RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-26 04:32:44 +01:00
Code Issues Projects Releases Wiki Activity
fca643847d
llvm-mirror/test/CodeGen/Hexagon/swp-carried-dep2.mir

71 lines
2.0 KiB
Plaintext
Raw Normal View History

[ModuloSchedule] Peel out prologs and epilogs, generate actual code Summary: This extends the PeelingModuloScheduleExpander to generate prolog and epilog code, and correctly stitch uses through the prolog, kernel, epilog DAG. The key concept in this patch is to ensure that all transforms are *local*; only a function of a block and its immediate predecessor and successor. By defining the problem in this way we can inductively rewrite the entire DAG using only local knowledge that is easy to reason about. For example, we assume that all prologs and epilogs are near-perfect clones of the steady-state kernel. This means that if a block has an instruction that is predicated out, we can redirect all users of that instruction to that equivalent instruction in our immediate predecessor. As all blocks are clones, every instruction must have an equivalent in every other block. Similarly we can make the assumption by construction that if a value defined in a block is used outside that block, the only possible user is its immediate successors. We maintain this even for values that are used outside the loop by creating a limited form of LCSSA. This code isn't small, but it isn't complex. Enabled a bunch of testing from Hexagon. There are a couple of tests not enabled yet; I'm about 80% sure there isn't buggy codegen but the tests are checking for patterns that we don't produce. Those still need a bit more investigation. In the meantime we (Google) are happy with the code produced by this on our downstream SMS implementation, and believe it generates correct code. Subscribers: mgorny, hiraditya, jsji, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D68205 llvm-svn: 373462
2019-10-02 14:46:44 +02:00
# RUN: llc -mtriple=hexagon -run-pass pipeliner -debug-only=pipeliner %s -o /dev/null 2>&1 -pipeliner-experimental-cg=true | FileCheck %s
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
# REQUIRES: asserts
# Test that the loop carried dependence check correctly identifies a recurrence
# when the loop variable decreases and the array index offset is negative.
# CHECK: Rec NodeSet
# CHECK: Rec NodeSet
# CHECK: SU(3)
# CHECK: SU(4)
# CHECK: SU(5)
--- |
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
define void @test() {
b0:
br label %b3
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
b3:
%lsr.iv = phi [9 x i32]* [ %0, %b3 ], [ undef, %b0 ]
%v0 = phi i32 [ %v8, %b3 ], [ 7, %b0 ]
%v1 = phi i32 [ %v6, %b3 ], [ undef, %b0 ]
%v2 = phi i32 [ %v1, %b3 ], [ undef, %b0 ]
%lsr.iv1 = bitcast [9 x i32]* %lsr.iv to i32*
%cgep = getelementptr i32, i32* %lsr.iv1, i32 -2
%v6 = load i32, i32* %cgep, align 4
%v7 = tail call i32 @llvm.hexagon.A2.subsat(i32 %v2, i32 %v6)
store i32 %v7, i32* %lsr.iv1, align 4
%v8 = add i32 %v0, -1
%cgep3 = getelementptr [9 x i32], [9 x i32]* %lsr.iv, i32 0, i32 -1
%0 = bitcast i32* %cgep3 to [9 x i32]*
%v9 = icmp sgt i32 %v8, 1
br i1 %v9, label %b3, label %b4
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
b4:
unreachable
}
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
declare i32 @llvm.hexagon.A2.subsat(i32, i32) #0
declare void @llvm.stackprotector(i8*, i8**) #1
...
---
name: test
tracksRegLiveness: true
body: |
bb.0:
successors: %bb.1
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
%10:intregs = IMPLICIT_DEF
%11:intregs = IMPLICIT_DEF
J2_loop0i %bb.1, 6, implicit-def $lc0, implicit-def $sa0, implicit-def $usr
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
bb.1 (address-taken):
successors: %bb.1, %bb.2
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
%0:intregs = PHI %11, %bb.0, %6, %bb.1
%2:intregs = PHI %10, %bb.0, %4, %bb.1
%3:intregs = PHI %10, %bb.0, %2, %bb.1
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
%4:intregs = L2_loadri_io %0, -8 :: (load (s32) from %ir.cgep)
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
%12:intregs = A2_subsat %3, %4, implicit-def dead $usr_ovf
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
S2_storeri_io %0, 0, %12 :: (store (s32) into %ir.lsr.iv1)
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
%6:intregs = A2_addi %0, -4
ENDLOOP0 %bb.1, implicit-def $pc, implicit-def $lc0, implicit $sa0, implicit $lc0
J2_jump %bb.2, implicit-def dead $pc
CodeGen: Print/parse LLTs in MachineMemOperands This will currently accept the old number of bytes syntax, and convert it to a scalar. This should be removed in the near future (I think I converted all of the tests already, but likely missed a few). Not sure what the exact syntax and policy should be. We can continue printing the number of bytes for non-generic instructions to avoid test churn and only allow non-scalar types for generic instructions. This will currently print the LLT in parentheses, but accept parsing the existing integers and implicitly converting to scalar. The parentheses are a bit ugly, but the parser logic seems unable to deal without either parentheses or some keyword to indicate the start of a type.
2021-05-20 04:25:51 +02:00
[Pipeliner] Fix incorrect loop carried dependence calculation The isLoopCarriedDep function does not correctly compute loop carried dependences when the array index offset is negative or the stride is smallar than the access size. Patch by Denis Antrushin. Differential Revision: https://reviews.llvm.org/D60135 llvm-svn: 358233
2019-04-11 23:57:51 +02:00
bb.2:
...
Reference in New Issue Copy Permalink