RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-24 19:52:54 +01:00
Code Issues Projects Releases Wiki Activity
6bc0aa2395
llvm-mirror/test/CodeGen/AArch64/cond-sel.ll

234 lines
7.3 KiB
LLVM
Raw Normal View History

AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 14:50:23 +02:00
; RUN: llc -verify-machineinstrs < %s -mtriple=aarch64-none-linux-gnu -mcpu=cyclone | FileCheck %s --check-prefix=CHECK
; RUN: llc -verify-machineinstrs < %s -mtriple=aarch64-none-linux-gnu -mattr=-fp-armv8 | FileCheck --check-prefix=CHECK-NOFP %s
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
@var32 = global i32 0
@var64 = global i64 0
ARM64: add patterns for csXYZ with reversed operands. AArch64 tests for this, and it's obviously a good idea. Have to invert the condition code, of course. llvm-svn: 206170
2014-04-14 14:51:02 +02:00
define void @test_csel(i32 %lhs32, i32 %rhs32, i64 %lhs64) minsize {
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; CHECK-LABEL: test_csel:
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
%tst1 = icmp ugt i32 %lhs32, %rhs32
%val1 = select i1 %tst1, i32 42, i32 52
store i32 %val1, i32* @var32
ARM64: use hex immediates for movz/movk instructions Since these are mostly used in "lsl #16", "lsl #32", "lsl #48" combinations to piece together an immediate in 16-bit chunks, hex is probably the most appropriate format. llvm-svn: 207635
2014-04-30 13:19:40 +02:00
; CHECK-DAG: movz [[W52:w[0-9]+]], #{{52|0x34}}
; CHECK-DAG: movz [[W42:w[0-9]+]], #{{42|0x2a}}
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
; CHECK: csel {{w[0-9]+}}, [[W42]], [[W52]], hi
%rhs64 = sext i32 %rhs32 to i64
%tst2 = icmp sle i64 %lhs64, %rhs64
%val2 = select i1 %tst2, i64 %lhs64, i64 %rhs64
store i64 %val2, i64* @var64
ARM64: add patterns for csXYZ with reversed operands. AArch64 tests for this, and it's obviously a good idea. Have to invert the condition code, of course. llvm-svn: 206170
2014-04-14 14:51:02 +02:00
; CHECK: sxtw [[EXT_RHS:x[0-9]+]], {{[wx]}}[[RHS:[0-9]+]]
; CHECK: cmp [[LHS:x[0-9]+]], w[[RHS]], sxtw
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
; CHECK: csel {{x[0-9]+}}, [[LHS]], [[EXT_RHS]], le
ret void
; CHECK: ret
}
define void @test_floatcsel(float %lhs32, float %rhs32, double %lhs64, double %rhs64) {
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; CHECK-LABEL: test_floatcsel:
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
%tst1 = fcmp one float %lhs32, %rhs32
; CHECK: fcmp {{s[0-9]+}}, {{s[0-9]+}}
[AArch64] Make the use of FP instructions optional, but enabled by default. This adds a new subtarget feature called FPARMv8 (implied by NEON), and predicates the support of the FP instructions and registers on this feature. llvm-svn: 193739
2013-10-31 10:32:11 +01:00
; CHECK-NOFP-NOT: fcmp
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
%val1 = select i1 %tst1, i32 42, i32 52
store i32 %val1, i32* @var32
ARM64: use hex immediates for movz/movk instructions Since these are mostly used in "lsl #16", "lsl #32", "lsl #48" combinations to piece together an immediate in 16-bit chunks, hex is probably the most appropriate format. llvm-svn: 207635
2014-04-30 13:19:40 +02:00
; CHECK: movz [[W52:w[0-9]+]], #{{52|0x34}}
; CHECK: movz [[W42:w[0-9]+]], #{{42|0x2a}}
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
; CHECK: csel [[MAYBETRUE:w[0-9]+]], [[W42]], [[W52]], mi
; CHECK: csel {{w[0-9]+}}, [[W42]], [[MAYBETRUE]], gt
%tst2 = fcmp ueq double %lhs64, %rhs64
; CHECK: fcmp {{d[0-9]+}}, {{d[0-9]+}}
[AArch64] Make the use of FP instructions optional, but enabled by default. This adds a new subtarget feature called FPARMv8 (implied by NEON), and predicates the support of the FP instructions and registers on this feature. llvm-svn: 193739
2013-10-31 10:32:11 +01:00
; CHECK-NOFP-NOT: fcmp
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
%val2 = select i1 %tst2, i64 9, i64 15
store i64 %val2, i64* @var64
AArch64/ARM64: move ARM64 into AArch64's place This commit starts with a "git mv ARM64 AArch64" and continues out from there, renaming the C++ classes, intrinsics, and other target-local objects for consistency. "ARM64" test directories are also moved, and tests that began their life in ARM64 use an arm64 triple, those from AArch64 use an aarch64 triple. Both should be equivalent though. This finishes the AArch64 merge, and everyone should feel free to continue committing as normal now. llvm-svn: 209577
2014-05-24 14:50:23 +02:00
; CHECK: orr w[[CONST15:[0-9]+]], wzr, #0xf
ARM64: use hex immediates for movz/movk instructions Since these are mostly used in "lsl #16", "lsl #32", "lsl #48" combinations to piece together an immediate in 16-bit chunks, hex is probably the most appropriate format. llvm-svn: 207635
2014-04-30 13:19:40 +02:00
; CHECK: movz {{[wx]}}[[CONST9:[0-9]+]], #{{9|0x9}}
ARM64: use 32-bit moves for constants where possible. If we know that a particular 64-bit constant has all high bits zero, then we can rely on the fact that 32-bit ARM64 instructions automatically zero out the high bits of an x-register. This gives the expansion logic less constraints to satisfy and so sometimes allows it to pick better sequences. Came up while porting test/CodeGen/AArch64/movw-consts.ll: this will allow a 32-bit MOVN to be used in @test8 soon. llvm-svn: 206379
2014-04-16 13:52:51 +02:00
; CHECK: csel [[MAYBETRUE:x[0-9]+]], x[[CONST9]], x[[CONST15]], eq
; CHECK: csel {{x[0-9]+}}, x[[CONST9]], [[MAYBETRUE]], vs
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
ret void
; CHECK: ret
}
ARM64: add patterns for csXYZ with reversed operands. AArch64 tests for this, and it's obviously a good idea. Have to invert the condition code, of course. llvm-svn: 206170
2014-04-14 14:51:02 +02:00
define void @test_csinc(i32 %lhs32, i32 %rhs32, i64 %lhs64) minsize {
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; CHECK-LABEL: test_csinc:
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
; Note that commuting rhs and lhs in the select changes ugt to ule (i.e. hi to ls).
%tst1 = icmp ugt i32 %lhs32, %rhs32
%inc1 = add i32 %rhs32, 1
%val1 = select i1 %tst1, i32 %inc1, i32 %lhs32
store volatile i32 %val1, i32* @var32
; CHECK: cmp [[LHS:w[0-9]+]], [[RHS:w[0-9]+]]
; CHECK: csinc {{w[0-9]+}}, [[LHS]], [[RHS]], ls
%rhs2 = add i32 %rhs32, 42
%tst2 = icmp sle i32 %lhs32, %rhs2
%inc2 = add i32 %rhs32, 1
%val2 = select i1 %tst2, i32 %lhs32, i32 %inc2
store volatile i32 %val2, i32* @var32
; CHECK: cmp [[LHS:w[0-9]+]], {{w[0-9]+}}
; CHECK: csinc {{w[0-9]+}}, [[LHS]], {{w[0-9]+}}, le
; Note that commuting rhs and lhs in the select changes ugt to ule (i.e. hi to ls).
%rhs3 = sext i32 %rhs32 to i64
%tst3 = icmp ugt i64 %lhs64, %rhs3
%inc3 = add i64 %rhs3, 1
%val3 = select i1 %tst3, i64 %inc3, i64 %lhs64
store volatile i64 %val3, i64* @var64
; CHECK: cmp [[LHS:x[0-9]+]], {{w[0-9]+}}
; CHECK: csinc {{x[0-9]+}}, [[LHS]], {{x[0-9]+}}, ls
%rhs4 = zext i32 %rhs32 to i64
%tst4 = icmp sle i64 %lhs64, %rhs4
%inc4 = add i64 %rhs4, 1
%val4 = select i1 %tst4, i64 %lhs64, i64 %inc4
store volatile i64 %val4, i64* @var64
; CHECK: cmp [[LHS:x[0-9]+]], {{w[0-9]+}}
; CHECK: csinc {{x[0-9]+}}, [[LHS]], {{x[0-9]+}}, le
ret void
; CHECK: ret
}
ARM64: add patterns for csXYZ with reversed operands. AArch64 tests for this, and it's obviously a good idea. Have to invert the condition code, of course. llvm-svn: 206170
2014-04-14 14:51:02 +02:00
define void @test_csinv(i32 %lhs32, i32 %rhs32, i64 %lhs64) minsize {
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; CHECK-LABEL: test_csinv:
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
; Note that commuting rhs and lhs in the select changes ugt to ule (i.e. hi to ls).
%tst1 = icmp ugt i32 %lhs32, %rhs32
%inc1 = xor i32 -1, %rhs32
%val1 = select i1 %tst1, i32 %inc1, i32 %lhs32
store volatile i32 %val1, i32* @var32
; CHECK: cmp [[LHS:w[0-9]+]], [[RHS:w[0-9]+]]
; CHECK: csinv {{w[0-9]+}}, [[LHS]], [[RHS]], ls
%rhs2 = add i32 %rhs32, 42
%tst2 = icmp sle i32 %lhs32, %rhs2
%inc2 = xor i32 -1, %rhs32
%val2 = select i1 %tst2, i32 %lhs32, i32 %inc2
store volatile i32 %val2, i32* @var32
; CHECK: cmp [[LHS:w[0-9]+]], {{w[0-9]+}}
; CHECK: csinv {{w[0-9]+}}, [[LHS]], {{w[0-9]+}}, le
; Note that commuting rhs and lhs in the select changes ugt to ule (i.e. hi to ls).
%rhs3 = sext i32 %rhs32 to i64
%tst3 = icmp ugt i64 %lhs64, %rhs3
%inc3 = xor i64 -1, %rhs3
%val3 = select i1 %tst3, i64 %inc3, i64 %lhs64
store volatile i64 %val3, i64* @var64
; CHECK: cmp [[LHS:x[0-9]+]], {{w[0-9]+}}
; CHECK: csinv {{x[0-9]+}}, [[LHS]], {{x[0-9]+}}, ls
%rhs4 = zext i32 %rhs32 to i64
%tst4 = icmp sle i64 %lhs64, %rhs4
%inc4 = xor i64 -1, %rhs4
%val4 = select i1 %tst4, i64 %lhs64, i64 %inc4
store volatile i64 %val4, i64* @var64
; CHECK: cmp [[LHS:x[0-9]+]], {{w[0-9]+}}
; CHECK: csinv {{x[0-9]+}}, [[LHS]], {{x[0-9]+}}, le
ret void
; CHECK: ret
}
ARM64: add patterns for csXYZ with reversed operands. AArch64 tests for this, and it's obviously a good idea. Have to invert the condition code, of course. llvm-svn: 206170
2014-04-14 14:51:02 +02:00
define void @test_csneg(i32 %lhs32, i32 %rhs32, i64 %lhs64) minsize {
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; CHECK-LABEL: test_csneg:
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
; Note that commuting rhs and lhs in the select changes ugt to ule (i.e. hi to ls).
%tst1 = icmp ugt i32 %lhs32, %rhs32
%inc1 = sub i32 0, %rhs32
%val1 = select i1 %tst1, i32 %inc1, i32 %lhs32
store volatile i32 %val1, i32* @var32
; CHECK: cmp [[LHS:w[0-9]+]], [[RHS:w[0-9]+]]
; CHECK: csneg {{w[0-9]+}}, [[LHS]], [[RHS]], ls
%rhs2 = add i32 %rhs32, 42
%tst2 = icmp sle i32 %lhs32, %rhs2
%inc2 = sub i32 0, %rhs32
%val2 = select i1 %tst2, i32 %lhs32, i32 %inc2
store volatile i32 %val2, i32* @var32
; CHECK: cmp [[LHS:w[0-9]+]], {{w[0-9]+}}
; CHECK: csneg {{w[0-9]+}}, [[LHS]], {{w[0-9]+}}, le
; Note that commuting rhs and lhs in the select changes ugt to ule (i.e. hi to ls).
%rhs3 = sext i32 %rhs32 to i64
%tst3 = icmp ugt i64 %lhs64, %rhs3
%inc3 = sub i64 0, %rhs3
%val3 = select i1 %tst3, i64 %inc3, i64 %lhs64
store volatile i64 %val3, i64* @var64
; CHECK: cmp [[LHS:x[0-9]+]], {{w[0-9]+}}
; CHECK: csneg {{x[0-9]+}}, [[LHS]], {{x[0-9]+}}, ls
%rhs4 = zext i32 %rhs32 to i64
%tst4 = icmp sle i64 %lhs64, %rhs4
%inc4 = sub i64 0, %rhs4
%val4 = select i1 %tst4, i64 %lhs64, i64 %inc4
store volatile i64 %val4, i64* @var64
; CHECK: cmp [[LHS:x[0-9]+]], {{w[0-9]+}}
; CHECK: csneg {{x[0-9]+}}, [[LHS]], {{x[0-9]+}}, le
ret void
; CHECK: ret
}
define void @test_cset(i32 %lhs, i32 %rhs, i64 %lhs64) {
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; CHECK-LABEL: test_cset:
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
; N.b. code is not optimal here (32-bit csinc would be better) but
; incoming DAG is too complex
%tst1 = icmp eq i32 %lhs, %rhs
%val1 = zext i1 %tst1 to i32
store i32 %val1, i32* @var32
; CHECK: cmp {{w[0-9]+}}, {{w[0-9]+}}
TableGen: use PrintMethods to print more aliases llvm-svn: 208607
2014-05-12 20:04:06 +02:00
; CHECK: cset {{w[0-9]+}}, eq
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
%rhs64 = sext i32 %rhs to i64
%tst2 = icmp ule i64 %lhs64, %rhs64
%val2 = zext i1 %tst2 to i64
store i64 %val2, i64* @var64
TableGen: use PrintMethods to print more aliases llvm-svn: 208607
2014-05-12 20:04:06 +02:00
; CHECK: cset {{w[0-9]+}}, ls
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
ret void
; CHECK: ret
}
define void @test_csetm(i32 %lhs, i32 %rhs, i64 %lhs64) {
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; CHECK-LABEL: test_csetm:
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
%tst1 = icmp eq i32 %lhs, %rhs
%val1 = sext i1 %tst1 to i32
store i32 %val1, i32* @var32
; CHECK: cmp {{w[0-9]+}}, {{w[0-9]+}}
TableGen: use PrintMethods to print more aliases llvm-svn: 208607
2014-05-12 20:04:06 +02:00
; CHECK: csetm {{w[0-9]+}}, eq
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
%rhs64 = sext i32 %rhs to i64
%tst2 = icmp ule i64 %lhs64, %rhs64
%val2 = sext i1 %tst2 to i64
store i64 %val2, i64* @var64
TableGen: use PrintMethods to print more aliases llvm-svn: 208607
2014-05-12 20:04:06 +02:00
; CHECK: csetm {{x[0-9]+}}, ls
Add AArch64 as an experimental target. This patch adds support for AArch64 (ARM's 64-bit architecture) to LLVM in the "experimental" category. Currently, it won't be built unless requested explicitly. This initial commit should have support for: + Assembly of all scalar (i.e. non-NEON, non-Crypto) instructions (except the late addition CRC instructions). + CodeGen features required for C++03 and C99. + Compilation for the "small" memory model: code+static data < 4GB. + Absolute and position-independent code. + GNU-style (i.e. "__thread") TLS. + Debugging information. The principal omission, currently, is performance tuning. This patch excludes the NEON support also reviewed due to an outbreak of batshit insanity in our legal department. That will be committed soon bringing the changes to precisely what has been approved. Further reviews would be gratefully received. llvm-svn: 174054
2013-01-31 13:12:40 +01:00
ret void
; CHECK: ret
}
AArch64: skip select/setcc combine in complex case. In an llvm-stress generated test, we were trying to create a v0iN type and asserting when that failed. This case could probably be handled by the function, but not without added complexity and the situation it arises in is sufficiently odd that there's probably no benefit anyway. Should fix PR20775. llvm-svn: 216725
2014-08-29 15:05:18 +02:00
define <1 x i1> @test_wide_comparison(i32 %in) {
; CHECK-LABEL: test_wide_comparison:
; CHECK: cmp w0, #1234
; CHECK: cset
%tmp = icmp sgt i32 %in, 1234
%res = select i1 %tmp, <1 x i1> <i1 1>, <1 x i1> zeroinitializer
ret <1 x i1> %res
}
AArch64: only try to get operand of a known node. A bug in r216725 meant we tried to discover the type of a SETCC before confirming the node actually was a SETCC. llvm-svn: 216734
2014-08-29 17:34:58 +02:00
define i32 @test_select_undef() {
; CHECK-LABEL: test_select_undef:
; CHECK: ret
%res = select i1 undef, i32 0, i32 42
ret i32 %res
}
Reference in New Issue Copy Permalink