maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Transforms/Attributor/nonnull.ll

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

1399 lines
53 KiB
LLVM
Raw Normal View History

[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --scrub-attributes --check-attributes
[Attributor] Implement AAPotentialValues This patch provides an implementation of `AAPotentialValues`. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D85632
2020-08-14 19:51:14 +08:00
; RUN: opt -attributor -enable-new-pm=0 -attributor-manifest-internal -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=15 -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=17 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_OPM,NOT_CGSCC_NPM,NOT_TUNIT_OPM,IS__TUNIT____,IS________NPM,IS__TUNIT_NPM
[NewPM][Attributor] Pin tests with -attributor to legacy PM All these tests already explicitly test against both legacy PM and NPM. $ sed -i 's/ -attributor / -attributor -enable-new-pm=0 /g' $(rg --path-separator // -l -- -passes=) $ sed -i 's/ -attributor-cgscc / -attributor-cgscc -enable-new-pm=0 /g' $(rg --path-separator // -l -- -passes=) Now all tests in Transforms/Attributor/ pass under NPM. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D84813
2020-07-29 08:57:21 +08:00
; RUN: opt -attributor-cgscc -enable-new-pm=0 -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
[Attributor][NFC] Remove obsolete option from tests Since D76871 it is sufficient to run `opt -atributor` or `-attributor-cgscc`.
2020-04-18 10:09:16 +08:00
; 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] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
[Attributor][NFC] Prepare some tests to be used with update test script
2020-02-20 16:06:14 +08:00
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
declare nonnull i8* @ret_nonnull()
[AssumeBundles] Use assume bundles in isKnownNonZero Summary: Use nonnull and dereferenceable from an assume bundle in isKnownNonZero Reviewers: jdoerfert, nikic, lebedev.ri, reames, fhahn, sstefan1 Reviewed By: jdoerfert Subscribers: fhahn, hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D76149
2020-04-25 01:46:18 +08:00
declare void @llvm.assume(i1)
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; Return a pointer trivially nonnull (call return attribute)
define i8* @test1() {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@test1()
; CHECK-NEXT: [[RET:%.*]] = call nonnull i8* @ret_nonnull()
; CHECK-NEXT: ret i8* [[RET]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%ret = call i8* @ret_nonnull()
ret i8* %ret
}
; Return a pointer trivially nonnull (argument attribute)
define i8* @test2(i8* nonnull %p) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test2
; IS__TUNIT____-SAME: (i8* nofree nonnull readnone returned "no-capture-maybe-returned" [[P:%.*]])
; IS__TUNIT____-NEXT: ret i8* [[P]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test2
; IS__CGSCC____-SAME: (i8* nofree nonnull readnone returned "no-capture-maybe-returned" [[P:%.*]])
; IS__CGSCC____-NEXT: ret 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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
ret i8* %p
}
[AssumeBundles] Use assume bundles in isKnownNonZero Summary: Use nonnull and dereferenceable from an assume bundle in isKnownNonZero Reviewers: jdoerfert, nikic, lebedev.ri, reames, fhahn, sstefan1 Reviewed By: jdoerfert Subscribers: fhahn, hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D76149
2020-04-25 01:46:18 +08:00
define i8* @test2A(i1 %c, i8* %ret) {
; ATTRIBUTOR: define nonnull i8* @test2A(i1 %c, i8* nofree nonnull readnone returned %ret)
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; NOT_CGSCC_OPM: Function Attrs: nounwind willreturn
[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 08:08:30 +08:00
; NOT_CGSCC_OPM-LABEL: define {{[^@]+}}@test2A
; NOT_CGSCC_OPM-SAME: (i1 [[C:%.*]], i8* nofree nonnull readnone returned "no-capture-maybe-returned" [[RET:%.*]])
; NOT_CGSCC_OPM-NEXT: br i1 [[C]], label [[A:%.*]], label [[B:%.*]]
; NOT_CGSCC_OPM: A:
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; NOT_CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR11:#.*]] [ "nonnull"(i8* [[RET]]) ]
[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 08:08:30 +08:00
; NOT_CGSCC_OPM-NEXT: ret i8* [[RET]]
; NOT_CGSCC_OPM: B:
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; NOT_CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR11]] [ "nonnull"(i8* [[RET]]) ]
[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 08:08:30 +08:00
; NOT_CGSCC_OPM-NEXT: ret i8* [[RET]]
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__CGSCC_OPM: Function Attrs: nounwind willreturn
[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 08:08:30 +08:00
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@test2A
; IS__CGSCC_OPM-SAME: (i1 [[C:%.*]], i8* nofree nonnull readnone returned "no-capture-maybe-returned" [[RET:%.*]])
; IS__CGSCC_OPM-NEXT: br i1 [[C]], label [[A:%.*]], label [[B:%.*]]
; IS__CGSCC_OPM: A:
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; IS__CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR12:#.*]] [ "nonnull"(i8* [[RET]]) ]
[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 08:08:30 +08:00
; IS__CGSCC_OPM-NEXT: ret i8* [[RET]]
; IS__CGSCC_OPM: B:
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; IS__CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR12]] [ "nonnull"(i8* [[RET]]) ]
[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 08:08:30 +08:00
; IS__CGSCC_OPM-NEXT: ret i8* [[RET]]
;
[AssumeBundles] Use assume bundles in isKnownNonZero Summary: Use nonnull and dereferenceable from an assume bundle in isKnownNonZero Reviewers: jdoerfert, nikic, lebedev.ri, reames, fhahn, sstefan1 Reviewed By: jdoerfert Subscribers: fhahn, hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D76149
2020-04-25 01:46:18 +08:00
br i1 %c, label %A, label %B
A:
call void @llvm.assume(i1 true) [ "nonnull"(i8* %ret) ]
ret i8* %ret
B:
call void @llvm.assume(i1 true) [ "nonnull"(i8* %ret) ]
ret i8* %ret
}
define i8* @test2B(i1 %c, i8* %ret) {
; ATTRIBUTOR: define nonnull dereferenceable(4) i8* @test2B(i1 %c, i8* nofree nonnull readnone returned dereferenceable(4) %ret)
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; NOT_CGSCC_OPM: Function Attrs: nounwind willreturn
[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 08:08:30 +08:00
; NOT_CGSCC_OPM-LABEL: define {{[^@]+}}@test2B
; NOT_CGSCC_OPM-SAME: (i1 [[C:%.*]], i8* nofree nonnull readnone returned dereferenceable(4) "no-capture-maybe-returned" [[RET:%.*]])
; NOT_CGSCC_OPM-NEXT: br i1 [[C]], label [[A:%.*]], label [[B:%.*]]
; NOT_CGSCC_OPM: A:
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; NOT_CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR11]] [ "dereferenceable"(i8* [[RET]], i32 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 08:08:30 +08:00
; NOT_CGSCC_OPM-NEXT: ret i8* [[RET]]
; NOT_CGSCC_OPM: B:
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; NOT_CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR11]] [ "dereferenceable"(i8* [[RET]], i32 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 08:08:30 +08:00
; NOT_CGSCC_OPM-NEXT: ret i8* [[RET]]
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__CGSCC_OPM: Function Attrs: nounwind willreturn
[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 08:08:30 +08:00
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@test2B
; IS__CGSCC_OPM-SAME: (i1 [[C:%.*]], i8* nofree nonnull readnone returned dereferenceable(4) "no-capture-maybe-returned" [[RET:%.*]])
; IS__CGSCC_OPM-NEXT: br i1 [[C]], label [[A:%.*]], label [[B:%.*]]
; IS__CGSCC_OPM: A:
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; IS__CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR12]] [ "dereferenceable"(i8* [[RET]], i32 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 08:08:30 +08:00
; IS__CGSCC_OPM-NEXT: ret i8* [[RET]]
; IS__CGSCC_OPM: B:
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; IS__CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR12]] [ "dereferenceable"(i8* [[RET]], i32 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 08:08:30 +08:00
; IS__CGSCC_OPM-NEXT: ret i8* [[RET]]
;
[AssumeBundles] Use assume bundles in isKnownNonZero Summary: Use nonnull and dereferenceable from an assume bundle in isKnownNonZero Reviewers: jdoerfert, nikic, lebedev.ri, reames, fhahn, sstefan1 Reviewed By: jdoerfert Subscribers: fhahn, hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D76149
2020-04-25 01:46:18 +08:00
br i1 %c, label %A, label %B
A:
call void @llvm.assume(i1 true) [ "dereferenceable"(i8* %ret, i32 4) ]
ret i8* %ret
B:
call void @llvm.assume(i1 true) [ "dereferenceable"(i8* %ret, i32 4) ]
ret i8* %ret
}
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; Given an SCC where one of the functions can not be marked nonnull,
; can we still mark the other one which is trivially nonnull
define i8* @scc_binder(i1 %c) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@scc_binder
; CHECK-SAME: (i1 [[C:%.*]])
; CHECK-NEXT: br i1 [[C]], label [[REC:%.*]], label [[END:%.*]]
; CHECK: rec:
; CHECK-NEXT: [[TMP1:%.*]] = call i8* @test3(i1 [[C]])
; CHECK-NEXT: br label [[END]]
; CHECK: end:
; CHECK-NEXT: ret i8* null
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
br i1 %c, label %rec, label %end
rec:
call i8* @test3(i1 %c)
br label %end
end:
ret i8* null
}
define i8* @test3(i1 %c) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@test3
; CHECK-SAME: (i1 [[C:%.*]])
; CHECK-NEXT: [[TMP1:%.*]] = call i8* @scc_binder(i1 [[C]])
; CHECK-NEXT: [[RET:%.*]] = call nonnull i8* @ret_nonnull()
; CHECK-NEXT: ret i8* [[RET]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call i8* @scc_binder(i1 %c)
%ret = call i8* @ret_nonnull()
ret i8* %ret
}
; Given a mutual recursive set of functions, we can mark them
; nonnull if neither can ever return null. (In this case, they
; just never return period.)
define i8* @test4_helper() {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; NOT_CGSCC_NPM: Function Attrs: nofree noreturn nosync nounwind readnone
; NOT_CGSCC_NPM-LABEL: define {{[^@]+}}@test4_helper()
; NOT_CGSCC_NPM-NEXT: unreachable
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse noreturn nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test4_helper()
; IS__CGSCC_NPM-NEXT: unreachable
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%ret = call i8* @test4()
ret i8* %ret
}
define i8* @test4() {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; NOT_CGSCC_NPM: Function Attrs: nofree noreturn nosync nounwind readnone
; NOT_CGSCC_NPM-LABEL: define {{[^@]+}}@test4()
; NOT_CGSCC_NPM-NEXT: unreachable
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse noreturn nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test4()
; IS__CGSCC_NPM-NEXT: unreachable
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%ret = call i8* @test4_helper()
ret i8* %ret
}
; Given a mutual recursive set of functions which *can* return null
; make sure we haven't marked them as nonnull.
define i8* @test5_helper(i1 %c) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; NOT_CGSCC_NPM: Function Attrs: nofree nosync nounwind readnone willreturn
; NOT_CGSCC_NPM-LABEL: define {{[^@]+}}@test5_helper
; NOT_CGSCC_NPM-SAME: (i1 [[C:%.*]])
; NOT_CGSCC_NPM-NEXT: br i1 [[C]], label [[REC:%.*]], label [[END:%.*]]
; NOT_CGSCC_NPM: rec:
; NOT_CGSCC_NPM-NEXT: br label [[END]]
; NOT_CGSCC_NPM: end:
; NOT_CGSCC_NPM-NEXT: ret i8* null
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test5_helper
; IS__CGSCC_NPM-SAME: (i1 [[C:%.*]])
; IS__CGSCC_NPM-NEXT: br i1 [[C]], label [[REC:%.*]], label [[END:%.*]]
; IS__CGSCC_NPM: rec:
; IS__CGSCC_NPM-NEXT: br label [[END]]
; IS__CGSCC_NPM: end:
; IS__CGSCC_NPM-NEXT: ret i8* null
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
br i1 %c, label %rec, label %end
rec:
%ret = call i8* @test5(i1 %c)
br label %end
end:
ret i8* null
}
define i8* @test5(i1 %c) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; NOT_CGSCC_NPM: Function Attrs: nofree nosync nounwind readnone willreturn
; NOT_CGSCC_NPM-LABEL: define {{[^@]+}}@test5
; NOT_CGSCC_NPM-SAME: (i1 [[C:%.*]])
; NOT_CGSCC_NPM-NEXT: ret i8* null
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test5
; IS__CGSCC_NPM-SAME: (i1 [[C:%.*]])
; IS__CGSCC_NPM-NEXT: ret i8* null
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%ret = call i8* @test5_helper(i1 %c)
ret i8* %ret
}
; Local analysis, but going through a self recursive phi
define i8* @test6a() {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: noreturn
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@test6a()
; CHECK-NEXT: entry:
; CHECK-NEXT: [[RET:%.*]] = call i8* @ret_nonnull()
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: unreachable
; CHECK: exit:
; CHECK-NEXT: unreachable
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
entry:
%ret = call i8* @ret_nonnull()
br label %loop
loop:
%phi = phi i8* [%ret, %entry], [%phi, %loop]
br i1 undef, label %loop, label %exit
exit:
ret i8* %phi
}
define i8* @test6b(i1 %c) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@test6b
; CHECK-SAME: (i1 [[C:%.*]])
; CHECK-NEXT: entry:
; CHECK-NEXT: [[RET:%.*]] = call nonnull i8* @ret_nonnull()
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: [[PHI:%.*]] = phi i8* [ [[RET]], [[ENTRY:%.*]] ], [ [[PHI]], [[LOOP]] ]
; CHECK-NEXT: br i1 [[C]], label [[LOOP]], label [[EXIT:%.*]]
; CHECK: exit:
; CHECK-NEXT: ret i8* [[PHI]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
entry:
%ret = call i8* @ret_nonnull()
br label %loop
loop:
%phi = phi i8* [%ret, %entry], [%phi, %loop]
br i1 %c, label %loop, label %exit
exit:
ret i8* %phi
}
define i8* @test7(i8* %a) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test7
; IS__TUNIT____-SAME: (i8* nofree readnone returned "no-capture-maybe-returned" [[A:%.*]])
; IS__TUNIT____-NEXT: [[B:%.*]] = getelementptr inbounds i8, i8* [[A]], i64 0
; IS__TUNIT____-NEXT: ret i8* [[B]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test7
; IS__CGSCC____-SAME: (i8* nofree readnone returned "no-capture-maybe-returned" [[A:%.*]])
; IS__CGSCC____-NEXT: [[B:%.*]] = getelementptr inbounds i8, i8* [[A]], i64 0
; IS__CGSCC____-NEXT: ret i8* [[B]]
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%b = getelementptr inbounds i8, i8* %a, i64 0
ret i8* %b
}
define i8* @test8(i8* %a) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test8
; IS__TUNIT____-SAME: (i8* nofree readnone "no-capture-maybe-returned" [[A:%.*]])
; IS__TUNIT____-NEXT: [[B:%.*]] = getelementptr inbounds i8, i8* [[A]], i64 1
; IS__TUNIT____-NEXT: ret i8* [[B]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test8
; IS__CGSCC____-SAME: (i8* nofree readnone "no-capture-maybe-returned" [[A:%.*]])
; IS__CGSCC____-NEXT: [[B:%.*]] = getelementptr inbounds i8, i8* [[A]], i64 1
; IS__CGSCC____-NEXT: ret i8* [[B]]
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%b = getelementptr inbounds i8, i8* %a, i64 1
ret i8* %b
}
define i8* @test9(i8* %a, i64 %n) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test9
; IS__TUNIT____-SAME: (i8* nofree readnone "no-capture-maybe-returned" [[A:%.*]], i64 [[N:%.*]])
; IS__TUNIT____-NEXT: [[B:%.*]] = getelementptr inbounds i8, i8* [[A]], i64 [[N]]
; IS__TUNIT____-NEXT: ret i8* [[B]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test9
; IS__CGSCC____-SAME: (i8* nofree readnone "no-capture-maybe-returned" [[A:%.*]], i64 [[N:%.*]])
; IS__CGSCC____-NEXT: [[B:%.*]] = getelementptr inbounds i8, i8* [[A]], i64 [[N]]
; IS__CGSCC____-NEXT: ret i8* [[B]]
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%b = getelementptr inbounds i8, i8* %a, i64 %n
ret i8* %b
}
[AssumeBundles] Use assume bundles in isKnownNonZero Summary: Use nonnull and dereferenceable from an assume bundle in isKnownNonZero Reviewers: jdoerfert, nikic, lebedev.ri, reames, fhahn, sstefan1 Reviewed By: jdoerfert Subscribers: fhahn, hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D76149
2020-04-25 01:46:18 +08:00
; ATTRIBUTOR_OPM: define i8* @test10
; ATTRIBUTOR_NPM: define nonnull i8* @test10
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
define i8* @test10(i8* %a, i64 %n) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: nounwind willreturn
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@test10
; CHECK-SAME: (i8* nofree readnone "no-capture-maybe-returned" [[A:%.*]], i64 [[N:%.*]])
[Attributor][FIX] Handle droppable uses when replacing values Since we use the fact that some uses are droppable in the Attributor we need to handle them explicitly when we replace uses. As an example, an assumed dead value can have live droppable users. In those we cannot replace the value simply by an undef. Instead, we either drop the uses (via `dropDroppableUses`) or keep them as they are. In this patch we do both, depending on the situation. For values that are dead but not necessarily removed we keep droppable uses around because they contain information we might be able to use later. For values that are removed we drop droppable uses explicitly to avoid replacement with undef.
2020-04-11 12:41:59 +08:00
; CHECK-NEXT: call void @llvm.assume(i1 true)
[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 08:04:57 +08:00
; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds i8, i8* [[A]], i64 [[N]]
; CHECK-NEXT: ret i8* [[B]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%cmp = icmp ne i64 %n, 0
call void @llvm.assume(i1 %cmp)
%b = getelementptr inbounds i8, i8* %a, i64 %n
ret i8* %b
}
; TEST 11
; char* test11(char *p) {
; return p? p: nonnull();
; }
; FIXME: missing nonnull
define i8* @test11(i8*) local_unnamed_addr {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@test11
; CHECK-SAME: (i8* [[TMP0:%.*]]) local_unnamed_addr
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i8* [[TMP0]], null
; CHECK-NEXT: br i1 [[TMP2]], label [[TMP3:%.*]], label [[TMP5:%.*]]
; CHECK: 3:
; CHECK-NEXT: [[TMP4:%.*]] = tail call i8* @ret_nonnull()
; CHECK-NEXT: br label [[TMP5]]
; CHECK: 5:
; CHECK-NEXT: [[TMP6:%.*]] = phi i8* [ [[TMP4]], [[TMP3]] ], [ [[TMP0]], [[TMP1:%.*]] ]
; CHECK-NEXT: ret i8* [[TMP6]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%2 = icmp eq i8* %0, null
br i1 %2, label %3, label %5
; <label>:3: ; preds = %1
%4 = tail call i8* @ret_nonnull()
br label %5
; <label>:5: ; preds = %3, %1
%6 = phi i8* [ %4, %3 ], [ %0, %1 ]
ret i8* %6
}
; TEST 12
; Simple CallSite Test
declare void @test12_helper(i8*)
define void @test12(i8* nonnull %a) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@test12
; CHECK-SAME: (i8* nonnull [[A:%.*]])
; CHECK-NEXT: tail call void @test12_helper(i8* nonnull [[A]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
tail call void @test12_helper(i8* %a)
ret void
}
; TEST 13
; Simple Argument Tests
declare i8* @unknown()
define void @test13_helper() {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@test13_helper()
[Attributor] fix AANoUndef initialization Currently, `AANoUndefImpl::initialize` mistakenly always indicates optimistic fixpoint for function returned position. This is because an associated value is `Function` in the case, and `isGuaranteedNotToBeUndefOrPoison` returns true for Function. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86361
2020-08-22 04:06:14 +08:00
; CHECK-NEXT: [[NONNULLPTR:%.*]] = tail call nonnull i8* @ret_nonnull()
; CHECK-NEXT: [[MAYBENULLPTR:%.*]] = tail call i8* @unknown()
; CHECK-NEXT: tail call void @test13(i8* noalias nocapture nofree nonnull readnone [[NONNULLPTR]], i8* noalias nocapture nofree nonnull readnone [[NONNULLPTR]], i8* noalias nocapture nofree readnone [[MAYBENULLPTR]])
; CHECK-NEXT: tail call void @test13(i8* noalias nocapture nofree nonnull readnone [[NONNULLPTR]], i8* noalias nocapture nofree readnone [[MAYBENULLPTR]], i8* noalias nocapture nofree nonnull readnone [[NONNULLPTR]])
[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 08:04:57 +08:00
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%nonnullptr = tail call i8* @ret_nonnull()
%maybenullptr = tail call i8* @unknown()
tail call void @test13(i8* %nonnullptr, i8* %nonnullptr, i8* %maybenullptr)
tail call void @test13(i8* %nonnullptr, i8* %maybenullptr, i8* %nonnullptr)
ret void
}
define internal void @test13(i8* %a, i8* %b, i8* %c) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nounwind
[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 08:04:57 +08:00
; IS__TUNIT____-LABEL: define {{[^@]+}}@test13
[Attributor] fix AANoUndef initialization Currently, `AANoUndefImpl::initialize` mistakenly always indicates optimistic fixpoint for function returned position. This is because an associated value is `Function` in the case, and `isGuaranteedNotToBeUndefOrPoison` returns true for Function. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86361
2020-08-22 04:06:14 +08:00
; IS__TUNIT____-SAME: (i8* noalias nocapture nofree nonnull readnone [[A:%.*]], i8* noalias nocapture nofree readnone [[B:%.*]], i8* noalias nocapture nofree readnone [[C:%.*]])
; IS__TUNIT____-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree nonnull readnone [[A]])
; IS__TUNIT____-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree readnone [[B]])
; IS__TUNIT____-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree readnone [[C]])
[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 08:04:57 +08:00
; IS__TUNIT____-NEXT: ret void
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__CGSCC____: Function Attrs: nounwind
[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 08:04:57 +08:00
; IS__CGSCC____-LABEL: define {{[^@]+}}@test13
; IS__CGSCC____-SAME: (i8* nocapture nofree readnone [[A:%.*]], i8* nocapture nofree readnone [[B:%.*]], i8* nocapture nofree readnone [[C:%.*]])
; IS__CGSCC____-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree readnone [[A]])
; IS__CGSCC____-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree readnone [[B]])
; IS__CGSCC____-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree readnone [[C]])
; IS__CGSCC____-NEXT: ret void
;
[Attributor] Function signature rewrite infrastructure As part of the Attributor manifest we want to change the signature of functions. This patch introduces a fairly generic interface to do so. As a first, very simple, use case, we remove unused arguments. A second use case, pointer privatization, will be committed with this patch as well. A lot of the code and ideas are taken from argument promotion and we run all argument promotion tests through this framework as well. Reviewed By: uenoku Differential Revision: https://reviews.llvm.org/D68765
2019-10-10 14:39:16 +08:00
call void @use_i8_ptr(i8* %a)
call void @use_i8_ptr(i8* %b)
call void @use_i8_ptr(i8* %c)
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
ret void
}
declare nonnull i8* @nonnull()
; TEST 14
; Complex propagation
; Argument of f1, f2, f3 can be marked with nonnull.
; * Argument
; 1. In f1:bb6, %arg can be marked with nonnull because of the comparison in bb1
; 2. Because f2 is internal function, f2(i32* %arg) -> @f2(i32* nonnull %arg)
; 3. In f1:bb4 %tmp5 is nonnull and f3 is internal function.
; Then, f3(i32* %arg) -> @f3(i32* nonnull %arg)
; 4. We get nonnull in whole f1 call sites so f1(i32* %arg) -> @f1(i32* nonnull %arg)
define internal i32* @f1(i32* %arg) {
; FIXME: missing nonnull It should be nonnull @f1(i32* nonnull readonly %arg)
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: argmemonly nofree nosync nounwind readonly
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@f1
; CHECK-SAME: (i32* nofree readonly [[ARG:%.*]])
; CHECK-NEXT: bb:
; CHECK-NEXT: [[TMP:%.*]] = icmp eq i32* [[ARG]], null
; CHECK-NEXT: br i1 [[TMP]], label [[BB9:%.*]], label [[BB1:%.*]]
; CHECK: bb1:
; CHECK-NEXT: [[TMP2:%.*]] = load i32, i32* [[ARG]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP2]], 0
; CHECK-NEXT: br i1 [[TMP3]], label [[BB6:%.*]], label [[BB4:%.*]]
; CHECK: bb4:
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, i32* [[ARG]], i64 1
[Attributor][FIX] Do not use cyclic arguments for `nonnull` `AADereferenceable::getAssumedDereferenceableBytes()` is actually deducing `dereferenceable_or_null`. We should not use that information to deduce `nonnull`, since it doesn't imply `nonnull`.
2020-08-18 23:01:37 +08:00
; CHECK-NEXT: [[TMP5B:%.*]] = tail call i32* @f3(i32* nofree nonnull readonly [[TMP5]])
[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 08:04:57 +08:00
; CHECK-NEXT: [[TMP5C:%.*]] = getelementptr inbounds i32, i32* [[TMP5B]], i64 -1
; CHECK-NEXT: br label [[BB9]]
; CHECK: bb6:
; CHECK-NEXT: [[TMP7:%.*]] = tail call nonnull i32* @f2(i32* nofree nonnull readonly align 4 dereferenceable(4) [[ARG]])
; CHECK-NEXT: ret i32* [[TMP7]]
; CHECK: bb9:
; CHECK-NEXT: [[TMP10:%.*]] = phi i32* [ [[TMP5C]], [[BB4]] ], [ inttoptr (i64 4 to i32*), [[BB:%.*]] ]
; CHECK-NEXT: ret i32* [[TMP10]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
bb:
%tmp = icmp eq i32* %arg, null
br i1 %tmp, label %bb9, label %bb1
bb1: ; preds = %bb
%tmp2 = load i32, i32* %arg, align 4
%tmp3 = icmp eq i32 %tmp2, 0
br i1 %tmp3, label %bb6, label %bb4
bb4: ; preds = %bb1
%tmp5 = getelementptr inbounds i32, i32* %arg, i64 1
%tmp5b = tail call i32* @f3(i32* %tmp5)
%tmp5c = getelementptr inbounds i32, i32* %tmp5b, i64 -1
br label %bb9
bb6: ; preds = %bb1
%tmp7 = tail call i32* @f2(i32* %arg)
ret i32* %tmp7
bb9: ; preds = %bb4, %bb
%tmp10 = phi i32* [ %tmp5c, %bb4 ], [ inttoptr (i64 4 to i32*), %bb ]
ret i32* %tmp10
}
define internal i32* @f2(i32* %arg) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: argmemonly nofree nosync nounwind readonly
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@f2
; CHECK-SAME: (i32* nofree nonnull readonly align 4 dereferenceable(4) [[ARG:%.*]])
; CHECK-NEXT: bb:
; CHECK-NEXT: [[TMP:%.*]] = tail call nonnull i32* @f1(i32* nofree nonnull readonly align 4 dereferenceable(4) [[ARG]])
; CHECK-NEXT: ret i32* [[TMP]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
bb:
%tmp = tail call i32* @f1(i32* %arg)
ret i32* %tmp
}
define dso_local noalias i32* @f3(i32* %arg) {
; FIXME: missing nonnull. It should be nonnull @f3(i32* nonnull readonly %arg)
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: argmemonly nofree nosync nounwind readonly
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@f3
; CHECK-SAME: (i32* nofree readonly [[ARG:%.*]])
; CHECK-NEXT: bb:
; CHECK-NEXT: [[TMP:%.*]] = call nonnull i32* @f1(i32* nofree readonly [[ARG]])
; CHECK-NEXT: ret i32* [[TMP]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
bb:
; FIXME: missing nonnull. It should be @f1(i32* nonnull readonly %arg)
%tmp = call i32* @f1(i32* %arg)
ret i32* %tmp
}
; TEST 15
define void @f15(i8* %arg) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@f15
; CHECK-SAME: (i8* nonnull dereferenceable(4) [[ARG:%.*]])
; CHECK-NEXT: tail call void @use1(i8* nonnull dereferenceable(4) [[ARG]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
tail call void @use1(i8* dereferenceable(4) %arg)
ret void
}
declare void @fun0() #1
declare void @fun1(i8*) #1
declare void @fun2(i8*, i8*) #1
declare void @fun3(i8*, i8*, i8*) #1
; TEST 16 simple path test
; if(..)
; fun2(nonnull %a, nonnull %b)
; else
; fun2(nonnull %a, %b)
; We can say that %a is nonnull but %b is not.
define void @f16(i8* %a, i8 * %b, i8 %c) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: nounwind willreturn
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@f16
; CHECK-SAME: (i8* nonnull [[A:%.*]], i8* [[B:%.*]], i8 [[C:%.*]])
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i8 [[C]], 0
; CHECK-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK: if.then:
; CHECK-NEXT: tail call void @fun2(i8* nonnull [[A]], i8* nonnull [[B]])
; CHECK-NEXT: ret void
; CHECK: if.else:
; CHECK-NEXT: tail call void @fun2(i8* nonnull [[A]], i8* [[B]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%cmp = icmp eq i8 %c, 0
br i1 %cmp, label %if.then, label %if.else
if.then:
tail call void @fun2(i8* nonnull %a, i8* nonnull %b)
ret void
if.else:
tail call void @fun2(i8* nonnull %a, i8* %b)
ret void
}
; TEST 17 explore child BB test
; if(..)
; ... (willreturn & nounwind)
; else
; ... (willreturn & nounwind)
; fun1(nonnull %a)
; We can say that %a is nonnull
define void @f17(i8* %a, i8 %c) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: nounwind willreturn
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@f17
; CHECK-SAME: (i8* nonnull [[A:%.*]], i8 [[C:%.*]])
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i8 [[C]], 0
; CHECK-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK: if.then:
; CHECK-NEXT: tail call void @fun0()
; CHECK-NEXT: br label [[CONT:%.*]]
; CHECK: if.else:
; CHECK-NEXT: tail call void @fun0()
; CHECK-NEXT: br label [[CONT]]
; CHECK: cont:
; CHECK-NEXT: tail call void @fun1(i8* nonnull [[A]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%cmp = icmp eq i8 %c, 0
br i1 %cmp, label %if.then, label %if.else
if.then:
tail call void @fun0()
br label %cont
if.else:
tail call void @fun0()
br label %cont
cont:
tail call void @fun1(i8* nonnull %a)
ret void
}
; TEST 18 More complex test
; if(..)
; ... (willreturn & nounwind)
; else
; ... (willreturn & nounwind)
; if(..)
; ... (willreturn & nounwind)
; else
; ... (willreturn & nounwind)
; fun1(nonnull %a)
define void @f18(i8* %a, i8* %b, i8 %c) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: nounwind willreturn
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@f18
; CHECK-SAME: (i8* nonnull [[A:%.*]], i8* [[B:%.*]], i8 [[C:%.*]])
; CHECK-NEXT: [[CMP1:%.*]] = icmp eq i8 [[C]], 0
; CHECK-NEXT: br i1 [[CMP1]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK: if.then:
; CHECK-NEXT: tail call void @fun0()
; CHECK-NEXT: br label [[CONT:%.*]]
; CHECK: if.else:
; CHECK-NEXT: tail call void @fun0()
; CHECK-NEXT: br label [[CONT]]
; CHECK: cont:
; CHECK-NEXT: [[CMP2:%.*]] = icmp eq i8 [[C]], 1
; CHECK-NEXT: br i1 [[CMP2]], label [[CONT_THEN:%.*]], label [[CONT_ELSE:%.*]]
; CHECK: cont.then:
; CHECK-NEXT: tail call void @fun1(i8* nonnull [[B]])
; CHECK-NEXT: br label [[CONT2:%.*]]
; CHECK: cont.else:
; CHECK-NEXT: tail call void @fun0()
; CHECK-NEXT: br label [[CONT2]]
; CHECK: cont2:
; CHECK-NEXT: tail call void @fun1(i8* nonnull [[A]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%cmp1 = icmp eq i8 %c, 0
br i1 %cmp1, label %if.then, label %if.else
if.then:
tail call void @fun0()
br label %cont
if.else:
tail call void @fun0()
br label %cont
cont:
%cmp2 = icmp eq i8 %c, 1
br i1 %cmp2, label %cont.then, label %cont.else
cont.then:
tail call void @fun1(i8* nonnull %b)
br label %cont2
cont.else:
tail call void @fun0()
br label %cont2
cont2:
tail call void @fun1(i8* nonnull %a)
ret void
}
; TEST 19: Loop
define void @f19(i8* %a, i8* %b, i8 %c) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: nounwind
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@f19
; CHECK-SAME: (i8* [[A:%.*]], i8* nonnull [[B:%.*]], i8 [[C:%.*]])
; CHECK-NEXT: br label [[LOOP_HEADER:%.*]]
; CHECK: loop.header:
; CHECK-NEXT: [[CMP2:%.*]] = icmp eq i8 [[C]], 0
; CHECK-NEXT: br i1 [[CMP2]], label [[LOOP_BODY:%.*]], label [[LOOP_EXIT:%.*]]
; CHECK: loop.body:
; CHECK-NEXT: tail call void @fun1(i8* nonnull [[B]])
; CHECK-NEXT: tail call void @fun1(i8* nonnull [[A]])
; CHECK-NEXT: br label [[LOOP_HEADER]]
; CHECK: loop.exit:
; CHECK-NEXT: tail call void @fun1(i8* nonnull [[B]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
br label %loop.header
loop.header:
%cmp2 = icmp eq i8 %c, 0
br i1 %cmp2, label %loop.body, label %loop.exit
loop.body:
tail call void @fun1(i8* nonnull %b)
tail call void @fun1(i8* nonnull %a)
br label %loop.header
loop.exit:
tail call void @fun1(i8* nonnull %b)
ret void
}
; Test propagation of nonnull callsite args back to caller.
declare void @use1(i8* %x)
declare void @use2(i8* %x, i8* %y);
declare void @use3(i8* %x, i8* %y, i8* %z);
declare void @use1nonnull(i8* nonnull %x);
declare void @use2nonnull(i8* nonnull %x, i8* nonnull %y);
declare void @use3nonnull(i8* nonnull %x, i8* nonnull %y, i8* nonnull %z);
declare i8 @use1safecall(i8* %x) readonly nounwind ; readonly+nounwind guarantees that execution continues to successor
; Can't extend non-null to parent for any argument because the 2nd call is not guaranteed to execute.
define void @parent1(i8* %a, i8* %b, i8* %c) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@parent1
; CHECK-SAME: (i8* [[A:%.*]], i8* [[B:%.*]], i8* [[C:%.*]])
; CHECK-NEXT: call void @use3(i8* [[C]], i8* [[A]], i8* [[B]])
; CHECK-NEXT: call void @use3nonnull(i8* nonnull [[B]], i8* nonnull [[C]], i8* nonnull [[A]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @use3(i8* %c, i8* %a, i8* %b)
call void @use3nonnull(i8* %b, i8* %c, i8* %a)
ret void
}
; Extend non-null to parent for all arguments.
define void @parent2(i8* %a, i8* %b, i8* %c) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@parent2
; CHECK-SAME: (i8* nonnull [[A:%.*]], i8* nonnull [[B:%.*]], i8* nonnull [[C:%.*]])
; CHECK-NEXT: call void @use3nonnull(i8* nonnull [[B]], i8* nonnull [[C]], i8* nonnull [[A]])
; CHECK-NEXT: call void @use3(i8* nonnull [[C]], i8* nonnull [[A]], i8* nonnull [[B]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @use3nonnull(i8* %b, i8* %c, i8* %a)
call void @use3(i8* %c, i8* %a, i8* %b)
ret void
}
; Extend non-null to parent for 1st argument.
define void @parent3(i8* %a, i8* %b, i8* %c) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@parent3
; CHECK-SAME: (i8* nonnull [[A:%.*]], i8* [[B:%.*]], i8* [[C:%.*]])
; CHECK-NEXT: call void @use1nonnull(i8* nonnull [[A]])
; CHECK-NEXT: call void @use3(i8* [[C]], i8* [[B]], i8* nonnull [[A]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @use1nonnull(i8* %a)
call void @use3(i8* %c, i8* %b, i8* %a)
ret void
}
; Extend non-null to parent for last 2 arguments.
define void @parent4(i8* %a, i8* %b, i8* %c) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@parent4
; CHECK-SAME: (i8* [[A:%.*]], i8* nonnull [[B:%.*]], i8* nonnull [[C:%.*]])
; CHECK-NEXT: call void @use2nonnull(i8* nonnull [[C]], i8* nonnull [[B]])
; CHECK-NEXT: call void @use2(i8* [[A]], i8* nonnull [[C]])
; CHECK-NEXT: call void @use1(i8* nonnull [[B]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @use2nonnull(i8* %c, i8* %b)
call void @use2(i8* %a, i8* %c)
call void @use1(i8* %b)
ret void
}
; The callsite must execute in order for the attribute to transfer to the parent.
; It appears benign to extend non-null to the parent in this case, but we can't do that
; because it would incorrectly propagate the wrong information to its callers.
define void @parent5(i8* %a, i1 %a_is_notnull) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@parent5
; CHECK-SAME: (i8* [[A:%.*]], i1 [[A_IS_NOTNULL:%.*]])
; CHECK-NEXT: br i1 [[A_IS_NOTNULL]], label [[T:%.*]], label [[F:%.*]]
; CHECK: t:
; CHECK-NEXT: call void @use1nonnull(i8* nonnull [[A]])
; CHECK-NEXT: ret void
; CHECK: f:
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
br i1 %a_is_notnull, label %t, label %f
t:
call void @use1nonnull(i8* %a)
ret void
f:
ret void
}
; The callsite must execute in order for the attribute to transfer to the parent.
; The volatile load can't trap, so we can guarantee that we'll get to the call.
define i8 @parent6(i8* %a, i8* %b) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@parent6
; CHECK-SAME: (i8* nonnull [[A:%.*]], i8* [[B:%.*]])
[Attributor] Use existing helpers to determine IR facts We now use getPointerDereferenceableBytes to determine `nonnull` and `dereferenceable` facts from the IR. We also use getPointerAlignment in AAAlign for the same reason. The latter can interfere with callbacks so we do restrict it to non-function-pointers for now.
2020-05-11 02:16:49 +08:00
; CHECK-NEXT: [[C:%.*]] = load volatile i8, i8* [[B]], align 1
[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 08:04:57 +08:00
; CHECK-NEXT: call void @use1nonnull(i8* nonnull [[A]])
; CHECK-NEXT: ret i8 [[C]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%c = load volatile i8, i8* %b
call void @use1nonnull(i8* %a)
ret i8 %c
}
; The nonnull callsite is guaranteed to execute, so the argument must be nonnull throughout the parent.
define i8 @parent7(i8* %a) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@parent7
; CHECK-SAME: (i8* nonnull [[A:%.*]])
; CHECK-NEXT: [[RET:%.*]] = call i8 @use1safecall(i8* nonnull readonly [[A]])
; CHECK-NEXT: call void @use1nonnull(i8* nonnull [[A]])
; CHECK-NEXT: ret i8 [[RET]]
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%ret = call i8 @use1safecall(i8* %a)
call void @use1nonnull(i8* %a)
ret i8 %ret
}
; Make sure that an invoke works similarly to a call.
declare i32 @esfp(...)
define i1 @parent8(i8* %a, i8* %bogus1, i8* %b) personality i8* bitcast (i32 (...)* @esfp to i8*){
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; CHECK: Function Attrs: nounwind
; CHECK-LABEL: define {{[^@]+}}@parent8
; CHECK-SAME: (i8* nonnull [[A:%.*]], i8* nocapture nofree readnone [[BOGUS1:%.*]], i8* nonnull [[B:%.*]]) [[ATTR4:#.*]] personality i8* bitcast (i32 (...)* @esfp to i8*)
; CHECK-NEXT: entry:
; CHECK-NEXT: invoke void @use2nonnull(i8* nonnull [[A]], i8* nonnull [[B]])
; CHECK-NEXT: to label [[CONT:%.*]] unwind label [[EXC:%.*]]
; CHECK: cont:
[Attributor] Simplify comparison against constant null pointer Comparison against null is a common pattern that usually is followed by error handling code and the likes. We now use AANonNull to simplify these comparisons optimistically in order to make more code dead early on. Reviewed By: uenoku Differential Revision: https://reviews.llvm.org/D86145
2020-08-18 23:03:25 +08:00
; CHECK-NEXT: ret i1 false
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; CHECK: exc:
; CHECK-NEXT: [[LP:%.*]] = landingpad { i8*, i32 }
; CHECK-NEXT: filter [0 x i8*] zeroinitializer
; CHECK-NEXT: unreachable
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
entry:
invoke void @use2nonnull(i8* %a, i8* %b)
to label %cont unwind label %exc
cont:
%null_check = icmp eq i8* %b, null
ret i1 %null_check
exc:
%lp = landingpad { i8*, i32 }
filter [0 x i8*] zeroinitializer
unreachable
}
define i32* @gep1(i32* %p) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@gep1
; IS__TUNIT____-SAME: (i32* nofree readnone "no-capture-maybe-returned" [[P:%.*]])
; IS__TUNIT____-NEXT: [[Q:%.*]] = getelementptr inbounds i32, i32* [[P]], i32 1
; IS__TUNIT____-NEXT: ret i32* [[Q]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@gep1
; IS__CGSCC____-SAME: (i32* nofree readnone "no-capture-maybe-returned" [[P:%.*]])
; IS__CGSCC____-NEXT: [[Q:%.*]] = getelementptr inbounds i32, i32* [[P]], i32 1
; IS__CGSCC____-NEXT: ret i32* [[Q]]
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%q = getelementptr inbounds i32, i32* %p, i32 1
ret i32* %q
}
define i32* @gep1_no_null_opt(i32* %p) #0 {
; Should't be able to derive nonnull based on gep.
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind null_pointer_is_valid readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@gep1_no_null_opt
; IS__TUNIT____-SAME: (i32* nofree readnone "no-capture-maybe-returned" [[P:%.*]])
; IS__TUNIT____-NEXT: [[Q:%.*]] = getelementptr inbounds i32, i32* [[P]], i32 1
; IS__TUNIT____-NEXT: ret i32* [[Q]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind null_pointer_is_valid readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@gep1_no_null_opt
; IS__CGSCC____-SAME: (i32* nofree readnone "no-capture-maybe-returned" [[P:%.*]])
; IS__CGSCC____-NEXT: [[Q:%.*]] = getelementptr inbounds i32, i32* [[P]], i32 1
; IS__CGSCC____-NEXT: ret i32* [[Q]]
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%q = getelementptr inbounds i32, i32* %p, i32 1
ret i32* %q
}
define i32 addrspace(3)* @gep2(i32 addrspace(3)* %p) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@gep2
; IS__TUNIT____-SAME: (i32 addrspace(3)* nofree readnone "no-capture-maybe-returned" [[P:%.*]])
; IS__TUNIT____-NEXT: [[Q:%.*]] = getelementptr inbounds i32, i32 addrspace(3)* [[P]], i32 1
; IS__TUNIT____-NEXT: ret i32 addrspace(3)* [[Q]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@gep2
; IS__CGSCC____-SAME: (i32 addrspace(3)* nofree readnone "no-capture-maybe-returned" [[P:%.*]])
; IS__CGSCC____-NEXT: [[Q:%.*]] = getelementptr inbounds i32, i32 addrspace(3)* [[P]], i32 1
; IS__CGSCC____-NEXT: ret i32 addrspace(3)* [[Q]]
[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
%q = getelementptr inbounds i32, i32 addrspace(3)* %p, i32 1
ret i32 addrspace(3)* %q
}
; FIXME: We should propagate dereferenceable here but *not* nonnull
define i32 addrspace(3)* @as(i32 addrspace(3)* dereferenceable(4) %p) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@as
; IS__TUNIT____-SAME: (i32 addrspace(3)* nofree nonnull readnone returned dereferenceable(4) "no-capture-maybe-returned" [[P:%.*]])
; IS__TUNIT____-NEXT: ret i32 addrspace(3)* [[P]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@as
; IS__CGSCC____-SAME: (i32 addrspace(3)* nofree nonnull readnone returned dereferenceable(4) "no-capture-maybe-returned" [[P:%.*]])
; IS__CGSCC____-NEXT: ret i32 addrspace(3)* [[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 08:04:57 +08:00
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
ret i32 addrspace(3)* %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 08:04:57 +08:00
; CHECK-NOT: @g2()
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
define internal i32* @g2() {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
[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 08:04:57 +08:00
; IS__CGSCC____-LABEL: define {{[^@]+}}@g2()
; IS__CGSCC____-NEXT: ret i32* inttoptr (i64 4 to i32*)
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
ret i32* inttoptr (i64 4 to i32*)
}
[Attributor][FIX] Remove duplicate, half-broken functionality The changeXXXAfterManifest functions are better suited to deal with changes so we should prefer them. These functions also recursively delete dead instructions which is why we see test changes.
2020-01-12 14:17:08 +08:00
define i32* @g1() {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@g1()
; IS__TUNIT____-NEXT: ret i32* inttoptr (i64 4 to i32*)
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@g1()
; IS__CGSCC____-NEXT: ret i32* inttoptr (i64 4 to 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 08:04:57 +08:00
;
%c = call i32* @g2()
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
ret i32* %c
}
[Attributor] Use assumed information to determine side-effects We relied on wouldInstructionBeTriviallyDead before but that functions does not take assumed information, especially for calls, into account. The replacement, AAIsDead::isAssumeSideEffectFree, does. This change makes AAIsDeadCallSiteReturn more complex as we can have a dead call or only dead users. The test have been modified to include a side effect where there was none in order to keep the coverage. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D73311
2020-01-12 14:34:38 +08:00
declare void @use_i32_ptr(i32* readnone nocapture) nounwind
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
define internal void @called_by_weak(i32* %a) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nounwind
[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 08:04:57 +08:00
; IS__TUNIT____-LABEL: define {{[^@]+}}@called_by_weak
; IS__TUNIT____-SAME: (i32* noalias nocapture nonnull readnone [[A:%.*]])
; IS__TUNIT____-NEXT: call void @use_i32_ptr(i32* noalias nocapture nonnull readnone [[A]])
; IS__TUNIT____-NEXT: ret void
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__CGSCC____: Function Attrs: nounwind
[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 08:04:57 +08:00
; IS__CGSCC____-LABEL: define {{[^@]+}}@called_by_weak
; IS__CGSCC____-SAME: (i32* nocapture nonnull readnone [[A:%.*]])
; IS__CGSCC____-NEXT: call void @use_i32_ptr(i32* noalias nocapture nonnull readnone [[A]])
; IS__CGSCC____-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @use_i32_ptr(i32* %a)
ret void
}
; Check we do not annotate the function interface of this weak function.
define weak_odr void @weak_caller(i32* nonnull %a) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@weak_caller
; CHECK-SAME: (i32* nonnull [[A:%.*]])
[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 08:08:30 +08:00
; CHECK-NEXT: call void @called_by_weak(i32* noalias nocapture nonnull readnone [[A]])
[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 08:04:57 +08:00
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @called_by_weak(i32* %a)
ret void
}
; Expect nonnull
define internal void @control(i32* dereferenceable(4) %a) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__TUNIT____: Function Attrs: nounwind
[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 08:04:57 +08:00
; IS__TUNIT____-LABEL: define {{[^@]+}}@control
; IS__TUNIT____-SAME: (i32* noalias nocapture nonnull readnone align 16 dereferenceable(8) [[A:%.*]])
; IS__TUNIT____-NEXT: call void @use_i32_ptr(i32* noalias nocapture nonnull readnone align 16 dereferenceable(8) [[A]])
; IS__TUNIT____-NEXT: ret void
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS__CGSCC____: Function Attrs: nounwind
[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 08:04:57 +08:00
; IS__CGSCC____-LABEL: define {{[^@]+}}@control
; IS__CGSCC____-SAME: (i32* nocapture nonnull readnone align 16 dereferenceable(8) [[A:%.*]])
; IS__CGSCC____-NEXT: call void @use_i32_ptr(i32* noalias nocapture nonnull readnone align 16 dereferenceable(8) [[A]])
; IS__CGSCC____-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @use_i32_ptr(i32* %a)
ret void
}
; Avoid nonnull as we do not touch naked functions
define internal void @naked(i32* dereferenceable(4) %a) naked {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: naked
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@naked
; CHECK-SAME: (i32* dereferenceable(4) [[A:%.*]])
; CHECK-NEXT: call void @use_i32_ptr(i32* [[A]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @use_i32_ptr(i32* %a)
ret void
}
; Avoid nonnull as we do not touch optnone
define internal void @optnone(i32* dereferenceable(4) %a) optnone noinline {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: noinline optnone
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@optnone
; CHECK-SAME: (i32* dereferenceable(4) [[A:%.*]])
; CHECK-NEXT: call void @use_i32_ptr(i32* [[A]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @use_i32_ptr(i32* %a)
ret void
}
define void @make_live(i32* nonnull dereferenceable(8) %a) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@make_live
; CHECK-SAME: (i32* nonnull align 16 dereferenceable(8) [[A:%.*]])
; CHECK-NEXT: call void @naked(i32* nonnull align 16 dereferenceable(8) [[A]])
[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 08:08:30 +08:00
; CHECK-NEXT: call void @control(i32* noalias nocapture nonnull readnone align 16 dereferenceable(8) [[A]])
[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 08:04:57 +08:00
; CHECK-NEXT: call void @optnone(i32* nonnull align 16 dereferenceable(8) [[A]])
; CHECK-NEXT: ret void
;
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
call void @naked(i32* nonnull dereferenceable(8) align 16 %a)
call void @control(i32* nonnull dereferenceable(8) align 16 %a)
call void @optnone(i32* nonnull dereferenceable(8) align 16 %a)
ret void
}
[Attributor][NFC] Prepare some tests to be used with update test script
2020-02-20 16:06:14 +08:00
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
;int f(int *u, int n){
; for(int i = 0;i<n;i++){
; h(u);
; }
; return g(nonnull u);
;}
declare void @h(i32*) willreturn nounwind
declare i32 @g(i32*) willreturn nounwind
define i32 @nonnull_exec_ctx_1(i32* %a, i32 %b) {
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS________OPM: Function Attrs: nounwind
; IS________OPM-LABEL: define {{[^@]+}}@nonnull_exec_ctx_1
; IS________OPM-SAME: (i32* [[A:%.*]], i32 [[B:%.*]])
; IS________OPM-NEXT: en:
; IS________OPM-NEXT: [[TMP3:%.*]] = icmp eq i32 [[B]], 0
; IS________OPM-NEXT: br i1 [[TMP3]], label [[EX:%.*]], label [[HD:%.*]]
; IS________OPM: ex:
; IS________OPM-NEXT: [[TMP5:%.*]] = tail call i32 @g(i32* nonnull [[A]])
; IS________OPM-NEXT: ret i32 [[TMP5]]
; IS________OPM: hd:
; IS________OPM-NEXT: [[TMP7:%.*]] = phi i32 [ [[TMP8:%.*]], [[HD]] ], [ 0, [[EN:%.*]] ]
; IS________OPM-NEXT: tail call void @h(i32* [[A]])
; IS________OPM-NEXT: [[TMP8]] = add nuw i32 [[TMP7]], 1
; IS________OPM-NEXT: [[TMP9:%.*]] = icmp eq i32 [[TMP8]], [[B]]
; IS________OPM-NEXT: br i1 [[TMP9]], label [[EX]], label [[HD]]
;
; IS________NPM: Function Attrs: nounwind willreturn
; IS________NPM-LABEL: define {{[^@]+}}@nonnull_exec_ctx_1
; IS________NPM-SAME: (i32* [[A:%.*]], i32 [[B:%.*]])
; IS________NPM-NEXT: en:
; IS________NPM-NEXT: [[TMP3:%.*]] = icmp eq i32 [[B]], 0
; IS________NPM-NEXT: br i1 [[TMP3]], label [[EX:%.*]], label [[HD:%.*]]
; IS________NPM: ex:
; IS________NPM-NEXT: [[TMP5:%.*]] = tail call i32 @g(i32* nonnull [[A]])
; IS________NPM-NEXT: ret i32 [[TMP5]]
; IS________NPM: hd:
; IS________NPM-NEXT: [[TMP7:%.*]] = phi i32 [ [[TMP8:%.*]], [[HD]] ], [ 0, [[EN:%.*]] ]
; IS________NPM-NEXT: tail call void @h(i32* [[A]])
; IS________NPM-NEXT: [[TMP8]] = add nuw i32 [[TMP7]], 1
; IS________NPM-NEXT: [[TMP9:%.*]] = icmp eq i32 [[TMP8]], [[B]]
; IS________NPM-NEXT: br i1 [[TMP9]], label [[EX]], label [[HD]]
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
;
en:
%tmp3 = icmp eq i32 %b, 0
br i1 %tmp3, label %ex, label %hd
ex:
%tmp5 = tail call i32 @g(i32* nonnull %a)
ret i32 %tmp5
hd:
%tmp7 = phi i32 [ %tmp8, %hd ], [ 0, %en ]
tail call void @h(i32* %a)
%tmp8 = add nuw i32 %tmp7, 1
%tmp9 = icmp eq i32 %tmp8, %b
br i1 %tmp9, label %ex, label %hd
}
define i32 @nonnull_exec_ctx_1b(i32* %a, i32 %b) {
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; IS________OPM: Function Attrs: nounwind
; IS________OPM-LABEL: define {{[^@]+}}@nonnull_exec_ctx_1b
; IS________OPM-SAME: (i32* [[A:%.*]], i32 [[B:%.*]])
; IS________OPM-NEXT: en:
; IS________OPM-NEXT: [[TMP3:%.*]] = icmp eq i32 [[B]], 0
; IS________OPM-NEXT: br i1 [[TMP3]], label [[EX:%.*]], label [[HD:%.*]]
; IS________OPM: ex:
; IS________OPM-NEXT: [[TMP5:%.*]] = tail call i32 @g(i32* nonnull [[A]])
; IS________OPM-NEXT: ret i32 [[TMP5]]
; IS________OPM: hd:
; IS________OPM-NEXT: [[TMP7:%.*]] = phi i32 [ [[TMP8:%.*]], [[HD2:%.*]] ], [ 0, [[EN:%.*]] ]
; IS________OPM-NEXT: tail call void @h(i32* [[A]])
; IS________OPM-NEXT: br label [[HD2]]
; IS________OPM: hd2:
; IS________OPM-NEXT: [[TMP8]] = add nuw i32 [[TMP7]], 1
; IS________OPM-NEXT: [[TMP9:%.*]] = icmp eq i32 [[TMP8]], [[B]]
; IS________OPM-NEXT: br i1 [[TMP9]], label [[EX]], label [[HD]]
;
; IS________NPM: Function Attrs: nounwind willreturn
; IS________NPM-LABEL: define {{[^@]+}}@nonnull_exec_ctx_1b
; IS________NPM-SAME: (i32* [[A:%.*]], i32 [[B:%.*]])
; IS________NPM-NEXT: en:
; IS________NPM-NEXT: [[TMP3:%.*]] = icmp eq i32 [[B]], 0
; IS________NPM-NEXT: br i1 [[TMP3]], label [[EX:%.*]], label [[HD:%.*]]
; IS________NPM: ex:
; IS________NPM-NEXT: [[TMP5:%.*]] = tail call i32 @g(i32* nonnull [[A]])
; IS________NPM-NEXT: ret i32 [[TMP5]]
; IS________NPM: hd:
; IS________NPM-NEXT: [[TMP7:%.*]] = phi i32 [ [[TMP8:%.*]], [[HD2:%.*]] ], [ 0, [[EN:%.*]] ]
; IS________NPM-NEXT: tail call void @h(i32* [[A]])
; IS________NPM-NEXT: br label [[HD2]]
; IS________NPM: hd2:
; IS________NPM-NEXT: [[TMP8]] = add nuw i32 [[TMP7]], 1
; IS________NPM-NEXT: [[TMP9:%.*]] = icmp eq i32 [[TMP8]], [[B]]
; IS________NPM-NEXT: br i1 [[TMP9]], label [[EX]], label [[HD]]
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
;
en:
%tmp3 = icmp eq i32 %b, 0
br i1 %tmp3, label %ex, label %hd
ex:
%tmp5 = tail call i32 @g(i32* nonnull %a)
ret i32 %tmp5
hd:
%tmp7 = phi i32 [ %tmp8, %hd2 ], [ 0, %en ]
tail call void @h(i32* %a)
br label %hd2
hd2:
%tmp8 = add nuw i32 %tmp7, 1
%tmp9 = icmp eq i32 %tmp8, %b
br i1 %tmp9, label %ex, label %hd
}
define i32 @nonnull_exec_ctx_2(i32* %a, i32 %b) willreturn nounwind {
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: nounwind willreturn
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@nonnull_exec_ctx_2
; CHECK-SAME: (i32* nonnull [[A:%.*]], i32 [[B:%.*]])
; CHECK-NEXT: en:
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[B]], 0
; CHECK-NEXT: br i1 [[TMP3]], label [[EX:%.*]], label [[HD:%.*]]
; CHECK: ex:
; CHECK-NEXT: [[TMP5:%.*]] = tail call i32 @g(i32* nonnull [[A]])
; CHECK-NEXT: ret i32 [[TMP5]]
; CHECK: hd:
; CHECK-NEXT: [[TMP7:%.*]] = phi i32 [ [[TMP8:%.*]], [[HD]] ], [ 0, [[EN:%.*]] ]
; CHECK-NEXT: tail call void @h(i32* nonnull [[A]])
; CHECK-NEXT: [[TMP8]] = add nuw i32 [[TMP7]], 1
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i32 [[TMP8]], [[B]]
; CHECK-NEXT: br i1 [[TMP9]], label [[EX]], label [[HD]]
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
;
en:
%tmp3 = icmp eq i32 %b, 0
br i1 %tmp3, label %ex, label %hd
ex:
%tmp5 = tail call i32 @g(i32* nonnull %a)
ret i32 %tmp5
hd:
%tmp7 = phi i32 [ %tmp8, %hd ], [ 0, %en ]
tail call void @h(i32* %a)
%tmp8 = add nuw i32 %tmp7, 1
%tmp9 = icmp eq i32 %tmp8, %b
br i1 %tmp9, label %ex, label %hd
}
define i32 @nonnull_exec_ctx_2b(i32* %a, i32 %b) willreturn nounwind {
;
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: nounwind willreturn
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@nonnull_exec_ctx_2b
; CHECK-SAME: (i32* nonnull [[A:%.*]], i32 [[B:%.*]])
; CHECK-NEXT: en:
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[B]], 0
; CHECK-NEXT: br i1 [[TMP3]], label [[EX:%.*]], label [[HD:%.*]]
; CHECK: ex:
; CHECK-NEXT: [[TMP5:%.*]] = tail call i32 @g(i32* nonnull [[A]])
; CHECK-NEXT: ret i32 [[TMP5]]
; CHECK: hd:
; CHECK-NEXT: [[TMP7:%.*]] = phi i32 [ [[TMP8:%.*]], [[HD2:%.*]] ], [ 0, [[EN:%.*]] ]
; CHECK-NEXT: tail call void @h(i32* nonnull [[A]])
; CHECK-NEXT: br label [[HD2]]
; CHECK: hd2:
; CHECK-NEXT: [[TMP8]] = add nuw i32 [[TMP7]], 1
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i32 [[TMP8]], [[B]]
; CHECK-NEXT: br i1 [[TMP9]], label [[EX]], label [[HD]]
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
;
en:
%tmp3 = icmp eq i32 %b, 0
br i1 %tmp3, label %ex, label %hd
ex:
%tmp5 = tail call i32 @g(i32* nonnull %a)
ret i32 %tmp5
hd:
%tmp7 = phi i32 [ %tmp8, %hd2 ], [ 0, %en ]
tail call void @h(i32* %a)
br label %hd2
hd2:
%tmp8 = add nuw i32 %tmp7, 1
%tmp9 = icmp eq i32 %tmp8, %b
br i1 %tmp9, label %ex, label %hd
}
; Original from PR43833
declare void @sink(i32*)
; FIXME: the sink argument should be marked nonnull as in @PR43833_simple.
define void @PR43833(i32* %0, i32 %1) {
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@PR43833
; CHECK-SAME: (i32* [[TMP0:%.*]], i32 [[TMP1:%.*]])
; CHECK-NEXT: [[TMP3:%.*]] = icmp sgt i32 [[TMP1]], 1
; CHECK-NEXT: br i1 [[TMP3]], label [[TMP4:%.*]], label [[TMP7:%.*]]
; CHECK: 4:
; CHECK-NEXT: [[TMP5:%.*]] = zext i32 [[TMP1]] to i64
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, i32* [[TMP0]], i64 [[TMP5]]
; CHECK-NEXT: br label [[TMP8:%.*]]
; CHECK: 7:
; CHECK-NEXT: ret void
; CHECK: 8:
; CHECK-NEXT: [[TMP9:%.*]] = phi i32 [ 1, [[TMP4]] ], [ [[TMP10:%.*]], [[TMP8]] ]
; CHECK-NEXT: tail call void @sink(i32* [[TMP6]])
; CHECK-NEXT: [[TMP10]] = add nuw nsw i32 [[TMP9]], 1
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i32 [[TMP10]], [[TMP1]]
; CHECK-NEXT: br i1 [[TMP11]], label [[TMP7]], label [[TMP8]]
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
;
%3 = icmp sgt i32 %1, 1
br i1 %3, label %4, label %7
4: ; preds = %2
%5 = zext i32 %1 to i64
%6 = getelementptr inbounds i32, i32* %0, i64 %5
br label %8
7: ; preds = %8, %2
ret void
8: ; preds = %8, %4
%9 = phi i32 [ 1, %4 ], [ %10, %8 ]
tail call void @sink(i32* %6)
%10 = add nuw nsw i32 %9, 1
%11 = icmp eq i32 %10, %1
br i1 %11, label %7, label %8
}
; Adjusted from PR43833
define void @PR43833_simple(i32* %0, i32 %1) {
[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 08:04:57 +08:00
; IS________OPM-LABEL: define {{[^@]+}}@PR43833_simple
; IS________OPM-SAME: (i32* [[TMP0:%.*]], i32 [[TMP1:%.*]])
; IS________OPM-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP1]], 0
; IS________OPM-NEXT: br i1 [[TMP3]], label [[TMP4:%.*]], label [[TMP7:%.*]]
; IS________OPM: 4:
; IS________OPM-NEXT: [[TMP5:%.*]] = zext i32 [[TMP1]] to i64
; IS________OPM-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, i32* [[TMP0]], i64 [[TMP5]]
; IS________OPM-NEXT: br label [[TMP8:%.*]]
; IS________OPM: 7:
; IS________OPM-NEXT: ret void
; IS________OPM: 8:
; IS________OPM-NEXT: [[TMP9:%.*]] = phi i32 [ 1, [[TMP4]] ], [ [[TMP10:%.*]], [[TMP8]] ]
; IS________OPM-NEXT: tail call void @sink(i32* [[TMP6]])
; IS________OPM-NEXT: [[TMP10]] = add nuw nsw i32 [[TMP9]], 1
; IS________OPM-NEXT: [[TMP11:%.*]] = icmp eq i32 [[TMP10]], [[TMP1]]
; IS________OPM-NEXT: br i1 [[TMP11]], label [[TMP7]], label [[TMP8]]
;
; IS________NPM-LABEL: define {{[^@]+}}@PR43833_simple
; IS________NPM-SAME: (i32* [[TMP0:%.*]], i32 [[TMP1:%.*]])
; IS________NPM-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP1]], 0
; IS________NPM-NEXT: br i1 [[TMP3]], label [[TMP4:%.*]], label [[TMP7:%.*]]
; IS________NPM: 4:
; IS________NPM-NEXT: [[TMP5:%.*]] = zext i32 [[TMP1]] to i64
; IS________NPM-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, i32* [[TMP0]], i64 [[TMP5]]
; IS________NPM-NEXT: br label [[TMP8:%.*]]
; IS________NPM: 7:
; IS________NPM-NEXT: ret void
; IS________NPM: 8:
; IS________NPM-NEXT: [[TMP9:%.*]] = phi i32 [ 1, [[TMP4]] ], [ [[TMP10:%.*]], [[TMP8]] ]
; IS________NPM-NEXT: tail call void @sink(i32* nonnull [[TMP6]])
; IS________NPM-NEXT: [[TMP10]] = add nuw nsw i32 [[TMP9]], 1
; IS________NPM-NEXT: [[TMP11:%.*]] = icmp eq i32 [[TMP10]], [[TMP1]]
; IS________NPM-NEXT: br i1 [[TMP11]], label [[TMP7]], label [[TMP8]]
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
;
%3 = icmp ne i32 %1, 0
br i1 %3, label %4, label %7
4: ; preds = %2
%5 = zext i32 %1 to i64
%6 = getelementptr inbounds i32, i32* %0, i64 %5
br label %8
7: ; preds = %8, %2
ret void
8: ; preds = %8, %4
%9 = phi i32 [ 1, %4 ], [ %10, %8 ]
tail call void @sink(i32* %6)
%10 = add nuw nsw i32 %9, 1
%11 = icmp eq i32 %10, %1
br i1 %11, label %7, label %8
}
[Attributor][FIX] Do not derive nonnull and dereferenceable w/o access An inbounds GEP results in poison if the value is not "inbounds", not in UB. We accidentally derived nonnull and dereferenceable from these inbounds GEPs even in the absence of accesses that would make the poison to UB.
2020-01-01 06:09:12 +08:00
declare i8* @strrchr(i8* %0, i32 %1) nofree nounwind readonly
; We should not mark the return of @strrchr as `nonnull`, it may well be NULL!
define i8* @mybasename(i8* nofree readonly %str) {
[Attributor][NFC] applying update_test_checks with --check-attributes Summary: All tests are updated, except wrapper.ll since it is not working nicely with newly created functions. Reviewers: jdoerfert, uenoku, baziotis, homerdin Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D84130
2020-07-20 03:55:43 +08:00
; CHECK: Function Attrs: nofree nounwind readonly
[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 08:04:57 +08:00
; CHECK-LABEL: define {{[^@]+}}@mybasename
; CHECK-SAME: (i8* nofree readonly [[STR:%.*]])
; CHECK-NEXT: [[CALL:%.*]] = call i8* @strrchr(i8* nofree readonly [[STR]], i32 47)
; CHECK-NEXT: [[TOBOOL:%.*]] = icmp ne i8* [[CALL]], null
; CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[CALL]], i64 1
; CHECK-NEXT: [[COND:%.*]] = select i1 [[TOBOOL]], i8* [[ADD_PTR]], i8* [[STR]]
; CHECK-NEXT: ret i8* [[COND]]
[Attributor][FIX] Do not derive nonnull and dereferenceable w/o access An inbounds GEP results in poison if the value is not "inbounds", not in UB. We accidentally derived nonnull and dereferenceable from these inbounds GEPs even in the absence of accesses that would make the poison to UB.
2020-01-01 06:09:12 +08:00
;
%call = call i8* @strrchr(i8* %str, i32 47)
%tobool = icmp ne i8* %call, null
%add.ptr = getelementptr inbounds i8, i8* %call, i64 1
%cond = select i1 %tobool, i8* %add.ptr, i8* %str
ret i8* %cond
}
[Attributor] Use knowledge retained in llvm.assume (operand bundles) This patch integrates operand bundle llvm.assumes [0] with the Attributor. Most IRAttributes will now look at uses of the associated value and if there are llvm.assume operand bundle uses with the right tag we will check if they are in the must-be-executed-context (around the context instruction). Droppable users, which is currently only llvm::assume, are handled special in some places now as well. [0] http://lists.llvm.org/pipermail/llvm-dev/2019-December/137632.html Reviewed By: uenoku Differential Revision: https://reviews.llvm.org/D74888
2020-02-20 16:06:48 +08:00
define void @nonnull_assume_pos(i8* %arg) {
; ATTRIBUTOR-LABEL: define {{[^@]+}}@nonnull_assume_pos
; ATTRIBUTOR-SAME: (i8* nocapture nofree nonnull readnone [[ARG:%.*]])
; ATTRIBUTOR-NEXT: call void @llvm.assume(i1 true) #11 [ "nonnull"(i8* [[ARG]]) ]
; ATTRIBUTOR-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; ATTRIBUTOR-NEXT: [[TMP1:%.*]] = call i8* @unknown()
; ATTRIBUTOR-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 08:04:57 +08:00
;
; NOT_CGSCC_OPM-LABEL: define {{[^@]+}}@nonnull_assume_pos
; NOT_CGSCC_OPM-SAME: (i8* nocapture nofree nonnull readnone [[ARG:%.*]])
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; NOT_CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR11]] [ "nonnull"(i8* [[ARG]]) ]
[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 08:04:57 +08:00
; NOT_CGSCC_OPM-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; NOT_CGSCC_OPM-NEXT: [[TMP1:%.*]] = call i8* @unknown()
; NOT_CGSCC_OPM-NEXT: ret void
;
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@nonnull_assume_pos
; IS__CGSCC_OPM-SAME: (i8* nocapture nofree nonnull readnone [[ARG:%.*]])
[Attributor][NFC] Reformat tests after D85099 Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D85700
2020-08-11 09:31:30 +08:00
; IS__CGSCC_OPM-NEXT: call void @llvm.assume(i1 true) [[ATTR12]] [ "nonnull"(i8* [[ARG]]) ]
[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 08:04:57 +08:00
; IS__CGSCC_OPM-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; IS__CGSCC_OPM-NEXT: [[TMP1:%.*]] = call i8* @unknown()
; IS__CGSCC_OPM-NEXT: ret void
[Attributor] Use knowledge retained in llvm.assume (operand bundles) This patch integrates operand bundle llvm.assumes [0] with the Attributor. Most IRAttributes will now look at uses of the associated value and if there are llvm.assume operand bundle uses with the right tag we will check if they are in the must-be-executed-context (around the context instruction). Droppable users, which is currently only llvm::assume, are handled special in some places now as well. [0] http://lists.llvm.org/pipermail/llvm-dev/2019-December/137632.html Reviewed By: uenoku Differential Revision: https://reviews.llvm.org/D74888
2020-02-20 16:06:48 +08:00
;
call void @llvm.assume(i1 true) ["nonnull"(i8* %arg)]
call void @use_i8_ptr(i8* %arg)
call i8* @unknown()
ret void
}
define void @nonnull_assume_neg(i8* %arg) {
; ATTRIBUTOR-LABEL: define {{[^@]+}}@nonnull_assume_neg
; ATTRIBUTOR-SAME: (i8* nocapture nofree readnone [[ARG:%.*]])
; ATTRIBUTOR-NEXT: [[TMP1:%.*]] = call i8* @unknown()
; ATTRIBUTOR-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree readnone [[ARG]])
; ATTRIBUTOR-NEXT: call void @llvm.assume(i1 true) [ "nonnull"(i8* [[ARG]]) ]
; ATTRIBUTOR-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; ATTRIBUTOR-NEXT: [[TMP2:%.*]] = call i8* @unknown()
; ATTRIBUTOR-NEXT: call void @use_i8_ptr_ret(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; ATTRIBUTOR-NEXT: call void @llvm.assume(i1 true) [ "nonnull"(i8* [[ARG]]) ]
; ATTRIBUTOR-NEXT: call void @use_i8_ptr_ret(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; ATTRIBUTOR-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 08:04:57 +08:00
;
; CHECK-LABEL: define {{[^@]+}}@nonnull_assume_neg
; CHECK-SAME: (i8* nocapture nofree readnone [[ARG:%.*]])
; CHECK-NEXT: [[TMP1:%.*]] = call i8* @unknown()
; CHECK-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree readnone [[ARG]])
; CHECK-NEXT: call void @llvm.assume(i1 true) [ "nonnull"(i8* [[ARG]]) ]
; CHECK-NEXT: call void @use_i8_ptr(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; CHECK-NEXT: [[TMP2:%.*]] = call i8* @unknown()
; CHECK-NEXT: call void @use_i8_ptr_ret(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; CHECK-NEXT: call void @llvm.assume(i1 true) [ "nonnull"(i8* [[ARG]]) ]
; CHECK-NEXT: call void @use_i8_ptr_ret(i8* noalias nocapture nofree nonnull readnone [[ARG]])
; CHECK-NEXT: ret void
[Attributor] Use knowledge retained in llvm.assume (operand bundles) This patch integrates operand bundle llvm.assumes [0] with the Attributor. Most IRAttributes will now look at uses of the associated value and if there are llvm.assume operand bundle uses with the right tag we will check if they are in the must-be-executed-context (around the context instruction). Droppable users, which is currently only llvm::assume, are handled special in some places now as well. [0] http://lists.llvm.org/pipermail/llvm-dev/2019-December/137632.html Reviewed By: uenoku Differential Revision: https://reviews.llvm.org/D74888
2020-02-20 16:06:48 +08:00
;
call i8* @unknown()
call void @use_i8_ptr(i8* %arg)
call void @llvm.assume(i1 true) ["nonnull"(i8* %arg)]
call void @use_i8_ptr(i8* %arg)
call i8* @unknown()
call void @use_i8_ptr_ret(i8* %arg)
call void @llvm.assume(i1 true) ["nonnull"(i8* %arg)]
call void @use_i8_ptr_ret(i8* %arg)
ret void
}
declare void @use_i8_ptr(i8* nofree nocapture readnone) nounwind
declare void @use_i8_ptr_ret(i8* nofree nocapture readnone) nounwind willreturn
[Attributor][FIX] Handle function pointers properly in AANonNull Before we tired to create a dominator tree for a declaration when we wanted to determine if the function pointer is `nonnull`. We now avoid looking at global values if `Value::getPointerDereferenceableBytes` not already determined `nonnull`.
2020-08-18 08:54:42 +08:00
define i8* @nonnull_function_ptr_1() {
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@nonnull_function_ptr_1()
; IS__TUNIT____-NEXT: [[BC:%.*]] = bitcast i8* ()* @nonnull_function_ptr_1 to i8*
; IS__TUNIT____-NEXT: ret i8* [[BC]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@nonnull_function_ptr_1()
; IS__CGSCC____-NEXT: [[BC:%.*]] = bitcast i8* ()* @nonnull_function_ptr_1 to i8*
; IS__CGSCC____-NEXT: ret i8* [[BC]]
;
%bc = bitcast i8*()* @nonnull_function_ptr_1 to i8*
ret i8* %bc
}
declare i8* @function_decl()
define i8* @nonnull_function_ptr_2() {
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@nonnull_function_ptr_2()
; IS__TUNIT____-NEXT: [[BC:%.*]] = bitcast i8* ()* @function_decl to i8*
; IS__TUNIT____-NEXT: ret i8* [[BC]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@nonnull_function_ptr_2()
; IS__CGSCC____-NEXT: [[BC:%.*]] = bitcast i8* ()* @function_decl to i8*
; IS__CGSCC____-NEXT: ret i8* [[BC]]
;
%bc = bitcast i8*()* @function_decl to i8*
ret i8* %bc
}
[IR] Convert null-pointer-is-valid into an enum attribute The "null-pointer-is-valid" attribute needs to be checked by many pointer-related combines. To make the check more efficient, convert it from a string into an enum attribute. In the future, this attribute may be replaced with data layout properties. Differential Revision: https://reviews.llvm.org/D78862
2020-04-25 18:57:07 +08:00
attributes #0 = { null_pointer_is_valid }
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
attributes #1 = { nounwind willreturn}