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/PowerPC/memCmpUsedInZeroEqualityCom...

222 lines
7.9 KiB
LLVM
Raw Normal View History

[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
[PowerPC] Recommit r314244 with refactoring and off by default This re-commits everything that was pulled in r314244. The transformation is off by default (patch to enable it to follow). The code is refactored to have a single entry-point and provide fine-grained control over patterns that it selects. This patch also fixes the bugs in the original code. Everything that failed with the original patch has been re-tested with this patch (with the transformation turned on). So the patch to turn this on is soon to follow. Differential Revision: https://reviews.llvm.org/D38575 llvm-svn: 319434
2017-11-30 21:39:10 +08:00
; RUN: llc -ppc-gpr-icmps=all -verify-machineinstrs -mcpu=pwr8 < %s | FileCheck %s
[PPC] Inline expansion of memcmp This patch does an inline expansion of memcmp. It changes the memcmp library call into an inline expansion when the size is known at compile time and is under a target specified threshold. This expansion is implemented in CodeGenPrepare and expands into straight line code. The target specifies a maximum load size and the expansion works by using this size to load the two sources, compare, and exit early if a difference is found. It also has a special case when the memcmp result is used in a compare to zero equality. Differential Revision: https://reviews.llvm.org/D28637 llvm-svn: 304313
2017-06-01 01:12:38 +08:00
target datalayout = "e-m:e-i64:64-n32:64"
target triple = "powerpc64le-unknown-linux-gnu"
@zeroEqualityTest01.buffer1 = private unnamed_addr constant [3 x i32] [i32 1, i32 2, i32 4], align 4
@zeroEqualityTest01.buffer2 = private unnamed_addr constant [3 x i32] [i32 1, i32 2, i32 3], align 4
@zeroEqualityTest02.buffer1 = private unnamed_addr constant [4 x i32] [i32 4, i32 0, i32 0, i32 0], align 4
@zeroEqualityTest02.buffer2 = private unnamed_addr constant [4 x i32] [i32 3, i32 0, i32 0, i32 0], align 4
@zeroEqualityTest03.buffer1 = private unnamed_addr constant [4 x i32] [i32 0, i32 0, i32 0, i32 3], align 4
@zeroEqualityTest03.buffer2 = private unnamed_addr constant [4 x i32] [i32 0, i32 0, i32 0, i32 4], align 4
@zeroEqualityTest04.buffer1 = private unnamed_addr constant [15 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14], align 4
@zeroEqualityTest04.buffer2 = private unnamed_addr constant [15 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 13], align 4
declare signext i32 @memcmp(i8* nocapture, i8* nocapture, i64) local_unnamed_addr #1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; Check 4 bytes - requires 1 load for each param.
define signext i32 @zeroEqualityTest02(i8* %x, i8* %y) {
; CHECK-LABEL: zeroEqualityTest02:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK: # %bb.0:
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: lwz 3, 0(3)
; CHECK-NEXT: lwz 4, 0(4)
[CGP / PowerPC] avoid multi-block overhead for simple memcmp expansion The test diff for PowerPC shows we can better optimize if this case is one block. For x86, there's would be a substantial difference if CGP expansion was enabled because branches are assumed cheap and SDAG can't optimize across blocks. Instead of this: _cmp_eq8: movq (%rdi), %rax cmpq (%rsi), %rax je LBB23_1 ## BB#2: ## %res_block movl $1, %ecx jmp LBB23_3 LBB23_1: xorl %ecx, %ecx LBB23_3: ## %endblock xorl %eax, %eax testl %ecx, %ecx sete %al retq We get this: cmp_eq8: movq (%rdi), %rcx xorl %eax, %eax cmpq (%rsi), %rcx sete %al retq And that matches the optimal codegen that we get from the current expansion in SelectionDAGBuilder::visitMemCmpCall(). If this looks right, then I just need to confirm that vector-sized expansion will work from here, and we can enable CGP memcmp() expansion for x86. Ie, we'll bypass the power-of-2 special cases currently optimized in SDAG because we can lower the IR produced here optimally. Differential Revision: https://reviews.llvm.org/D34005 llvm-svn: 304987
2017-06-09 00:53:18 +08:00
; CHECK-NEXT: xor 3, 3, 4
; CHECK-NEXT: cntlzw 3, 3
; CHECK-NEXT: srwi 3, 3, 5
; CHECK-NEXT: xori 3, 3, 1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: blr
%call = tail call signext i32 @memcmp(i8* %x, i8* %y, i64 4)
%not.cmp = icmp ne i32 %call, 0
%. = zext i1 %not.cmp to i32
[PPC] Inline expansion of memcmp This patch does an inline expansion of memcmp. It changes the memcmp library call into an inline expansion when the size is known at compile time and is under a target specified threshold. This expansion is implemented in CodeGenPrepare and expands into straight line code. The target specifies a maximum load size and the expansion works by using this size to load the two sources, compare, and exit early if a difference is found. It also has a special case when the memcmp result is used in a compare to zero equality. Differential Revision: https://reviews.llvm.org/D28637 llvm-svn: 304313
2017-06-01 01:12:38 +08:00
ret i32 %.
}
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; Check 16 bytes - requires 2 loads for each param (or use vectors?).
define signext i32 @zeroEqualityTest01(i8* %x, i8* %y) {
; CHECK-LABEL: zeroEqualityTest01:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK: # %bb.0:
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: ld 5, 0(3)
; CHECK-NEXT: ld 6, 0(4)
[CGP / PowerPC] use direct compares if there's only one load per block in memcmp() expansion I'd like to enable CGP memcmp expansion for x86, but the output from CGP would regress the special cases (memcmp(x,y,N) != 0 for N=1,2,4,8,16,32 bytes) that we already handle. I'm not sure if we'll actually be able to produce the optimal code given the block-at-a-time limitation in the DAG. We might have to just avoid those special-cases here in CGP. But regardless of that, I think this is a win for the more general cases. http://rise4fun.com/Alive/cbQ Differential Revision: https://reviews.llvm.org/D33963 llvm-svn: 304849
2017-06-07 08:17:08 +08:00
; CHECK-NEXT: cmpld 5, 6
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: bne 0, .LBB1_2
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK-NEXT: # %bb.1: # %loadbb1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: ld 3, 8(3)
; CHECK-NEXT: ld 4, 8(4)
[CGP / PowerPC] use direct compares if there's only one load per block in memcmp() expansion I'd like to enable CGP memcmp expansion for x86, but the output from CGP would regress the special cases (memcmp(x,y,N) != 0 for N=1,2,4,8,16,32 bytes) that we already handle. I'm not sure if we'll actually be able to produce the optimal code given the block-at-a-time limitation in the DAG. We might have to just avoid those special-cases here in CGP. But regardless of that, I think this is a win for the more general cases. http://rise4fun.com/Alive/cbQ Differential Revision: https://reviews.llvm.org/D33963 llvm-svn: 304849
2017-06-07 08:17:08 +08:00
; CHECK-NEXT: cmpld 3, 4
; CHECK-NEXT: li 3, 0
[PowerPC] Recommit r314244 with refactoring and off by default This re-commits everything that was pulled in r314244. The transformation is off by default (patch to enable it to follow). The code is refactored to have a single entry-point and provide fine-grained control over patterns that it selects. This patch also fixes the bugs in the original code. Everything that failed with the original patch has been re-tested with this patch (with the transformation turned on). So the patch to turn this on is soon to follow. Differential Revision: https://reviews.llvm.org/D38575 llvm-svn: 319434
2017-11-30 21:39:10 +08:00
; CHECK-NEXT: beq 0, .LBB1_3
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: .LBB1_2: # %res_block
; CHECK-NEXT: li 3, 1
[PowerPC] Recommit r314244 with refactoring and off by default This re-commits everything that was pulled in r314244. The transformation is off by default (patch to enable it to follow). The code is refactored to have a single entry-point and provide fine-grained control over patterns that it selects. This patch also fixes the bugs in the original code. Everything that failed with the original patch has been re-tested with this patch (with the transformation turned on). So the patch to turn this on is soon to follow. Differential Revision: https://reviews.llvm.org/D38575 llvm-svn: 319434
2017-11-30 21:39:10 +08:00
; CHECK-NEXT: .LBB1_3: # %endblock
; CHECK-NEXT: clrldi 3, 3, 32
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: blr
%call = tail call signext i32 @memcmp(i8* %x, i8* %y, i64 16)
%not.tobool = icmp ne i32 %call, 0
%. = zext i1 %not.tobool to i32
[PPC] Inline expansion of memcmp This patch does an inline expansion of memcmp. It changes the memcmp library call into an inline expansion when the size is known at compile time and is under a target specified threshold. This expansion is implemented in CodeGenPrepare and expands into straight line code. The target specifies a maximum load size and the expansion works by using this size to load the two sources, compare, and exit early if a difference is found. It also has a special case when the memcmp result is used in a compare to zero equality. Differential Revision: https://reviews.llvm.org/D28637 llvm-svn: 304313
2017-06-01 01:12:38 +08:00
ret i32 %.
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
}
[PPC] Inline expansion of memcmp This patch does an inline expansion of memcmp. It changes the memcmp library call into an inline expansion when the size is known at compile time and is under a target specified threshold. This expansion is implemented in CodeGenPrepare and expands into straight line code. The target specifies a maximum load size and the expansion works by using this size to load the two sources, compare, and exit early if a difference is found. It also has a special case when the memcmp result is used in a compare to zero equality. Differential Revision: https://reviews.llvm.org/D28637 llvm-svn: 304313
2017-06-01 01:12:38 +08:00
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; Check 7 bytes - requires 3 loads for each param.
define signext i32 @zeroEqualityTest03(i8* %x, i8* %y) {
; CHECK-LABEL: zeroEqualityTest03:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK: # %bb.0:
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: lwz 5, 0(3)
; CHECK-NEXT: lwz 6, 0(4)
[CGP] avoid zext/trunc of a memcmp expansion compare This could be viewed as another shortcoming of the DAGCombiner: when both operands of a compare are zexted from the same source type, we should be able to compare the original types. The effect on PowerPC perf is likely unnoticeable, but there's a visible regression for x86 if we feed the suboptimal IR for memcmp expansion to the DAG: _cmp_eq4_zexted_to_i64: movl (%rdi), %ecx movl (%rsi), %edx xorl %eax, %eax cmpq %rdx, %rcx sete %al _cmp_eq4_better: movl (%rdi), %ecx xorl %eax, %eax cmpl (%rsi), %ecx sete %al llvm-svn: 304923
2017-06-08 00:16:45 +08:00
; CHECK-NEXT: cmplw 5, 6
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: bne 0, .LBB2_3
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK-NEXT: # %bb.1: # %loadbb1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: lhz 5, 4(3)
; CHECK-NEXT: lhz 6, 4(4)
[CGP] avoid zext/trunc of a memcmp expansion compare This could be viewed as another shortcoming of the DAGCombiner: when both operands of a compare are zexted from the same source type, we should be able to compare the original types. The effect on PowerPC perf is likely unnoticeable, but there's a visible regression for x86 if we feed the suboptimal IR for memcmp expansion to the DAG: _cmp_eq4_zexted_to_i64: movl (%rdi), %ecx movl (%rsi), %edx xorl %eax, %eax cmpq %rdx, %rcx sete %al _cmp_eq4_better: movl (%rdi), %ecx xorl %eax, %eax cmpl (%rsi), %ecx sete %al llvm-svn: 304923
2017-06-08 00:16:45 +08:00
; CHECK-NEXT: cmplw 5, 6
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: bne 0, .LBB2_3
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK-NEXT: # %bb.2: # %loadbb2
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: lbz 3, 6(3)
; CHECK-NEXT: lbz 4, 6(4)
[CGP] avoid zext/trunc of a memcmp expansion compare This could be viewed as another shortcoming of the DAGCombiner: when both operands of a compare are zexted from the same source type, we should be able to compare the original types. The effect on PowerPC perf is likely unnoticeable, but there's a visible regression for x86 if we feed the suboptimal IR for memcmp expansion to the DAG: _cmp_eq4_zexted_to_i64: movl (%rdi), %ecx movl (%rsi), %edx xorl %eax, %eax cmpq %rdx, %rcx sete %al _cmp_eq4_better: movl (%rdi), %ecx xorl %eax, %eax cmpl (%rsi), %ecx sete %al llvm-svn: 304923
2017-06-08 00:16:45 +08:00
; CHECK-NEXT: cmplw 3, 4
[CGP / PowerPC] use direct compares if there's only one load per block in memcmp() expansion I'd like to enable CGP memcmp expansion for x86, but the output from CGP would regress the special cases (memcmp(x,y,N) != 0 for N=1,2,4,8,16,32 bytes) that we already handle. I'm not sure if we'll actually be able to produce the optimal code given the block-at-a-time limitation in the DAG. We might have to just avoid those special-cases here in CGP. But regardless of that, I think this is a win for the more general cases. http://rise4fun.com/Alive/cbQ Differential Revision: https://reviews.llvm.org/D33963 llvm-svn: 304849
2017-06-07 08:17:08 +08:00
; CHECK-NEXT: li 3, 0
[PowerPC] Recommit r314244 with refactoring and off by default This re-commits everything that was pulled in r314244. The transformation is off by default (patch to enable it to follow). The code is refactored to have a single entry-point and provide fine-grained control over patterns that it selects. This patch also fixes the bugs in the original code. Everything that failed with the original patch has been re-tested with this patch (with the transformation turned on). So the patch to turn this on is soon to follow. Differential Revision: https://reviews.llvm.org/D38575 llvm-svn: 319434
2017-11-30 21:39:10 +08:00
; CHECK-NEXT: beq 0, .LBB2_4
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: .LBB2_3: # %res_block
; CHECK-NEXT: li 3, 1
[PowerPC] Recommit r314244 with refactoring and off by default This re-commits everything that was pulled in r314244. The transformation is off by default (patch to enable it to follow). The code is refactored to have a single entry-point and provide fine-grained control over patterns that it selects. This patch also fixes the bugs in the original code. Everything that failed with the original patch has been re-tested with this patch (with the transformation turned on). So the patch to turn this on is soon to follow. Differential Revision: https://reviews.llvm.org/D38575 llvm-svn: 319434
2017-11-30 21:39:10 +08:00
; CHECK-NEXT: .LBB2_4: # %endblock
; CHECK-NEXT: clrldi 3, 3, 32
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: blr
%call = tail call signext i32 @memcmp(i8* %x, i8* %y, i64 7)
%not.lnot = icmp ne i32 %call, 0
%cond = zext i1 %not.lnot to i32
ret i32 %cond
[PPC] Inline expansion of memcmp This patch does an inline expansion of memcmp. It changes the memcmp library call into an inline expansion when the size is known at compile time and is under a target specified threshold. This expansion is implemented in CodeGenPrepare and expands into straight line code. The target specifies a maximum load size and the expansion works by using this size to load the two sources, compare, and exit early if a difference is found. It also has a special case when the memcmp result is used in a compare to zero equality. Differential Revision: https://reviews.llvm.org/D28637 llvm-svn: 304313
2017-06-01 01:12:38 +08:00
}
; Validate with > 0
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
define signext i32 @zeroEqualityTest04() {
; CHECK-LABEL: zeroEqualityTest04:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK: # %bb.0:
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: addis 3, 2, .LzeroEqualityTest02.buffer1@toc@ha
; CHECK-NEXT: addis 4, 2, .LzeroEqualityTest02.buffer2@toc@ha
; CHECK-NEXT: addi 6, 3, .LzeroEqualityTest02.buffer1@toc@l
; CHECK-NEXT: addi 5, 4, .LzeroEqualityTest02.buffer2@toc@l
; CHECK-NEXT: ldbrx 3, 0, 6
; CHECK-NEXT: ldbrx 4, 0, 5
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: cmpld 3, 4
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: bne 0, .LBB3_2
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK-NEXT: # %bb.1: # %loadbb1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: li 4, 8
; CHECK-NEXT: ldbrx 3, 6, 4
; CHECK-NEXT: ldbrx 4, 5, 4
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: li 5, 0
; CHECK-NEXT: cmpld 3, 4
; CHECK-NEXT: beq 0, .LBB3_3
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: .LBB3_2: # %res_block
; CHECK-NEXT: cmpld 3, 4
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: li 11, 1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: li 12, -1
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: isel 5, 12, 11, 0
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: .LBB3_3: # %endblock
[PowerPC] Recommit r314244 with refactoring and off by default This re-commits everything that was pulled in r314244. The transformation is off by default (patch to enable it to follow). The code is refactored to have a single entry-point and provide fine-grained control over patterns that it selects. This patch also fixes the bugs in the original code. Everything that failed with the original patch has been re-tested with this patch (with the transformation turned on). So the patch to turn this on is soon to follow. Differential Revision: https://reviews.llvm.org/D38575 llvm-svn: 319434
2017-11-30 21:39:10 +08:00
; CHECK-NEXT: extsw 3, 5
; CHECK-NEXT: neg 3, 3
[PowerPC] Eliminate compares - add i32 sext/zext handling for SETLT/SETGT As mentioned in https://reviews.llvm.org/D33718, this simply adds another pattern to the compare elimination sequence and is committed without a differential revision. llvm-svn: 314055
2017-09-23 12:41:34 +08:00
; CHECK-NEXT: rldicl 3, 3, 1, 63
; CHECK-NEXT: xori 3, 3, 1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: blr
[PPC] Inline expansion of memcmp This patch does an inline expansion of memcmp. It changes the memcmp library call into an inline expansion when the size is known at compile time and is under a target specified threshold. This expansion is implemented in CodeGenPrepare and expands into straight line code. The target specifies a maximum load size and the expansion works by using this size to load the two sources, compare, and exit early if a difference is found. It also has a special case when the memcmp result is used in a compare to zero equality. Differential Revision: https://reviews.llvm.org/D28637 llvm-svn: 304313
2017-06-01 01:12:38 +08:00
%call = tail call signext i32 @memcmp(i8* bitcast ([4 x i32]* @zeroEqualityTest02.buffer1 to i8*), i8* bitcast ([4 x i32]* @zeroEqualityTest02.buffer2 to i8*), i64 16)
%not.cmp = icmp slt i32 %call, 1
%. = zext i1 %not.cmp to i32
ret i32 %.
}
; Validate with < 0
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
define signext i32 @zeroEqualityTest05() {
; CHECK-LABEL: zeroEqualityTest05:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK: # %bb.0:
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: addis 3, 2, .LzeroEqualityTest03.buffer1@toc@ha
; CHECK-NEXT: addis 4, 2, .LzeroEqualityTest03.buffer2@toc@ha
; CHECK-NEXT: addi 6, 3, .LzeroEqualityTest03.buffer1@toc@l
; CHECK-NEXT: addi 5, 4, .LzeroEqualityTest03.buffer2@toc@l
; CHECK-NEXT: ldbrx 3, 0, 6
; CHECK-NEXT: ldbrx 4, 0, 5
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: cmpld 3, 4
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: bne 0, .LBB4_2
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK-NEXT: # %bb.1: # %loadbb1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: li 4, 8
; CHECK-NEXT: ldbrx 3, 6, 4
; CHECK-NEXT: ldbrx 4, 5, 4
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: li 5, 0
; CHECK-NEXT: cmpld 3, 4
; CHECK-NEXT: beq 0, .LBB4_3
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: .LBB4_2: # %res_block
; CHECK-NEXT: cmpld 3, 4
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: li 11, 1
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: li 12, -1
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: isel 5, 12, 11, 0
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: .LBB4_3: # %endblock
[CGP] eliminate a sub instruction in memcmp expansion As noted in D34071, there are some IR optimization opportunities that could be handled by normal IR passes if this expansion wasn't happening so late in CGP. Regardless of that, it seems wasteful to knowingly produce suboptimal IR here, so I'm proposing this change: %s = sub i32 %x, %y %r = icmp ne %s, 0 => %r = icmp ne %x, %y Changing the predicate to 'eq' mimics what InstCombine would do, so that's just an efficiency improvement if we decide this expansion should happen sooner. The fact that the PowerPC backend doesn't eliminate the 'subf.' might be something for PPC folks to investigate separately. Differential Revision: https://reviews.llvm.org/D34416 llvm-svn: 306471
2017-06-28 05:46:34 +08:00
; CHECK-NEXT: srwi 3, 5, 31
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: xori 3, 3, 1
; CHECK-NEXT: clrldi 3, 3, 32
; CHECK-NEXT: blr
[PPC] Inline expansion of memcmp This patch does an inline expansion of memcmp. It changes the memcmp library call into an inline expansion when the size is known at compile time and is under a target specified threshold. This expansion is implemented in CodeGenPrepare and expands into straight line code. The target specifies a maximum load size and the expansion works by using this size to load the two sources, compare, and exit early if a difference is found. It also has a special case when the memcmp result is used in a compare to zero equality. Differential Revision: https://reviews.llvm.org/D28637 llvm-svn: 304313
2017-06-01 01:12:38 +08:00
%call = tail call signext i32 @memcmp(i8* bitcast ([4 x i32]* @zeroEqualityTest03.buffer1 to i8*), i8* bitcast ([4 x i32]* @zeroEqualityTest03.buffer2 to i8*), i64 16)
%call.lobit = lshr i32 %call, 31
%call.lobit.not = xor i32 %call.lobit, 1
ret i32 %call.lobit.not
}
; Validate with memcmp()?:
[PowerPC] add memcmp test with one constant operand and equality cmp; NFC llvm-svn: 305131
2017-06-10 07:15:14 +08:00
define signext i32 @equalityFoldTwoConstants() {
; CHECK-LABEL: equalityFoldTwoConstants:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK: # %bb.0: # %endblock
[PowerPC] auto-generate full checks and increase test coverage 3 of the tests were testing exactly the same thing: memcmp(x, y, 16) != 0. I changed that to test 4, 7, and 16 bytes, so we can see how those differ. llvm-svn: 304838
2017-06-07 06:06:07 +08:00
; CHECK-NEXT: li 3, 1
; CHECK-NEXT: blr
[PPC] Inline expansion of memcmp This patch does an inline expansion of memcmp. It changes the memcmp library call into an inline expansion when the size is known at compile time and is under a target specified threshold. This expansion is implemented in CodeGenPrepare and expands into straight line code. The target specifies a maximum load size and the expansion works by using this size to load the two sources, compare, and exit early if a difference is found. It also has a special case when the memcmp result is used in a compare to zero equality. Differential Revision: https://reviews.llvm.org/D28637 llvm-svn: 304313
2017-06-01 01:12:38 +08:00
%call = tail call signext i32 @memcmp(i8* bitcast ([15 x i32]* @zeroEqualityTest04.buffer1 to i8*), i8* bitcast ([15 x i32]* @zeroEqualityTest04.buffer2 to i8*), i64 16)
%not.tobool = icmp eq i32 %call, 0
%cond = zext i1 %not.tobool to i32
ret i32 %cond
}
[PowerPC] add memcmp test with one constant operand and equality cmp; NFC llvm-svn: 305131
2017-06-10 07:15:14 +08:00
define signext i32 @equalityFoldOneConstant(i8* %X) {
; CHECK-LABEL: equalityFoldOneConstant:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK: # %bb.0:
[PowerPC] Recommit r314244 with refactoring and off by default This re-commits everything that was pulled in r314244. The transformation is off by default (patch to enable it to follow). The code is refactored to have a single entry-point and provide fine-grained control over patterns that it selects. This patch also fixes the bugs in the original code. Everything that failed with the original patch has been re-tested with this patch (with the transformation turned on). So the patch to turn this on is soon to follow. Differential Revision: https://reviews.llvm.org/D38575 llvm-svn: 319434
2017-11-30 21:39:10 +08:00
; CHECK-NEXT: li 5, 1
; CHECK-NEXT: ld 4, 0(3)
; CHECK-NEXT: sldi 5, 5, 32
; CHECK-NEXT: cmpld 4, 5
[PowerPC] add memcmp test with one constant operand and equality cmp; NFC llvm-svn: 305131
2017-06-10 07:15:14 +08:00
; CHECK-NEXT: bne 0, .LBB6_2
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK-NEXT: # %bb.1: # %loadbb1
[CGP, PowerPC] try to constant fold before creating loads for memcmp expansion This is the last step needed to avoid regressions for x86 before we flip the switch to allow expansion of the smallest set of memcpy() via CGP. The DAG version checks for constant strings, so we need to do that here too. FWIW, the 2 constant test is not handled by LibCallSimplifier::optimizeMemCmp() because that code is limited to 8-bit constant arrays. LibCallSimplifier will also fail to optimize some 1 constant tests because its alignment requirements are too strict (shouldn't require alignment for a constant operand). Differential Revision: https://reviews.llvm.org/D34071 llvm-svn: 305734
2017-06-20 03:48:35 +08:00
; CHECK-NEXT: li 4, 3
[PowerPC] add memcmp test with one constant operand and equality cmp; NFC llvm-svn: 305131
2017-06-10 07:15:14 +08:00
; CHECK-NEXT: ld 3, 8(3)
[CGP, PowerPC] try to constant fold before creating loads for memcmp expansion This is the last step needed to avoid regressions for x86 before we flip the switch to allow expansion of the smallest set of memcpy() via CGP. The DAG version checks for constant strings, so we need to do that here too. FWIW, the 2 constant test is not handled by LibCallSimplifier::optimizeMemCmp() because that code is limited to 8-bit constant arrays. LibCallSimplifier will also fail to optimize some 1 constant tests because its alignment requirements are too strict (shouldn't require alignment for a constant operand). Differential Revision: https://reviews.llvm.org/D34071 llvm-svn: 305734
2017-06-20 03:48:35 +08:00
; CHECK-NEXT: sldi 4, 4, 32
; CHECK-NEXT: ori 4, 4, 2
; CHECK-NEXT: cmpld 3, 4
[PowerPC] add memcmp test with one constant operand and equality cmp; NFC llvm-svn: 305131
2017-06-10 07:15:14 +08:00
; CHECK-NEXT: li 3, 0
; CHECK-NEXT: beq 0, .LBB6_3
; CHECK-NEXT: .LBB6_2: # %res_block
; CHECK-NEXT: li 3, 1
; CHECK-NEXT: .LBB6_3: # %endblock
; CHECK-NEXT: cntlzw 3, 3
; CHECK-NEXT: srwi 3, 3, 5
; CHECK-NEXT: blr
%call = tail call signext i32 @memcmp(i8* bitcast ([15 x i32]* @zeroEqualityTest04.buffer1 to i8*), i8* %X, i64 16)
%not.tobool = icmp eq i32 %call, 0
%cond = zext i1 %not.tobool to i32
ret i32 %cond
}
[PowerPC] add memcmp test with nobuiltin attr; NFC In SDAG, we don't expand libcalls with a nobuiltin attribute. It's not clear if that's correct from the existing code comment: "Don't do the check if marked as nobuiltin for some reason." ...adding a test here either way to show that there is currently a different behavior implemented in the CGP-based expansion. llvm-svn: 304991
2017-06-09 01:09:18 +08:00
define i1 @length2_eq_nobuiltin_attr(i8* %X, i8* %Y) {
; CHECK-LABEL: length2_eq_nobuiltin_attr:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; CHECK: # %bb.0:
[CGP] don't expand a memcmp with nobuiltin attribute This matches the behavior used in the SDAG when expanding memcmp. For reference, we're intentionally treating the earlier fortified call transforms differently after: https://bugs.llvm.org/show_bug.cgi?id=23093 https://reviews.llvm.org/rL233776 One motivation for not transforming nobuiltin calls is that it can interfere with sanitizers: https://reviews.llvm.org/D19781 https://reviews.llvm.org/D19801 Differential Revision: https://reviews.llvm.org/D34043 llvm-svn: 305007
2017-06-09 03:47:25 +08:00
; CHECK-NEXT: mflr 0
; CHECK-NEXT: std 0, 16(1)
; CHECK-NEXT: stdu 1, -32(1)
; CHECK-NEXT: .cfi_def_cfa_offset 32
; CHECK-NEXT: .cfi_offset lr, 16
; CHECK-NEXT: li 5, 2
; CHECK-NEXT: bl memcmp
; CHECK-NEXT: nop
; CHECK-NEXT: cntlzw 3, 3
; CHECK-NEXT: rlwinm 3, 3, 27, 31, 31
; CHECK-NEXT: addi 1, 1, 32
; CHECK-NEXT: ld 0, 16(1)
; CHECK-NEXT: mtlr 0
[PowerPC] add memcmp test with nobuiltin attr; NFC In SDAG, we don't expand libcalls with a nobuiltin attribute. It's not clear if that's correct from the existing code comment: "Don't do the check if marked as nobuiltin for some reason." ...adding a test here either way to show that there is currently a different behavior implemented in the CGP-based expansion. llvm-svn: 304991
2017-06-09 01:09:18 +08:00
; CHECK-NEXT: blr
%m = tail call signext i32 @memcmp(i8* %X, i8* %Y, i64 2) nobuiltin
%c = icmp eq i32 %m, 0
ret i1 %c
}