RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-26 12:43:36 +01:00
Code Issues Projects Releases Wiki Activity
c70f94a2a1
llvm-mirror/test/Transforms/Attributor/heap_to_stack.ll

750 lines
30 KiB
LLVM
Raw Normal View History

[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --scrub-attributes
[Attributor][NFC] Remove obsolete option from tests Since D76871 it is sufficient to run `opt -atributor` or `-attributor-cgscc`.
2020-04-18 04:09:16 +02:00
; RUN: opt -attributor -attributor-manifest-internal -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=2 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_NPM,NOT_CGSCC_OPM,NOT_TUNIT_NPM,IS__TUNIT____,IS________OPM,IS__TUNIT_OPM
; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=2 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_OPM,NOT_CGSCC_NPM,NOT_TUNIT_OPM,IS__TUNIT____,IS________NPM,IS__TUNIT_NPM
; RUN: opt -attributor-cgscc -attributor-manifest-internal -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_NPM,IS__CGSCC____,IS________OPM,IS__CGSCC_OPM
; RUN: opt -aa-pipeline=basic-aa -passes=attributor-cgscc -attributor-manifest-internal -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_OPM,IS__CGSCC____,IS________NPM,IS__CGSCC_NPM
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
declare noalias i8* @malloc(i64)
declare void @nocapture_func_frees_pointer(i8* nocapture)
declare void @func_throws(...)
declare void @sync_func(i8* %p)
[Attributor] Add "free"-based heap2stack deduction Summary: If there is a unique free of the allocated that has to be reached from the malloc, we can apply the heap-2-stack transformation even if the pointer escapes. Reviewers: hfinkel, sstefan1, uenoku Subscribers: hiraditya, bollu, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D68958
2019-10-15 00:29:05 +02:00
declare void @sync_will_return(i8* %p) willreturn nounwind
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
declare void @no_sync_func(i8* nocapture %p) nofree nosync willreturn
declare void @nofree_func(i8* nocapture %p) nofree nosync willreturn
declare void @foo(i32* %p)
declare void @foo_nounw(i32* %p) nounwind nofree
declare i32 @no_return_call() noreturn
declare void @free(i8* nocapture)
declare void @llvm.lifetime.start.p0i8(i64, i8* nocapture) nounwind
[Attributor] Use nofree argument attribute for heap-to-stack conversion Reviewers: jdoerfert, uenoku Subscribers: Differential Revision: https://reviews.llvm.org/D70140
2019-11-17 21:35:04 +01:00
define void @nofree_arg_only(i8* %p1, i8* %p2) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@nofree_arg_only
; CHECK-SAME: (i8* nocapture nofree [[P1:%.*]], i8* nocapture [[P2:%.*]])
; CHECK-NEXT: tail call void @free(i8* nocapture [[P2]])
; CHECK-NEXT: tail call void @nofree_func(i8* nocapture nofree [[P1]])
; CHECK-NEXT: ret void
;
[Attributor] Use nofree argument attribute for heap-to-stack conversion Reviewers: jdoerfert, uenoku Subscribers: Differential Revision: https://reviews.llvm.org/D70140
2019-11-17 21:35:04 +01:00
tail call void @free(i8* %p2)
tail call void @nofree_func(i8* %p1)
ret void
}
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
; TEST 1 - negative, pointer freed in another function.
define void @test1() {
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-LABEL: define {{[^@]+}}@test1()
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; CHECK-NEXT: tail call void @nocapture_func_frees_pointer(i8* noalias nocapture [[TMP1]])
; CHECK-NEXT: tail call void (...) @func_throws()
; CHECK-NEXT: tail call void @free(i8* noalias nocapture [[TMP1]])
; CHECK-NEXT: ret void
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @nocapture_func_frees_pointer(i8* %1)
tail call void (...) @func_throws()
tail call void @free(i8* %1)
ret void
}
; TEST 2 - negative, call to a sync function.
define void @test2() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test2()
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; CHECK-NEXT: tail call void @sync_func(i8* [[TMP1]])
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @free(i8* nocapture [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @sync_func(i8* %1)
tail call void @free(i8* %1)
ret void
}
; TEST 3 - 1 malloc, 1 free
define void @test3() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test3()
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: ret void
;
; IS________NPM-LABEL: define {{[^@]+}}@test3()
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; IS________NPM-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @no_sync_func(i8* %1)
tail call void @free(i8* %1)
ret void
}
[Attributor] Use nofree argument attribute for heap-to-stack conversion Reviewers: jdoerfert, uenoku Subscribers: Differential Revision: https://reviews.llvm.org/D70140
2019-11-17 21:35:04 +01:00
define void @test3a(i8* %p) {
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test3a
; IS________OPM-SAME: (i8* nocapture [[P:%.*]])
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: tail call void @nofree_arg_only(i8* nocapture nofree [[TMP1]], i8* nocapture [[P]])
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture [[TMP1]])
; IS________OPM-NEXT: ret void
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
;
; IS________NPM-LABEL: define {{[^@]+}}@test3a
; IS________NPM-SAME: (i8* nocapture [[P:%.*]])
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: tail call void @nofree_arg_only(i8* noalias nocapture nofree [[TMP1]], i8* nocapture [[P]])
; IS________NPM-NEXT: ret void
;
[Attributor] Use nofree argument attribute for heap-to-stack conversion Reviewers: jdoerfert, uenoku Subscribers: Differential Revision: https://reviews.llvm.org/D70140
2019-11-17 21:35:04 +01:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @nofree_arg_only(i8* %1, i8* %p)
tail call void @free(i8* %1)
ret void
}
[Attributor] Make attributor aware of aligned_alloc for heap to stack conversion Make the attributor pass aware of aligned_alloc for converting heap allocations to stack ones. Depends on D76971. Differential Revision: https://reviews.llvm.org/D76974
2020-03-28 08:15:59 +01:00
declare noalias i8* @aligned_alloc(i64, i64)
define void @test3b(i8* %p) {
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test3b
; IS________OPM-SAME: (i8* nocapture [[P:%.*]])
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @aligned_alloc(i64 32, i64 128)
; IS________OPM-NEXT: tail call void @nofree_arg_only(i8* nocapture nofree [[TMP1]], i8* nocapture [[P]])
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture [[TMP1]])
; IS________OPM-NEXT: ret void
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
;
; IS________NPM-LABEL: define {{[^@]+}}@test3b
; IS________NPM-SAME: (i8* nocapture [[P:%.*]])
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 128, align 32
; IS________NPM-NEXT: tail call void @nofree_arg_only(i8* noalias nocapture nofree [[TMP1]], i8* nocapture [[P]])
; IS________NPM-NEXT: ret void
;
[Attributor] Make attributor aware of aligned_alloc for heap to stack conversion Make the attributor pass aware of aligned_alloc for converting heap allocations to stack ones. Depends on D76971. Differential Revision: https://reviews.llvm.org/D76974
2020-03-28 08:15:59 +01:00
%1 = tail call noalias i8* @aligned_alloc(i64 32, i64 128)
tail call void @nofree_arg_only(i8* %1, i8* %p)
tail call void @free(i8* %1)
ret void
}
; leave alone non-constant alignments.
define void @test3c(i64 %alignment) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test3c
; CHECK-SAME: (i64 [[ALIGNMENT:%.*]])
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @aligned_alloc(i64 [[ALIGNMENT]], i64 128)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @free(i8* noalias nocapture [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret void
;
[Attributor] Make attributor aware of aligned_alloc for heap to stack conversion Make the attributor pass aware of aligned_alloc for converting heap allocations to stack ones. Depends on D76971. Differential Revision: https://reviews.llvm.org/D76974
2020-03-28 08:15:59 +01:00
%1 = tail call noalias i8* @aligned_alloc(i64 %alignment, i64 128)
tail call void @free(i8* %1)
ret void
}
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
declare noalias i8* @calloc(i64, i64)
define void @test0() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test0()
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @calloc(i64 2, i64 4)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: ret void
;
; IS________NPM-LABEL: define {{[^@]+}}@test0()
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 8
; IS________NPM-NEXT: [[CALLOC_BC:%.*]] = bitcast i8* [[TMP1]] to i8*
; IS________NPM-NEXT: call void @llvm.memset.p0i8.i64(i8* [[CALLOC_BC]], i8 0, i64 8, i1 false)
; IS________NPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; IS________NPM-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @calloc(i64 2, i64 4)
tail call void @no_sync_func(i8* %1)
tail call void @free(i8* %1)
ret void
}
[Attributor] Make attributor aware of aligned_alloc for heap to stack conversion Make the attributor pass aware of aligned_alloc for converting heap allocations to stack ones. Depends on D76971. Differential Revision: https://reviews.llvm.org/D76974
2020-03-28 08:15:59 +01:00
; TEST 4
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
define void @test4() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test4()
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @nofree_func(i8* noalias nocapture nofree [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: ret void
;
; IS________NPM-LABEL: define {{[^@]+}}@test4()
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: tail call void @nofree_func(i8* noalias nocapture nofree [[TMP1]])
; IS________NPM-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @nofree_func(i8* %1)
ret void
}
; TEST 5 - not all exit paths have a call to free, but all uses of malloc
; are in nofree functions and are not captured
[Attributor] Use nofree argument attribute for heap-to-stack conversion Reviewers: jdoerfert, uenoku Subscribers: Differential Revision: https://reviews.llvm.org/D70140
2019-11-17 21:35:04 +01:00
define void @test5(i32, i8* %p) {
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test5
; IS________OPM-SAME: (i32 [[TMP0:%.*]], i8* nocapture [[P:%.*]])
; IS________OPM-NEXT: [[TMP2:%.*]] = tail call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP0]], 0
; IS________OPM-NEXT: br i1 [[TMP3]], label [[TMP5:%.*]], label [[TMP4:%.*]]
; IS________OPM: 4:
; IS________OPM-NEXT: tail call void @nofree_func(i8* noalias nocapture nofree [[TMP2]])
; IS________OPM-NEXT: br label [[TMP6:%.*]]
; IS________OPM: 5:
; IS________OPM-NEXT: tail call void @nofree_arg_only(i8* nocapture nofree [[TMP2]], i8* nocapture [[P]])
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture [[TMP2]])
; IS________OPM-NEXT: br label [[TMP6]]
; IS________OPM: 6:
; IS________OPM-NEXT: ret void
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
;
; IS________NPM-LABEL: define {{[^@]+}}@test5
; IS________NPM-SAME: (i32 [[TMP0:%.*]], i8* nocapture [[P:%.*]])
; IS________NPM-NEXT: [[TMP2:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP0]], 0
; IS________NPM-NEXT: br i1 [[TMP3]], label [[TMP5:%.*]], label [[TMP4:%.*]]
; IS________NPM: 4:
; IS________NPM-NEXT: tail call void @nofree_func(i8* noalias nocapture nofree [[TMP2]])
; IS________NPM-NEXT: br label [[TMP6:%.*]]
; IS________NPM: 5:
; IS________NPM-NEXT: tail call void @nofree_arg_only(i8* noalias nocapture nofree [[TMP2]], i8* nocapture [[P]])
; IS________NPM-NEXT: br label [[TMP6]]
; IS________NPM: 6:
; IS________NPM-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%2 = tail call noalias i8* @malloc(i64 4)
%3 = icmp eq i32 %0, 0
br i1 %3, label %5, label %4
4: ; preds = %1
tail call void @nofree_func(i8* %2)
br label %6
5: ; preds = %1
[Attributor] Use nofree argument attribute for heap-to-stack conversion Reviewers: jdoerfert, uenoku Subscribers: Differential Revision: https://reviews.llvm.org/D70140
2019-11-17 21:35:04 +01:00
tail call void @nofree_arg_only(i8* %2, i8* %p)
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
tail call void @free(i8* %2)
br label %6
6: ; preds = %5, %4
ret void
}
; TEST 6 - all exit paths have a call to free
define void @test6(i32) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test6
; IS________OPM-SAME: (i32 [[TMP0:%.*]])
; IS________OPM-NEXT: [[TMP2:%.*]] = tail call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP0]], 0
; IS________OPM-NEXT: br i1 [[TMP3]], label [[TMP5:%.*]], label [[TMP4:%.*]]
; IS________OPM: 4:
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @nofree_func(i8* noalias nocapture nofree [[TMP2]])
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture [[TMP2]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: br label [[TMP6:%.*]]
; IS________OPM: 5:
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture [[TMP2]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: br label [[TMP6]]
; IS________OPM: 6:
; IS________OPM-NEXT: ret void
;
; IS________NPM-LABEL: define {{[^@]+}}@test6
; IS________NPM-SAME: (i32 [[TMP0:%.*]])
; IS________NPM-NEXT: [[TMP2:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP0]], 0
; IS________NPM-NEXT: br i1 [[TMP3]], label [[TMP5:%.*]], label [[TMP4:%.*]]
; IS________NPM: 4:
; IS________NPM-NEXT: tail call void @nofree_func(i8* noalias nocapture nofree [[TMP2]])
; IS________NPM-NEXT: br label [[TMP6:%.*]]
; IS________NPM: 5:
; IS________NPM-NEXT: br label [[TMP6]]
; IS________NPM: 6:
; IS________NPM-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%2 = tail call noalias i8* @malloc(i64 4)
%3 = icmp eq i32 %0, 0
br i1 %3, label %5, label %4
4: ; preds = %1
tail call void @nofree_func(i8* %2)
tail call void @free(i8* %2)
br label %6
5: ; preds = %1
tail call void @free(i8* %2)
br label %6
6: ; preds = %5, %4
ret void
}
; TEST 7 - free is dead.
define void @test7() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test7()
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: [[TMP2:%.*]] = tail call i32 @no_return_call()
; IS________OPM-NEXT: unreachable
;
; IS________NPM-LABEL: define {{[^@]+}}@test7()
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: [[TMP2:%.*]] = tail call i32 @no_return_call()
; IS________NPM-NEXT: unreachable
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call i32 @no_return_call()
tail call void @free(i8* %1)
ret void
}
; TEST 8 - Negative: bitcast pointer used in capture function
define void @test8() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test8()
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; CHECK-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; CHECK-NEXT: store i32 10, i32* [[TMP2]], align 4
; CHECK-NEXT: tail call void @foo(i32* align 4 [[TMP2]])
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @free(i8* nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @no_sync_func(i8* %1)
%2 = bitcast i8* %1 to i32*
store i32 10, i32* %2
%3 = load i32, i32* %2
tail call void @foo(i32* %2)
tail call void @free(i8* %1)
ret void
}
; TEST 9 - FIXME: malloc should be converted.
define void @test9() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test9()
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; CHECK-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; CHECK-NEXT: store i32 10, i32* [[TMP2]], align 4
; CHECK-NEXT: tail call void @foo_nounw(i32* nofree align 4 [[TMP2]])
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @free(i8* nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @no_sync_func(i8* %1)
%2 = bitcast i8* %1 to i32*
store i32 10, i32* %2
%3 = load i32, i32* %2
tail call void @foo_nounw(i32* %2)
tail call void @free(i8* %1)
ret void
}
; TEST 10 - 1 malloc, 1 free
define i32 @test10() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test10()
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; IS________OPM-NEXT: store i32 10, i32* [[TMP2]], align 4
; IS________OPM-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: ret i32 [[TMP3]]
;
; IS________NPM-LABEL: define {{[^@]+}}@test10()
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; IS________NPM-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; IS________NPM-NEXT: store i32 10, i32* [[TMP2]], align 4
; IS________NPM-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
; IS________NPM-NEXT: ret i32 [[TMP3]]
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @no_sync_func(i8* %1)
%2 = bitcast i8* %1 to i32*
store i32 10, i32* %2
%3 = load i32, i32* %2
tail call void @free(i8* %1)
ret i32 %3
}
define i32 @test_lifetime() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test_lifetime()
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; IS________OPM-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* noalias nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; IS________OPM-NEXT: store i32 10, i32* [[TMP2]], align 4
; IS________OPM-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: ret i32 [[TMP3]]
;
; IS________NPM-LABEL: define {{[^@]+}}@test_lifetime()
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________NPM-NEXT: call void @llvm.lifetime.start.p0i8(i64 4, i8* noalias nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________NPM-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; IS________NPM-NEXT: store i32 10, i32* [[TMP2]], align 4
; IS________NPM-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
; IS________NPM-NEXT: ret i32 [[TMP3]]
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @no_sync_func(i8* %1)
call void @llvm.lifetime.start.p0i8(i64 4, i8* %1)
%2 = bitcast i8* %1 to i32*
store i32 10, i32* %2
%3 = load i32, i32* %2
tail call void @free(i8* %1)
ret i32 %3
}
[Attributor] Make attributor aware of aligned_alloc for heap to stack conversion Make the attributor pass aware of aligned_alloc for converting heap allocations to stack ones. Depends on D76971. Differential Revision: https://reviews.llvm.org/D76974
2020-03-28 08:15:59 +01:00
; TEST 11
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
define void @test11() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test11()
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: tail call void @sync_will_return(i8* [[TMP1]])
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @free(i8* nocapture [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: ret void
;
; IS________NPM-LABEL: define {{[^@]+}}@test11()
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: tail call void @sync_will_return(i8* [[TMP1]])
; IS________NPM-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
tail call void @sync_will_return(i8* %1)
tail call void @free(i8* %1)
ret void
}
; TEST 12
define i32 @irreducible_cfg(i32 %0) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@irreducible_cfg
; IS________OPM-SAME: (i32 [[TMP0:%.*]])
; IS________OPM-NEXT: [[TMP2:%.*]] = call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32*
; IS________OPM-NEXT: store i32 10, i32* [[TMP3]], align 4
; IS________OPM-NEXT: [[TMP4:%.*]] = icmp eq i32 [[TMP0]], 1
; IS________OPM-NEXT: br i1 [[TMP4]], label [[TMP5:%.*]], label [[TMP7:%.*]]
; IS________OPM: 5:
; IS________OPM-NEXT: [[TMP6:%.*]] = add nsw i32 [[TMP0]], 5
; IS________OPM-NEXT: br label [[TMP13:%.*]]
; IS________OPM: 7:
; IS________OPM-NEXT: br label [[TMP8:%.*]]
; IS________OPM: 8:
; IS________OPM-NEXT: [[DOT0:%.*]] = phi i32 [ [[TMP14:%.*]], [[TMP13]] ], [ 1, [[TMP7]] ]
; IS________OPM-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP3]], align 4
; IS________OPM-NEXT: [[TMP10:%.*]] = add nsw i32 [[TMP9]], -1
; IS________OPM-NEXT: store i32 [[TMP10]], i32* [[TMP3]], align 4
; IS________OPM-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP9]], 0
; IS________OPM-NEXT: br i1 [[TMP11]], label [[TMP12:%.*]], label [[TMP15:%.*]]
; IS________OPM: 12:
; IS________OPM-NEXT: br label [[TMP13]]
; IS________OPM: 13:
; IS________OPM-NEXT: [[DOT1:%.*]] = phi i32 [ [[TMP6]], [[TMP5]] ], [ [[DOT0]], [[TMP12]] ]
; IS________OPM-NEXT: [[TMP14]] = add nsw i32 [[DOT1]], 1
; IS________OPM-NEXT: br label [[TMP8]]
; IS________OPM: 15:
; IS________OPM-NEXT: [[TMP16:%.*]] = bitcast i32* [[TMP3]] to i8*
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: call void @free(i8* nocapture [[TMP16]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: [[TMP17:%.*]] = load i32, i32* [[TMP3]], align 4
; IS________OPM-NEXT: ret i32 [[TMP17]]
;
; IS________NPM-LABEL: define {{[^@]+}}@irreducible_cfg
; IS________NPM-SAME: (i32 [[TMP0:%.*]])
; IS________NPM-NEXT: [[TMP2:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32*
; IS________NPM-NEXT: store i32 10, i32* [[TMP3]], align 4
; IS________NPM-NEXT: [[TMP4:%.*]] = icmp eq i32 [[TMP0]], 1
; IS________NPM-NEXT: br i1 [[TMP4]], label [[TMP5:%.*]], label [[TMP7:%.*]]
; IS________NPM: 5:
; IS________NPM-NEXT: [[TMP6:%.*]] = add nsw i32 [[TMP0]], 5
; IS________NPM-NEXT: br label [[TMP13:%.*]]
; IS________NPM: 7:
; IS________NPM-NEXT: br label [[TMP8:%.*]]
; IS________NPM: 8:
; IS________NPM-NEXT: [[DOT0:%.*]] = phi i32 [ [[TMP14:%.*]], [[TMP13]] ], [ 1, [[TMP7]] ]
; IS________NPM-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP3]], align 4
; IS________NPM-NEXT: [[TMP10:%.*]] = add nsw i32 [[TMP9]], -1
; IS________NPM-NEXT: store i32 [[TMP10]], i32* [[TMP3]], align 4
; IS________NPM-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP9]], 0
; IS________NPM-NEXT: br i1 [[TMP11]], label [[TMP12:%.*]], label [[TMP15:%.*]]
; IS________NPM: 12:
; IS________NPM-NEXT: br label [[TMP13]]
; IS________NPM: 13:
; IS________NPM-NEXT: [[DOT1:%.*]] = phi i32 [ [[TMP6]], [[TMP5]] ], [ [[DOT0]], [[TMP12]] ]
; IS________NPM-NEXT: [[TMP14]] = add nsw i32 [[DOT1]], 1
; IS________NPM-NEXT: br label [[TMP8]]
; IS________NPM: 15:
; IS________NPM-NEXT: [[TMP16:%.*]] = bitcast i32* [[TMP3]] to i8*
; IS________NPM-NEXT: [[TMP17:%.*]] = load i32, i32* [[TMP3]], align 4
; IS________NPM-NEXT: ret i32 [[TMP17]]
;
[Attributor][FIX] Ensure h2s doesn't trigger on escaped pointers We do not yet perform h2s because we know something is free'ed but we do it because we know the pointer does not escape. Storing the pointer allows it to escape so we have to prevent that. llvm-svn: 374699
2019-10-13 06:14:15 +02:00
%2 = call noalias i8* @malloc(i64 4)
%3 = bitcast i8* %2 to i32*
store i32 10, i32* %3, align 4
%4 = icmp eq i32 %0, 1
br i1 %4, label %5, label %7
5: ; preds = %1
%6 = add nsw i32 %0, 5
br label %13
7: ; preds = %1
br label %8
8: ; preds = %13, %7
%.0 = phi i32 [ %14, %13 ], [ 1, %7 ]
%9 = load i32, i32* %3, align 4
%10 = add nsw i32 %9, -1
store i32 %10, i32* %3, align 4
%11 = icmp ne i32 %9, 0
br i1 %11, label %12, label %15
12: ; preds = %8
br label %13
13: ; preds = %12, %5
%.1 = phi i32 [ %6, %5 ], [ %.0, %12 ]
%14 = add nsw i32 %.1, 1
br label %8
15: ; preds = %8
%16 = load i32, i32* %3, align 4
%17 = bitcast i32* %3 to i8*
call void @free(i8* %17)
%18 = load i32, i32* %3, align 4
ret i32 %18
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
}
[Attributor][FIX] Ensure h2s doesn't trigger on escaped pointers We do not yet perform h2s because we know something is free'ed but we do it because we know the pointer does not escape. Storing the pointer allows it to escape so we have to prevent that. llvm-svn: 374699
2019-10-13 06:14:15 +02:00
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
define i32 @malloc_in_loop(i32 %0) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@malloc_in_loop
; IS________OPM-SAME: (i32 [[TMP0:%.*]])
; IS________OPM-NEXT: [[TMP2:%.*]] = alloca i32, align 4
; IS________OPM-NEXT: [[TMP3:%.*]] = alloca i32*, align 8
; IS________OPM-NEXT: store i32 [[TMP0]], i32* [[TMP2]], align 4
; IS________OPM-NEXT: br label [[TMP4:%.*]]
; IS________OPM: 4:
; IS________OPM-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP2]], align 4
; IS________OPM-NEXT: [[TMP6:%.*]] = add nsw i32 [[TMP5]], -1
; IS________OPM-NEXT: store i32 [[TMP6]], i32* [[TMP2]], align 4
; IS________OPM-NEXT: [[TMP7:%.*]] = icmp sgt i32 [[TMP6]], 0
; IS________OPM-NEXT: br i1 [[TMP7]], label [[TMP8:%.*]], label [[TMP11:%.*]]
; IS________OPM: 8:
; IS________OPM-NEXT: [[TMP9:%.*]] = call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: [[TMP10:%.*]] = bitcast i8* [[TMP9]] to i32*
; IS________OPM-NEXT: store i32 1, i32* [[TMP10]], align 8
; IS________OPM-NEXT: br label [[TMP4]]
; IS________OPM: 11:
; IS________OPM-NEXT: ret i32 5
;
; IS________NPM-LABEL: define {{[^@]+}}@malloc_in_loop
; IS________NPM-SAME: (i32 [[TMP0:%.*]])
; IS________NPM-NEXT: [[TMP2:%.*]] = alloca i32, align 4
; IS________NPM-NEXT: [[TMP3:%.*]] = alloca i32*, align 8
; IS________NPM-NEXT: store i32 [[TMP0]], i32* [[TMP2]], align 4
; IS________NPM-NEXT: br label [[TMP4:%.*]]
; IS________NPM: 4:
; IS________NPM-NEXT: [[TMP5:%.*]] = load i32, i32* [[TMP2]], align 4
; IS________NPM-NEXT: [[TMP6:%.*]] = add nsw i32 [[TMP5]], -1
; IS________NPM-NEXT: store i32 [[TMP6]], i32* [[TMP2]], align 4
; IS________NPM-NEXT: [[TMP7:%.*]] = icmp sgt i32 [[TMP6]], 0
; IS________NPM-NEXT: br i1 [[TMP7]], label [[TMP8:%.*]], label [[TMP11:%.*]]
; IS________NPM: 8:
; IS________NPM-NEXT: [[TMP9:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: [[TMP10:%.*]] = bitcast i8* [[TMP9]] to i32*
; IS________NPM-NEXT: store i32 1, i32* [[TMP10]], align 8
; IS________NPM-NEXT: br label [[TMP4]]
; IS________NPM: 11:
; IS________NPM-NEXT: ret i32 5
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%2 = alloca i32, align 4
%3 = alloca i32*, align 8
store i32 %0, i32* %2, align 4
br label %4
4: ; preds = %8, %1
%5 = load i32, i32* %2, align 4
%6 = add nsw i32 %5, -1
store i32 %6, i32* %2, align 4
%7 = icmp sgt i32 %6, 0
br i1 %7, label %8, label %11
8: ; preds = %4
%9 = call noalias i8* @malloc(i64 4)
%10 = bitcast i8* %9 to i32*
[Attributor][FIX] Ensure h2s doesn't trigger on escaped pointers We do not yet perform h2s because we know something is free'ed but we do it because we know the pointer does not escape. Storing the pointer allows it to escape so we have to prevent that. llvm-svn: 374699
2019-10-13 06:14:15 +02:00
store i32 1, i32* %10, align 8
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
br label %4
11: ; preds = %4
ret i32 5
}
; Malloc/Calloc too large
define i32 @test13() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test13()
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 256)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; CHECK-NEXT: store i32 10, i32* [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @free(i8* noalias nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret i32 [[TMP3]]
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @malloc(i64 256)
tail call void @no_sync_func(i8* %1)
%2 = bitcast i8* %1 to i32*
store i32 10, i32* %2
%3 = load i32, i32* %2
tail call void @free(i8* %1)
ret i32 %3
}
[Attributor][NFC] Add test for sle comparison in h2s.
2019-12-17 20:36:10 +01:00
define i32 @test_sle() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test_sle()
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 -1)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; CHECK-NEXT: store i32 10, i32* [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @free(i8* noalias nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret i32 [[TMP3]]
;
[Attributor][NFC] Add test for sle comparison in h2s.
2019-12-17 20:36:10 +01:00
%1 = tail call noalias i8* @malloc(i64 -1)
tail call void @no_sync_func(i8* %1)
%2 = bitcast i8* %1 to i32*
store i32 10, i32* %2
%3 = load i32, i32* %2
tail call void @free(i8* %1)
[Attributor] H2S fix. Summary: Fixing issues that were noticed in D71521 Reviewers: jdoerfert, lebedev.ri, uenoku Subscribers: Differential Revision: https://reviews.llvm.org/D71564
2019-12-17 20:41:09 +01:00
ret i32 %3
}
define i32 @test_overflow() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test_overflow()
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @calloc(i64 65537, i64 65537)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i32*
; CHECK-NEXT: store i32 10, i32* [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @free(i8* noalias nocapture nonnull align 4 dereferenceable(4) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret i32 [[TMP3]]
;
[Attributor] H2S fix. Summary: Fixing issues that were noticed in D71521 Reviewers: jdoerfert, lebedev.ri, uenoku Subscribers: Differential Revision: https://reviews.llvm.org/D71564
2019-12-17 20:41:09 +01:00
%1 = tail call noalias i8* @calloc(i64 65537, i64 65537)
tail call void @no_sync_func(i8* %1)
%2 = bitcast i8* %1 to i32*
store i32 10, i32* %2
%3 = load i32, i32* %2
tail call void @free(i8* %1)
[Attributor][NFC] Add test for sle comparison in h2s.
2019-12-17 20:36:10 +01:00
ret i32 %3
}
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
define void @test14() {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test14()
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @calloc(i64 64, i64 4)
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; CHECK-NEXT: tail call void @free(i8* noalias nocapture [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret void
;
[Attributor] Heap-To-Stack Conversion D53362 gives a prototype heap-to-stack conversion pass. With addition of new attributes in the attributor, this can now be revisted and improved. This will place it in the Attributor to make it easier to use new attributes (eg. nofree, nosync, willreturn, etc.) and other attributor features. Reviewers: jdoerfert, uenoku, hfinkel, efriedma Subscribers: lebedev.ri, xbolva00, hiraditya, llvm-commits Differential Revision: https://reviews.llvm.org/D65408 llvm-svn: 371942
2019-09-15 23:47:41 +02:00
%1 = tail call noalias i8* @calloc(i64 64, i64 4)
tail call void @no_sync_func(i8* %1)
tail call void @free(i8* %1)
ret void
}
[Attributor][FIX] Do not apply h2s for arbitrary mallocs H2S did apply to mallocs of non-constant sizes if the uses were OK. This is now forbidden through reording of the "good" and "bad" cases in the conditional. llvm-svn: 374698
2019-10-13 05:54:08 +02:00
define void @test15(i64 %S) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test15
; CHECK-SAME: (i64 [[S:%.*]])
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 [[S]])
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree [[TMP1]])
; CHECK-NEXT: tail call void @free(i8* noalias nocapture [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-NEXT: ret void
;
[Attributor][FIX] Do not apply h2s for arbitrary mallocs H2S did apply to mallocs of non-constant sizes if the uses were OK. This is now forbidden through reording of the "good" and "bad" cases in the conditional. llvm-svn: 374698
2019-10-13 05:54:08 +02:00
%1 = tail call noalias i8* @malloc(i64 %S)
[Attributor][FIX] Ensure h2s doesn't trigger on escaped pointers We do not yet perform h2s because we know something is free'ed but we do it because we know the pointer does not escape. Storing the pointer allows it to escape so we have to prevent that. llvm-svn: 374699
2019-10-13 06:14:15 +02:00
tail call void @no_sync_func(i8* %1)
tail call void @free(i8* %1)
ret void
}
define void @test16a(i8 %v, i8** %P) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test16a
; IS________OPM-SAME: (i8 [[V:%.*]], i8** nocapture nofree readnone [[P:%.*]])
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: store i8 [[V]], i8* [[TMP1]]
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree nonnull dereferenceable(1) [[TMP1]])
; IS________OPM-NEXT: tail call void @free(i8* noalias nocapture nonnull dereferenceable(1) [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: ret void
;
; IS________NPM-LABEL: define {{[^@]+}}@test16a
; IS________NPM-SAME: (i8 [[V:%.*]], i8** nocapture nofree readnone [[P:%.*]])
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: store i8 [[V]], i8* [[TMP1]]
; IS________NPM-NEXT: tail call void @no_sync_func(i8* noalias nocapture nofree nonnull dereferenceable(1) [[TMP1]])
; IS________NPM-NEXT: ret void
;
[Attributor][FIX] Ensure h2s doesn't trigger on escaped pointers We do not yet perform h2s because we know something is free'ed but we do it because we know the pointer does not escape. Storing the pointer allows it to escape so we have to prevent that. llvm-svn: 374699
2019-10-13 06:14:15 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
store i8 %v, i8* %1
tail call void @no_sync_func(i8* %1)
[MustExecute] Add backward exploration for must-be-executed-context Summary: As mentioned in D71974, it is useful for must-be-executed-context to explore CFG backwardly. This patch is ported from parts of D64975. We use a dominator tree to find the previous context if a dominator tree is available. Reviewers: jdoerfert, hfinkel, baziotis, sstefan1 Reviewed By: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74817
2020-02-20 05:00:43 +01:00
tail call void @free(i8* nonnull dereferenceable(1) %1)
[Attributor][FIX] Ensure h2s doesn't trigger on escaped pointers We do not yet perform h2s because we know something is free'ed but we do it because we know the pointer does not escape. Storing the pointer allows it to escape so we have to prevent that. llvm-svn: 374699
2019-10-13 06:14:15 +02:00
ret void
}
define void @test16b(i8 %v, i8** %P) {
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; CHECK-LABEL: define {{[^@]+}}@test16b
; CHECK-SAME: (i8 [[V:%.*]], i8** nocapture writeonly [[P:%.*]])
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; CHECK-NEXT: store i8* [[TMP1]], i8** [[P]]
; CHECK-NEXT: tail call void @no_sync_func(i8* nocapture nofree [[TMP1]])
; CHECK-NEXT: tail call void @free(i8* nocapture [[TMP1]])
; CHECK-NEXT: ret void
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
;
[Attributor][FIX] Ensure h2s doesn't trigger on escaped pointers We do not yet perform h2s because we know something is free'ed but we do it because we know the pointer does not escape. Storing the pointer allows it to escape so we have to prevent that. llvm-svn: 374699
2019-10-13 06:14:15 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
store i8* %1, i8** %P
tail call void @no_sync_func(i8* %1)
tail call void @free(i8* %1)
[Attributor][FIX] Do not apply h2s for arbitrary mallocs H2S did apply to mallocs of non-constant sizes if the uses were OK. This is now forbidden through reording of the "good" and "bad" cases in the conditional. llvm-svn: 374698
2019-10-13 05:54:08 +02:00
ret void
}
[Attributor] Add "free"-based heap2stack deduction Summary: If there is a unique free of the allocated that has to be reached from the malloc, we can apply the heap-2-stack transformation even if the pointer escapes. Reviewers: hfinkel, sstefan1, uenoku Subscribers: hiraditya, bollu, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D68958
2019-10-15 00:29:05 +02:00
define void @test16c(i8 %v, i8** %P) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-LABEL: define {{[^@]+}}@test16c
; IS________OPM-SAME: (i8 [[V:%.*]], i8** nocapture writeonly [[P:%.*]])
; IS________OPM-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; IS________OPM-NEXT: store i8* [[TMP1]], i8** [[P]]
[Attributor][NFC] Proactively ask for `nocapure` on call site arguments This minimizes test noise later on and is in line with other attributes we derive proactively.
2020-05-04 02:08:30 +02:00
; IS________OPM-NEXT: tail call void @no_sync_func(i8* nocapture nofree [[TMP1]])
; IS________OPM-NEXT: tail call void @free(i8* nocapture [[TMP1]])
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; IS________OPM-NEXT: ret void
;
; IS________NPM-LABEL: define {{[^@]+}}@test16c
; IS________NPM-SAME: (i8 [[V:%.*]], i8** nocapture writeonly [[P:%.*]])
; IS________NPM-NEXT: [[TMP1:%.*]] = alloca i8, i64 4
; IS________NPM-NEXT: store i8* [[TMP1]], i8** [[P]]
; IS________NPM-NEXT: tail call void @no_sync_func(i8* nocapture nofree [[TMP1]])
; IS________NPM-NEXT: ret void
;
[Attributor] Add "free"-based heap2stack deduction Summary: If there is a unique free of the allocated that has to be reached from the malloc, we can apply the heap-2-stack transformation even if the pointer escapes. Reviewers: hfinkel, sstefan1, uenoku Subscribers: hiraditya, bollu, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D68958
2019-10-15 00:29:05 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
store i8* %1, i8** %P
tail call void @no_sync_func(i8* %1) nounwind
tail call void @free(i8* %1)
ret void
}
define void @test16d(i8 %v, i8** %P) {
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 02:04:57 +02:00
; CHECK-LABEL: define {{[^@]+}}@test16d
; CHECK-SAME: (i8 [[V:%.*]], i8** nocapture writeonly [[P:%.*]])
; CHECK-NEXT: [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 4)
; CHECK-NEXT: store i8* [[TMP1]], i8** [[P]]
; CHECK-NEXT: ret void
;
[Attributor] Add "free"-based heap2stack deduction Summary: If there is a unique free of the allocated that has to be reached from the malloc, we can apply the heap-2-stack transformation even if the pointer escapes. Reviewers: hfinkel, sstefan1, uenoku Subscribers: hiraditya, bollu, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D68958
2019-10-15 00:29:05 +02:00
%1 = tail call noalias i8* @malloc(i64 4)
store i8* %1, i8** %P
ret void
}
Reference in New Issue Copy Permalink