maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/AArch64/cxx-tlscc.ll

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

225 lines
6.4 KiB
LLVM
Raw Normal View History

[CXX TLS calling convention] Add support for AArch64. rdar://9001553 llvm-svn: 254978
2015-12-08 08:14:38 +08:00
; RUN: llc < %s -mtriple=aarch64-apple-ios | FileCheck %s
[LLVM] Remove unwanted --check-prefix=CHECK from unit tests. NFC. Summary: Removed unwanted --check-prefix=CHECK from numerous unit tests. Reviewers: t.p.northover, dblaikie, uweigand, MatzeB, tstellarAMD, mcrosier Subscribers: mcrosier, dsanders Differential Revision: http://reviews.llvm.org/D19279 llvm-svn: 266834
2016-04-20 07:51:52 +08:00
; RUN: llc < %s -mtriple=aarch64-apple-ios -enable-shrink-wrap=true | FileCheck %s
[CXX TLS calling convention] Add support for AArch64. rdar://9001553 llvm-svn: 254978
2015-12-08 08:14:38 +08:00
; Shrink wrapping currently does not kick in because we have a TLS CALL
; in the entry block and it will clobber the link register.
[AArch64][GlobalISel] Enable GlobalISel at -O0 by default Tests updated to explicitly use fast-isel at -O0 instead of implicitly. This change also allows an explicit -fast-isel option to override an implicitly enabled global-isel. Otherwise -fast-isel would have no effect at -O0. Differential Revision: https://reviews.llvm.org/D41362 llvm-svn: 321655
2018-01-03 00:30:47 +08:00
; RUN: llc < %s -mtriple=aarch64-apple-ios -O0 -fast-isel | FileCheck --check-prefix=CHECK-O0 %s
[CXX_FAST_TLS] fix issues with O0 on ARM, AArch64 and X86. Since at O0, explicit copies via SplitCSR may not be removed even if they are unnecessary, we choose not to use SplitCSR at O0. llvm-svn: 263855
2016-03-19 07:38:49 +08:00
[CXX TLS calling convention] Add support for AArch64. rdar://9001553 llvm-svn: 254978
2015-12-08 08:14:38 +08:00
%struct.S = type { i8 }
@sg = internal thread_local global %struct.S zeroinitializer, align 1
@__dso_handle = external global i8
@__tls_guard = internal thread_local unnamed_addr global i1 false
CXX_FAST_TLS calling convention: fix issue on AArch64. When we have a single basic block, the explicit copy-back instructions should be inserted right before the terminator. Before this fix, they were wrongly placed at the beginning of the basic block. I will commit fixes to other platforms as well. PR26136 llvm-svn: 257929
2016-01-16 04:13:28 +08:00
@sum1 = internal thread_local global i32 0, align 4
[CXX TLS calling convention] Add support for AArch64. rdar://9001553 llvm-svn: 254978
2015-12-08 08:14:38 +08:00
declare %struct.S* @_ZN1SC1Ev(%struct.S* returned)
declare %struct.S* @_ZN1SD1Ev(%struct.S* returned)
declare i32 @_tlv_atexit(void (i8*)*, i8*, i8*)
CXX_FAST_TLS calling convention: performance improvement for AArch64. The access function has a short entry and a short exit, the initialization block is only run the first time. To improve the performance, we want to have a short frame at the entry and exit. We explicitly handle most of the CSRs via copies. Only the CSRs that are not handled via copies will be in CSR_SaveList. Frame lowering and prologue/epilogue insertion will generate a short frame in the entry and exit according to CSR_SaveList. The majority of the CSRs will be handled by register allcoator. Register allocator will try to spill and reload them in the initialization block. We add CSRsViaCopy, it will be explicitly handled during lowering. 1> we first set FunctionLoweringInfo->SplitCSR if conditions are met (the target supports it for the given machine function and the function has only return exits). We also call TLI->initializeSplitCSR to perform initialization. 2> we call TLI->insertCopiesSplitCSR to insert copies from CSRsViaCopy to virtual registers at beginning of the entry block and copies from virtual registers to CSRsViaCopy at beginning of the exit blocks. 3> we also need to make sure the explicit copies will not be eliminated. The target independent portion was committed as r255353. rdar://problem/23557469 Differential Revision: http://reviews.llvm.org/D15341 llvm-svn: 255821
2015-12-17 05:04:19 +08:00
define cxx_fast_tlscc nonnull %struct.S* @_ZTW2sg() nounwind {
[CXX TLS calling convention] Add support for AArch64. rdar://9001553 llvm-svn: 254978
2015-12-08 08:14:38 +08:00
%.b.i = load i1, i1* @__tls_guard, align 1
br i1 %.b.i, label %__tls_init.exit, label %init.i
init.i:
store i1 true, i1* @__tls_guard, align 1
%call.i.i = tail call %struct.S* @_ZN1SC1Ev(%struct.S* nonnull @sg)
%1 = tail call i32 @_tlv_atexit(void (i8*)* nonnull bitcast (%struct.S* (%struct.S*)* @_ZN1SD1Ev to void (i8*)*), i8* nonnull getelementptr inbounds (%struct.S, %struct.S* @sg, i64 0, i32 0), i8* nonnull @__dso_handle)
br label %__tls_init.exit
__tls_init.exit:
ret %struct.S* @sg
}
; CHECK-LABEL: _ZTW2sg
CXX_FAST_TLS calling convention: performance improvement for AArch64. The access function has a short entry and a short exit, the initialization block is only run the first time. To improve the performance, we want to have a short frame at the entry and exit. We explicitly handle most of the CSRs via copies. Only the CSRs that are not handled via copies will be in CSR_SaveList. Frame lowering and prologue/epilogue insertion will generate a short frame in the entry and exit according to CSR_SaveList. The majority of the CSRs will be handled by register allcoator. Register allocator will try to spill and reload them in the initialization block. We add CSRsViaCopy, it will be explicitly handled during lowering. 1> we first set FunctionLoweringInfo->SplitCSR if conditions are met (the target supports it for the given machine function and the function has only return exits). We also call TLI->initializeSplitCSR to perform initialization. 2> we call TLI->insertCopiesSplitCSR to insert copies from CSRsViaCopy to virtual registers at beginning of the entry block and copies from virtual registers to CSRsViaCopy at beginning of the exit blocks. 3> we also need to make sure the explicit copies will not be eliminated. The target independent portion was committed as r255353. rdar://problem/23557469 Differential Revision: http://reviews.llvm.org/D15341 llvm-svn: 255821
2015-12-17 05:04:19 +08:00
; CHECK-NOT: stp d31, d30
; CHECK-NOT: stp d29, d28
; CHECK-NOT: stp d27, d26
; CHECK-NOT: stp d25, d24
; CHECK-NOT: stp d23, d22
; CHECK-NOT: stp d21, d20
; CHECK-NOT: stp d19, d18
; CHECK-NOT: stp d17, d16
; CHECK-NOT: stp d7, d6
; CHECK-NOT: stp d5, d4
; CHECK-NOT: stp d3, d2
; CHECK-NOT: stp d1, d0
; CHECK-NOT: stp x20, x19
MachineScheduler: Fully compare top/bottom candidates In bidirectional scheduling this gives more stable results than just comparing the "reason" fields of the top/bottom node because the reason field may be higher depending on what other nodes are in the queue. Differential Revision: http://reviews.llvm.org/D19401 llvm-svn: 273755
2016-06-25 08:23:00 +08:00
; FIXME: The splitting logic in the register allocator fails to split along
; control flow here, we used to get this right by accident before...
; CHECK-NOTXX: stp x14, x13
CXX_FAST_TLS calling convention: performance improvement for AArch64. The access function has a short entry and a short exit, the initialization block is only run the first time. To improve the performance, we want to have a short frame at the entry and exit. We explicitly handle most of the CSRs via copies. Only the CSRs that are not handled via copies will be in CSR_SaveList. Frame lowering and prologue/epilogue insertion will generate a short frame in the entry and exit according to CSR_SaveList. The majority of the CSRs will be handled by register allcoator. Register allocator will try to spill and reload them in the initialization block. We add CSRsViaCopy, it will be explicitly handled during lowering. 1> we first set FunctionLoweringInfo->SplitCSR if conditions are met (the target supports it for the given machine function and the function has only return exits). We also call TLI->initializeSplitCSR to perform initialization. 2> we call TLI->insertCopiesSplitCSR to insert copies from CSRsViaCopy to virtual registers at beginning of the entry block and copies from virtual registers to CSRsViaCopy at beginning of the exit blocks. 3> we also need to make sure the explicit copies will not be eliminated. The target independent portion was committed as r255353. rdar://problem/23557469 Differential Revision: http://reviews.llvm.org/D15341 llvm-svn: 255821
2015-12-17 05:04:19 +08:00
; CHECK-NOT: stp x12, x11
; CHECK-NOT: stp x10, x9
; CHECK-NOT: stp x8, x7
; CHECK-NOT: stp x6, x5
; CHECK-NOT: stp x4, x3
; CHECK-NOT: stp x2, x1
[CXX TLS calling convention] Add support for AArch64. rdar://9001553 llvm-svn: 254978
2015-12-08 08:14:38 +08:00
; CHECK: blr
; CHECK: tbnz w{{.*}}, #0, [[BB_end:.?LBB0_[0-9]+]]
; CHECK: blr
; CHECK: tlv_atexit
; CHECK: [[BB_end]]:
; CHECK: blr
CXX_FAST_TLS calling convention: performance improvement for AArch64. The access function has a short entry and a short exit, the initialization block is only run the first time. To improve the performance, we want to have a short frame at the entry and exit. We explicitly handle most of the CSRs via copies. Only the CSRs that are not handled via copies will be in CSR_SaveList. Frame lowering and prologue/epilogue insertion will generate a short frame in the entry and exit according to CSR_SaveList. The majority of the CSRs will be handled by register allcoator. Register allocator will try to spill and reload them in the initialization block. We add CSRsViaCopy, it will be explicitly handled during lowering. 1> we first set FunctionLoweringInfo->SplitCSR if conditions are met (the target supports it for the given machine function and the function has only return exits). We also call TLI->initializeSplitCSR to perform initialization. 2> we call TLI->insertCopiesSplitCSR to insert copies from CSRsViaCopy to virtual registers at beginning of the entry block and copies from virtual registers to CSRsViaCopy at beginning of the exit blocks. 3> we also need to make sure the explicit copies will not be eliminated. The target independent portion was committed as r255353. rdar://problem/23557469 Differential Revision: http://reviews.llvm.org/D15341 llvm-svn: 255821
2015-12-17 05:04:19 +08:00
; CHECK-NOT: ldp x2, x1
; CHECK-NOT: ldp x4, x3
; CHECK-NOT: ldp x6, x5
; CHECK-NOT: ldp x8, x7
; CHECK-NOT: ldp x10, x9
; CHECK-NOT: ldp x12, x11
MachineScheduler: Fully compare top/bottom candidates In bidirectional scheduling this gives more stable results than just comparing the "reason" fields of the top/bottom node because the reason field may be higher depending on what other nodes are in the queue. Differential Revision: http://reviews.llvm.org/D19401 llvm-svn: 273755
2016-06-25 08:23:00 +08:00
; CHECK-NOTXX: ldp x14, x13
CXX_FAST_TLS calling convention: performance improvement for AArch64. The access function has a short entry and a short exit, the initialization block is only run the first time. To improve the performance, we want to have a short frame at the entry and exit. We explicitly handle most of the CSRs via copies. Only the CSRs that are not handled via copies will be in CSR_SaveList. Frame lowering and prologue/epilogue insertion will generate a short frame in the entry and exit according to CSR_SaveList. The majority of the CSRs will be handled by register allcoator. Register allocator will try to spill and reload them in the initialization block. We add CSRsViaCopy, it will be explicitly handled during lowering. 1> we first set FunctionLoweringInfo->SplitCSR if conditions are met (the target supports it for the given machine function and the function has only return exits). We also call TLI->initializeSplitCSR to perform initialization. 2> we call TLI->insertCopiesSplitCSR to insert copies from CSRsViaCopy to virtual registers at beginning of the entry block and copies from virtual registers to CSRsViaCopy at beginning of the exit blocks. 3> we also need to make sure the explicit copies will not be eliminated. The target independent portion was committed as r255353. rdar://problem/23557469 Differential Revision: http://reviews.llvm.org/D15341 llvm-svn: 255821
2015-12-17 05:04:19 +08:00
; CHECK-NOT: ldp x20, x19
; CHECK-NOT: ldp d1, d0
; CHECK-NOT: ldp d3, d2
; CHECK-NOT: ldp d5, d4
; CHECK-NOT: ldp d7, d6
; CHECK-NOT: ldp d17, d16
; CHECK-NOT: ldp d19, d18
; CHECK-NOT: ldp d21, d20
; CHECK-NOT: ldp d23, d22
; CHECK-NOT: ldp d25, d24
; CHECK-NOT: ldp d27, d26
; CHECK-NOT: ldp d29, d28
; CHECK-NOT: ldp d31, d30
CXX_FAST_TLS calling convention: fix issue on AArch64. When we have a single basic block, the explicit copy-back instructions should be inserted right before the terminator. Before this fix, they were wrongly placed at the beginning of the basic block. I will commit fixes to other platforms as well. PR26136 llvm-svn: 257929
2016-01-16 04:13:28 +08:00
[CXX_FAST_TLS] fix issues with O0 on ARM, AArch64 and X86. Since at O0, explicit copies via SplitCSR may not be removed even if they are unnecessary, we choose not to use SplitCSR at O0. llvm-svn: 263855
2016-03-19 07:38:49 +08:00
; CHECK-O0-LABEL: _ZTW2sg
; CHECK-O0: stp d31, d30
; CHECK-O0: stp d29, d28
; CHECK-O0: stp d27, d26
; CHECK-O0: stp d25, d24
; CHECK-O0: stp d23, d22
; CHECK-O0: stp d21, d20
; CHECK-O0: stp d19, d18
; CHECK-O0: stp d17, d16
; CHECK-O0: stp d7, d6
; CHECK-O0: stp d5, d4
; CHECK-O0: stp d3, d2
; CHECK-O0: stp d1, d0
; CHECK-O0: stp x14, x13
; CHECK-O0: stp x12, x11
; CHECK-O0: stp x10, x9
; CHECK-O0: stp x8, x7
; CHECK-O0: stp x6, x5
; CHECK-O0: stp x4, x3
; CHECK-O0: stp x2, x1
; CHECK-O0: blr
; CHECK-O0: tbnz w{{.*}}, #0, [[BB_end:.?LBB0_[0-9]+]]
; CHECK-O0: blr
; CHECK-O0: tlv_atexit
; CHECK-O0: [[BB_end]]:
; CHECK-O0: blr
; CHECK-O0: ldp x2, x1
; CHECK-O0: ldp x4, x3
; CHECK-O0: ldp x6, x5
; CHECK-O0: ldp x8, x7
; CHECK-O0: ldp x10, x9
; CHECK-O0: ldp x12, x11
; CHECK-O0: ldp x14, x13
; CHECK-O0: ldp d1, d0
; CHECK-O0: ldp d3, d2
; CHECK-O0: ldp d5, d4
; CHECK-O0: ldp d7, d6
; CHECK-O0: ldp d17, d16
; CHECK-O0: ldp d19, d18
; CHECK-O0: ldp d21, d20
; CHECK-O0: ldp d23, d22
; CHECK-O0: ldp d25, d24
; CHECK-O0: ldp d27, d26
; CHECK-O0: ldp d29, d28
; CHECK-O0: ldp d31, d30
CXX_FAST_TLS calling convention: fix issue on AArch64. When we have a single basic block, the explicit copy-back instructions should be inserted right before the terminator. Before this fix, they were wrongly placed at the beginning of the basic block. I will commit fixes to other platforms as well. PR26136 llvm-svn: 257929
2016-01-16 04:13:28 +08:00
; CHECK-LABEL: _ZTW4sum1
; CHECK-NOT: stp d31, d30
; CHECK-NOT: stp d29, d28
; CHECK-NOT: stp d27, d26
; CHECK-NOT: stp d25, d24
; CHECK-NOT: stp d23, d22
; CHECK-NOT: stp d21, d20
; CHECK-NOT: stp d19, d18
; CHECK-NOT: stp d17, d16
; CHECK-NOT: stp d7, d6
; CHECK-NOT: stp d5, d4
; CHECK-NOT: stp d3, d2
; CHECK-NOT: stp d1, d0
; CHECK-NOT: stp x20, x19
; CHECK-NOT: stp x14, x13
; CHECK-NOT: stp x12, x11
; CHECK-NOT: stp x10, x9
; CHECK-NOT: stp x8, x7
; CHECK-NOT: stp x6, x5
; CHECK-NOT: stp x4, x3
; CHECK-NOT: stp x2, x1
; CHECK: blr
[CXX_FAST_TLS] fix issues with O0 on ARM, AArch64 and X86. Since at O0, explicit copies via SplitCSR may not be removed even if they are unnecessary, we choose not to use SplitCSR at O0. llvm-svn: 263855
2016-03-19 07:38:49 +08:00
; CHECK-O0-LABEL: _ZTW4sum1
; CHECK-O0-NOT: vstr
; CHECK-O0-NOT: vldr
CXX_FAST_TLS calling convention: fix issue on AArch64. When we have a single basic block, the explicit copy-back instructions should be inserted right before the terminator. Before this fix, they were wrongly placed at the beginning of the basic block. I will commit fixes to other platforms as well. PR26136 llvm-svn: 257929
2016-01-16 04:13:28 +08:00
define cxx_fast_tlscc nonnull i32* @_ZTW4sum1() nounwind {
ret i32* @sum1
}
[CXX_FAST_TLS] fix issues with O0 on ARM, AArch64 and X86. Since at O0, explicit copies via SplitCSR may not be removed even if they are unnecessary, we choose not to use SplitCSR at O0. llvm-svn: 263855
2016-03-19 07:38:49 +08:00
; Make sure at O0, we don't generate spilling/reloading of the CSRs.
; CHECK-O0-LABEL: tls_test2
; CHECK-O0-NOT: stp d31, d30
; CHECK-O0-NOT: stp d29, d28
; CHECK-O0-NOT: stp d27, d26
; CHECK-O0-NOT: stp d25, d24
; CHECK-O0-NOT: stp d23, d22
; CHECK-O0-NOT: stp d21, d20
; CHECK-O0-NOT: stp d19, d18
; CHECK-O0-NOT: stp d17, d16
; CHECK-O0-NOT: stp d7, d6
; CHECK-O0-NOT: stp d5, d4
; CHECK-O0-NOT: stp d3, d2
; CHECK-O0-NOT: stp d1, d0
; CHECK-O0-NOT: stp x20, x19
; CHECK-O0-NOT: stp x14, x13
; CHECK-O0-NOT: stp x12, x11
; CHECK-O0-NOT: stp x10, x9
; CHECK-O0-NOT: stp x8, x7
; CHECK-O0-NOT: stp x6, x5
; CHECK-O0-NOT: stp x4, x3
; CHECK-O0-NOT: stp x2, x1
; CHECK-O0: bl {{.*}}tls_helper
; CHECK-O0-NOT: ldp x2, x1
; CHECK-O0-NOT: ldp x4, x3
; CHECK-O0-NOT: ldp x6, x5
; CHECK-O0-NOT: ldp x8, x7
; CHECK-O0-NOT: ldp x10, x9
; CHECK-O0-NOT: ldp x12, x11
; CHECK-O0-NOT: ldp x14, x13
; CHECK-O0-NOT: ldp x20, x19
; CHECK-O0-NOT: ldp d1, d0
; CHECK-O0-NOT: ldp d3, d2
; CHECK-O0-NOT: ldp d5, d4
; CHECK-O0-NOT: ldp d7, d6
; CHECK-O0-NOT: ldp d17, d16
; CHECK-O0-NOT: ldp d19, d18
; CHECK-O0-NOT: ldp d21, d20
; CHECK-O0-NOT: ldp d23, d22
; CHECK-O0-NOT: ldp d25, d24
; CHECK-O0-NOT: ldp d27, d26
; CHECK-O0-NOT: ldp d29, d28
; CHECK-O0-NOT: ldp d31, d30
; CHECK-O0: ret
%class.C = type { i32 }
@tC = internal thread_local global %class.C zeroinitializer, align 4
declare cxx_fast_tlscc void @tls_helper()
define cxx_fast_tlscc %class.C* @tls_test2() #1 {
call cxx_fast_tlscc void @tls_helper()
ret %class.C* @tC
}
[CXX_FAST_TLS] Disable tail call when calling conventions are mismatched. Since CXX_FAST_TLS has a bigger set of CSRs, we don't tail call when caller and callee have mismatched calling conventions. llvm-svn: 263856
2016-03-19 07:41:51 +08:00
; Make sure we do not allow tail call when caller and callee have different
; calling conventions.
declare %class.C* @_ZN1CD1Ev(%class.C* readnone returned %this)
; CHECK-LABEL: tls_test
; CHECK: bl __tlv_atexit
define cxx_fast_tlscc void @__tls_test() {
entry:
store i32 0, i32* getelementptr inbounds (%class.C, %class.C* @tC, i64 0, i32 0), align 4
%0 = tail call i32 @_tlv_atexit(void (i8*)* bitcast (%class.C* (%class.C*)* @_ZN1CD1Ev to void (i8*)*), i8* bitcast (%class.C* @tC to i8*), i8* nonnull @__dso_handle) #1
ret void
}
[CXX_FAST_TLS] fix issues with O0 on ARM, AArch64 and X86. Since at O0, explicit copies via SplitCSR may not be removed even if they are unnecessary, we choose not to use SplitCSR at O0. llvm-svn: 263855
2016-03-19 07:38:49 +08:00
attributes #0 = { nounwind "no-frame-pointer-elim"="true" }
attributes #1 = { nounwind }