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/FunctionAttrs/nocapture.ll

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

345 lines
9.2 KiB
LLVM
Raw Normal View History

[FunctionAttrs] Rename functionattrs -> function-attrs To match NewPM pass name, and also for readability. Also rename rpo-functionattrs -> rpo-function-attrs while we're here. Reviewed By: arsenm Differential Revision: https://reviews.llvm.org/D84694
2020-07-29 00:08:08 +08:00
; RUN: opt -function-attrs -S < %s | FileCheck %s --check-prefixes=FNATTR
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; RUN: opt -passes=function-attrs -S < %s | FileCheck %s --check-prefixes=FNATTR
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
@g = global i32* null ; <i32**> [#uses=1]
[Attributor] Deduce "no-capture" argument attribute Add the no-capture argument attribute deduction to the Attributor fixpoint framework. The new string attributed "no-capture-maybe-returned" is introduced to allow deduction of no-capture through functions that "capture" an argument but only by "returning" it. It is only used by the Attributor for testing. Differential Revision: https://reviews.llvm.org/D59922 llvm-svn: 370817
2019-09-04 04:37:24 +08:00
; FNATTR: define i32* @c1(i32* readnone returned %q)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define i32* @c1(i32* %q) {
ret i32* %q
}
[funcattrs] Fix a bug in recently introduced writeonly argument inference This fixes a bug in 740057d. There's two ways to describe the issue: * One caller hasn't yet proven nocapture on the argument. Given that, the inference routine is responsible for bailing out on a potential capture. * Even if we know the argument is nocapture, the access inference needs to traverse the exact set of users the capture tracking would (or exit conservatively). Even if capture tracking can prove a store is non-capturing (e.g. to a local alloc which doesn't escape), we still need to track the copy of the pointer to see if it's later reloaded and accessed again. Note that all the test changes except the newly added ones appear to be false negatives. That is, cases where we could prove writeonly, but the current code isn't strong enough. That's why I didn't spot this originally.
2021-12-04 00:52:40 +08:00
; FNATTR: define void @c2(i32* %q)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
; It would also be acceptable to mark %q as readnone. Update @c3 too.
define void @c2(i32* %q) {
store i32* %q, i32** @g
ret void
}
[Attributor][MemBehavior] Fallback to the function state for arguments Even if an argument is captured, we cannot have an effect the function does not have. This is fine except for the special case of `inalloca` as it does not behave by the rules. TODO: Maybe the special rule for `inalloca` is wrong after all. llvm-svn: 374736
2019-10-14 04:47:16 +08:00
; FNATTR: define void @c3(i32* %q)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @c3(i32* %q) {
call void @c2(i32* %q)
ret void
}
[Attributor] Deduce memory behavior of functions and arguments Deduce the memory behavior, aka "read-none", "read-only", or "write-only", for functions and arguments. Reviewers: sstefan1, uenoku Subscribers: hiraditya, bollu, jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67384 llvm-svn: 373965
2019-10-08 05:07:57 +08:00
; FNATTR: define i1 @c4(i32* %q, i32 %bitno)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define i1 @c4(i32* %q, i32 %bitno) {
%tmp = ptrtoint i32* %q to i32
%tmp2 = lshr i32 %tmp, %bitno
%bit = trunc i32 %tmp2 to i1
br i1 %bit, label %l1, label %l0
l0:
ret i1 0 ; escaping value not caught by def-use chaining.
l1:
ret i1 1 ; escaping value not caught by def-use chaining.
}
[Attributor] Teach AANoCapture to use information in-flight more aggressively AAReturnedValues, AAMemoryBehavior, and AANoUnwind, can provide information that helps during the tracking or even justifies no-capture. We now use this information and enable no-capture in some test cases designed a long while a ago for these cases. llvm-svn: 375382
2019-10-21 08:48:42 +08:00
; c4b is c4 but without the escaping part
; FNATTR: define i1 @c4b(i32* %q, i32 %bitno)
define i1 @c4b(i32* %q, i32 %bitno) {
%tmp = ptrtoint i32* %q to i32
%tmp2 = lshr i32 %tmp, %bitno
%bit = trunc i32 %tmp2 to i1
br i1 %bit, label %l1, label %l0
l0:
ret i1 0 ; not escaping!
l1:
ret i1 0 ; not escaping!
}
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
@lookup_table = global [2 x i1] [ i1 0, i1 1 ]
[Attributor][MemBehavior] Fallback to the function state for arguments Even if an argument is captured, we cannot have an effect the function does not have. This is fine except for the special case of `inalloca` as it does not behave by the rules. TODO: Maybe the special rule for `inalloca` is wrong after all. llvm-svn: 374736
2019-10-14 04:47:16 +08:00
; FNATTR: define i1 @c5(i32* %q, i32 %bitno)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define i1 @c5(i32* %q, i32 %bitno) {
%tmp = ptrtoint i32* %q to i32
%tmp2 = lshr i32 %tmp, %bitno
%bit = and i32 %tmp2, 1
; subtle escape mechanism follows
%lookup = getelementptr [2 x i1], [2 x i1]* @lookup_table, i32 0, i32 %bit
%val = load i1, i1* %lookup
ret i1 %val
}
declare void @throw_if_bit_set(i8*, i8) readonly
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; FNATTR: define i1 @c6(i8* readonly %q, i8 %bit)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define i1 @c6(i8* %q, i8 %bit) personality i32 (...)* @__gxx_personality_v0 {
invoke void @throw_if_bit_set(i8* %q, i8 %bit)
to label %ret0 unwind label %ret1
ret0:
ret i1 0
ret1:
%exn = landingpad {i8*, i32}
cleanup
ret i1 1
}
declare i32 @__gxx_personality_v0(...)
define i1* @lookup_bit(i32* %q, i32 %bitno) readnone nounwind {
%tmp = ptrtoint i32* %q to i32
%tmp2 = lshr i32 %tmp, %bitno
%bit = and i32 %tmp2, 1
%lookup = getelementptr [2 x i1], [2 x i1]* @lookup_table, i32 0, i32 %bit
ret i1* %lookup
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define i1 @c7(i32* readonly %q, i32 %bitno)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define i1 @c7(i32* %q, i32 %bitno) {
%ptr = call i1* @lookup_bit(i32* %q, i32 %bitno)
%val = load i1, i1* %ptr
ret i1 %val
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define i32 @nc1(i32* %q, i32* nocapture %p, i1 %b)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define i32 @nc1(i32* %q, i32* %p, i1 %b) {
e:
br label %l
l:
%x = phi i32* [ %p, %e ]
%y = phi i32* [ %q, %e ]
%tmp = bitcast i32* %x to i32* ; <i32*> [#uses=2]
%tmp2 = select i1 %b, i32* %tmp, i32* %y
%val = load i32, i32* %tmp2 ; <i32> [#uses=1]
store i32 0, i32* %tmp
store i32* %y, i32** @g
ret i32 %val
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define i32 @nc1_addrspace(i32* %q, i32 addrspace(1)* nocapture %p, i1 %b)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define i32 @nc1_addrspace(i32* %q, i32 addrspace(1)* %p, i1 %b) {
e:
br label %l
l:
%x = phi i32 addrspace(1)* [ %p, %e ]
%y = phi i32* [ %q, %e ]
%tmp = addrspacecast i32 addrspace(1)* %x to i32* ; <i32*> [#uses=2]
%tmp2 = select i1 %b, i32* %tmp, i32* %y
%val = load i32, i32* %tmp2 ; <i32> [#uses=1]
store i32 0, i32* %tmp
store i32* %y, i32** @g
ret i32 %val
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define void @nc2(i32* nocapture %p, i32* %q)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @nc2(i32* %p, i32* %q) {
%1 = call i32 @nc1(i32* %q, i32* %p, i1 0) ; <i32> [#uses=0]
ret void
}
[Attributor][MustExec] Deduce dereferenceable and nonnull attribute using MustBeExecutedContextExplorer Summary: In D65186 and related patches, MustBeExecutedContextExplorer is introduced. This enables us to traverse instructions guaranteed to execute from function entry. If we can know the argument is used as `dereferenceable` or `nonnull` in these instructions, we can mark `dereferenceable` or `nonnull` in the argument definition: 1. Memory instruction (similar to D64258) Trace memory instruction pointer operand. Currently, only inbounds GEPs are traced. ``` define i64* @f(i64* %a) { entry: %add.ptr = getelementptr inbounds i64, i64* %a, i64 1 ; (because of inbounds GEP we can know that %a is at least dereferenceable(16)) store i64 1, i64* %add.ptr, align 8 ret i64* %add.ptr ; dereferenceable 8 (because above instruction stores into it) } ``` 2. Propagation from callsite (similar to D27855) If `deref` or `nonnull` are known in call site parameter attributes we can also say that argument also that attribute. ``` declare void @use3(i8* %x, i8* %y, i8* %z); declare void @use3nonnull(i8* nonnull %x, i8* nonnull %y, i8* nonnull %z); define void @parent1(i8* %a, i8* %b, i8* %c) { call void @use3nonnull(i8* %b, i8* %c, i8* %a) ; Above instruction is always executed so we can say that@parent1(i8* nonnnull %a, i8* nonnull %b, i8* nonnull %c) call void @use3(i8* %c, i8* %a, i8* %b) ret void } ``` Reviewers: jdoerfert, sstefan1, spatel, reames Reviewed By: jdoerfert Subscribers: xbolva00, hiraditya, jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D65402 llvm-svn: 374063
2019-10-08 23:25:56 +08:00
[funcattrs] Consistently treat calling a function pointer as a non-capturing read We were being wildly inconsistent about what memory access was implied by an indirect function call. Depending on the call site attributes, you could get anything from a read, to unknown, to none at all. (The last was a miscompile.) We were also always traversing the uses of a readonly indirect call. This is entirely unneeded as the indirect call does not capture. The callee might capture itself internally, but that has no implications for this caller. (See the nice explanation in the CaptureTracking comments if that case is confusing.) Note that elsewhere in the same file, we were correctly computing the nocapture attribute for indirect calls. The changed case only resulted in conservatism when computing memory attributes if say the return value was written to. Differential Revision: https://reviews.llvm.org/D115916
2021-12-18 00:32:43 +08:00
; FNATTR: define void @nc3(void ()* nocapture readonly %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @nc3(void ()* %p) {
call void %p()
ret void
}
declare void @external(i8*) readonly nounwind
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; FNATTR: define void @nc4(i8* nocapture readonly %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @nc4(i8* %p) {
call void @external(i8* %p)
ret void
}
[funcattrs] Consistently treat calling a function pointer as a non-capturing read We were being wildly inconsistent about what memory access was implied by an indirect function call. Depending on the call site attributes, you could get anything from a read, to unknown, to none at all. (The last was a miscompile.) We were also always traversing the uses of a readonly indirect call. This is entirely unneeded as the indirect call does not capture. The callee might capture itself internally, but that has no implications for this caller. (See the nice explanation in the CaptureTracking comments if that case is confusing.) Note that elsewhere in the same file, we were correctly computing the nocapture attribute for indirect calls. The changed case only resulted in conservatism when computing memory attributes if say the return value was written to. Differential Revision: https://reviews.llvm.org/D115916
2021-12-18 00:32:43 +08:00
; FNATTR: define void @nc5(void (i8*)* nocapture readonly %f, i8* nocapture %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @nc5(void (i8*)* %f, i8* %p) {
call void %f(i8* %p) readonly nounwind
call void %f(i8* nocapture %p)
ret void
}
[Attributor] Deduce "no-capture" argument attribute Add the no-capture argument attribute deduction to the Attributor fixpoint framework. The new string attributed "no-capture-maybe-returned" is introduced to allow deduction of no-capture through functions that "capture" an argument but only by "returning" it. It is only used by the Attributor for testing. Differential Revision: https://reviews.llvm.org/D59922 llvm-svn: 370817
2019-09-04 04:37:24 +08:00
; FNATTR: define void @test1_1(i8* nocapture readnone %x1_1, i8* %y1_1, i1 %c)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
; It would be acceptable to add readnone to %y1_1 and %y1_2.
[Attributor] Deduce "no-capture" argument attribute Add the no-capture argument attribute deduction to the Attributor fixpoint framework. The new string attributed "no-capture-maybe-returned" is introduced to allow deduction of no-capture through functions that "capture" an argument but only by "returning" it. It is only used by the Attributor for testing. Differential Revision: https://reviews.llvm.org/D59922 llvm-svn: 370817
2019-09-04 04:37:24 +08:00
define void @test1_1(i8* %x1_1, i8* %y1_1, i1 %c) {
call i8* @test1_2(i8* %x1_1, i8* %y1_1, i1 %c)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
store i32* null, i32** @g
ret void
}
[Attributor] Deduce "no-capture" argument attribute Add the no-capture argument attribute deduction to the Attributor fixpoint framework. The new string attributed "no-capture-maybe-returned" is introduced to allow deduction of no-capture through functions that "capture" an argument but only by "returning" it. It is only used by the Attributor for testing. Differential Revision: https://reviews.llvm.org/D59922 llvm-svn: 370817
2019-09-04 04:37:24 +08:00
; FNATTR: define i8* @test1_2(i8* nocapture readnone %x1_2, i8* returned %y1_2, i1 %c)
define i8* @test1_2(i8* %x1_2, i8* %y1_2, i1 %c) {
br i1 %c, label %t, label %f
t:
call void @test1_1(i8* %x1_2, i8* %y1_2, i1 %c)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
store i32* null, i32** @g
[Attributor] Deduce "no-capture" argument attribute Add the no-capture argument attribute deduction to the Attributor fixpoint framework. The new string attributed "no-capture-maybe-returned" is introduced to allow deduction of no-capture through functions that "capture" an argument but only by "returning" it. It is only used by the Attributor for testing. Differential Revision: https://reviews.llvm.org/D59922 llvm-svn: 370817
2019-09-04 04:37:24 +08:00
br label %f
f:
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
ret i8* %y1_2
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define void @test2(i8* nocapture readnone %x2)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @test2(i8* %x2) {
call void @test2(i8* %x2)
store i32* null, i32** @g
ret void
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define void @test3(i8* nocapture readnone %x3, i8* nocapture readnone %y3, i8* nocapture readnone %z3)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @test3(i8* %x3, i8* %y3, i8* %z3) {
call void @test3(i8* %z3, i8* %y3, i8* %x3)
store i32* null, i32** @g
ret void
}
[Attributor] Deduce "no-capture" argument attribute Add the no-capture argument attribute deduction to the Attributor fixpoint framework. The new string attributed "no-capture-maybe-returned" is introduced to allow deduction of no-capture through functions that "capture" an argument but only by "returning" it. It is only used by the Attributor for testing. Differential Revision: https://reviews.llvm.org/D59922 llvm-svn: 370817
2019-09-04 04:37:24 +08:00
; FNATTR: define void @test4_1(i8* %x4_1, i1 %c)
define void @test4_1(i8* %x4_1, i1 %c) {
call i8* @test4_2(i8* %x4_1, i8* %x4_1, i8* %x4_1, i1 %c)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
store i32* null, i32** @g
ret void
}
[Attributor] Deduce "no-capture" argument attribute Add the no-capture argument attribute deduction to the Attributor fixpoint framework. The new string attributed "no-capture-maybe-returned" is introduced to allow deduction of no-capture through functions that "capture" an argument but only by "returning" it. It is only used by the Attributor for testing. Differential Revision: https://reviews.llvm.org/D59922 llvm-svn: 370817
2019-09-04 04:37:24 +08:00
; FNATTR: define i8* @test4_2(i8* nocapture readnone %x4_2, i8* readnone returned %y4_2, i8* nocapture readnone %z4_2, i1 %c)
define i8* @test4_2(i8* %x4_2, i8* %y4_2, i8* %z4_2, i1 %c) {
br i1 %c, label %t, label %f
t:
call void @test4_1(i8* null, i1 %c)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
store i32* null, i32** @g
[Attributor] Deduce "no-capture" argument attribute Add the no-capture argument attribute deduction to the Attributor fixpoint framework. The new string attributed "no-capture-maybe-returned" is introduced to allow deduction of no-capture through functions that "capture" an argument but only by "returning" it. It is only used by the Attributor for testing. Differential Revision: https://reviews.llvm.org/D59922 llvm-svn: 370817
2019-09-04 04:37:24 +08:00
br label %f
f:
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
ret i8* %y4_2
}
declare i8* @test5_1(i8* %x5_1)
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; FNATTR: define void @test5_2(i8* %x5_2)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @test5_2(i8* %x5_2) {
call i8* @test5_1(i8* %x5_2)
store i32* null, i32** @g
ret void
}
declare void @test6_1(i8* %x6_1, i8* nocapture %y6_1, ...)
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; FNATTR: define void @test6_2(i8* %x6_2, i8* nocapture %y6_2, i8* %z6_2)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @test6_2(i8* %x6_2, i8* %y6_2, i8* %z6_2) {
call void (i8*, i8*, ...) @test6_1(i8* %x6_2, i8* %y6_2, i8* %z6_2)
store i32* null, i32** @g
ret void
}
[Attributor][MustExec] Deduce dereferenceable and nonnull attribute using MustBeExecutedContextExplorer Summary: In D65186 and related patches, MustBeExecutedContextExplorer is introduced. This enables us to traverse instructions guaranteed to execute from function entry. If we can know the argument is used as `dereferenceable` or `nonnull` in these instructions, we can mark `dereferenceable` or `nonnull` in the argument definition: 1. Memory instruction (similar to D64258) Trace memory instruction pointer operand. Currently, only inbounds GEPs are traced. ``` define i64* @f(i64* %a) { entry: %add.ptr = getelementptr inbounds i64, i64* %a, i64 1 ; (because of inbounds GEP we can know that %a is at least dereferenceable(16)) store i64 1, i64* %add.ptr, align 8 ret i64* %add.ptr ; dereferenceable 8 (because above instruction stores into it) } ``` 2. Propagation from callsite (similar to D27855) If `deref` or `nonnull` are known in call site parameter attributes we can also say that argument also that attribute. ``` declare void @use3(i8* %x, i8* %y, i8* %z); declare void @use3nonnull(i8* nonnull %x, i8* nonnull %y, i8* nonnull %z); define void @parent1(i8* %a, i8* %b, i8* %c) { call void @use3nonnull(i8* %b, i8* %c, i8* %a) ; Above instruction is always executed so we can say that@parent1(i8* nonnnull %a, i8* nonnull %b, i8* nonnull %c) call void @use3(i8* %c, i8* %a, i8* %b) ret void } ``` Reviewers: jdoerfert, sstefan1, spatel, reames Reviewed By: jdoerfert Subscribers: xbolva00, hiraditya, jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D65402 llvm-svn: 374063
2019-10-08 23:25:56 +08:00
; FNATTR: define void @test_cmpxchg(i32* nocapture %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @test_cmpxchg(i32* %p) {
cmpxchg i32* %p, i32 0, i32 1 acquire monotonic
ret void
}
[Attributor][MustExec] Deduce dereferenceable and nonnull attribute using MustBeExecutedContextExplorer Summary: In D65186 and related patches, MustBeExecutedContextExplorer is introduced. This enables us to traverse instructions guaranteed to execute from function entry. If we can know the argument is used as `dereferenceable` or `nonnull` in these instructions, we can mark `dereferenceable` or `nonnull` in the argument definition: 1. Memory instruction (similar to D64258) Trace memory instruction pointer operand. Currently, only inbounds GEPs are traced. ``` define i64* @f(i64* %a) { entry: %add.ptr = getelementptr inbounds i64, i64* %a, i64 1 ; (because of inbounds GEP we can know that %a is at least dereferenceable(16)) store i64 1, i64* %add.ptr, align 8 ret i64* %add.ptr ; dereferenceable 8 (because above instruction stores into it) } ``` 2. Propagation from callsite (similar to D27855) If `deref` or `nonnull` are known in call site parameter attributes we can also say that argument also that attribute. ``` declare void @use3(i8* %x, i8* %y, i8* %z); declare void @use3nonnull(i8* nonnull %x, i8* nonnull %y, i8* nonnull %z); define void @parent1(i8* %a, i8* %b, i8* %c) { call void @use3nonnull(i8* %b, i8* %c, i8* %a) ; Above instruction is always executed so we can say that@parent1(i8* nonnnull %a, i8* nonnull %b, i8* nonnull %c) call void @use3(i8* %c, i8* %a, i8* %b) ret void } ``` Reviewers: jdoerfert, sstefan1, spatel, reames Reviewed By: jdoerfert Subscribers: xbolva00, hiraditya, jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D65402 llvm-svn: 374063
2019-10-08 23:25:56 +08:00
; FNATTR: define void @test_cmpxchg_ptr(i32** nocapture %p, i32* %q)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @test_cmpxchg_ptr(i32** %p, i32* %q) {
cmpxchg i32** %p, i32* null, i32* %q acquire monotonic
ret void
}
[Attributor][MustExec] Deduce dereferenceable and nonnull attribute using MustBeExecutedContextExplorer Summary: In D65186 and related patches, MustBeExecutedContextExplorer is introduced. This enables us to traverse instructions guaranteed to execute from function entry. If we can know the argument is used as `dereferenceable` or `nonnull` in these instructions, we can mark `dereferenceable` or `nonnull` in the argument definition: 1. Memory instruction (similar to D64258) Trace memory instruction pointer operand. Currently, only inbounds GEPs are traced. ``` define i64* @f(i64* %a) { entry: %add.ptr = getelementptr inbounds i64, i64* %a, i64 1 ; (because of inbounds GEP we can know that %a is at least dereferenceable(16)) store i64 1, i64* %add.ptr, align 8 ret i64* %add.ptr ; dereferenceable 8 (because above instruction stores into it) } ``` 2. Propagation from callsite (similar to D27855) If `deref` or `nonnull` are known in call site parameter attributes we can also say that argument also that attribute. ``` declare void @use3(i8* %x, i8* %y, i8* %z); declare void @use3nonnull(i8* nonnull %x, i8* nonnull %y, i8* nonnull %z); define void @parent1(i8* %a, i8* %b, i8* %c) { call void @use3nonnull(i8* %b, i8* %c, i8* %a) ; Above instruction is always executed so we can say that@parent1(i8* nonnnull %a, i8* nonnull %b, i8* nonnull %c) call void @use3(i8* %c, i8* %a, i8* %b) ret void } ``` Reviewers: jdoerfert, sstefan1, spatel, reames Reviewed By: jdoerfert Subscribers: xbolva00, hiraditya, jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D65402 llvm-svn: 374063
2019-10-08 23:25:56 +08:00
; FNATTR: define void @test_atomicrmw(i32* nocapture %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @test_atomicrmw(i32* %p) {
atomicrmw add i32* %p, i32 1 seq_cst
ret void
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define void @test_volatile(i32* %x)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @test_volatile(i32* %x) {
entry:
%gep = getelementptr i32, i32* %x, i64 1
store volatile i32 0, i32* %gep, align 4
ret void
}
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; FNATTR: nocaptureLaunder(i8* nocapture %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @nocaptureLaunder(i8* %p) {
entry:
%b = call i8* @llvm.launder.invariant.group.p0i8(i8* %p)
store i8 42, i8* %b
ret void
}
@g2 = global i8* null
[Attributor] Copy or port test cases related to Attributor to` Attributor` test folder Summary: This patch moves the test cases related to Attributor to `Transforms/Attributor` folder. We have used `Transforms/FunctionAttrs` as the primary folder for Attributor test but we need to change testing way now. For the test cases which I think functionattrs doesn't infer anything something like (willreturn, nosync, value-simplify, h2s ..etc), I moved them with the command `git mv`. For the test cases in which functoinattrs and attributor are tested, I copied the test to the folder and remove the check only used by functoinattrs. Reviewers: jdoerfert, sstefan1 Reviewed By: jdoerfert Subscribers: jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D70843
2019-12-02 21:40:09 +08:00
; FNATTR: define void @captureLaunder(i8* %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @captureLaunder(i8* %p) {
%b = call i8* @llvm.launder.invariant.group.p0i8(i8* %p)
store i8* %b, i8** @g2
ret void
}
[funcattrs] Infer writeonly argument attribute This change extends the current logic for inferring readonly and readnone argument attributes to also infer writeonly. This change is deliberately minimal; there's a couple of areas for follow up. * I left out all call handling and thus any benefit from the SCC walk. When examining the test changes, I realized the existing code is imprecise, and am going to fix that in it's own revision before adding in the writeonly handling. (Mostly because updating the tests is hard when I, the human, can't figure out whether the result is correct.) * I left out handling for storing a value (as opposed to storing to a pointer). This should benefit readonly/readnone as well, and applies to a bunch of other instructions. Seemed worth having as a separate review. Differential Revision: https://reviews.llvm.org/D114963
2021-12-03 04:56:09 +08:00
; FNATTR: @nocaptureStrip(i8* nocapture writeonly %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @nocaptureStrip(i8* %p) {
entry:
%b = call i8* @llvm.strip.invariant.group.p0i8(i8* %p)
store i8 42, i8* %b
ret void
}
@g3 = global i8* null
[funcattrs] Fix a bug in recently introduced writeonly argument inference This fixes a bug in 740057d. There's two ways to describe the issue: * One caller hasn't yet proven nocapture on the argument. Given that, the inference routine is responsible for bailing out on a potential capture. * Even if we know the argument is nocapture, the access inference needs to traverse the exact set of users the capture tracking would (or exit conservatively). Even if capture tracking can prove a store is non-capturing (e.g. to a local alloc which doesn't escape), we still need to track the copy of the pointer to see if it's later reloaded and accessed again. Note that all the test changes except the newly added ones appear to be false negatives. That is, cases where we could prove writeonly, but the current code isn't strong enough. That's why I didn't spot this originally.
2021-12-04 00:52:40 +08:00
; FNATTR: define void @captureStrip(i8* %p)
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
define void @captureStrip(i8* %p) {
%b = call i8* @llvm.strip.invariant.group.p0i8(i8* %p)
store i8* %b, i8** @g3
ret void
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define i1 @captureICmp(i32* readnone %x)
[CaptureTracking] Don't let comparisons against null escape inbounds pointers Pointers that are in-bounds (either through dereferenceable_or_null or thorough a getelementptr inbounds) cannot be captured with a comparison against null. There is no way to construct a pointer that is still in bounds but also NULL. This helps safe languages that insert null checks before load/store instructions. Without this patch, almost all pointers would be considered captured even for simple loads. With this patch, an icmp with null will not be seen as escaping as long as certain conditions are met. There was a lot of discussion about this patch. See the Phabricator thread for detals. Differential Revision: https://reviews.llvm.org/D60047 llvm-svn: 362900
2019-06-09 18:20:33 +08:00
define i1 @captureICmp(i32* %x) {
%1 = icmp eq i32* %x, null
ret i1 %1
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define i1 @captureICmpRev(i32* readnone %x)
[CaptureTracking] Allow null to be in either icmp operand Summary: Before we required the comparison against null to be "canonical", hence null to be operand #1. This patch allows null to be in either operand, similar to the handling of loaded globals that follows. Reviewers: sanjoy, hfinkel, aykevl, sstefan1, uenoku Subscribers: hiraditya, bollu, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D66321 llvm-svn: 369158
2019-08-17 05:53:49 +08:00
define i1 @captureICmpRev(i32* %x) {
%1 = icmp eq i32* null, %x
ret i1 %1
}
[Attributor] Really use the executed-context Before we did not follow casts and geps when we looked at the users of a pointer in the pointers must-be-executed-context. This caused us to fail to determine if it was accessed for sure. With this change we follow such users now. The above extension exposed problems in getKnownNonNullAndDerefBytesForUse which did not always check what the base pointer was. We also did not handle negative offsets as conservative as we have to without explicit loop handling. Finally, we should not derive a huge number if we access a pointer that was traversed backwards first. The problems exposed by this functional change are already tested in the existing test cases as is the functional change. Differential Revision: https://reviews.llvm.org/D69647
2019-10-31 06:21:53 +08:00
; FNATTR: define i1 @nocaptureInboundsGEPICmp(i32* nocapture readnone %x)
[CaptureTracking] Don't let comparisons against null escape inbounds pointers Pointers that are in-bounds (either through dereferenceable_or_null or thorough a getelementptr inbounds) cannot be captured with a comparison against null. There is no way to construct a pointer that is still in bounds but also NULL. This helps safe languages that insert null checks before load/store instructions. Without this patch, almost all pointers would be considered captured even for simple loads. With this patch, an icmp with null will not be seen as escaping as long as certain conditions are met. There was a lot of discussion about this patch. See the Phabricator thread for detals. Differential Revision: https://reviews.llvm.org/D60047 llvm-svn: 362900
2019-06-09 18:20:33 +08:00
define i1 @nocaptureInboundsGEPICmp(i32* %x) {
%1 = getelementptr inbounds i32, i32* %x, i32 5
%2 = bitcast i32* %1 to i8*
%3 = icmp eq i8* %2, null
ret i1 %3
}
[Attributor] Really use the executed-context Before we did not follow casts and geps when we looked at the users of a pointer in the pointers must-be-executed-context. This caused us to fail to determine if it was accessed for sure. With this change we follow such users now. The above extension exposed problems in getKnownNonNullAndDerefBytesForUse which did not always check what the base pointer was. We also did not handle negative offsets as conservative as we have to without explicit loop handling. Finally, we should not derive a huge number if we access a pointer that was traversed backwards first. The problems exposed by this functional change are already tested in the existing test cases as is the functional change. Differential Revision: https://reviews.llvm.org/D69647
2019-10-31 06:21:53 +08:00
; FNATTR: define i1 @nocaptureInboundsGEPICmpRev(i32* nocapture readnone %x)
[CaptureTracking] Allow null to be in either icmp operand Summary: Before we required the comparison against null to be "canonical", hence null to be operand #1. This patch allows null to be in either operand, similar to the handling of loaded globals that follows. Reviewers: sanjoy, hfinkel, aykevl, sstefan1, uenoku Subscribers: hiraditya, bollu, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D66321 llvm-svn: 369158
2019-08-17 05:53:49 +08:00
define i1 @nocaptureInboundsGEPICmpRev(i32* %x) {
%1 = getelementptr inbounds i32, i32* %x, i32 5
%2 = bitcast i32* %1 to i8*
%3 = icmp eq i8* null, %2
ret i1 %3
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define i1 @nocaptureDereferenceableOrNullICmp(i32* nocapture readnone dereferenceable_or_null(4) %x)
[CaptureTracking] Don't let comparisons against null escape inbounds pointers Pointers that are in-bounds (either through dereferenceable_or_null or thorough a getelementptr inbounds) cannot be captured with a comparison against null. There is no way to construct a pointer that is still in bounds but also NULL. This helps safe languages that insert null checks before load/store instructions. Without this patch, almost all pointers would be considered captured even for simple loads. With this patch, an icmp with null will not be seen as escaping as long as certain conditions are met. There was a lot of discussion about this patch. See the Phabricator thread for detals. Differential Revision: https://reviews.llvm.org/D60047 llvm-svn: 362900
2019-06-09 18:20:33 +08:00
define i1 @nocaptureDereferenceableOrNullICmp(i32* dereferenceable_or_null(4) %x) {
%1 = bitcast i32* %x to i8*
%2 = icmp eq i8* %1, null
ret i1 %2
}
NoFree argument attribute. Summary: Deducing nofree atrribute for function arguments. Reviewers: jdoerfert Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D67886
2019-11-02 23:35:38 +08:00
; FNATTR: define i1 @captureDereferenceableOrNullICmp(i32* readnone dereferenceable_or_null(4) %x)
[IR] Convert null-pointer-is-valid into an enum attribute The "null-pointer-is-valid" attribute needs to be checked by many pointer-related combines. To make the check more efficient, convert it from a string into an enum attribute. In the future, this attribute may be replaced with data layout properties. Differential Revision: https://reviews.llvm.org/D78862
2020-04-25 18:57:07 +08:00
define i1 @captureDereferenceableOrNullICmp(i32* dereferenceable_or_null(4) %x) null_pointer_is_valid {
[CaptureTracking] Don't let comparisons against null escape inbounds pointers Pointers that are in-bounds (either through dereferenceable_or_null or thorough a getelementptr inbounds) cannot be captured with a comparison against null. There is no way to construct a pointer that is still in bounds but also NULL. This helps safe languages that insert null checks before load/store instructions. Without this patch, almost all pointers would be considered captured even for simple loads. With this patch, an icmp with null will not be seen as escaping as long as certain conditions are met. There was a lot of discussion about this patch. See the Phabricator thread for detals. Differential Revision: https://reviews.llvm.org/D60047 llvm-svn: 362900
2019-06-09 18:20:33 +08:00
%1 = bitcast i32* %x to i8*
%2 = icmp eq i8* %1, null
ret i1 %2
}
[funcattrs] Add some additional indirect call tests
2021-12-17 08:38:40 +08:00
declare void @capture(i8*)
[funcattrs] Consistently treat calling a function pointer as a non-capturing read We were being wildly inconsistent about what memory access was implied by an indirect function call. Depending on the call site attributes, you could get anything from a read, to unknown, to none at all. (The last was a miscompile.) We were also always traversing the uses of a readonly indirect call. This is entirely unneeded as the indirect call does not capture. The callee might capture itself internally, but that has no implications for this caller. (See the nice explanation in the CaptureTracking comments if that case is confusing.) Note that elsewhere in the same file, we were correctly computing the nocapture attribute for indirect calls. The changed case only resulted in conservatism when computing memory attributes if say the return value was written to. Differential Revision: https://reviews.llvm.org/D115916
2021-12-18 00:32:43 +08:00
; FNATTR: define void @nocapture_fptr(i8* (i8*)* nocapture readonly %f, i8* %p)
[funcattrs] Add some additional indirect call tests
2021-12-17 08:38:40 +08:00
define void @nocapture_fptr(i8* (i8*)* %f, i8* %p) {
%res = call i8* %f(i8* %p)
call void @capture(i8* %res)
ret void
}
[funcattrs] Consistently treat calling a function pointer as a non-capturing read We were being wildly inconsistent about what memory access was implied by an indirect function call. Depending on the call site attributes, you could get anything from a read, to unknown, to none at all. (The last was a miscompile.) We were also always traversing the uses of a readonly indirect call. This is entirely unneeded as the indirect call does not capture. The callee might capture itself internally, but that has no implications for this caller. (See the nice explanation in the CaptureTracking comments if that case is confusing.) Note that elsewhere in the same file, we were correctly computing the nocapture attribute for indirect calls. The changed case only resulted in conservatism when computing memory attributes if say the return value was written to. Differential Revision: https://reviews.llvm.org/D115916
2021-12-18 00:32:43 +08:00
; FNATTR: define void @recurse_fptr(i8* (i8*)* nocapture readonly %f, i8* %p)
[funcattrs] Add some additional indirect call tests
2021-12-17 08:38:40 +08:00
define void @recurse_fptr(i8* (i8*)* %f, i8* %p) {
%res = call i8* %f(i8* %p)
store i8 0, i8* %res
ret void
}
[funcattrs] Consistently treat calling a function pointer as a non-capturing read We were being wildly inconsistent about what memory access was implied by an indirect function call. Depending on the call site attributes, you could get anything from a read, to unknown, to none at all. (The last was a miscompile.) We were also always traversing the uses of a readonly indirect call. This is entirely unneeded as the indirect call does not capture. The callee might capture itself internally, but that has no implications for this caller. (See the nice explanation in the CaptureTracking comments if that case is confusing.) Note that elsewhere in the same file, we were correctly computing the nocapture attribute for indirect calls. The changed case only resulted in conservatism when computing memory attributes if say the return value was written to. Differential Revision: https://reviews.llvm.org/D115916
2021-12-18 00:32:43 +08:00
; FNATTR: define void @readnone_indirec(void (i8*)* nocapture readonly %f, i8* readnone %p)
[funcattrs] Add some additional indirect call tests
2021-12-17 08:38:40 +08:00
define void @readnone_indirec(void (i8*)* %f, i8* %p) {
call void %f(i8* %p) readnone
ret void
}
Revert "Temporarily Revert "Add basic loop fusion pass."" The reversion apparently deleted the test/Transforms directory. Will be re-reverting again. llvm-svn: 358552
2019-04-17 12:52:47 +08:00
declare i8* @llvm.launder.invariant.group.p0i8(i8*)
declare i8* @llvm.strip.invariant.group.p0i8(i8*)