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/range.ll

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

1959 lines
87 KiB
LLVM
Raw Normal View History

[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
[Attributor] Introduce a helper function to deal with undef + none We often need to deal with the value lattice that contains none and undef as special values. A simple helper makes this much nicer. Differential Revision: https://reviews.llvm.org/D103857
2021-05-08 12:18:24 +08:00
; RUN: opt -attributor -enable-new-pm=0 -attributor-manifest-internal -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=25 -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=26 -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][Tests] Run the CGSCC versions on the range.ll test
2020-02-10 09:08:19 +08:00
; FIXME: CGSCC is not looking at callees and calleers even though it could be allowed.
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
define i32 @test0(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: argmemonly nofree nosync nounwind readonly willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test0
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT____-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0:[0-9]+]] {
; IS__TUNIT____-NEXT: [[A:%.*]] = load i32, i32* [[P]], align 4, !range [[RNG0:![0-9]+]]
[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____-NEXT: ret i32 [[A]]
;
; IS__CGSCC____: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test0
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC____-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0:[0-9]+]] {
; IS__CGSCC____-NEXT: [[A:%.*]] = load i32, i32* [[P]], align 4, !range [[RNG0:![0-9]+]]
[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____-NEXT: ret i32 [[A]]
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
%a = load i32, i32* %p, !range !0
ret i32 %a
}
define i32 @test0-range-check(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_OPM: Function Attrs: argmemonly nofree nosync nounwind readonly willreturn
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@test0-range-check
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_OPM-SAME: (i32* nocapture nofree readonly align 4 [[P:%.*]]) #[[ATTR0]] {
; IS__TUNIT_OPM-NEXT: [[A:%.*]] = tail call i32 @test0(i32* nocapture nofree readonly align 4 [[P]]) #[[ATTR3:[0-9]+]], !range [[RNG0]]
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: ret 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_NPM: Function Attrs: argmemonly nofree nosync nounwind readonly willreturn
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@test0-range-check
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-SAME: (i32* nocapture nofree readonly align 4 [[P:%.*]]) #[[ATTR0]] {
; IS__TUNIT_NPM-NEXT: [[A:%.*]] = tail call i32 @test0(i32* nocapture nofree readonly align 4 [[P]]) #[[ATTR2:[0-9]+]], !range [[RNG0]]
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: ret i32 [[A]]
[Attributor][Tests] Run the CGSCC versions on the range.ll test
2020-02-10 09:08:19 +08:00
;
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@test0-range-check
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0]] {
; IS__CGSCC_OPM-NEXT: [[A:%.*]] = tail call i32 @test0(i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]]) #[[ATTR3:[0-9]+]], !range [[RNG0]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: ret i32 [[A]]
;
; IS__CGSCC_NPM: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test0-range-check
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0]] {
; IS__CGSCC_NPM-NEXT: [[A:%.*]] = tail call i32 @test0(i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]]) #[[ATTR2:[0-9]+]], !range [[RNG0]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: ret i32 [[A]]
[Attributor][FIX] Make check lines explicit There is a bug in `update_test_checks.py` that combines check lines it should not. For now we unbreak the bots by making all possibilities explicit.
2020-02-10 15:31:20 +08:00
;
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
%a = tail call i32 @test0(i32* %p)
ret i32 %a
}
declare void @use3-dummy(i1, i1, i1)
define void @use3(i1, i1, i1) {
; CHECK-LABEL: define {{[^@]+}}@use3
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; CHECK-SAME: (i1 [[TMP0:%.*]], i1 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
; CHECK-NEXT: tail call void @use3-dummy(i1 [[TMP0]], i1 [[TMP1]], i1 [[TMP2]])
; CHECK-NEXT: ret void
;
tail call void @use3-dummy(i1 %0, i1 %1, i1 %2)
ret void
}
; TEST0 icmp test
define void @test0-icmp-check(i32* %p){
Temporarily revert "[Attributor] Unify testing (=updates,...)" This patch reverts the 2 patches below, as on most systems the disabled tests actually pass and that causes most bots to be red, including http://green.lab.llvm.org/green/job/clang-stage1-RA/8541/ http://green.lab.llvm.org/green/job/clang-stage1-cmake-RA-expensive/15646/ http://lab.llvm.org:8011/builders/clang-ppc64le-linux-lnt/builds/23690 http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-bootstrap/builds/16751 * [Attributor] Disable three tests until the SCC update bug was fixed commit 2ae1a76c27295370d552f57352508b7af4dba791. * [Attributor] Unify testing (=updates,prefixes,run configurations,...) 2bcf5793e14577ed3c9f15a77a35179ae544fff4.
2020-04-09 18:07:36 +08:00
; ret = [0, 10)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@test0-icmp-check
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__TUNIT_OPM-SAME: (i32* nocapture nofree readonly align 4 [[P:%.*]]) {
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_OPM-NEXT: [[RET:%.*]] = tail call i32 @test0(i32* nocapture nofree readonly align 4 [[P]]) #[[ATTR3]], !range [[RNG0]]
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP_EQ_2:%.*]] = icmp eq i32 [[RET]], 9
; IS__TUNIT_OPM-NEXT: [[CMP_EQ_3:%.*]] = icmp eq i32 [[RET]], 8
; IS__TUNIT_OPM-NEXT: [[CMP_EQ_4:%.*]] = icmp eq i32 [[RET]], 1
; IS__TUNIT_OPM-NEXT: [[CMP_EQ_5:%.*]] = icmp eq i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_EQ_2]], i1 [[CMP_EQ_3]])
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 [[CMP_EQ_4]], i1 [[CMP_EQ_5]], i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP_NE_2:%.*]] = icmp ne i32 [[RET]], 9
; IS__TUNIT_OPM-NEXT: [[CMP_NE_3:%.*]] = icmp ne i32 [[RET]], 8
; IS__TUNIT_OPM-NEXT: [[CMP_NE_4:%.*]] = icmp ne i32 [[RET]], 1
; IS__TUNIT_OPM-NEXT: [[CMP_NE_5:%.*]] = icmp ne i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 noundef true, i1 [[CMP_NE_2]], i1 [[CMP_NE_3]])
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 [[CMP_NE_4]], i1 [[CMP_NE_5]], i1 noundef true)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP_UGT_3:%.*]] = icmp ugt i32 [[RET]], 8
; IS__TUNIT_OPM-NEXT: [[CMP_UGT_4:%.*]] = icmp ugt i32 [[RET]], 1
; IS__TUNIT_OPM-NEXT: [[CMP_UGT_5:%.*]] = icmp ugt i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 noundef false, i1 noundef false, i1 [[CMP_UGT_3]])
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 [[CMP_UGT_4]], i1 [[CMP_UGT_5]], i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP_UGE_2:%.*]] = icmp uge i32 [[RET]], 9
; IS__TUNIT_OPM-NEXT: [[CMP_UGE_3:%.*]] = icmp uge i32 [[RET]], 8
; IS__TUNIT_OPM-NEXT: [[CMP_UGE_4:%.*]] = icmp uge i32 [[RET]], 1
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_UGE_2]], i1 [[CMP_UGE_3]])
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 [[CMP_UGE_4]], i1 noundef true, i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP_SGT_3:%.*]] = icmp sgt i32 [[RET]], 8
; IS__TUNIT_OPM-NEXT: [[CMP_SGT_4:%.*]] = icmp sgt i32 [[RET]], 1
; IS__TUNIT_OPM-NEXT: [[CMP_SGT_5:%.*]] = icmp sgt i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 noundef false, i1 noundef false, i1 [[CMP_SGT_3]])
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 [[CMP_SGT_4]], i1 [[CMP_SGT_5]], i1 noundef true)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP_GTE_2:%.*]] = icmp sge i32 [[RET]], 9
; IS__TUNIT_OPM-NEXT: [[CMP_GTE_3:%.*]] = icmp sge i32 [[RET]], 8
; IS__TUNIT_OPM-NEXT: [[CMP_GTE_4:%.*]] = icmp sge i32 [[RET]], 1
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_GTE_2]], i1 [[CMP_GTE_3]])
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 [[CMP_GTE_4]], i1 noundef true, i1 noundef true)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP_SLT_2:%.*]] = icmp slt i32 [[RET]], 9
; IS__TUNIT_OPM-NEXT: [[CMP_SLT_3:%.*]] = icmp slt i32 [[RET]], 8
; IS__TUNIT_OPM-NEXT: [[CMP_SLT_4:%.*]] = icmp slt i32 [[RET]], 1
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 noundef true, i1 [[CMP_SLT_2]], i1 [[CMP_SLT_3]])
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 [[CMP_SLT_4]], i1 noundef false, i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP_LTE_3:%.*]] = icmp sle i32 [[RET]], 8
; IS__TUNIT_OPM-NEXT: [[CMP_LTE_4:%.*]] = icmp sle i32 [[RET]], 1
; IS__TUNIT_OPM-NEXT: [[CMP_LTE_5:%.*]] = icmp sle i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 noundef true, i1 noundef true, i1 [[CMP_LTE_3]])
; IS__TUNIT_OPM-NEXT: tail call void @use3(i1 [[CMP_LTE_4]], i1 [[CMP_LTE_5]], i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: ret void
;
; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@test0-icmp-check
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__TUNIT_NPM-SAME: (i32* nocapture nofree readonly align 4 [[P:%.*]]) {
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-NEXT: [[RET:%.*]] = tail call i32 @test0(i32* nocapture nofree readonly align 4 [[P]]) #[[ATTR2]], !range [[RNG0]]
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP_EQ_2:%.*]] = icmp eq i32 [[RET]], 9
; IS__TUNIT_NPM-NEXT: [[CMP_EQ_3:%.*]] = icmp eq i32 [[RET]], 8
; IS__TUNIT_NPM-NEXT: [[CMP_EQ_4:%.*]] = icmp eq i32 [[RET]], 1
; IS__TUNIT_NPM-NEXT: [[CMP_EQ_5:%.*]] = icmp eq i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_EQ_2]], i1 [[CMP_EQ_3]])
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 [[CMP_EQ_4]], i1 [[CMP_EQ_5]], i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP_NE_2:%.*]] = icmp ne i32 [[RET]], 9
; IS__TUNIT_NPM-NEXT: [[CMP_NE_3:%.*]] = icmp ne i32 [[RET]], 8
; IS__TUNIT_NPM-NEXT: [[CMP_NE_4:%.*]] = icmp ne i32 [[RET]], 1
; IS__TUNIT_NPM-NEXT: [[CMP_NE_5:%.*]] = icmp ne i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 noundef true, i1 [[CMP_NE_2]], i1 [[CMP_NE_3]])
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 [[CMP_NE_4]], i1 [[CMP_NE_5]], i1 noundef true)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP_UGT_3:%.*]] = icmp ugt i32 [[RET]], 8
; IS__TUNIT_NPM-NEXT: [[CMP_UGT_4:%.*]] = icmp ugt i32 [[RET]], 1
; IS__TUNIT_NPM-NEXT: [[CMP_UGT_5:%.*]] = icmp ugt i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 noundef false, i1 noundef false, i1 [[CMP_UGT_3]])
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 [[CMP_UGT_4]], i1 [[CMP_UGT_5]], i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP_UGE_2:%.*]] = icmp uge i32 [[RET]], 9
; IS__TUNIT_NPM-NEXT: [[CMP_UGE_3:%.*]] = icmp uge i32 [[RET]], 8
; IS__TUNIT_NPM-NEXT: [[CMP_UGE_4:%.*]] = icmp uge i32 [[RET]], 1
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_UGE_2]], i1 [[CMP_UGE_3]])
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 [[CMP_UGE_4]], i1 noundef true, i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP_SGT_3:%.*]] = icmp sgt i32 [[RET]], 8
; IS__TUNIT_NPM-NEXT: [[CMP_SGT_4:%.*]] = icmp sgt i32 [[RET]], 1
; IS__TUNIT_NPM-NEXT: [[CMP_SGT_5:%.*]] = icmp sgt i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 noundef false, i1 noundef false, i1 [[CMP_SGT_3]])
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 [[CMP_SGT_4]], i1 [[CMP_SGT_5]], i1 noundef true)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP_GTE_2:%.*]] = icmp sge i32 [[RET]], 9
; IS__TUNIT_NPM-NEXT: [[CMP_GTE_3:%.*]] = icmp sge i32 [[RET]], 8
; IS__TUNIT_NPM-NEXT: [[CMP_GTE_4:%.*]] = icmp sge i32 [[RET]], 1
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_GTE_2]], i1 [[CMP_GTE_3]])
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 [[CMP_GTE_4]], i1 noundef true, i1 noundef true)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP_SLT_2:%.*]] = icmp slt i32 [[RET]], 9
; IS__TUNIT_NPM-NEXT: [[CMP_SLT_3:%.*]] = icmp slt i32 [[RET]], 8
; IS__TUNIT_NPM-NEXT: [[CMP_SLT_4:%.*]] = icmp slt i32 [[RET]], 1
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 noundef true, i1 [[CMP_SLT_2]], i1 [[CMP_SLT_3]])
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 [[CMP_SLT_4]], i1 noundef false, i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP_LTE_3:%.*]] = icmp sle i32 [[RET]], 8
; IS__TUNIT_NPM-NEXT: [[CMP_LTE_4:%.*]] = icmp sle i32 [[RET]], 1
; IS__TUNIT_NPM-NEXT: [[CMP_LTE_5:%.*]] = icmp sle i32 [[RET]], 0
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 noundef true, i1 noundef true, i1 [[CMP_LTE_3]])
; IS__TUNIT_NPM-NEXT: tail call void @use3(i1 [[CMP_LTE_4]], i1 [[CMP_LTE_5]], i1 noundef false)
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-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
;
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@test0-icmp-check
[ValueTracking] Dereferenced pointers are noundef This is a follow-up of D95238's LangRef update. This patch updates `programUndefinedIfUndefOrPoison(V)` to return true if `V` is used by any memory-accessing instruction. Interestingly, this affected many tests in Attributors, mainly about adding noundefs. The tests are updated using llvm/utils/update_test_checks.py. I checked that the diffs are about updating noundefs. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D96642
2021-02-13 15:22:25 +08:00
; IS__CGSCC_OPM-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) {
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-NEXT: [[RET:%.*]] = tail call i32 @test0(i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]]) #[[ATTR3]], !range [[RNG0]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP_EQ_2:%.*]] = icmp eq i32 [[RET]], 9
; IS__CGSCC_OPM-NEXT: [[CMP_EQ_3:%.*]] = icmp eq i32 [[RET]], 8
; IS__CGSCC_OPM-NEXT: [[CMP_EQ_4:%.*]] = icmp eq i32 [[RET]], 1
; IS__CGSCC_OPM-NEXT: [[CMP_EQ_5:%.*]] = icmp eq i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_EQ_2]], i1 [[CMP_EQ_3]])
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 [[CMP_EQ_4]], i1 [[CMP_EQ_5]], i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP_NE_2:%.*]] = icmp ne i32 [[RET]], 9
; IS__CGSCC_OPM-NEXT: [[CMP_NE_3:%.*]] = icmp ne i32 [[RET]], 8
; IS__CGSCC_OPM-NEXT: [[CMP_NE_4:%.*]] = icmp ne i32 [[RET]], 1
; IS__CGSCC_OPM-NEXT: [[CMP_NE_5:%.*]] = icmp ne i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 noundef true, i1 [[CMP_NE_2]], i1 [[CMP_NE_3]])
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 [[CMP_NE_4]], i1 [[CMP_NE_5]], i1 noundef true)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP_UGT_3:%.*]] = icmp ugt i32 [[RET]], 8
; IS__CGSCC_OPM-NEXT: [[CMP_UGT_4:%.*]] = icmp ugt i32 [[RET]], 1
; IS__CGSCC_OPM-NEXT: [[CMP_UGT_5:%.*]] = icmp ugt i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 noundef false, i1 noundef false, i1 [[CMP_UGT_3]])
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 [[CMP_UGT_4]], i1 [[CMP_UGT_5]], i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP_UGE_2:%.*]] = icmp uge i32 [[RET]], 9
; IS__CGSCC_OPM-NEXT: [[CMP_UGE_3:%.*]] = icmp uge i32 [[RET]], 8
; IS__CGSCC_OPM-NEXT: [[CMP_UGE_4:%.*]] = icmp uge i32 [[RET]], 1
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_UGE_2]], i1 [[CMP_UGE_3]])
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 [[CMP_UGE_4]], i1 noundef true, i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP_SGT_3:%.*]] = icmp sgt i32 [[RET]], 8
; IS__CGSCC_OPM-NEXT: [[CMP_SGT_4:%.*]] = icmp sgt i32 [[RET]], 1
; IS__CGSCC_OPM-NEXT: [[CMP_SGT_5:%.*]] = icmp sgt i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 noundef false, i1 noundef false, i1 [[CMP_SGT_3]])
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 [[CMP_SGT_4]], i1 [[CMP_SGT_5]], i1 noundef true)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP_GTE_2:%.*]] = icmp sge i32 [[RET]], 9
; IS__CGSCC_OPM-NEXT: [[CMP_GTE_3:%.*]] = icmp sge i32 [[RET]], 8
; IS__CGSCC_OPM-NEXT: [[CMP_GTE_4:%.*]] = icmp sge i32 [[RET]], 1
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_GTE_2]], i1 [[CMP_GTE_3]])
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 [[CMP_GTE_4]], i1 noundef true, i1 noundef true)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP_SLT_2:%.*]] = icmp slt i32 [[RET]], 9
; IS__CGSCC_OPM-NEXT: [[CMP_SLT_3:%.*]] = icmp slt i32 [[RET]], 8
; IS__CGSCC_OPM-NEXT: [[CMP_SLT_4:%.*]] = icmp slt i32 [[RET]], 1
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 noundef true, i1 [[CMP_SLT_2]], i1 [[CMP_SLT_3]])
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 [[CMP_SLT_4]], i1 noundef false, i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP_LTE_3:%.*]] = icmp sle i32 [[RET]], 8
; IS__CGSCC_OPM-NEXT: [[CMP_LTE_4:%.*]] = icmp sle i32 [[RET]], 1
; IS__CGSCC_OPM-NEXT: [[CMP_LTE_5:%.*]] = icmp sle i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 noundef true, i1 noundef true, i1 [[CMP_LTE_3]])
; IS__CGSCC_OPM-NEXT: tail call void @use3(i1 [[CMP_LTE_4]], i1 [[CMP_LTE_5]], i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: ret void
;
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test0-icmp-check
[ValueTracking] Dereferenced pointers are noundef This is a follow-up of D95238's LangRef update. This patch updates `programUndefinedIfUndefOrPoison(V)` to return true if `V` is used by any memory-accessing instruction. Interestingly, this affected many tests in Attributors, mainly about adding noundefs. The tests are updated using llvm/utils/update_test_checks.py. I checked that the diffs are about updating noundefs. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D96642
2021-02-13 15:22:25 +08:00
; IS__CGSCC_NPM-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) {
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-NEXT: [[RET:%.*]] = tail call i32 @test0(i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]]) #[[ATTR2]], !range [[RNG0]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP_EQ_2:%.*]] = icmp eq i32 [[RET]], 9
; IS__CGSCC_NPM-NEXT: [[CMP_EQ_3:%.*]] = icmp eq i32 [[RET]], 8
; IS__CGSCC_NPM-NEXT: [[CMP_EQ_4:%.*]] = icmp eq i32 [[RET]], 1
; IS__CGSCC_NPM-NEXT: [[CMP_EQ_5:%.*]] = icmp eq i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_EQ_2]], i1 [[CMP_EQ_3]])
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 [[CMP_EQ_4]], i1 [[CMP_EQ_5]], i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP_NE_2:%.*]] = icmp ne i32 [[RET]], 9
; IS__CGSCC_NPM-NEXT: [[CMP_NE_3:%.*]] = icmp ne i32 [[RET]], 8
; IS__CGSCC_NPM-NEXT: [[CMP_NE_4:%.*]] = icmp ne i32 [[RET]], 1
; IS__CGSCC_NPM-NEXT: [[CMP_NE_5:%.*]] = icmp ne i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 noundef true, i1 [[CMP_NE_2]], i1 [[CMP_NE_3]])
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 [[CMP_NE_4]], i1 [[CMP_NE_5]], i1 noundef true)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP_UGT_3:%.*]] = icmp ugt i32 [[RET]], 8
; IS__CGSCC_NPM-NEXT: [[CMP_UGT_4:%.*]] = icmp ugt i32 [[RET]], 1
; IS__CGSCC_NPM-NEXT: [[CMP_UGT_5:%.*]] = icmp ugt i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 noundef false, i1 noundef false, i1 [[CMP_UGT_3]])
[Attributor] Fix AANoUndef identification Even though `noundef` IR attribute might be attached to non-void type values, AANoUndef is mistakenly identified for pointer type values only. This patch fixes that. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86737
2020-08-30 04:34:24 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 [[CMP_UGT_4]], i1 [[CMP_UGT_5]], i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP_UGE_2:%.*]] = icmp uge i32 [[RET]], 9
; IS__CGSCC_NPM-NEXT: [[CMP_UGE_3:%.*]] = icmp uge i32 [[RET]], 8
; IS__CGSCC_NPM-NEXT: [[CMP_UGE_4:%.*]] = icmp uge i32 [[RET]], 1
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_UGE_2]], i1 [[CMP_UGE_3]])
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 [[CMP_UGE_4]], i1 noundef true, i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP_SGT_3:%.*]] = icmp sgt i32 [[RET]], 8
; IS__CGSCC_NPM-NEXT: [[CMP_SGT_4:%.*]] = icmp sgt i32 [[RET]], 1
; IS__CGSCC_NPM-NEXT: [[CMP_SGT_5:%.*]] = icmp sgt i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 noundef false, i1 noundef false, i1 [[CMP_SGT_3]])
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 [[CMP_SGT_4]], i1 [[CMP_SGT_5]], i1 noundef true)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP_GTE_2:%.*]] = icmp sge i32 [[RET]], 9
; IS__CGSCC_NPM-NEXT: [[CMP_GTE_3:%.*]] = icmp sge i32 [[RET]], 8
; IS__CGSCC_NPM-NEXT: [[CMP_GTE_4:%.*]] = icmp sge i32 [[RET]], 1
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 noundef false, i1 [[CMP_GTE_2]], i1 [[CMP_GTE_3]])
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 [[CMP_GTE_4]], i1 noundef true, i1 noundef true)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP_SLT_2:%.*]] = icmp slt i32 [[RET]], 9
; IS__CGSCC_NPM-NEXT: [[CMP_SLT_3:%.*]] = icmp slt i32 [[RET]], 8
; IS__CGSCC_NPM-NEXT: [[CMP_SLT_4:%.*]] = icmp slt i32 [[RET]], 1
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 noundef true, i1 [[CMP_SLT_2]], i1 [[CMP_SLT_3]])
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 [[CMP_SLT_4]], i1 noundef false, i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP_LTE_3:%.*]] = icmp sle i32 [[RET]], 8
; IS__CGSCC_NPM-NEXT: [[CMP_LTE_4:%.*]] = icmp sle i32 [[RET]], 1
; IS__CGSCC_NPM-NEXT: [[CMP_LTE_5:%.*]] = icmp sle i32 [[RET]], 0
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 noundef true, i1 noundef true, i1 [[CMP_LTE_3]])
; IS__CGSCC_NPM-NEXT: tail call void @use3(i1 [[CMP_LTE_4]], i1 [[CMP_LTE_5]], i1 noundef false)
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-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
;
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
%ret = tail call i32 @test0(i32 *%p)
; ret = [0, 10), eq
%cmp-eq-1 = icmp eq i32 %ret, 10
%cmp-eq-2 = icmp eq i32 %ret, 9
%cmp-eq-3 = icmp eq i32 %ret, 8
%cmp-eq-4 = icmp eq i32 %ret, 1
%cmp-eq-5 = icmp eq i32 %ret, 0
%cmp-eq-6 = icmp eq i32 %ret, -1
tail call void @use3(i1 %cmp-eq-1, i1 %cmp-eq-2, i1 %cmp-eq-3)
tail call void @use3(i1 %cmp-eq-4, i1 %cmp-eq-5, i1 %cmp-eq-6)
; ret = [0, 10), ne
%cmp-ne-1 = icmp ne i32 %ret, 10
%cmp-ne-2 = icmp ne i32 %ret, 9
%cmp-ne-3 = icmp ne i32 %ret, 8
%cmp-ne-4 = icmp ne i32 %ret, 1
%cmp-ne-5 = icmp ne i32 %ret, 0
%cmp-ne-6 = icmp ne i32 %ret, -1
tail call void @use3(i1 %cmp-ne-1, i1 %cmp-ne-2, i1 %cmp-ne-3)
tail call void @use3(i1 %cmp-ne-4, i1 %cmp-ne-5, i1 %cmp-ne-6)
; ret = [0, 10), ugt
%cmp-ugt-1 = icmp ugt i32 %ret, 10
%cmp-ugt-2 = icmp ugt i32 %ret, 9
%cmp-ugt-3 = icmp ugt i32 %ret, 8
%cmp-ugt-4 = icmp ugt i32 %ret, 1
%cmp-ugt-5 = icmp ugt i32 %ret, 0
%cmp-ugt-6 = icmp ugt i32 %ret, -1
tail call void @use3(i1 %cmp-ugt-1, i1 %cmp-ugt-2, i1 %cmp-ugt-3)
tail call void @use3(i1 %cmp-ugt-4, i1 %cmp-ugt-5, i1 %cmp-ugt-6)
; ret = [0, 10), uge
%cmp-uge-1 = icmp uge i32 %ret, 10
%cmp-uge-2 = icmp uge i32 %ret, 9
%cmp-uge-3 = icmp uge i32 %ret, 8
%cmp-uge-4 = icmp uge i32 %ret, 1
%cmp-uge-5 = icmp uge i32 %ret, 0
%cmp-uge-6 = icmp uge i32 %ret, -1
tail call void @use3(i1 %cmp-uge-1, i1 %cmp-uge-2, i1 %cmp-uge-3)
tail call void @use3(i1 %cmp-uge-4, i1 %cmp-uge-5, i1 %cmp-uge-6)
; ret = [0, 10), sgt
%cmp-sgt-1 = icmp sgt i32 %ret, 10
%cmp-sgt-2 = icmp sgt i32 %ret, 9
%cmp-sgt-3 = icmp sgt i32 %ret, 8
%cmp-sgt-4 = icmp sgt i32 %ret, 1
%cmp-sgt-5 = icmp sgt i32 %ret, 0
%cmp-sgt-6 = icmp sgt i32 %ret, -1
tail call void @use3(i1 %cmp-sgt-1, i1 %cmp-sgt-2, i1 %cmp-sgt-3)
tail call void @use3(i1 %cmp-sgt-4, i1 %cmp-sgt-5, i1 %cmp-sgt-6)
; ret = [0, 10), sge
%cmp-gte-1 = icmp sge i32 %ret, 10
%cmp-gte-2 = icmp sge i32 %ret, 9
%cmp-gte-3 = icmp sge i32 %ret, 8
%cmp-gte-4 = icmp sge i32 %ret, 1
%cmp-gte-5 = icmp sge i32 %ret, 0
%cmp-gte-6 = icmp sge i32 %ret, -1
tail call void @use3(i1 %cmp-gte-1, i1 %cmp-gte-2, i1 %cmp-gte-3)
tail call void @use3(i1 %cmp-gte-4, i1 %cmp-gte-5, i1 %cmp-gte-6)
; ret = [0, 10), slt
%cmp-slt-1 = icmp slt i32 %ret, 10
%cmp-slt-2 = icmp slt i32 %ret, 9
%cmp-slt-3 = icmp slt i32 %ret, 8
%cmp-slt-4 = icmp slt i32 %ret, 1
%cmp-slt-5 = icmp slt i32 %ret, 0
%cmp-slt-6 = icmp slt i32 %ret, -1
tail call void @use3(i1 %cmp-slt-1, i1 %cmp-slt-2, i1 %cmp-slt-3)
tail call void @use3(i1 %cmp-slt-4, i1 %cmp-slt-5, i1 %cmp-slt-6)
; ret = [0, 10), sle
%cmp-lte-1 = icmp sle i32 %ret, 10
%cmp-lte-2 = icmp sle i32 %ret, 9
%cmp-lte-3 = icmp sle i32 %ret, 8
%cmp-lte-4 = icmp sle i32 %ret, 1
%cmp-lte-5 = icmp sle i32 %ret, 0
%cmp-lte-6 = icmp sle i32 %ret, -1
tail call void @use3(i1 %cmp-lte-1, i1 %cmp-lte-2, i1 %cmp-lte-3)
tail call void @use3(i1 %cmp-lte-4, i1 %cmp-lte-5, i1 %cmp-lte-6)
ret void
}
define i32 @test1(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: argmemonly nofree nosync nounwind readonly willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test1
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT____-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0]] {
; IS__TUNIT____-NEXT: [[LOAD_10_100:%.*]] = load i32, i32* [[P]], align 4, !range [[RNG1:![0-9]+]]
[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____-NEXT: [[ADD_10_THEN_20_110:%.*]] = add i32 [[LOAD_10_100]], 10
; IS__TUNIT____-NEXT: [[MUL_10_THEN_200_1091:%.*]] = mul i32 [[ADD_10_THEN_20_110]], 10
; IS__TUNIT____-NEXT: ret i32 [[MUL_10_THEN_200_1091]]
;
; IS__CGSCC____: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test1
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC____-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0]] {
; IS__CGSCC____-NEXT: [[LOAD_10_100:%.*]] = load i32, i32* [[P]], align 4, !range [[RNG1:![0-9]+]]
[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____-NEXT: [[ADD_10_THEN_20_110:%.*]] = add i32 [[LOAD_10_100]], 10
; IS__CGSCC____-NEXT: [[MUL_10_THEN_200_1091:%.*]] = mul i32 [[ADD_10_THEN_20_110]], 10
; IS__CGSCC____-NEXT: ret i32 [[MUL_10_THEN_200_1091]]
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
%load-10-100 = load i32, i32* %p, !range !1
%add-10-then-20-110 = add i32 %load-10-100, 10
%mul-10-then-200-1091 = mul i32 %add-10-then-20-110, 10
ret i32 %mul-10-then-200-1091
}
define i1 @test1-check(i32* %p) {
[Attributor][FIX] Make check lines explicit There is a bug in `update_test_checks.py` that combines check lines it should not. For now we unbreak the bots by making all possibilities explicit.
2020-02-10 15:31:20 +08:00
;
[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_OPM: Function Attrs: argmemonly nofree nosync nounwind readonly willreturn
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@test1-check
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_OPM-SAME: (i32* nocapture nofree readonly align 4 [[P:%.*]]) #[[ATTR0]] {
; IS__TUNIT_OPM-NEXT: [[RES:%.*]] = tail call i32 @test1(i32* nocapture nofree readonly align 4 [[P]]) #[[ATTR3]], !range [[RNG2:![0-9]+]]
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_OPM-NEXT: [[CMP:%.*]] = icmp eq i32 [[RES]], 500
; IS__TUNIT_OPM-NEXT: ret i1 [[CMP]]
;
[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_NPM: Function Attrs: argmemonly nofree nosync nounwind readonly willreturn
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@test1-check
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-SAME: (i32* nocapture nofree readonly align 4 [[P:%.*]]) #[[ATTR0]] {
; IS__TUNIT_NPM-NEXT: [[RES:%.*]] = tail call i32 @test1(i32* nocapture nofree readonly align 4 [[P]]) #[[ATTR2]], !range [[RNG2:![0-9]+]]
[Attributor] Remember only necessary dependences Before we eagerly put dependences into the QueryMap as soon as we encountered them (via `Attributor::getAAFor<>` or `Attributor::recordDependence`). Now we will wait to see if the dependence is useful, that is if the target is not already in a fixpoint state at the end of the update. If so, there is no need to record the dependence at all. Due to the abstraction via `Attributor::updateAA` we will now also treat the very first update (during attribute creation) as we do subsequent updates. Finally this resolves the problematic usage of QueriedNonFixAA. --- Single run of the Attributor module and then CGSCC pass (oldPM) for SPASS/clause.c (~10k LLVM-IR loc): Before: ``` calls to allocation functions: 554675 (389245/s) temporary memory allocations: 101574 (71280/s) peak heap memory consumption: 28.46MB peak RSS (including heaptrack overhead): 116.26MB total memory leaked: 269.10KB ``` After: ``` calls to allocation functions: 512465 (345559/s) temporary memory allocations: 98832 (66643/s) peak heap memory consumption: 22.54MB peak RSS (including heaptrack overhead): 106.58MB total memory leaked: 269.10KB ``` Difference: ``` calls to allocation functions: -42210 (-727758/s) temporary memory allocations: -2742 (-47275/s) peak heap memory consumption: -5.92MB peak RSS (including heaptrack overhead): 0B total memory leaked: 0B ```
2020-05-01 01:12:22 +08:00
; IS__TUNIT_NPM-NEXT: [[CMP:%.*]] = icmp eq i32 [[RES]], 500
; IS__TUNIT_NPM-NEXT: ret i1 [[CMP]]
[Attributor][FIX] Make check lines explicit There is a bug in `update_test_checks.py` that combines check lines it should not. For now we unbreak the bots by making all possibilities explicit.
2020-02-10 15:31:20 +08:00
;
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@test1-check
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0]] {
; IS__CGSCC_OPM-NEXT: [[RES:%.*]] = tail call i32 @test1(i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]]) #[[ATTR3]], !range [[RNG2:![0-9]+]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[CMP:%.*]] = icmp eq i32 [[RES]], 500
; IS__CGSCC_OPM-NEXT: ret i1 [[CMP]]
;
; IS__CGSCC_NPM: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test1-check
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0]] {
; IS__CGSCC_NPM-NEXT: [[RES:%.*]] = tail call i32 @test1(i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]]) #[[ATTR2]], !range [[RNG2:![0-9]+]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[CMP:%.*]] = icmp eq i32 [[RES]], 500
; IS__CGSCC_NPM-NEXT: ret i1 [[CMP]]
[Attributor][FIX] Make check lines explicit There is a bug in `update_test_checks.py` that combines check lines it should not. For now we unbreak the bots by making all possibilities explicit.
2020-02-10 15:31:20 +08:00
;
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
%res = tail call i32 @test1(i32* %p)
%cmp = icmp eq i32 %res, 500
ret i1 %cmp
}
; TEST2
; int test2(int *p) { return *p == 0 ? 4 : 3; }
; int test2_check(int *p) {
; int call = test2(p);
; if (call == 5) {
; // dead block
; return 2;
; } else {
; return 3;
; }
; }
define i32 @test2(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: argmemonly nofree nosync nounwind readonly willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test2
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT____-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0]] {
[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____-NEXT: entry:
; IS__TUNIT____-NEXT: [[TMP0:%.*]] = load i32, i32* [[P]], align 4
; IS__TUNIT____-NEXT: [[TOBOOL:%.*]] = icmp eq i32 [[TMP0]], 0
; IS__TUNIT____-NEXT: [[COND:%.*]] = select i1 [[TOBOOL]], i32 4, i32 3
; IS__TUNIT____-NEXT: ret i32 [[COND]]
;
; IS__CGSCC____: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test2
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC____-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR0]] {
[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____-NEXT: entry:
; IS__CGSCC____-NEXT: [[TMP0:%.*]] = load i32, i32* [[P]], align 4
; IS__CGSCC____-NEXT: [[TOBOOL:%.*]] = icmp eq i32 [[TMP0]], 0
; IS__CGSCC____-NEXT: [[COND:%.*]] = select i1 [[TOBOOL]], i32 4, i32 3
; IS__CGSCC____-NEXT: ret i32 [[COND]]
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
entry:
%0 = load i32, i32* %p, align 4
%tobool = icmp eq i32 %0, 0
%cond = select i1 %tobool, i32 4, i32 3
ret i32 %cond
}
define i32 @test2_check(i32* %p) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree 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__TUNIT____-LABEL: define {{[^@]+}}@test2_check
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____-SAME: (i32* nocapture nofree readnone align 4 [[P:%.*]]) #[[ATTR1:[0-9]+]] {
[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: entry:
; IS__TUNIT____-NEXT: br label [[IF_THEN:%.*]]
; IS__TUNIT____: if.then:
; IS__TUNIT____-NEXT: br label [[RETURN:%.*]]
; IS__TUNIT____: if.end:
; IS__TUNIT____-NEXT: unreachable
; IS__TUNIT____: return:
; IS__TUNIT____-NEXT: ret i32 2
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC____-LABEL: define {{[^@]+}}@test2_check
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____-SAME: (i32* nocapture nofree noundef nonnull readnone align 4 dereferenceable(4) [[P:%.*]]) #[[ATTR1:[0-9]+]] {
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC____-NEXT: entry:
; IS__CGSCC____-NEXT: br label [[IF_THEN:%.*]]
; IS__CGSCC____: if.then:
; IS__CGSCC____-NEXT: br label [[RETURN:%.*]]
; IS__CGSCC____: if.end:
; IS__CGSCC____-NEXT: unreachable
; IS__CGSCC____: return:
; IS__CGSCC____-NEXT: ret i32 2
[Attributor][FIX] Make check lines explicit There is a bug in `update_test_checks.py` that combines check lines it should not. For now we unbreak the bots by making all possibilities explicit.
2020-02-10 15:31:20 +08:00
;
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
entry:
%call = tail call i32 @test2(i32* %p)
%cmp = icmp slt i32 %call, 5
br i1 %cmp, label %if.then, label %if.end
if.then: ; preds = %entry
br label %return
if.end: ; preds = %entry
br label %return
return: ; preds = %if.end, %if.then
%retval.0 = phi i32 [ 2, %if.then ], [ 3, %if.end ]
ret i32 %retval.0
}
; TEST 3 SECV test
; void unkown();
; int r1(unsigned int u){
; int sum = 0;
; for(int i = 0; i<100;i++){
; sum += i;
; }
; // sum = 50 * 49 / 2
; if(sum > 10000){
; // dead block
; return 20;
; }else {
; return 10;
; }
; }
; void f1(int u){
; if(r1(u) > 15){
; // deadblock
; unkown();
; }else {
; return;
; }
; }
declare dso_local void @unkown()
define internal i32 @r1(i32) local_unnamed_addr {
[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_OPM: Function Attrs: nofree nosync nounwind readnone
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@r1
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-SAME: () local_unnamed_addr #[[ATTR2:[0-9]+]] {
[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_OPM-NEXT: br label [[TMP4:%.*]]
; IS__TUNIT_OPM: 1:
; IS__TUNIT_OPM-NEXT: [[TMP2:%.*]] = icmp sgt i32 [[TMP7:%.*]], 10000
; IS__TUNIT_OPM-NEXT: br i1 [[TMP2]], label [[TMP3:%.*]], label [[F:%.*]]
; IS__TUNIT_OPM: 3:
; IS__TUNIT_OPM-NEXT: ret i32 20
; IS__TUNIT_OPM: f:
; IS__TUNIT_OPM-NEXT: ret i32 10
; IS__TUNIT_OPM: 4:
; IS__TUNIT_OPM-NEXT: [[TMP5:%.*]] = phi i32 [ 0, [[TMP0:%.*]] ], [ [[TMP8:%.*]], [[TMP4]] ]
; IS__TUNIT_OPM-NEXT: [[TMP6:%.*]] = phi i32 [ 0, [[TMP0]] ], [ [[TMP7]], [[TMP4]] ]
; IS__TUNIT_OPM-NEXT: [[TMP7]] = add nuw nsw i32 [[TMP5]], [[TMP6]]
; IS__TUNIT_OPM-NEXT: [[TMP8]] = add nuw nsw i32 [[TMP5]], 1
; IS__TUNIT_OPM-NEXT: [[TMP9:%.*]] = icmp eq i32 [[TMP8]], 100
; IS__TUNIT_OPM-NEXT: br i1 [[TMP9]], label [[TMP1:%.*]], label [[TMP4]]
[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][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: nofree norecurse nosync nounwind readnone
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@r1
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM-SAME: () local_unnamed_addr #[[ATTR2:[0-9]+]] {
[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-NEXT: br label [[TMP4:%.*]]
; IS__CGSCC_OPM: 1:
; IS__CGSCC_OPM-NEXT: [[TMP2:%.*]] = icmp sgt i32 [[TMP7:%.*]], 10000
; IS__CGSCC_OPM-NEXT: br i1 [[TMP2]], label [[TMP3:%.*]], label [[F:%.*]]
; IS__CGSCC_OPM: 3:
; IS__CGSCC_OPM-NEXT: ret i32 20
; IS__CGSCC_OPM: f:
; IS__CGSCC_OPM-NEXT: ret i32 10
; IS__CGSCC_OPM: 4:
; IS__CGSCC_OPM-NEXT: [[TMP5:%.*]] = phi i32 [ 0, [[TMP0:%.*]] ], [ [[TMP8:%.*]], [[TMP4]] ]
; IS__CGSCC_OPM-NEXT: [[TMP6:%.*]] = phi i32 [ 0, [[TMP0]] ], [ [[TMP7]], [[TMP4]] ]
; IS__CGSCC_OPM-NEXT: [[TMP7]] = add nuw nsw i32 [[TMP5]], [[TMP6]]
; IS__CGSCC_OPM-NEXT: [[TMP8]] = add nuw nsw i32 [[TMP5]], 1
; IS__CGSCC_OPM-NEXT: [[TMP9:%.*]] = icmp eq i32 [[TMP8]], 100
; IS__CGSCC_OPM-NEXT: br i1 [[TMP9]], label [[TMP1:%.*]], label [[TMP4]]
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@r1
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_NPM-SAME: () local_unnamed_addr #[[ATTR1]] {
[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_NPM-NEXT: br label [[TMP3:%.*]]
; IS__CGSCC_NPM: 1:
; IS__CGSCC_NPM-NEXT: br label [[F:%.*]]
; IS__CGSCC_NPM: 2:
; IS__CGSCC_NPM-NEXT: unreachable
; IS__CGSCC_NPM: f:
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; IS__CGSCC_NPM-NEXT: ret i32 undef
[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_NPM: 3:
; IS__CGSCC_NPM-NEXT: [[TMP4:%.*]] = phi i32 [ 0, [[TMP0:%.*]] ], [ [[TMP7:%.*]], [[TMP3]] ]
; IS__CGSCC_NPM-NEXT: [[TMP5:%.*]] = phi i32 [ 0, [[TMP0]] ], [ [[TMP6:%.*]], [[TMP3]] ]
; IS__CGSCC_NPM-NEXT: [[TMP6]] = add nuw nsw i32 [[TMP4]], [[TMP5]]
; IS__CGSCC_NPM-NEXT: [[TMP7]] = add nuw nsw i32 [[TMP4]], 1
; IS__CGSCC_NPM-NEXT: [[TMP8:%.*]] = icmp eq i32 [[TMP7]], 100
; IS__CGSCC_NPM-NEXT: br i1 [[TMP8]], label [[TMP1:%.*]], label [[TMP3]]
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
br label %5
2: ; preds = %5
%3 = icmp sgt i32 %8, 10000
br i1 %3, label %4, label %f
4:
ret i32 20
f:
ret i32 10
5: ; preds = %5, %1
%6 = phi i32 [ 0, %1 ], [ %9, %5 ]
%7 = phi i32 [ 0, %1 ], [ %8, %5 ]
%8 = add nuw nsw i32 %6, %7
%9 = add nuw nsw i32 %6, 1
%10 = icmp eq i32 %9, 100
br i1 %10, label %2, label %5
}
define void @f1(i32){
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@f1
; IS__TUNIT_OPM-SAME: (i32 [[TMP0:%.*]]) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-NEXT: [[TMP2:%.*]] = tail call i32 @r1() #[[ATTR2]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__TUNIT_OPM-NEXT: [[TMP3:%.*]] = icmp sgt i32 [[TMP2]], 15
; IS__TUNIT_OPM-NEXT: br i1 [[TMP3]], label [[TMP4:%.*]], label [[TMP5:%.*]]
; IS__TUNIT_OPM: 4:
; IS__TUNIT_OPM-NEXT: tail call void @unkown()
; IS__TUNIT_OPM-NEXT: br label [[TMP5]]
; IS__TUNIT_OPM: 5:
; IS__TUNIT_OPM-NEXT: ret void
[Attributor] Unify testing (=updates,prefixes,run configurations,...) When the Attributor was created the test update scripts were not well suited to deal with the challenges of IR attribute checking. This partially improved. Since then we also added three additional configurations that need testing; in total we now have the following four: { TUNIT, CGSCC } x { old pass manager (OPM), new pass manager (NPM) } Finally, the number of developers and tests grew rapidly (partially due to the addition of ArgumentPromotion and IPConstantProp tests), which resulted in tests only being run in some configurations, different prefixes being used, and different "styles" of checks being used. Due to the above reasons I believed we needed to take another look at the test update scripts. While we started to use them, via UTC_ARGS: --enable/disable, the other problems remained. To improve the testing situation for *all* configurations, to simplify future updates to the test, and to help identify subtle effects of future changes, we now use the test update scripts for (almost) all Attributor tests. An exhaustive prefix list minimizes the number of check lines and makes it easy to identify and compare configurations. Tests have been adjusted in the process but we tried to keep their intend unchanged. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D76588
2020-04-09 08:04:57 +08:00
;
[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_NPM: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@f1
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_NPM-SAME: (i32 [[TMP0:%.*]]) #[[ATTR1]] {
[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_NPM-NEXT: br label [[TMP3:%.*]]
; IS__TUNIT_NPM: 2:
; IS__TUNIT_NPM-NEXT: unreachable
; IS__TUNIT_NPM: 3:
; IS__TUNIT_NPM-NEXT: ret void
;
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@f1
; IS__CGSCC_OPM-SAME: (i32 [[TMP0:%.*]]) {
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-NEXT: [[TMP2:%.*]] = tail call i32 @r1() #[[ATTR4:[0-9]+]], !range [[RNG3:![0-9]+]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[TMP3:%.*]] = icmp sgt i32 [[TMP2]], 15
; IS__CGSCC_OPM-NEXT: br i1 [[TMP3]], label [[TMP4:%.*]], label [[TMP5:%.*]]
; IS__CGSCC_OPM: 4:
; IS__CGSCC_OPM-NEXT: tail call void @unkown()
; IS__CGSCC_OPM-NEXT: br label [[TMP5]]
; IS__CGSCC_OPM: 5:
; IS__CGSCC_OPM-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_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@f1
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i32 [[TMP0:%.*]]) #[[ATTR1]] {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_NPM-NEXT: br label [[TMP3:%.*]]
; IS__CGSCC_NPM: 2:
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
; IS__CGSCC_NPM-NEXT: unreachable
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_NPM: 3:
[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_NPM-NEXT: ret void
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
%2 = tail call i32 @r1(i32 %0)
%3 = icmp sgt i32 %2, 15
br i1 %3, label %4, label %5
4: ; preds = %1
tail call void @unkown()
br label %5
5: ; preds = %1, %4
ret void
}
; TEST4 LVI test
; f1
; int test4-f1(int u){
; if(u>=0) {
; return u;
; }else{
; return 0;
; }
; }
define dso_local i32 @test4-f1(i32 %u) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test4-f1
; IS__TUNIT____-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: entry:
; IS__TUNIT____-NEXT: [[CMP:%.*]] = icmp sgt i32 [[U]], -1
; IS__TUNIT____-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[RETURN:%.*]]
; IS__TUNIT____: if.then:
; IS__TUNIT____-NEXT: br label [[RETURN]]
; IS__TUNIT____: return:
; IS__TUNIT____-NEXT: [[RETVAL_0:%.*]] = phi i32 [ [[U]], [[IF_THEN]] ], [ 0, [[ENTRY:%.*]] ]
; IS__TUNIT____-NEXT: ret i32 [[RETVAL_0]]
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test4-f1
; IS__CGSCC____-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: entry:
; IS__CGSCC____-NEXT: [[CMP:%.*]] = icmp sgt i32 [[U]], -1
; IS__CGSCC____-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[RETURN:%.*]]
; IS__CGSCC____: if.then:
; IS__CGSCC____-NEXT: br label [[RETURN]]
; IS__CGSCC____: return:
; IS__CGSCC____-NEXT: [[RETVAL_0:%.*]] = phi i32 [ [[U]], [[IF_THEN]] ], [ 0, [[ENTRY:%.*]] ]
; IS__CGSCC____-NEXT: ret i32 [[RETVAL_0]]
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
; FIXME: RETVAL_0 >= 0
entry:
%cmp = icmp sgt i32 %u, -1
br i1 %cmp, label %if.then, label %return
if.then: ; preds = %entry
br label %return
return: ; preds = %entry, %if.then
%retval.0 = phi i32 [ %u, %if.then ], [ 0, %entry ]
ret i32 %retval.0
}
define dso_local i32 @test4-g1(i32 %u) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test4-g1
; IS__TUNIT____-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: entry:
; IS__TUNIT____-NEXT: [[CALL:%.*]] = tail call i32 @test4-f1(i32 [[U]]) #[[ATTR1]]
; IS__TUNIT____-NEXT: ret i32 [[CALL]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
;
; IS__CGSCC_OPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@test4-g1
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: entry:
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-NEXT: [[CALL:%.*]] = tail call i32 @test4-f1(i32 [[U]]) #[[ATTR5:[0-9]+]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: ret i32 [[CALL]]
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test4-g1
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: entry:
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-NEXT: [[CALL:%.*]] = tail call i32 @test4-f1(i32 [[U]]) #[[ATTR3:[0-9]+]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: ret i32 [[CALL]]
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
; FIXME: %call should have range [0, inf]
entry:
%call = tail call i32 @test4-f1(i32 %u)
ret i32 %call
}
; f2
; int test4-f1(int u){
; if(u>-1) {
; return u+1;
; }else{
; return 1;
; }
; }
define dso_local i32 @test4-f2(i32 %u) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@test4-f2
; IS__TUNIT____-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: entry:
; IS__TUNIT____-NEXT: [[CMP:%.*]] = icmp sgt i32 [[U]], -1
; IS__TUNIT____-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; IS__TUNIT____: if.then:
; IS__TUNIT____-NEXT: [[ADD:%.*]] = add nuw nsw i32 [[U]], 1
; IS__TUNIT____-NEXT: br label [[RETURN:%.*]]
; IS__TUNIT____: if.else:
; IS__TUNIT____-NEXT: br label [[RETURN]]
; IS__TUNIT____: return:
; IS__TUNIT____-NEXT: [[RETVAL_0:%.*]] = phi i32 [ [[ADD]], [[IF_THEN]] ], [ 1, [[IF_ELSE]] ]
; IS__TUNIT____-NEXT: ret i32 [[RETVAL_0]]
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@test4-f2
; IS__CGSCC____-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: entry:
; IS__CGSCC____-NEXT: [[CMP:%.*]] = icmp sgt i32 [[U]], -1
; IS__CGSCC____-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; IS__CGSCC____: if.then:
; IS__CGSCC____-NEXT: [[ADD:%.*]] = add nuw nsw i32 [[U]], 1
; IS__CGSCC____-NEXT: br label [[RETURN:%.*]]
; IS__CGSCC____: if.else:
; IS__CGSCC____-NEXT: br label [[RETURN]]
; IS__CGSCC____: return:
; IS__CGSCC____-NEXT: [[RETVAL_0:%.*]] = phi i32 [ [[ADD]], [[IF_THEN]] ], [ 1, [[IF_ELSE]] ]
; IS__CGSCC____-NEXT: ret i32 [[RETVAL_0]]
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
entry:
%cmp = icmp sgt i32 %u, -1
br i1 %cmp, label %if.then, label %if.else
if.then: ; preds = %entry
%add = add nuw nsw i32 %u, 1
br label %return
if.else: ; preds = %entry
br label %return
return: ; preds = %if.else, %if.then
%retval.0 = phi i32 [ %add, %if.then ], [ 1, %if.else ]
ret i32 %retval.0
}
define dso_local i32 @test4-g2(i32 %u) {
[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_OPM: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@test4-g2
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
[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_OPM-NEXT: entry:
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-NEXT: [[CALL:%.*]] = tail call i32 @test4-f2(i32 [[U]]) #[[ATTR1]]
[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_OPM-NEXT: ret i32 [[CALL]]
[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][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_NPM: Function Attrs: nofree 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__TUNIT_NPM-LABEL: define {{[^@]+}}@test4-g2
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
[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_NPM-NEXT: entry:
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-NEXT: [[CALL:%.*]] = tail call i32 @test4-f2(i32 [[U]]) #[[ATTR1]], !range [[RNG3:![0-9]+]]
[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_NPM-NEXT: ret i32 [[CALL]]
[Attributor][FIX] Make check lines explicit There is a bug in `update_test_checks.py` that combines check lines it should not. For now we unbreak the bots by making all possibilities explicit.
2020-02-10 15:31:20 +08:00
;
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@test4-g2
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: entry:
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-NEXT: [[CALL:%.*]] = tail call i32 @test4-f2(i32 [[U]]) #[[ATTR5]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: ret i32 [[CALL]]
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@test4-g2
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i32 [[U:%.*]]) #[[ATTR1]] {
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: entry:
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-NEXT: [[CALL:%.*]] = tail call i32 @test4-f2(i32 [[U]]) #[[ATTR3]], !range [[RNG3:![0-9]+]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: ret i32 [[CALL]]
[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
;
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
entry:
%call = tail call i32 @test4-f2(i32 %u)
ret i32 %call
}
define dso_local i32 @test-5() {
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@test-5() {
[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_OPM-NEXT: entry:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT_OPM-NEXT: [[CALL:%.*]] = call noundef i32 @rec(i32 noundef 0), !range [[RNG3:![0-9]+]]
[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_OPM-NEXT: ret i32 [[CALL]]
[Attributor][FIX] Call right base method in AAValueConstantRangeFloating We now call the base class method as we should.
2020-02-10 09:05:15 +08:00
;
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; NOT_TUNIT_OPM-LABEL: define {{[^@]+}}@test-5() {
; NOT_TUNIT_OPM-NEXT: entry:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; NOT_TUNIT_OPM-NEXT: [[CALL:%.*]] = call noundef i32 @rec(i32 noundef 0), !range [[RNG4:![0-9]+]]
[Attributor] Introduce module slice. Summary: The module slice describes which functions we can analyze and transform while working on an SCC as part of the Attributor-CGSCC pass. So far we simply restricted it to the SCC. Reviewers: jdoerfert Differential Revision: https://reviews.llvm.org/D86319
2020-08-30 16:29:07 +08:00
; NOT_TUNIT_OPM-NEXT: ret i32 [[CALL]]
[Attributor][Tests] Run the CGSCC versions on the range.ll test
2020-02-10 09:08:19 +08:00
;
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
entry:
%call = call i32 @rec(i32 0)
ret i32 %call
}
define internal i32 @rec(i32 %depth) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@rec
; IS__TUNIT_OPM-SAME: (i32 [[DEPTH:%.*]]) {
; IS__TUNIT_OPM-NEXT: entry:
; IS__TUNIT_OPM-NEXT: [[CALL:%.*]] = call i32 @foo(i32 [[DEPTH]])
; IS__TUNIT_OPM-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[CALL]], 0
; IS__TUNIT_OPM-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
; IS__TUNIT_OPM: if.then:
; IS__TUNIT_OPM-NEXT: br label [[RETURN:%.*]]
; IS__TUNIT_OPM: if.end:
; IS__TUNIT_OPM-NEXT: [[CMP:%.*]] = icmp slt i32 [[DEPTH]], 10
; IS__TUNIT_OPM-NEXT: br i1 [[CMP]], label [[IF_THEN1:%.*]], label [[IF_END3:%.*]]
; IS__TUNIT_OPM: if.then1:
; IS__TUNIT_OPM-NEXT: [[ADD:%.*]] = add nsw i32 [[DEPTH]], 1
; IS__TUNIT_OPM-NEXT: [[CALL2:%.*]] = call i32 @rec(i32 [[ADD]]), !range [[RNG3]]
; IS__TUNIT_OPM-NEXT: br label [[IF_END3]]
; IS__TUNIT_OPM: if.end3:
; IS__TUNIT_OPM-NEXT: br label [[RETURN]]
; IS__TUNIT_OPM: return:
; IS__TUNIT_OPM-NEXT: [[RETVAL_0:%.*]] = phi i32 [ 0, [[IF_THEN]] ], [ 1, [[IF_END3]] ]
; IS__TUNIT_OPM-NEXT: ret i32 [[RETVAL_0]]
;
; NOT_TUNIT_OPM-LABEL: define {{[^@]+}}@rec
; NOT_TUNIT_OPM-SAME: (i32 [[DEPTH:%.*]]) {
; NOT_TUNIT_OPM-NEXT: entry:
; NOT_TUNIT_OPM-NEXT: [[CALL:%.*]] = call i32 @foo(i32 [[DEPTH]])
; NOT_TUNIT_OPM-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[CALL]], 0
; NOT_TUNIT_OPM-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
; NOT_TUNIT_OPM: if.then:
; NOT_TUNIT_OPM-NEXT: br label [[RETURN:%.*]]
; NOT_TUNIT_OPM: if.end:
; NOT_TUNIT_OPM-NEXT: [[CMP:%.*]] = icmp slt i32 [[DEPTH]], 10
; NOT_TUNIT_OPM-NEXT: br i1 [[CMP]], label [[IF_THEN1:%.*]], label [[IF_END3:%.*]]
; NOT_TUNIT_OPM: if.then1:
; NOT_TUNIT_OPM-NEXT: [[ADD:%.*]] = add nsw i32 [[DEPTH]], 1
; NOT_TUNIT_OPM-NEXT: [[CALL2:%.*]] = call i32 @rec(i32 [[ADD]]), !range [[RNG4]]
; NOT_TUNIT_OPM-NEXT: br label [[IF_END3]]
; NOT_TUNIT_OPM: if.end3:
; NOT_TUNIT_OPM-NEXT: br label [[RETURN]]
; NOT_TUNIT_OPM: return:
; NOT_TUNIT_OPM-NEXT: [[RETVAL_0:%.*]] = phi i32 [ 0, [[IF_THEN]] ], [ 1, [[IF_END3]] ]
; NOT_TUNIT_OPM-NEXT: ret i32 [[RETVAL_0]]
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
;
entry:
%call = call i32 @foo(i32 %depth)
%tobool = icmp ne i32 %call, 0
br i1 %tobool, label %if.then, label %if.end
if.then: ; preds = %entry
br label %return
if.end: ; preds = %entry
%cmp = icmp slt i32 %depth, 10
br i1 %cmp, label %if.then1, label %if.end3
if.then1: ; preds = %if.end
%add = add nsw i32 %depth, 1
%call2 = call i32 @rec(i32 %add)
br label %if.end3
if.end3: ; preds = %if.then1, %if.end
br label %return
return: ; preds = %if.end3, %if.then
%retval.0 = phi i32 [ 0, %if.then ], [ 1, %if.end3 ]
ret i32 %retval.0
}
declare dso_local i32 @foo(i32)
[Attributor][Tests][NFC] Add more range tests Inspired by https://llvm.discourse.group/t/impossible-condition-optimization/461
2020-02-10 09:06:09 +08:00
; Examples taken from https://llvm.discourse.group/t/impossible-condition-optimization/461/1
;
; The important part is that we return a constant (false)
;
; {
; FIXME: All but the return is not needed anymore
define dso_local zeroext i1 @phi(i32 %arg) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@phi
; IS__TUNIT____-SAME: (i32 [[ARG:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: bb:
; IS__TUNIT____-NEXT: [[TMP:%.*]] = icmp sgt i32 [[ARG]], 5
; IS__TUNIT____-NEXT: br i1 [[TMP]], label [[BB1:%.*]], label [[BB2:%.*]]
; IS__TUNIT____: bb1:
; IS__TUNIT____-NEXT: br label [[BB3:%.*]]
; IS__TUNIT____: bb2:
; IS__TUNIT____-NEXT: br label [[BB3]]
; IS__TUNIT____: bb3:
; IS__TUNIT____-NEXT: [[TMP4:%.*]] = icmp sgt i32 [[ARG]], 10
; IS__TUNIT____-NEXT: br i1 [[TMP4]], label [[BB5:%.*]], label [[BB7:%.*]]
; IS__TUNIT____: bb5:
; IS__TUNIT____-NEXT: br label [[BB9:%.*]]
; IS__TUNIT____: bb7:
; IS__TUNIT____-NEXT: br label [[BB9]]
; IS__TUNIT____: bb9:
; IS__TUNIT____-NEXT: br label [[BB12:%.*]]
; IS__TUNIT____: bb11:
; IS__TUNIT____-NEXT: unreachable
; IS__TUNIT____: bb12:
; IS__TUNIT____-NEXT: br label [[BB13:%.*]]
; IS__TUNIT____: bb13:
; IS__TUNIT____-NEXT: ret i1 false
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@phi
; IS__CGSCC____-SAME: (i32 [[ARG:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: bb:
; IS__CGSCC____-NEXT: [[TMP:%.*]] = icmp sgt i32 [[ARG]], 5
; IS__CGSCC____-NEXT: br i1 [[TMP]], label [[BB1:%.*]], label [[BB2:%.*]]
; IS__CGSCC____: bb1:
; IS__CGSCC____-NEXT: br label [[BB3:%.*]]
; IS__CGSCC____: bb2:
; IS__CGSCC____-NEXT: br label [[BB3]]
; IS__CGSCC____: bb3:
; IS__CGSCC____-NEXT: [[TMP4:%.*]] = icmp sgt i32 [[ARG]], 10
; IS__CGSCC____-NEXT: br i1 [[TMP4]], label [[BB5:%.*]], label [[BB7:%.*]]
; IS__CGSCC____: bb5:
; IS__CGSCC____-NEXT: br label [[BB9:%.*]]
; IS__CGSCC____: bb7:
; IS__CGSCC____-NEXT: br label [[BB9]]
; IS__CGSCC____: bb9:
; IS__CGSCC____-NEXT: br label [[BB12:%.*]]
; IS__CGSCC____: bb11:
; IS__CGSCC____-NEXT: unreachable
; IS__CGSCC____: bb12:
; IS__CGSCC____-NEXT: br label [[BB13:%.*]]
; IS__CGSCC____: bb13:
; IS__CGSCC____-NEXT: ret i1 false
[Attributor][Tests][NFC] Add more range tests Inspired by https://llvm.discourse.group/t/impossible-condition-optimization/461
2020-02-10 09:06:09 +08:00
;
bb:
%tmp = icmp sgt i32 %arg, 5
br i1 %tmp, label %bb1, label %bb2
bb1: ; preds = %bb
br label %bb3
bb2: ; preds = %bb
br label %bb3
bb3: ; preds = %bb2, %bb1
%.02 = phi i32 [ 1, %bb1 ], [ 2, %bb2 ]
%tmp4 = icmp sgt i32 %arg, 10
br i1 %tmp4, label %bb5, label %bb7
bb5: ; preds = %bb3
%tmp6 = add nsw i32 %.02, 1
br label %bb9
bb7: ; preds = %bb3
%tmp8 = add nsw i32 %.02, 2
br label %bb9
bb9: ; preds = %bb7, %bb5
%.01 = phi i32 [ %tmp6, %bb5 ], [ %tmp8, %bb7 ]
%tmp10 = icmp eq i32 %.01, 5
br i1 %tmp10, label %bb11, label %bb12
bb11: ; preds = %bb9
br label %bb13
bb12: ; preds = %bb9
br label %bb13
bb13: ; preds = %bb12, %bb11
%.0 = phi i1 [ true, %bb11 ], [ false, %bb12 ]
ret i1 %.0
}
define dso_local i1 @select(i32 %a) local_unnamed_addr #0 {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@select
; IS__TUNIT____-SAME: (i32 [[A:%.*]]) local_unnamed_addr #[[ATTR1]] {
; IS__TUNIT____-NEXT: entry:
; IS__TUNIT____-NEXT: ret i1 false
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@select
; IS__CGSCC____-SAME: (i32 [[A:%.*]]) local_unnamed_addr #[[ATTR1]] {
; IS__CGSCC____-NEXT: entry:
; IS__CGSCC____-NEXT: ret i1 false
[Attributor][Tests][NFC] Add more range tests Inspired by https://llvm.discourse.group/t/impossible-condition-optimization/461
2020-02-10 09:06:09 +08:00
;
entry:
%cmp = icmp sgt i32 %a, 5
%. = select i1 %cmp, i32 1, i32 2
%cmp1 = icmp sgt i32 %a, 10
%y.0.v = select i1 %cmp1, i32 1, i32 2
%y.0 = add nuw nsw i32 %., %y.0.v
%cmp6 = icmp eq i32 %y.0, 5
ret i1 %cmp6
}
define dso_local i32 @select_zext(i32 %a) local_unnamed_addr #0 {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@select_zext
; IS__TUNIT____-SAME: (i32 [[A:%.*]]) local_unnamed_addr #[[ATTR1]] {
; IS__TUNIT____-NEXT: entry:
; IS__TUNIT____-NEXT: ret i32 0
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@select_zext
; IS__CGSCC____-SAME: (i32 [[A:%.*]]) local_unnamed_addr #[[ATTR1]] {
; IS__CGSCC____-NEXT: entry:
; IS__CGSCC____-NEXT: ret i32 0
[Attributor] Look through (some) casts in AAValueConstantRangeFloating Casts can be handled natively by the ConstantRange class. We do limit it to extends for now as we assume an integer type in different locations. A TODO and a test case with a FIXME was added to remove that restriction in the future.
2020-02-10 09:07:30 +08:00
;
entry:
%cmp = icmp sgt i32 %a, 5
%. = select i1 %cmp, i32 1, i32 2
%cmp1 = icmp sgt i32 %a, 10
%y.0.v = select i1 %cmp1, i32 1, i32 2
%y.0 = add nuw nsw i32 %., %y.0.v
%cmp6 = icmp eq i32 %y.0, 5
%.13 = zext i1 %cmp6 to i32
ret i32 %.13
}
; FIXME: We do not look through the ptr casts here.
define dso_local i64 @select_int2ptr_bitcast_ptr2int(i32 %a) local_unnamed_addr #0 {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@select_int2ptr_bitcast_ptr2int
; IS__TUNIT____-SAME: (i32 [[A:%.*]]) local_unnamed_addr #[[ATTR1]] {
; IS__TUNIT____-NEXT: entry:
; IS__TUNIT____-NEXT: [[I2P:%.*]] = inttoptr i1 false to i1*
; IS__TUNIT____-NEXT: [[BC:%.*]] = bitcast i1* [[I2P]] to i32*
; IS__TUNIT____-NEXT: [[P2I:%.*]] = ptrtoint i32* [[BC]] to i64
; IS__TUNIT____-NEXT: ret i64 [[P2I]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@select_int2ptr_bitcast_ptr2int
; IS__CGSCC____-SAME: (i32 [[A:%.*]]) local_unnamed_addr #[[ATTR1]] {
; IS__CGSCC____-NEXT: entry:
; IS__CGSCC____-NEXT: [[I2P:%.*]] = inttoptr i1 false to i1*
; IS__CGSCC____-NEXT: [[BC:%.*]] = bitcast i1* [[I2P]] to i32*
; IS__CGSCC____-NEXT: [[P2I:%.*]] = ptrtoint i32* [[BC]] to i64
; IS__CGSCC____-NEXT: ret i64 [[P2I]]
[Attributor][Tests][NFC] Add more range tests Inspired by https://llvm.discourse.group/t/impossible-condition-optimization/461
2020-02-10 09:06:09 +08:00
;
entry:
%cmp = icmp sgt i32 %a, 5
%. = select i1 %cmp, i32 1, i32 2
%cmp1 = icmp sgt i32 %a, 10
%y.0.v = select i1 %cmp1, i32 1, i32 2
%y.0 = add nuw nsw i32 %., %y.0.v
%cmp6 = icmp eq i32 %y.0, 5
[Attributor] Look through (some) casts in AAValueConstantRangeFloating Casts can be handled natively by the ConstantRange class. We do limit it to extends for now as we assume an integer type in different locations. A TODO and a test case with a FIXME was added to remove that restriction in the future.
2020-02-10 09:07:30 +08:00
%i2p = inttoptr i1 %cmp6 to i1*
%bc = bitcast i1* %i2p to i32*
%p2i = ptrtoint i32* %bc to i64
ret i64 %p2i
[Attributor][Tests][NFC] Add more range tests Inspired by https://llvm.discourse.group/t/impossible-condition-optimization/461
2020-02-10 09:06:09 +08:00
}
; }
[Attributor][FIX] Validate the type for AAValueConstantRange as needed Due to the genericValueTraversal we might visit values for which we did not create an AAValueConstantRange object, e.g., as they are behind a PHI or select or call with `returned` argument. As a consequence we need to validate the types as we are about to query AAValueConstantRange for operands.
2020-02-14 04:22:26 +08:00
define i1 @f_fcmp(float %a, float %b) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@f_fcmp
; IS__TUNIT____-SAME: (float [[A:%.*]], float [[B:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: [[R:%.*]] = fcmp uge float [[A]], [[B]]
; IS__TUNIT____-NEXT: [[S:%.*]] = select i1 [[R]], i1 [[R]], i1 false
; IS__TUNIT____-NEXT: ret i1 [[S]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@f_fcmp
; IS__CGSCC____-SAME: (float [[A:%.*]], float [[B:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[R:%.*]] = fcmp uge float [[A]], [[B]]
; IS__CGSCC____-NEXT: [[S:%.*]] = select i1 [[R]], i1 [[R]], i1 false
; IS__CGSCC____-NEXT: ret i1 [[S]]
[Attributor][NFC] Add check lines for tests
2020-02-15 09:27:41 +08:00
;
[Attributor][FIX] Validate the type for AAValueConstantRange as needed Due to the genericValueTraversal we might visit values for which we did not create an AAValueConstantRange object, e.g., as they are behind a PHI or select or call with `returned` argument. As a consequence we need to validate the types as we are about to query AAValueConstantRange for operands.
2020-02-14 04:22:26 +08:00
%r = fcmp uge float %a, %b
%s = select i1 %r, i1 %r, i1 0
ret i1 %s
}
define i1 @d_fcmp(double %a, double %b) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@d_fcmp
; IS__TUNIT____-SAME: (double [[A:%.*]], double [[B:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: [[R:%.*]] = fcmp oeq double [[A]], [[B]]
; IS__TUNIT____-NEXT: [[S:%.*]] = select i1 [[R]], i1 [[R]], i1 false
; IS__TUNIT____-NEXT: ret i1 [[S]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@d_fcmp
; IS__CGSCC____-SAME: (double [[A:%.*]], double [[B:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[R:%.*]] = fcmp oeq double [[A]], [[B]]
; IS__CGSCC____-NEXT: [[S:%.*]] = select i1 [[R]], i1 [[R]], i1 false
; IS__CGSCC____-NEXT: ret i1 [[S]]
[Attributor][NFC] Add check lines for tests
2020-02-15 09:27:41 +08:00
;
[Attributor][FIX] Validate the type for AAValueConstantRange as needed Due to the genericValueTraversal we might visit values for which we did not create an AAValueConstantRange object, e.g., as they are behind a PHI or select or call with `returned` argument. As a consequence we need to validate the types as we are about to query AAValueConstantRange for operands.
2020-02-14 04:22:26 +08:00
%r = fcmp oeq double %a, %b
%s = select i1 %r, i1 %r, i1 0
ret i1 %s
}
define i1 @dp_icmp(double* %a, double* %b) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@dp_icmp
; IS__TUNIT____-SAME: (double* nofree readnone [[A:%.*]], double* nofree readnone [[B:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: [[R:%.*]] = icmp sge double* [[A]], [[B]]
; IS__TUNIT____-NEXT: [[S:%.*]] = select i1 [[R]], i1 [[R]], i1 false
; IS__TUNIT____-NEXT: ret i1 [[S]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@dp_icmp
; IS__CGSCC____-SAME: (double* nofree readnone [[A:%.*]], double* nofree readnone [[B:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[R:%.*]] = icmp sge double* [[A]], [[B]]
; IS__CGSCC____-NEXT: [[S:%.*]] = select i1 [[R]], i1 [[R]], i1 false
; IS__CGSCC____-NEXT: ret i1 [[S]]
[Attributor][NFC] Add check lines for tests
2020-02-15 09:27:41 +08:00
;
[Attributor][FIX] Validate the type for AAValueConstantRange as needed Due to the genericValueTraversal we might visit values for which we did not create an AAValueConstantRange object, e.g., as they are behind a PHI or select or call with `returned` argument. As a consequence we need to validate the types as we are about to query AAValueConstantRange for operands.
2020-02-14 04:22:26 +08:00
%r = icmp sge double* %a, %b
%s = select i1 %r, i1 %r, i1 0
ret i1 %s
}
define i1 @ip_icmp(i8* %a, i8* %b) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@ip_icmp
; IS__TUNIT____-SAME: (i8* nofree readnone [[A:%.*]], i8* nofree readnone [[B:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: [[R:%.*]] = icmp ult i8* [[A]], [[B]]
; IS__TUNIT____-NEXT: [[S:%.*]] = select i1 [[R]], i1 [[R]], i1 false
; IS__TUNIT____-NEXT: ret i1 [[S]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@ip_icmp
; IS__CGSCC____-SAME: (i8* nofree readnone [[A:%.*]], i8* nofree readnone [[B:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[R:%.*]] = icmp ult i8* [[A]], [[B]]
; IS__CGSCC____-NEXT: [[S:%.*]] = select i1 [[R]], i1 [[R]], i1 false
; IS__CGSCC____-NEXT: ret i1 [[S]]
[Attributor][NFC] Add check lines for tests
2020-02-15 09:27:41 +08:00
;
[Attributor][FIX] Validate the type for AAValueConstantRange as needed Due to the genericValueTraversal we might visit values for which we did not create an AAValueConstantRange object, e.g., as they are behind a PHI or select or call with `returned` argument. As a consequence we need to validate the types as we are about to query AAValueConstantRange for operands.
2020-02-14 04:22:26 +08:00
%r = icmp ult i8* %a, %b
%s = select i1 %r, i1 %r, i1 0
ret i1 %s
}
define i1 @fcmp_caller(float %fa, float %fb, double %da, double %db, double* %dpa, double* %dpb, i8* %ipa, i8* %ipb) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@fcmp_caller
; IS__TUNIT____-SAME: (float [[FA:%.*]], float [[FB:%.*]], double [[DA:%.*]], double [[DB:%.*]], double* nofree readnone [[DPA:%.*]], double* nofree readnone [[DPB:%.*]], i8* nofree readnone [[IPA:%.*]], i8* nofree readnone [[IPB:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: [[R1:%.*]] = call i1 @f_fcmp(float [[FA]], float [[FB]]) #[[ATTR1]]
; IS__TUNIT____-NEXT: [[R2:%.*]] = call i1 @d_fcmp(double [[DA]], double [[DB]]) #[[ATTR1]]
; IS__TUNIT____-NEXT: [[R3:%.*]] = call i1 @dp_icmp(double* noalias nofree readnone [[DPA]], double* noalias nofree readnone [[DPB]]) #[[ATTR1]]
; IS__TUNIT____-NEXT: [[R4:%.*]] = call i1 @ip_icmp(i8* noalias nofree readnone [[IPA]], i8* noalias nofree readnone [[IPB]]) #[[ATTR1]]
; IS__TUNIT____-NEXT: [[O1:%.*]] = or i1 [[R1]], [[R2]]
; IS__TUNIT____-NEXT: [[O2:%.*]] = or i1 [[R3]], [[R4]]
; IS__TUNIT____-NEXT: [[O3:%.*]] = or i1 [[O1]], [[O2]]
; IS__TUNIT____-NEXT: ret i1 [[O3]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
;
; IS__CGSCC_OPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@fcmp_caller
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM-SAME: (float [[FA:%.*]], float [[FB:%.*]], double [[DA:%.*]], double [[DB:%.*]], double* nofree readnone [[DPA:%.*]], double* nofree readnone [[DPB:%.*]], i8* nofree readnone [[IPA:%.*]], i8* nofree readnone [[IPB:%.*]]) #[[ATTR1]] {
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-NEXT: [[R1:%.*]] = call i1 @f_fcmp(float [[FA]], float [[FB]]) #[[ATTR5]]
; IS__CGSCC_OPM-NEXT: [[R2:%.*]] = call i1 @d_fcmp(double [[DA]], double [[DB]]) #[[ATTR5]]
; IS__CGSCC_OPM-NEXT: [[R3:%.*]] = call i1 @dp_icmp(double* noalias nofree readnone [[DPA]], double* noalias nofree readnone [[DPB]]) #[[ATTR5]]
; IS__CGSCC_OPM-NEXT: [[R4:%.*]] = call i1 @ip_icmp(i8* noalias nofree readnone [[IPA]], i8* noalias nofree readnone [[IPB]]) #[[ATTR5]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[O1:%.*]] = or i1 [[R1]], [[R2]]
; IS__CGSCC_OPM-NEXT: [[O2:%.*]] = or i1 [[R3]], [[R4]]
; IS__CGSCC_OPM-NEXT: [[O3:%.*]] = or i1 [[O1]], [[O2]]
; IS__CGSCC_OPM-NEXT: ret i1 [[O3]]
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@fcmp_caller
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (float [[FA:%.*]], float [[FB:%.*]], double [[DA:%.*]], double [[DB:%.*]], double* nofree readnone [[DPA:%.*]], double* nofree readnone [[DPB:%.*]], i8* nofree readnone [[IPA:%.*]], i8* nofree readnone [[IPB:%.*]]) #[[ATTR1]] {
; IS__CGSCC_NPM-NEXT: [[R1:%.*]] = call i1 @f_fcmp(float [[FA]], float [[FB]]) #[[ATTR3]]
; IS__CGSCC_NPM-NEXT: [[R2:%.*]] = call i1 @d_fcmp(double [[DA]], double [[DB]]) #[[ATTR3]]
; IS__CGSCC_NPM-NEXT: [[R3:%.*]] = call i1 @dp_icmp(double* noalias nofree readnone [[DPA]], double* noalias nofree readnone [[DPB]]) #[[ATTR3]]
; IS__CGSCC_NPM-NEXT: [[R4:%.*]] = call i1 @ip_icmp(i8* noalias nofree readnone [[IPA]], i8* noalias nofree readnone [[IPB]]) #[[ATTR3]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[O1:%.*]] = or i1 [[R1]], [[R2]]
; IS__CGSCC_NPM-NEXT: [[O2:%.*]] = or i1 [[R3]], [[R4]]
; IS__CGSCC_NPM-NEXT: [[O3:%.*]] = or i1 [[O1]], [[O2]]
; IS__CGSCC_NPM-NEXT: ret i1 [[O3]]
[Attributor][NFC] Add check lines for tests
2020-02-15 09:27:41 +08:00
;
[Attributor][FIX] Validate the type for AAValueConstantRange as needed Due to the genericValueTraversal we might visit values for which we did not create an AAValueConstantRange object, e.g., as they are behind a PHI or select or call with `returned` argument. As a consequence we need to validate the types as we are about to query AAValueConstantRange for operands.
2020-02-14 04:22:26 +08:00
%r1 = call i1 @f_fcmp(float %fa, float %fb)
%r2 = call i1 @d_fcmp(double %da, double %db)
%r3 = call i1 @dp_icmp(double* %dpa, double* %dpb)
%r4 = call i1 @ip_icmp(i8* %ipa, i8* %ipb)
%o1 = or i1 %r1, %r2
%o2 = or i1 %r3, %r4
%o3 = or i1 %o1, %o2
ret i1 %o3
}
[Attributor][FIX] Collapse `undef` to a proper value If we see an undef we cannot assume it's the same as "no value". For now we just collapse it to 0.
2020-02-15 10:08:20 +08:00
define i8 @ret_two() {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@ret_two
; IS__TUNIT____-SAME: () #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i8 2
[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
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@ret_two
; IS__CGSCC____-SAME: () #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i8 2
[Attributor][FIX] Collapse `undef` to a proper value If we see an undef we cannot assume it's the same as "no value". For now we just collapse it to 0.
2020-02-15 10:08:20 +08:00
;
ret i8 2
}
define i8 @ret_undef() {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@ret_undef
; IS__TUNIT____-SAME: () #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i8 undef
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@ret_undef
; IS__CGSCC____-SAME: () #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i8 undef
[Attributor][FIX] Collapse `undef` to a proper value If we see an undef we cannot assume it's the same as "no value". For now we just collapse it to 0.
2020-02-15 10:08:20 +08:00
;
ret i8 undef
}
; Verify we collapse undef to a value and return something non-undef here.
define i8 @undef_collapse_1() {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@undef_collapse_1
; IS__TUNIT____-SAME: () #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i8 0
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@undef_collapse_1
; IS__CGSCC____-SAME: () #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i8 0
[Attributor][FIX] Collapse `undef` to a proper value If we see an undef we cannot assume it's the same as "no value". For now we just collapse it to 0.
2020-02-15 10:08:20 +08:00
;
%c = call i8 @ret_undef()
%s = shl i8 %c, 2
ret i8 %s
}
; Verify we collapse undef to a value and return something non-undef here.
define i8 @undef_collapse_2() {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@undef_collapse_2
; IS__TUNIT____-SAME: () #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i8 0
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@undef_collapse_2
; IS__CGSCC____-SAME: () #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i8 0
[Attributor][FIX] Collapse `undef` to a proper value If we see an undef we cannot assume it's the same as "no value". For now we just collapse it to 0.
2020-02-15 10:08:20 +08:00
;
%c = call i8 @ret_two()
%s = shl i8 undef, %c
ret i8 %s
}
define i8 @undef_collapse_caller() {
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@undef_collapse_caller
; IS__TUNIT____-SAME: () #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i8 0
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@undef_collapse_caller
; IS__CGSCC____-SAME: () #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i8 0
[Attributor][FIX] Collapse `undef` to a proper value If we see an undef we cannot assume it's the same as "no value". For now we just collapse it to 0.
2020-02-15 10:08:20 +08:00
;
%c1 = call i8 @undef_collapse_1()
%c2 = call i8 @undef_collapse_2()
%a = add i8 %c1, %c2
ret i8 %a
}
[Attributor] Handle calls in AAValueConstantRange properly We did handle calls that were operands of certain instructions but not standalone calls we visit via indirection, e.g., selects.
2020-03-20 14:37:27 +08:00
define i32 @ret1or2(i1 %c) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@ret1or2
; IS__TUNIT____-SAME: (i1 [[C:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: [[S:%.*]] = select i1 [[C]], i32 1, i32 2
; IS__TUNIT____-NEXT: ret i32 [[S]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@ret1or2
; IS__CGSCC____-SAME: (i1 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[S:%.*]] = select i1 [[C]], i32 1, i32 2
; IS__CGSCC____-NEXT: ret i32 [[S]]
[Attributor] Handle calls in AAValueConstantRange properly We did handle calls that were operands of certain instructions but not standalone calls we visit via indirection, e.g., selects.
2020-03-20 14:37:27 +08:00
;
%s = select i1 %c, i32 1, i32 2
ret i32 %s
}
define i1 @callee_range_1(i1 %c1, i1 %c2, i1 %c3) {
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@callee_range_1
; IS__TUNIT____-SAME: (i1 [[C1:%.*]], i1 [[C2:%.*]], i1 [[C3:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i1 true
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@callee_range_1
; IS__CGSCC____-SAME: (i1 [[C1:%.*]], i1 [[C2:%.*]], i1 [[C3:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i1 true
[Attributor] Handle calls in AAValueConstantRange properly We did handle calls that were operands of certain instructions but not standalone calls we visit via indirection, e.g., selects.
2020-03-20 14:37:27 +08:00
;
%r1 = call i32 @ret1or2(i1 %c1)
%r2 = call i32 @ret1or2(i1 %c2)
%indirection = select i1 %c3, i32 %r1, i32 %r2
%a = add i32 %r1, %indirection
%i1 = icmp sle i32 %a, 4
%i2 = icmp sge i32 %a, 2
%f = and i1 %i1, %i2
ret i1 %f
}
define i1 @callee_range_2(i1 %c1, i1 %c2) {
;
[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_OPM: Function Attrs: nofree 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__TUNIT_OPM-LABEL: define {{[^@]+}}@callee_range_2
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-SAME: (i1 [[C1:%.*]], i1 [[C2:%.*]]) #[[ATTR1]] {
; IS__TUNIT_OPM-NEXT: [[R1:%.*]] = call i32 @ret1or2(i1 [[C1]]) #[[ATTR1]], !range [[RNG4:![0-9]+]]
; IS__TUNIT_OPM-NEXT: [[R2:%.*]] = call i32 @ret1or2(i1 [[C2]]) #[[ATTR1]], !range [[RNG4]]
[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_OPM-NEXT: [[A:%.*]] = add i32 [[R1]], [[R2]]
; IS__TUNIT_OPM-NEXT: [[I1:%.*]] = icmp sle i32 [[A]], 3
[Attributor] Simplify operands inside of simplification AAs first When we do simplification via AAPotentialValues or AAValueConstantRange we need to simplify the operands of an instruction we deconstruct first. This does not only improve the result, see for example range.ll, but is required as we allow outside AAs to provide simplification rules via callbacks. If we do ignore the simplification rules and base other simplifications on the IR instead we can create an inconsistent state.
2021-06-23 02:19:39 +08:00
; IS__TUNIT_OPM-NEXT: [[F:%.*]] = and i1 [[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
; IS__TUNIT_OPM-NEXT: ret i1 [[F]]
;
[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_NPM: Function Attrs: nofree 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__TUNIT_NPM-LABEL: define {{[^@]+}}@callee_range_2
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-SAME: (i1 [[C1:%.*]], i1 [[C2:%.*]]) #[[ATTR1]] {
; IS__TUNIT_NPM-NEXT: [[R1:%.*]] = call i32 @ret1or2(i1 [[C1]]) #[[ATTR1]], !range [[RNG5:![0-9]+]]
; IS__TUNIT_NPM-NEXT: [[R2:%.*]] = call i32 @ret1or2(i1 [[C2]]) #[[ATTR1]], !range [[RNG5]]
[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_NPM-NEXT: [[A:%.*]] = add i32 [[R1]], [[R2]]
; IS__TUNIT_NPM-NEXT: [[I1:%.*]] = icmp sle i32 [[A]], 3
[Attributor] Simplify operands inside of simplification AAs first When we do simplification via AAPotentialValues or AAValueConstantRange we need to simplify the operands of an instruction we deconstruct first. This does not only improve the result, see for example range.ll, but is required as we allow outside AAs to provide simplification rules via callbacks. If we do ignore the simplification rules and base other simplifications on the IR instead we can create an inconsistent state.
2021-06-23 02:19:39 +08:00
; IS__TUNIT_NPM-NEXT: [[F:%.*]] = and i1 [[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
; IS__TUNIT_NPM-NEXT: ret i1 [[F]]
;
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@callee_range_2
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM-SAME: (i1 [[C1:%.*]], i1 [[C2:%.*]]) #[[ATTR1]] {
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-NEXT: [[R1:%.*]] = call i32 @ret1or2(i1 [[C1]]) #[[ATTR5]], !range [[RNG5:![0-9]+]]
; IS__CGSCC_OPM-NEXT: [[R2:%.*]] = call i32 @ret1or2(i1 [[C2]]) #[[ATTR5]], !range [[RNG5]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: [[A:%.*]] = add i32 [[R1]], [[R2]]
; IS__CGSCC_OPM-NEXT: [[I1:%.*]] = icmp sle i32 [[A]], 3
[Attributor] Simplify operands inside of simplification AAs first When we do simplification via AAPotentialValues or AAValueConstantRange we need to simplify the operands of an instruction we deconstruct first. This does not only improve the result, see for example range.ll, but is required as we allow outside AAs to provide simplification rules via callbacks. If we do ignore the simplification rules and base other simplifications on the IR instead we can create an inconsistent state.
2021-06-23 02:19:39 +08:00
; IS__CGSCC_OPM-NEXT: [[F:%.*]] = and i1 [[I1]], true
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_OPM-NEXT: ret i1 [[F]]
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@callee_range_2
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i1 [[C1:%.*]], i1 [[C2:%.*]]) #[[ATTR1]] {
; IS__CGSCC_NPM-NEXT: [[R1:%.*]] = call i32 @ret1or2(i1 [[C1]]) #[[ATTR3]], !range [[RNG5:![0-9]+]]
; IS__CGSCC_NPM-NEXT: [[R2:%.*]] = call i32 @ret1or2(i1 [[C2]]) #[[ATTR3]], !range [[RNG5]]
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: [[A:%.*]] = add i32 [[R1]], [[R2]]
; IS__CGSCC_NPM-NEXT: [[I1:%.*]] = icmp sle i32 [[A]], 3
[Attributor] Simplify operands inside of simplification AAs first When we do simplification via AAPotentialValues or AAValueConstantRange we need to simplify the operands of an instruction we deconstruct first. This does not only improve the result, see for example range.ll, but is required as we allow outside AAs to provide simplification rules via callbacks. If we do ignore the simplification rules and base other simplifications on the IR instead we can create an inconsistent state.
2021-06-23 02:19:39 +08:00
; IS__CGSCC_NPM-NEXT: [[F:%.*]] = and i1 [[I1]], true
[Attributor][NFC] rerun update_test_checks without --scrub-attributes
2020-08-30 01:31:13 +08:00
; IS__CGSCC_NPM-NEXT: ret i1 [[F]]
[Attributor] Handle calls in AAValueConstantRange properly We did handle calls that were operands of certain instructions but not standalone calls we visit via indirection, e.g., selects.
2020-03-20 14:37:27 +08:00
;
%r1 = call i32 @ret1or2(i1 %c1)
%r2 = call i32 @ret1or2(i1 %c2)
%a = add i32 %r1, %r2
%i1 = icmp sle i32 %a, 3
%i2 = icmp sge i32 %a, 2
%f = and i1 %i1, %i2
ret i1 %f
}
[Attributor][FIX] Collapse `undef` to a proper value If we see an undef we cannot assume it's the same as "no value". For now we just collapse it to 0.
2020-02-15 10:08:20 +08:00
[Attributor] Use the proper context instruction in genericValueTraversal There was a TODO in genericValueTraversal to provide the context instruction and due to the lack of it users that wanted one just used something available. Unfortunately, using a fixed instruction is wrong in the presence of PHIs so we need to update the context instruction properly. Reviewed By: uenoku Differential Revision: https://reviews.llvm.org/D76870
2020-04-02 11:05:24 +08:00
define i32 @ret100() {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@ret100
; IS__TUNIT____-SAME: () #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i32 100
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@ret100
; IS__CGSCC____-SAME: () #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i32 100
[Attributor] Use the proper context instruction in genericValueTraversal There was a TODO in genericValueTraversal to provide the context instruction and due to the lack of it users that wanted one just used something available. Unfortunately, using a fixed instruction is wrong in the presence of PHIs so we need to update the context instruction properly. Reviewed By: uenoku Differential Revision: https://reviews.llvm.org/D76870
2020-04-02 11:05:24 +08:00
;
ret i32 100
}
define i1 @ctx_adjustment(i32 %V) {
;
[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_OPM: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@ctx_adjustment
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-SAME: (i32 [[V:%.*]]) #[[ATTR1]] {
[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_OPM-NEXT: [[C1:%.*]] = icmp sge i32 [[V]], 100
; IS__TUNIT_OPM-NEXT: br i1 [[C1]], label [[IF_TRUE:%.*]], label [[IF_FALSE:%.*]]
; IS__TUNIT_OPM: if.true:
; IS__TUNIT_OPM-NEXT: br label [[END:%.*]]
; IS__TUNIT_OPM: if.false:
; IS__TUNIT_OPM-NEXT: br label [[END]]
; IS__TUNIT_OPM: end:
; IS__TUNIT_OPM-NEXT: [[PHI:%.*]] = phi i32 [ [[V]], [[IF_TRUE]] ], [ 100, [[IF_FALSE]] ]
; IS__TUNIT_OPM-NEXT: [[C2:%.*]] = icmp sge i32 [[PHI]], 100
; IS__TUNIT_OPM-NEXT: ret i1 [[C2]]
;
; IS__TUNIT_NPM: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@ctx_adjustment
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-SAME: (i32 [[V:%.*]]) #[[ATTR1]] {
[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_NPM-NEXT: [[C1:%.*]] = icmp sge i32 [[V]], 100
; IS__TUNIT_NPM-NEXT: br i1 [[C1]], label [[IF_TRUE:%.*]], label [[IF_FALSE:%.*]]
; IS__TUNIT_NPM: if.true:
; IS__TUNIT_NPM-NEXT: br label [[END:%.*]]
; IS__TUNIT_NPM: if.false:
; IS__TUNIT_NPM-NEXT: br label [[END]]
; IS__TUNIT_NPM: end:
; IS__TUNIT_NPM-NEXT: ret i1 true
;
; IS__CGSCC_OPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@ctx_adjustment
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM-SAME: (i32 [[V:%.*]]) #[[ATTR1]] {
[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-NEXT: [[C1:%.*]] = icmp sge i32 [[V]], 100
; IS__CGSCC_OPM-NEXT: br i1 [[C1]], label [[IF_TRUE:%.*]], label [[IF_FALSE:%.*]]
; IS__CGSCC_OPM: if.true:
; IS__CGSCC_OPM-NEXT: br label [[END:%.*]]
; IS__CGSCC_OPM: if.false:
; IS__CGSCC_OPM-NEXT: br label [[END]]
; IS__CGSCC_OPM: end:
; IS__CGSCC_OPM-NEXT: [[PHI:%.*]] = phi i32 [ [[V]], [[IF_TRUE]] ], [ 100, [[IF_FALSE]] ]
; IS__CGSCC_OPM-NEXT: [[C2:%.*]] = icmp sge i32 [[PHI]], 100
; IS__CGSCC_OPM-NEXT: ret i1 [[C2]]
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@ctx_adjustment
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i32 [[V:%.*]]) #[[ATTR1]] {
[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_NPM-NEXT: [[C1:%.*]] = icmp sge i32 [[V]], 100
; IS__CGSCC_NPM-NEXT: br i1 [[C1]], label [[IF_TRUE:%.*]], label [[IF_FALSE:%.*]]
; IS__CGSCC_NPM: if.true:
; IS__CGSCC_NPM-NEXT: br label [[END:%.*]]
; IS__CGSCC_NPM: if.false:
; IS__CGSCC_NPM-NEXT: br label [[END]]
; IS__CGSCC_NPM: end:
; IS__CGSCC_NPM-NEXT: ret i1 true
[Attributor] Use the proper context instruction in genericValueTraversal There was a TODO in genericValueTraversal to provide the context instruction and due to the lack of it users that wanted one just used something available. Unfortunately, using a fixed instruction is wrong in the presence of PHIs so we need to update the context instruction properly. Reviewed By: uenoku Differential Revision: https://reviews.llvm.org/D76870
2020-04-02 11:05:24 +08:00
;
%c1 = icmp sge i32 %V, 100
br i1 %c1, label %if.true, label %if.false
if.true:
br label %end
if.false:
%call = call i32 @ret100()
br label %end
end:
%phi = phi i32 [ %V, %if.true ], [ %call, %if.false ]
%c2 = icmp sge i32 %phi, 100
ret i1 %c2
}
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
define i32 @func(i1 %c) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@func
; IS__TUNIT____-SAME: (i1 [[C:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: [[RET:%.*]] = select i1 [[C]], i32 0, i32 1
; IS__TUNIT____-NEXT: ret i32 [[RET]]
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@func
; IS__CGSCC____-SAME: (i1 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[RET:%.*]] = select i1 [[C]], i32 0, i32 1
; IS__CGSCC____-NEXT: ret i32 [[RET]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
;
%ret = select i1 %c, i32 0, i32 1
ret i32 %ret
}
define i32 @simplify_callsite_argument(i1 %d) {
[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_OPM: Function Attrs: nofree nosync nounwind readnone willreturn
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@simplify_callsite_argument
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-SAME: (i1 [[D:%.*]]) #[[ATTR1]] {
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__TUNIT_OPM-NEXT: [[C:%.*]] = select i1 [[D]], i1 true, i1 false
; IS__TUNIT_OPM-NEXT: br i1 [[C]], label [[T:%.*]], label [[F:%.*]]
; IS__TUNIT_OPM: t:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT_OPM-NEXT: [[RET1:%.*]] = call noundef i32 @func(i1 noundef [[C]]) #[[ATTR1]], !range [[RNG3]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__TUNIT_OPM-NEXT: ret i32 [[RET1]]
; IS__TUNIT_OPM: f:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT_OPM-NEXT: [[RET2:%.*]] = call noundef i32 @func(i1 noundef false) #[[ATTR1]], !range [[RNG3]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__TUNIT_OPM-NEXT: ret i32 [[RET2]]
;
[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_NPM: Function Attrs: nofree nosync nounwind readnone willreturn
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@simplify_callsite_argument
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-SAME: (i1 [[D:%.*]]) #[[ATTR1]] {
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__TUNIT_NPM-NEXT: [[C:%.*]] = select i1 [[D]], i1 true, i1 false
; IS__TUNIT_NPM-NEXT: br i1 [[C]], label [[T:%.*]], label [[F:%.*]]
; IS__TUNIT_NPM: t:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT_NPM-NEXT: [[RET1:%.*]] = call noundef i32 @func(i1 noundef true) #[[ATTR1]], !range [[RNG4]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__TUNIT_NPM-NEXT: ret i32 [[RET1]]
; IS__TUNIT_NPM: f:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT_NPM-NEXT: [[RET2:%.*]] = call noundef i32 @func(i1 noundef false) #[[ATTR1]], !range [[RNG4]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__TUNIT_NPM-NEXT: ret i32 [[RET2]]
;
[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: nofree norecurse nosync nounwind readnone willreturn
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@simplify_callsite_argument
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM-SAME: (i1 [[D:%.*]]) #[[ATTR1]] {
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__CGSCC_OPM-NEXT: [[C:%.*]] = select i1 [[D]], i1 true, i1 false
; IS__CGSCC_OPM-NEXT: br i1 [[C]], label [[T:%.*]], label [[F:%.*]]
; IS__CGSCC_OPM: t:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__CGSCC_OPM-NEXT: [[RET1:%.*]] = call noundef i32 @func(i1 noundef [[C]]) #[[ATTR5]], !range [[RNG4]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__CGSCC_OPM-NEXT: ret i32 [[RET1]]
; IS__CGSCC_OPM: f:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__CGSCC_OPM-NEXT: [[RET2:%.*]] = call noundef i32 @func(i1 noundef false) #[[ATTR5]], !range [[RNG4]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__CGSCC_OPM-NEXT: ret i32 [[RET2]]
;
[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_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@simplify_callsite_argument
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i1 [[D:%.*]]) #[[ATTR1]] {
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__CGSCC_NPM-NEXT: [[C:%.*]] = select i1 [[D]], i1 true, i1 false
; IS__CGSCC_NPM-NEXT: br i1 [[C]], label [[T:%.*]], label [[F:%.*]]
; IS__CGSCC_NPM: t:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__CGSCC_NPM-NEXT: [[RET1:%.*]] = call noundef i32 @func(i1 noundef true) #[[ATTR3]], !range [[RNG4]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__CGSCC_NPM-NEXT: ret i32 [[RET1]]
; IS__CGSCC_NPM: f:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__CGSCC_NPM-NEXT: [[RET2:%.*]] = call noundef i32 @func(i1 noundef false) #[[ATTR3]], !range [[RNG4]]
[Attributor] Add AAValueSimplifyCallSiteArgument::manifest Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D82861
2020-07-12 18:11:49 +08:00
; IS__CGSCC_NPM-NEXT: ret i32 [[RET2]]
;
%c = select i1 %d, i1 true, i1 false
br i1 %c, label %t, label %f
t:
%ret1 = call i32 @func(i1 %c)
ret i32 %ret1
f:
%ret2 = call i32 @func(i1 false)
ret i32 %ret2
}
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
define internal i32 @less_than_65536(i32 %arg) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@less_than_65536
; IS__CGSCC____-SAME: (i32 [[ARG:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[SHRINKED:%.*]] = udiv i32 undef, 65536
; IS__CGSCC____-NEXT: ret i32 undef
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
%shrinked = udiv i32 %arg, 65536
ret i32 %shrinked
}
define internal i1 @is_less_than_65536(i32 %arg) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@is_less_than_65536
; IS__CGSCC____-SAME: (i32 [[ARG:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[CMP:%.*]] = icmp ult i32 undef, 65536
; IS__CGSCC____-NEXT: ret i1 undef
;
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
%cmp = icmp ult i32 %arg, 65536
ret i1 %cmp
}
define i1 @check_divided_range(i32 %arg) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@check_divided_range
; IS__TUNIT____-SAME: (i32 [[ARG:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i1 true
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@check_divided_range
; IS__CGSCC____-SAME: (i32 [[ARG:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i1 true
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
%csret1 = call i32 @less_than_65536(i32 0)
%csret2 = call i32 @less_than_65536(i32 %arg)
%true1 = call i1 @is_less_than_65536(i32 %csret1)
%true2 = call i1 @is_less_than_65536(i32 %csret2)
%ret = and i1 %true1, %true2
ret i1 %ret
}
define internal i32 @cast_and_return(i1 %c) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@cast_and_return
; IS__CGSCC____-SAME: (i1 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[RET:%.*]] = zext i1 undef to i32
; IS__CGSCC____-NEXT: ret i32 undef
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
%ret = zext i1 %c to i32
ret i32 %ret
}
define internal i1 @is_less_than_3(i32 %c) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@is_less_than_3
; IS__CGSCC____-SAME: () #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i1 undef
;
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
%cmp = icmp slt i32 %c, 3
ret i1 %cmp
}
define i1 @check_casted_range(i1 %c) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@check_casted_range
; IS__TUNIT____-SAME: (i1 [[C:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i1 true
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@check_casted_range
; IS__CGSCC____-SAME: (i1 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i1 true
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
%csret1 = call i32 @cast_and_return(i1 true)
%csret2 = call i32 @cast_and_return(i1 %c)
%add = add i32 %csret1, %csret2
%ret = call i1 @is_less_than_3(i32 %add)
ret i1 %ret
}
define internal i32 @less_than_100_1(i32 %c) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@less_than_100_1
; IS__CGSCC____-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: switch i32 [[C]], label [[OTHERWISE:%.*]] [
; IS__CGSCC____-NEXT: i32 0, label [[ONZERO:%.*]]
; IS__CGSCC____-NEXT: i32 1, label [[ONONE:%.*]]
; IS__CGSCC____-NEXT: i32 2, label [[ONTWO:%.*]]
; IS__CGSCC____-NEXT: i32 3, label [[ONTHREE:%.*]]
; IS__CGSCC____-NEXT: i32 4, label [[ONFOUR:%.*]]
; IS__CGSCC____-NEXT: i32 5, label [[ONFIVE:%.*]]
; IS__CGSCC____-NEXT: i32 6, label [[ONSIX:%.*]]
; IS__CGSCC____-NEXT: ]
; IS__CGSCC____: onzero:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onone:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: ontwo:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onthree:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onfour:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onfive:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onsix:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: otherwise:
; IS__CGSCC____-NEXT: ret i32 undef
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
switch i32 %c, label %otherwise [ i32 0, label %onzero
i32 1, label %onone
i32 2, label %ontwo
i32 3, label %onthree
i32 4, label %onfour
i32 5, label %onfive
i32 6, label %onsix]
onzero:
ret i32 0
onone:
ret i32 1
ontwo:
ret i32 2
onthree:
ret i32 3
onfour:
ret i32 4
onfive:
ret i32 5
onsix:
ret i32 6
otherwise:
ret i32 99
}
define internal i1 @is_less_than_100_1(i32 %c) {
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@is_less_than_100_1
; IS__CGSCC____-SAME: () #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i1 undef
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
%cmp = icmp slt i32 %c, 100
ret i1 %cmp
}
define i1 @propagate_range1(i32 %c){
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@propagate_range1
; IS__TUNIT____-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i1 true
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@propagate_range1
; IS__CGSCC____-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i1 true
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
%csret = call i32 @less_than_100_1(i32 %c)
%true = call i1 @is_less_than_100_1(i32 %csret)
ret i1 %true
}
define internal i32 @less_than_100_2(i32 %c) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@less_than_100_2
; IS__CGSCC____-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: switch i32 [[C]], label [[OTHERWISE:%.*]] [
; IS__CGSCC____-NEXT: i32 0, label [[ONZERO:%.*]]
; IS__CGSCC____-NEXT: i32 1, label [[ONONE:%.*]]
; IS__CGSCC____-NEXT: i32 2, label [[ONTWO:%.*]]
; IS__CGSCC____-NEXT: i32 3, label [[ONTHREE:%.*]]
; IS__CGSCC____-NEXT: i32 4, label [[ONFOUR:%.*]]
; IS__CGSCC____-NEXT: i32 5, label [[ONFIVE:%.*]]
; IS__CGSCC____-NEXT: i32 6, label [[ONSIX:%.*]]
; IS__CGSCC____-NEXT: ]
; IS__CGSCC____: onzero:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onone:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: ontwo:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onthree:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onfour:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onfive:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: onsix:
; IS__CGSCC____-NEXT: ret i32 undef
; IS__CGSCC____: otherwise:
; IS__CGSCC____-NEXT: ret i32 undef
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
switch i32 %c, label %otherwise [ i32 0, label %onzero
i32 1, label %onone
i32 2, label %ontwo
i32 3, label %onthree
i32 4, label %onfour
i32 5, label %onfive
i32 6, label %onsix]
onzero:
ret i32 0
onone:
ret i32 1
ontwo:
ret i32 2
onthree:
ret i32 3
onfour:
ret i32 4
onfive:
ret i32 5
onsix:
ret i32 6
otherwise:
ret i32 99
}
define internal i1 @is_less_than_100_2(i32 %c) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
;
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@is_less_than_100_2
; IS__CGSCC____-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: [[CMP:%.*]] = icmp slt i32 undef, 100
; IS__CGSCC____-NEXT: ret i1 undef
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
%cmp = icmp slt i32 %c, 100
ret i1 %cmp
}
define i1 @propagate_range2(i32 %c) {
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT____: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT____-LABEL: define {{[^@]+}}@propagate_range2
; IS__TUNIT____-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; IS__TUNIT____-NEXT: ret i1 true
Revert Attributor patch series Broke check-clang, see https://reviews.llvm.org/D102307#2869065 Ran `git revert -n ebbe149a6f08535ede848a531a601ae6591cfbc5..269416d41908bb670f67af689155d5ab8eea689a`
2021-07-11 04:15:55 +08:00
;
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC____: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC____-LABEL: define {{[^@]+}}@propagate_range2
; IS__CGSCC____-SAME: (i32 [[C:%.*]]) #[[ATTR1]] {
; IS__CGSCC____-NEXT: ret i1 true
[Attributor][NFC] Add tests to range.ll Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D86128
2020-08-19 14:01:14 +08:00
;
%csret1 = call i32 @less_than_100_2(i32 0)
%true1 = call i1 @is_less_than_100_2(i32 %csret1)
%csret2 = call i32 @less_than_100_2(i32 %c)
%true2 = call i1 @is_less_than_100_2(i32 %csret2)
%true = and i1 %true1, %true2
ret i1 %true
}
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
define internal i1 @non_zero(i8 %v) {
; IS__TUNIT_OPM: Function Attrs: nofree nosync nounwind readnone willreturn
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@non_zero
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-SAME: (i8 [[V:%.*]]) #[[ATTR1]] {
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__TUNIT_OPM-NEXT: [[R:%.*]] = icmp ne i8 [[V]], 0
; IS__TUNIT_OPM-NEXT: ret i1 [[R]]
;
; IS__CGSCC_OPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@non_zero
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM-SAME: (i8 [[V:%.*]]) #[[ATTR1]] {
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__CGSCC_OPM-NEXT: [[R:%.*]] = icmp ne i8 [[V]], 0
; IS__CGSCC_OPM-NEXT: ret i1 [[R]]
;
; IS__CGSCC_NPM: Function Attrs: nofree norecurse nosync nounwind readnone willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@non_zero
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__CGSCC_NPM-SAME: () #[[ATTR1]] {
; IS__CGSCC_NPM-NEXT: ret i1 undef
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
;
%r = icmp ne i8 %v, 0
ret i1 %r
}
; Avoid range metadata for %l below
define i1 @context(i8* %p) {
; IS__TUNIT_OPM: Function Attrs: argmemonly nofree nosync nounwind readonly willreturn
; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@context
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_OPM-SAME: (i8* nocapture nofree noundef nonnull readonly dereferenceable(1) [[P:%.*]]) #[[ATTR0]] {
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__TUNIT_OPM-NEXT: [[L:%.*]] = load i8, i8* [[P]], align 1
; IS__TUNIT_OPM-NEXT: [[C:%.*]] = icmp slt i8 0, [[L]]
; IS__TUNIT_OPM-NEXT: br i1 [[C]], label [[T:%.*]], label [[F:%.*]]
; IS__TUNIT_OPM: t:
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM-NEXT: [[R:%.*]] = call i1 @non_zero(i8 [[L]]) #[[ATTR1]]
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__TUNIT_OPM-NEXT: ret i1 [[R]]
; IS__TUNIT_OPM: f:
; IS__TUNIT_OPM-NEXT: ret i1 false
;
; IS__TUNIT_NPM: Function Attrs: argmemonly nofree nosync nounwind readonly willreturn
; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@context
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_NPM-SAME: (i8* nocapture nofree noundef nonnull readonly dereferenceable(1) [[P:%.*]]) #[[ATTR0]] {
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__TUNIT_NPM-NEXT: [[L:%.*]] = load i8, i8* [[P]], align 1
; IS__TUNIT_NPM-NEXT: [[C:%.*]] = icmp slt i8 0, [[L]]
; IS__TUNIT_NPM-NEXT: br i1 [[C]], label [[T:%.*]], label [[F:%.*]]
; IS__TUNIT_NPM: t:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__TUNIT_NPM-NEXT: ret i1 true
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__TUNIT_NPM: f:
; IS__TUNIT_NPM-NEXT: ret i1 false
;
; IS__CGSCC_OPM: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@context
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-SAME: (i8* nocapture nofree noundef nonnull readonly dereferenceable(1) [[P:%.*]]) #[[ATTR0]] {
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__CGSCC_OPM-NEXT: [[L:%.*]] = load i8, i8* [[P]], align 1
; IS__CGSCC_OPM-NEXT: [[C:%.*]] = icmp slt i8 0, [[L]]
; IS__CGSCC_OPM-NEXT: br i1 [[C]], label [[T:%.*]], label [[F:%.*]]
; IS__CGSCC_OPM: t:
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM-NEXT: [[R:%.*]] = call i1 @non_zero(i8 [[L]]) #[[ATTR5]]
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__CGSCC_OPM-NEXT: ret i1 [[R]]
; IS__CGSCC_OPM: f:
; IS__CGSCC_OPM-NEXT: ret i1 false
;
; IS__CGSCC_NPM: Function Attrs: argmemonly nofree norecurse nosync nounwind readonly willreturn
; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@context
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_NPM-SAME: (i8* nocapture nofree noundef nonnull readonly dereferenceable(1) [[P:%.*]]) #[[ATTR0]] {
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__CGSCC_NPM-NEXT: [[L:%.*]] = load i8, i8* [[P]], align 1
; IS__CGSCC_NPM-NEXT: [[C:%.*]] = icmp slt i8 0, [[L]]
; IS__CGSCC_NPM-NEXT: br i1 [[C]], label [[T:%.*]], label [[F:%.*]]
; IS__CGSCC_NPM: t:
[Attributor] Use AAValueSimplify to simplify returned values We should use AAValueSimplify for all value simplification, however there was some leftover logic that predates AAValueSimplify in AAReturnedValues. This remove the AAReturnedValues part and provides a replacement by making AAValueSimplifyReturned strong enough to handle all previously covered cases. Further, this improve AAValueSimplifyCallSiteReturned to handle returned arguments. AAReturnedValues is now much easier and the collected returned values/instructions are now from the associated function only, making it much more sane. We also do not have the brittle logic anymore that looks for unresolved calls. Instead, we use AAValueSimplify to handle recursion. Useful code has been split into helper functions, e.g., an Attributor interface to get a simplified value. Differential Revision: https://reviews.llvm.org/D103860
2021-05-10 09:16:50 +08:00
; IS__CGSCC_NPM-NEXT: ret i1 true
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; IS__CGSCC_NPM: f:
; IS__CGSCC_NPM-NEXT: ret i1 false
;
%l = load i8, i8* %p
%c = icmp slt i8 0, %l
br i1 %c, label %t, label %f
t:
%r = call i1 @non_zero(i8 %l)
ret i1 %r
f:
ret i1 false
}
define void @spam(i32* %arg, i32* %arg1) {
; CHECK-LABEL: define {{[^@]+}}@spam
[Attributor] Set floating point loads and stores as nofree in AANoFreeFloating Summary: The current implementation of AANoFreeFloating will incorrectly list floating point loads and stores as may-free. This prevents other attributor instances like HeapToStack from pushing some allocations to the stack. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D103975
2021-06-10 02:43:59 +08:00
; CHECK-SAME: (i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[ARG:%.*]], i32* nocapture nofree readnone [[ARG1:%.*]]) {
[Attributor][FIX] Do not attach range metadata to the wrong Instruction If we are looking at a call site argument it might be a load or call which is in a different context than the call site argument. We cannot simply use the call site argument range for the call or load. Bug reported and reduced by Whitney Tsang <whitneyt@ca.ibm.com>.
2020-10-10 23:06:03 +08:00
; CHECK-NEXT: bb:
; CHECK-NEXT: [[TMP:%.*]] = load i32, i32* [[ARG]], align 8
; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 [[TMP]], 4
; CHECK-NEXT: br i1 [[TMP2]], label [[BB3:%.*]], label [[BB4:%.*]]
; CHECK: bb3:
; CHECK-NEXT: call fastcc void @wobble(i32 signext [[TMP]])
; CHECK-NEXT: br label [[BB5:%.*]]
; CHECK: bb4:
; CHECK-NEXT: call void @ham(i32 [[TMP]])
; CHECK-NEXT: br label [[BB5]]
; CHECK: bb5:
; CHECK-NEXT: ret void
;
bb:
%tmp = load i32, i32* %arg, align 8
%tmp2 = icmp ult i32 %tmp, 4
br i1 %tmp2, label %bb3, label %bb4
bb3: ; preds = %bb
call fastcc void @wobble(i32 signext %tmp)
br label %bb5
bb4: ; preds = %bb
call void @ham(i32 %tmp)
br label %bb5
bb5: ; preds = %bb4, %bb3
ret void
}
define internal fastcc void @wobble(i32 signext %arg) {
; CHECK-LABEL: define {{[^@]+}}@wobble
; CHECK-SAME: (i32 signext [[ARG:%.*]]) {
; CHECK-NEXT: bb:
; CHECK-NEXT: [[TMP:%.*]] = icmp ult i32 [[ARG]], 2
; CHECK-NEXT: br i1 [[TMP]], label [[BB1:%.*]], label [[BB2:%.*]]
; CHECK: bb1:
; CHECK-NEXT: call void @barney(i32 noundef signext 32, i32 noundef signext 0)
; CHECK-NEXT: br label [[BB3:%.*]]
; CHECK: bb2:
; CHECK-NEXT: br label [[BB3]]
; CHECK: bb3:
; CHECK-NEXT: ret void
;
bb:
%tmp = icmp ult i32 %arg, 2
br i1 %tmp, label %bb1, label %bb2
bb1: ; preds = %bb
call void @barney(i32 signext 32, i32 signext 0)
br label %bb3
bb2: ; preds = %bb
br label %bb3
bb3: ; preds = %bb2, %bb1
ret void
}
declare void @ham(i32)
declare void @barney(i32 signext, i32 signext)
[Attributor] AAValueConstantRange: Value range analysis using constant range Summary: This patch introduces `AAValueConstantRange`, which answers a possible range for integer value in a specific program point. One of the motivations is propagating existing `range` metadata. (I think we need to change the situation that `range` metadata cannot be put to Argument). The state is a tuple of `ConstantRange` and it is initialized to (known, assumed) = ([-∞, +∞], empty). Currently, AAValueConstantRange is created in `getAssumedConstant` method when `AAValueSimplify` returns `nullptr`(worst state). Supported - BinaryOperator(add, sub, ...) - CmpInst(icmp eq, ...) - !range metadata `AAValueConstantRange` is not intended to extend to polyhedral range value analysis. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: phosek, davezarzycki, baziotis, hiraditya, javed.absar, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D71620
2020-01-15 14:25:52 +08:00
!0 = !{i32 0, i32 10}
!1 = !{i32 10, i32 100}
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
;.
; IS__TUNIT_OPM: attributes #[[ATTR0]] = { argmemonly nofree nosync nounwind readonly willreturn }
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__TUNIT_OPM: attributes #[[ATTR1]] = { nofree nosync nounwind readnone willreturn }
; IS__TUNIT_OPM: attributes #[[ATTR2]] = { nofree nosync nounwind readnone }
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__TUNIT_OPM: attributes #[[ATTR3]] = { nofree nosync nounwind readonly willreturn }
;.
; IS__TUNIT_NPM: attributes #[[ATTR0]] = { argmemonly nofree nosync nounwind readonly willreturn }
; IS__TUNIT_NPM: attributes #[[ATTR1]] = { nofree nosync nounwind readnone willreturn }
; IS__TUNIT_NPM: attributes #[[ATTR2]] = { nofree nosync nounwind readonly willreturn }
;.
; IS__CGSCC_OPM: attributes #[[ATTR0]] = { argmemonly nofree norecurse nosync nounwind readonly willreturn }
[Attributor][FIX] Traverse uses even if a value is assumed constant Not all attributes are able to handle the interprocedural step and follow the uses into a call site. Let them be able to combine call site uses instead. This might result in some unused values/arguments being leftover but it removes problems where we misused "is dead" even though it was actually "is simplified/replaced". We explicitly check for dead values due to constant propagation in `AAIsDeadValueImpl::areAllUsesAssumedDead` instead. Differential Revision: https://reviews.llvm.org/D103858
2021-05-08 12:06:44 +08:00
; IS__CGSCC_OPM: attributes #[[ATTR1]] = { nofree norecurse nosync nounwind readnone willreturn }
; IS__CGSCC_OPM: attributes #[[ATTR2]] = { nofree norecurse nosync nounwind readnone }
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
; IS__CGSCC_OPM: attributes #[[ATTR3]] = { readonly willreturn }
; IS__CGSCC_OPM: attributes #[[ATTR4]] = { nounwind readnone }
[Attributor] Derive `willreturn` based on `mustprogress` Since D86233 we have `mustprogress` which, in combination with `readonly`, implies `willreturn`. The idea is that every side-effect has to be modeled as a "write". Consequently, `readonly` means there is no side-effect, and `mustprogress` guarantees that we cannot "loop" forever without side-effect. Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D94125
2021-01-06 06:55:54 +08:00
; IS__CGSCC_OPM: attributes #[[ATTR5]] = { readnone willreturn }
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
;.
; IS__CGSCC_NPM: attributes #[[ATTR0]] = { argmemonly nofree norecurse nosync nounwind readonly willreturn }
; IS__CGSCC_NPM: attributes #[[ATTR1]] = { nofree norecurse nosync nounwind readnone willreturn }
; IS__CGSCC_NPM: attributes #[[ATTR2]] = { readonly willreturn }
[Attributor] Derive `willreturn` based on `mustprogress` Since D86233 we have `mustprogress` which, in combination with `readonly`, implies `willreturn`. The idea is that every side-effect has to be modeled as a "write". Consequently, `readonly` means there is no side-effect, and `mustprogress` guarantees that we cannot "loop" forever without side-effect. Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D94125
2021-01-06 06:55:54 +08:00
; IS__CGSCC_NPM: attributes #[[ATTR3]] = { readnone willreturn }
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
;.
; IS__TUNIT_OPM: [[RNG0]] = !{i32 0, i32 10}
; IS__TUNIT_OPM: [[RNG1]] = !{i32 10, i32 100}
; IS__TUNIT_OPM: [[RNG2]] = !{i32 200, i32 1091}
; IS__TUNIT_OPM: [[RNG3]] = !{i32 0, i32 2}
; IS__TUNIT_OPM: [[RNG4]] = !{i32 1, i32 3}
;.
[Attributor] Simplify operands inside of simplification AAs first When we do simplification via AAPotentialValues or AAValueConstantRange we need to simplify the operands of an instruction we deconstruct first. This does not only improve the result, see for example range.ll, but is required as we allow outside AAs to provide simplification rules via callbacks. If we do ignore the simplification rules and base other simplifications on the IR instead we can create an inconsistent state.
2021-06-23 02:19:39 +08:00
; IS________NPM: [[RNG0]] = !{i32 0, i32 10}
; IS________NPM: [[RNG1]] = !{i32 10, i32 100}
; IS________NPM: [[META2:![0-9]+]] = !{i32 200, i32 1091}
; IS________NPM: [[META3:![0-9]+]] = !{i32 1, i32 -2147483648}
; IS________NPM: [[RNG4]] = !{i32 0, i32 2}
; IS________NPM: [[META5:![0-9]+]] = !{i32 1, i32 3}
[Attributor][NFC] Update tests after D94741 The update_test_checks script can now check for global symbols and is able to handle them properly when they differ across prefixes, e.g., attribute #0 might be different in different runs. This patch simply updates all the Attributor tests with the new script. Reviewed By: sstefan1 Differential Revision: https://reviews.llvm.org/D97906
2021-03-12 13:21:32 +08:00
;.
; IS__CGSCC_OPM: [[RNG0]] = !{i32 0, i32 10}
; IS__CGSCC_OPM: [[RNG1]] = !{i32 10, i32 100}
; IS__CGSCC_OPM: [[RNG2]] = !{i32 200, i32 1091}
; IS__CGSCC_OPM: [[RNG3]] = !{i32 10, i32 21}
; IS__CGSCC_OPM: [[RNG4]] = !{i32 0, i32 2}
; IS__CGSCC_OPM: [[RNG5]] = !{i32 1, i32 3}
;.