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/AArch64/GlobalISel/legalize-combines.mir

105 lines
2.4 KiB
Plaintext
Raw Normal View History

MIRTests: Remove unnecessary 2>&1 redirection llc mir output goes to stdout nowadays, so the 2>&1 is not necessary anymore for most tests. llvm-svn: 295859
2017-02-23 02:47:41 +08:00
# RUN: llc -O0 -run-pass=legalizer -global-isel %s -o - | FileCheck %s
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
--- |
target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
[GlobalISel] Get the AArch64 tests to work on Linux Mostly this just means changing the triple from aarch64-apple-ios to the generic aarch64--. Only one test needs more significant changes, but GlobalISel already does the right thing so it's ok to just change the checks. Differential Revision: https://reviews.llvm.org/D25532 llvm-svn: 284223
2016-10-14 18:19:40 +08:00
target triple = "aarch64--"
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
define void @test_combines_2() { ret void }
define void @test_combines_3() { ret void }
define void @test_combines_4() { ret void }
define void @test_combines_5() { ret void }
[GlobalISel] Replace all combined G_EXTRACT uses. Iterating on the use-list we're modifying doesn't work: after the first iteration, the use-list iterator will point to a MachineOperand referencing the new register. This caused us to skip the other uses to replace. Instead, use MRI.replaceRegWith(), which accounts for this behavior. llvm-svn: 296551
2017-03-01 08:43:39 +08:00
define void @test_combines_6() { ret void }
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
...
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
---
name: test_combines_2
body: |
bb.0:
liveins: %w0
%0:_(s32) = COPY %w0
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
; Similarly, here the types don't match.
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
; CHECK-LABEL: name: test_combines_2
GlobalISel: remove G_SEQUENCE instruction. It was trying to do too many things. The basic lumping together of values for legalization purposes is now handled by G_MERGE_VALUES. More complex things involving gaps and odd sizes are handled by G_INSERT sequences. llvm-svn: 306120
2017-06-24 00:15:55 +08:00
; CHECK: %2(s64) = G_MERGE_VALUES %0(s32), %1(s32)
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
; CHECK: %3(s1) = G_EXTRACT %2(s64), 0
; CHECK: %4(s64) = G_EXTRACT %2(s64), 0
%1:_(s32) = G_ADD %0, %0
GlobalISel: remove G_SEQUENCE instruction. It was trying to do too many things. The basic lumping together of values for legalization purposes is now handled by G_MERGE_VALUES. More complex things involving gaps and odd sizes are handled by G_INSERT sequences. llvm-svn: 306120
2017-06-24 00:15:55 +08:00
%2:_(s64) = G_MERGE_VALUES %0, %1
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
%3:_(s1) = G_EXTRACT %2, 0
%4:_(s64) = G_EXTRACT %2, 0
...
---
name: test_combines_3
body: |
bb.0:
liveins: %w0
%0:_(s32) = COPY %w0
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
; CHECK-LABEL: name: test_combines_3
; CHECK: %1(s32) = G_ADD %0, %0
; CHECK-NOT: G_SEQUENCE
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
; CHECK-NOT: G_EXTRACT
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
; CHECK: %5(s32) = G_ADD %0, %1
%1:_(s32) = G_ADD %0, %0
GlobalISel: remove G_SEQUENCE instruction. It was trying to do too many things. The basic lumping together of values for legalization purposes is now handled by G_MERGE_VALUES. More complex things involving gaps and odd sizes are handled by G_INSERT sequences. llvm-svn: 306120
2017-06-24 00:15:55 +08:00
%2:_(s64) = G_MERGE_VALUES %0, %1
%3:_(s32), %4:_(s32) = G_UNMERGE_VALUES %2
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
%5:_(s32) = G_ADD %3, %4
...
---
name: test_combines_4
body: |
bb.0:
liveins: %x0
%0:_(s64) = COPY %x0
; CHECK-LABEL: name: test_combines_4
AArch64: legalize G_EXTRACT operations. This is the dual problem to legalizing G_INSERTs so most of the code and testing was cribbed from there. llvm-svn: 306328
2017-06-27 04:34:13 +08:00
; CHECK: %2(s64) = COPY %0(s64)
; CHECK: %3(s64) = G_ADD %2, %2
GlobalISel: remove G_SEQUENCE instruction. It was trying to do too many things. The basic lumping together of values for legalization purposes is now handled by G_MERGE_VALUES. More complex things involving gaps and odd sizes are handled by G_INSERT sequences. llvm-svn: 306120
2017-06-24 00:15:55 +08:00
%1:_(s128) = G_MERGE_VALUES %0, %0
AArch64: legalize G_EXTRACT operations. This is the dual problem to legalizing G_INSERTs so most of the code and testing was cribbed from there. llvm-svn: 306328
2017-06-27 04:34:13 +08:00
%2:_(s64) = G_EXTRACT %1, 0
%3:_(s64) = G_ADD %2, %2
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
...
---
name: test_combines_5
body: |
bb.0:
liveins: %w0
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
%0:_(s32) = COPY %w0
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
; CHECK-LABEL: name: test_combines_5
GlobalISel: remove G_SEQUENCE instruction. It was trying to do too many things. The basic lumping together of values for legalization purposes is now handled by G_MERGE_VALUES. More complex things involving gaps and odd sizes are handled by G_INSERT sequences. llvm-svn: 306120
2017-06-24 00:15:55 +08:00
; CHECK-NOT: G_MERGE_VALUES
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
; CHECK-NOT: G_EXTRACT
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
; CHECK: %5(s32) = G_ADD %0, %1
%1:_(s32) = G_ADD %0, %0
GlobalISel: remove G_SEQUENCE instruction. It was trying to do too many things. The basic lumping together of values for legalization purposes is now handled by G_MERGE_VALUES. More complex things involving gaps and odd sizes are handled by G_INSERT sequences. llvm-svn: 306120
2017-06-24 00:15:55 +08:00
%2:_(s64) = G_MERGE_VALUES %0, %1
%3:_(s32), %4:_(s32) = G_UNMERGE_VALUES %2
[AArch64][GlobalISel] Split legalizer combine tests. NFC. Big functions with large vreg # are quite unwieldy to update. This test also relied on legal s8 operations which we're considering removing. Change it to have one function per test (it does increase boilerplate, but makes the core hopefully more readable and maintanable), and use 100% legal operations throughout. llvm-svn: 292551
2017-01-20 08:30:06 +08:00
%5:_(s32) = G_ADD %3, %4
GlobalISel: combine extracts & sequences created for legalization Legalization ends up creating many G_SEQUENCE/G_EXTRACT pairs which leads to inefficient codegen (even for -O0), so add a quick pass over the function to remove them again. llvm-svn: 280155
2016-08-31 04:51:25 +08:00
...
[GlobalISel] Replace all combined G_EXTRACT uses. Iterating on the use-list we're modifying doesn't work: after the first iteration, the use-list iterator will point to a MachineOperand referencing the new register. This caused us to skip the other uses to replace. Instead, use MRI.replaceRegWith(), which accounts for this behavior. llvm-svn: 296551
2017-03-01 08:43:39 +08:00
---
name: test_combines_6
body: |
bb.0:
liveins: %w0
; CHECK-LABEL: name: test_combines_6
; CHECK: %0(s32) = COPY %w0
%0:_(s32) = COPY %w0
; Check that we replace all the uses of a G_EXTRACT.
GlobalISel: remove G_SEQUENCE instruction. It was trying to do too many things. The basic lumping together of values for legalization purposes is now handled by G_MERGE_VALUES. More complex things involving gaps and odd sizes are handled by G_INSERT sequences. llvm-svn: 306120
2017-06-24 00:15:55 +08:00
; CHECK-NOT: G_MERGE_VALUES
[GlobalISel] Replace all combined G_EXTRACT uses. Iterating on the use-list we're modifying doesn't work: after the first iteration, the use-list iterator will point to a MachineOperand referencing the new register. This caused us to skip the other uses to replace. Instead, use MRI.replaceRegWith(), which accounts for this behavior. llvm-svn: 296551
2017-03-01 08:43:39 +08:00
; CHECK-NOT: G_EXTRACT
; CHECK: %3(s32) = G_MUL %0, %0
; CHECK: %4(s32) = G_ADD %0, %3
GlobalISel: remove G_SEQUENCE instruction. It was trying to do too many things. The basic lumping together of values for legalization purposes is now handled by G_MERGE_VALUES. More complex things involving gaps and odd sizes are handled by G_INSERT sequences. llvm-svn: 306120
2017-06-24 00:15:55 +08:00
%1:_(s32) = G_MERGE_VALUES %0
%2:_(s32) = G_UNMERGE_VALUES %1
[GlobalISel] Replace all combined G_EXTRACT uses. Iterating on the use-list we're modifying doesn't work: after the first iteration, the use-list iterator will point to a MachineOperand referencing the new register. This caused us to skip the other uses to replace. Instead, use MRI.replaceRegWith(), which accounts for this behavior. llvm-svn: 296551
2017-03-01 08:43:39 +08:00
%3:_(s32) = G_MUL %2, %2
%4:_(s32) = G_ADD %2, %3
...