: 5, class: grx32bit }
- { id: 6, class: grx32bit }
- { id: 7, class: gr64bit }
- { id: 8, class: gr64bit }
- { id: 9, class: gr64bit }
- { id: 10, class: gr64bit }
- { id: 11, class: gr32bit }
frameInfo:
adjustsStack: true
hasCalls: true
body: |
bb.0 (%ir-block.1):
%3 = LHIMux 0
%2 = LHIMux 4
%5 = LHIMux 10
bb.1 (%ir-block.2):
CHIMux %3, 0, implicit-def $cc
%0 = LOCRMux undef %0, %5, 14, 6, implicit $cc
%0 = LOCRMux %0, %2, 14, 6, implicit killed $cc
%7 = LGFR %0
$r3d = LGHI 0
$r4d = COPY %7
CallBRASL @foo, undef $r2d, killed $r3d, killed $r4d, csr_systemz_elf, implicit-def dead $r14d, implicit-def dead $cc, implicit-def dead $r2d
J %bb.1
...