github
/
tinygo
mirror of https://github.com/tinygo-org/tinygo.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
2997 Commits
99 Branches
41 Tags
26 MiB
Tree: 91104b2f27
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '91104b2f27'
${ noResults }
tinygo/compiler/testdata/channel.ll

131 lines
7.3 KiB
Raw Normal View History

compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
; ModuleID = 'channel.go'
source_filename = "channel.go"
all: update to LLVM 14 Switch over to LLVM 14 for static builds. Keep using LLVM 13 for regular builds for now. This uses a branch of the upstream Espressif branch to fix an issue, see: https://github.com/espressif/llvm-project/pull/59
3 years ago
target datalayout = "e-m:e-p:32:32-p10:8:8-p20:8:8-i64:64-n32:64-S128-ni:1:10:20"
build: normalize target triples to match Clang This commit changes a target triple like "armv6m-none-eabi" to "armv6m-unknown-unknow-eabi". The reason is that while the former is correctly parsed in Clang (due to normalization), it wasn't parsed correctly in LLVM meaning that the environment wasn't set to EABI. This change normalizes all target triples and uses the EABI environment (-eabi in the triple) for Cortex-M targets. This change also drops the `--target=` flag in the target JSON files, the flag is now added implicitly in `(*compileopts.Config).CFlags()`. This removes some duplication in target JSON files. Unfortunately, this change also increases code size for Cortex-M targets. It looks like LLVM now emits calls like __aeabi_memmove instead of memmove, which pull in slightly more code (they basically just call the regular C functions) and the calls themself don't seem to be as efficient as they could be. Perhaps this is a LLVM bug that will be fixed in the future, as this is a very common occurrence.
3 years ago
target triple = "wasm32-unknown-wasi"
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
%runtime.channel = type { i32, i32, i8, %runtime.channelBlockedList*, i32, i32, i32, i8* }
%runtime.channelBlockedList = type { %runtime.channelBlockedList*, %"internal/task.Task"*, %runtime.chanSelectState*, { %runtime.channelBlockedList*, i32, i32 } }
runtime: move *task.DeferFrame here This is a refactor that makes the next commit simpler. Perhaps it should have been like this from the beginning but I didn't like all the casts.
2 years ago
%"internal/task.Task" = type { %"internal/task.Task"*, i8*, i64, %"internal/task.gcData", %"internal/task.state", i8* }
internal/task: swap stack chain when switching goroutines This change swaps the stack chain when switching goroutines, ensuring that the chain is maintained consistently. This is only really currently necessary with asyncify on wasm.
3 years ago
%"internal/task.gcData" = type { i8* }
internal/task: use asyncify on webassembly This change implements a new "scheduler" for WebAssembly using binaryen's asyncify transform. This is more reliable than the current "coroutines" transform, and works with non-Go code in the call stack. runtime (js/wasm): handle scheduler nesting If WASM calls into JS which calls back into WASM, it is possible for the scheduler to nest. The event from the callback must be handled immediately, so the task cannot simply be deferred to the outer scheduler. This creates a minimal scheduler loop which is used to handle such nesting.
3 years ago
%"internal/task.state" = type { i32, i8*, %"internal/task.stackState", i1 }
%"internal/task.stackState" = type { i32, i32 }
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
%runtime.chanSelectState = type { %runtime.channel*, i8* }
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*) #0
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
declare void @runtime.trackPointer(i8* nocapture readonly, i8*) #0
compiler: predeclare runtime.trackPointer When a package only uses runtime.trackPointer to create interface packs, the compiler fails to find runtime.trackPointer. This change predeclares it alongside runtime.alloc and updates the tests to use runtime.trackPointer when the test's target uses it.
3 years ago
compiler: add nounwind attribute This attribute is also set by Clang when it compiles C source files (unless -fexceptions is set). The advantage is that no unwind tables are emitted on Linux (and perhaps other systems). It also avoids __aeabi_unwind_cpp_pr0 on ARM when using the musl libc.
3 years ago
; Function Attrs: nounwind
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
define hidden void @main.init(i8* %context) unnamed_addr #1 {
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
entry:
ret void
}
compiler: add nounwind attribute This attribute is also set by Clang when it compiles C source files (unless -fexceptions is set). The advantage is that no unwind tables are emitted on Linux (and perhaps other systems). It also avoids __aeabi_unwind_cpp_pr0 on ARM when using the musl libc.
3 years ago
; Function Attrs: nounwind
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
define hidden void @main.chanIntSend(%runtime.channel* dereferenceable_or_null(32) %ch, i8* %context) unnamed_addr #1 {
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
entry:
%chan.blockedList = alloca %runtime.channelBlockedList, align 8
%chan.value = alloca i32, align 4
%chan.value.bitcast = bitcast i32* %chan.value to i8*
call void @llvm.lifetime.start.p0i8(i64 4, i8* nonnull %chan.value.bitcast)
store i32 3, i32* %chan.value, align 4
%chan.blockedList.bitcast = bitcast %runtime.channelBlockedList* %chan.blockedList to i8*
call void @llvm.lifetime.start.p0i8(i64 24, i8* nonnull %chan.blockedList.bitcast)
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
call void @runtime.chanSend(%runtime.channel* %ch, i8* nonnull %chan.value.bitcast, %runtime.channelBlockedList* nonnull %chan.blockedList, i8* undef) #3
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
call void @llvm.lifetime.end.p0i8(i64 24, i8* nonnull %chan.blockedList.bitcast)
call void @llvm.lifetime.end.p0i8(i64 4, i8* nonnull %chan.value.bitcast)
ret void
}
all: add LLVM 12 support Originally based on a PR by @QuLogic, but extended a lot to get all tests to pass.
4 years ago
; Function Attrs: argmemonly nofree nosync nounwind willreturn
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture) #2
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
declare void @runtime.chanSend(%runtime.channel* dereferenceable_or_null(32), i8*, %runtime.channelBlockedList* dereferenceable_or_null(24), i8*) #0
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
all: add LLVM 12 support Originally based on a PR by @QuLogic, but extended a lot to get all tests to pass.
4 years ago
; Function Attrs: argmemonly nofree nosync nounwind willreturn
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture) #2
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
compiler: add nounwind attribute This attribute is also set by Clang when it compiles C source files (unless -fexceptions is set). The advantage is that no unwind tables are emitted on Linux (and perhaps other systems). It also avoids __aeabi_unwind_cpp_pr0 on ARM when using the musl libc.
3 years ago
; Function Attrs: nounwind
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
define hidden void @main.chanIntRecv(%runtime.channel* dereferenceable_or_null(32) %ch, i8* %context) unnamed_addr #1 {
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
entry:
%chan.blockedList = alloca %runtime.channelBlockedList, align 8
%chan.value = alloca i32, align 4
%chan.value.bitcast = bitcast i32* %chan.value to i8*
call void @llvm.lifetime.start.p0i8(i64 4, i8* nonnull %chan.value.bitcast)
%chan.blockedList.bitcast = bitcast %runtime.channelBlockedList* %chan.blockedList to i8*
call void @llvm.lifetime.start.p0i8(i64 24, i8* nonnull %chan.blockedList.bitcast)
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
%0 = call i1 @runtime.chanRecv(%runtime.channel* %ch, i8* nonnull %chan.value.bitcast, %runtime.channelBlockedList* nonnull %chan.blockedList, i8* undef) #3
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
call void @llvm.lifetime.end.p0i8(i64 4, i8* nonnull %chan.value.bitcast)
call void @llvm.lifetime.end.p0i8(i64 24, i8* nonnull %chan.blockedList.bitcast)
ret void
}
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
declare i1 @runtime.chanRecv(%runtime.channel* dereferenceable_or_null(32), i8*, %runtime.channelBlockedList* dereferenceable_or_null(24), i8*) #0
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
compiler: add nounwind attribute This attribute is also set by Clang when it compiles C source files (unless -fexceptions is set). The advantage is that no unwind tables are emitted on Linux (and perhaps other systems). It also avoids __aeabi_unwind_cpp_pr0 on ARM when using the musl libc.
3 years ago
; Function Attrs: nounwind
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
define hidden void @main.chanZeroSend(%runtime.channel* dereferenceable_or_null(32) %ch, i8* %context) unnamed_addr #1 {
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
entry:
compiler: predeclare runtime.trackPointer When a package only uses runtime.trackPointer to create interface packs, the compiler fails to find runtime.trackPointer. This change predeclares it alongside runtime.alloc and updates the tests to use runtime.trackPointer when the test's target uses it.
3 years ago
%complit = alloca {}, align 8
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
%chan.blockedList = alloca %runtime.channelBlockedList, align 8
compiler: predeclare runtime.trackPointer When a package only uses runtime.trackPointer to create interface packs, the compiler fails to find runtime.trackPointer. This change predeclares it alongside runtime.alloc and updates the tests to use runtime.trackPointer when the test's target uses it.
3 years ago
%0 = bitcast {}* %complit to i8*
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
call void @runtime.trackPointer(i8* nonnull %0, i8* undef) #3
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
%chan.blockedList.bitcast = bitcast %runtime.channelBlockedList* %chan.blockedList to i8*
call void @llvm.lifetime.start.p0i8(i64 24, i8* nonnull %chan.blockedList.bitcast)
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
call void @runtime.chanSend(%runtime.channel* %ch, i8* null, %runtime.channelBlockedList* nonnull %chan.blockedList, i8* undef) #3
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
call void @llvm.lifetime.end.p0i8(i64 24, i8* nonnull %chan.blockedList.bitcast)
ret void
}
compiler: add nounwind attribute This attribute is also set by Clang when it compiles C source files (unless -fexceptions is set). The advantage is that no unwind tables are emitted on Linux (and perhaps other systems). It also avoids __aeabi_unwind_cpp_pr0 on ARM when using the musl libc.
3 years ago
; Function Attrs: nounwind
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
define hidden void @main.chanZeroRecv(%runtime.channel* dereferenceable_or_null(32) %ch, i8* %context) unnamed_addr #1 {
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
entry:
%chan.blockedList = alloca %runtime.channelBlockedList, align 8
%chan.blockedList.bitcast = bitcast %runtime.channelBlockedList* %chan.blockedList to i8*
call void @llvm.lifetime.start.p0i8(i64 24, i8* nonnull %chan.blockedList.bitcast)
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
%0 = call i1 @runtime.chanRecv(%runtime.channel* %ch, i8* null, %runtime.channelBlockedList* nonnull %chan.blockedList, i8* undef) #3
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
call void @llvm.lifetime.end.p0i8(i64 24, i8* nonnull %chan.blockedList.bitcast)
ret void
}
compiler: add nounwind attribute This attribute is also set by Clang when it compiles C source files (unless -fexceptions is set). The advantage is that no unwind tables are emitted on Linux (and perhaps other systems). It also avoids __aeabi_unwind_cpp_pr0 on ARM when using the musl libc.
3 years ago
; Function Attrs: nounwind
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
define hidden void @main.selectZeroRecv(%runtime.channel* dereferenceable_or_null(32) %ch1, %runtime.channel* dereferenceable_or_null(32) %ch2, i8* %context) unnamed_addr #1 {
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
entry:
%select.states.alloca = alloca [2 x %runtime.chanSelectState], align 8
%select.send.value = alloca i32, align 4
store i32 1, i32* %select.send.value, align 4
%select.states.alloca.bitcast = bitcast [2 x %runtime.chanSelectState]* %select.states.alloca to i8*
call void @llvm.lifetime.start.p0i8(i64 16, i8* nonnull %select.states.alloca.bitcast)
%.repack = getelementptr inbounds [2 x %runtime.chanSelectState], [2 x %runtime.chanSelectState]* %select.states.alloca, i32 0, i32 0, i32 0
store %runtime.channel* %ch1, %runtime.channel** %.repack, align 8
%.repack1 = getelementptr inbounds [2 x %runtime.chanSelectState], [2 x %runtime.chanSelectState]* %select.states.alloca, i32 0, i32 0, i32 1
%0 = bitcast i8** %.repack1 to i32**
store i32* %select.send.value, i32** %0, align 4
%.repack3 = getelementptr inbounds [2 x %runtime.chanSelectState], [2 x %runtime.chanSelectState]* %select.states.alloca, i32 0, i32 1, i32 0
store %runtime.channel* %ch2, %runtime.channel** %.repack3, align 8
%.repack4 = getelementptr inbounds [2 x %runtime.chanSelectState], [2 x %runtime.chanSelectState]* %select.states.alloca, i32 0, i32 1, i32 1
store i8* null, i8** %.repack4, align 4
%select.states = getelementptr inbounds [2 x %runtime.chanSelectState], [2 x %runtime.chanSelectState]* %select.states.alloca, i32 0, i32 0
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
%select.result = call { i32, i1 } @runtime.tryChanSelect(i8* undef, %runtime.chanSelectState* nonnull %select.states, i32 2, i32 2, i8* undef) #3
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
call void @llvm.lifetime.end.p0i8(i64 16, i8* nonnull %select.states.alloca.bitcast)
%1 = extractvalue { i32, i1 } %select.result, 0
%2 = icmp eq i32 %1, 0
br i1 %2, label %select.done, label %select.next
select.done: ; preds = %select.body, %select.next, %entry
ret void
select.next: ; preds = %entry
%3 = icmp eq i32 %1, 1
br i1 %3, label %select.body, label %select.done
select.body: ; preds = %select.next
br label %select.done
}
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
declare { i32, i1 } @runtime.tryChanSelect(i8*, %runtime.chanSelectState*, i32, i32, i8*) #0
compiler: avoid zero-sized alloca in channel operations This works around a bug in LLVM (https://bugs.llvm.org/show_bug.cgi?id=49916) but seems like a good change in general.
3 years ago
compiler: update tests after adding new wasm features I don't understand why this wasn't caught in CI. It should have. In any case, because the llvm-features string was updated, these IR outputs were updated.
2 years ago
attributes #0 = { "target-features"="+bulk-memory,+nontrapping-fptoint,+sign-ext" }
attributes #1 = { nounwind "target-features"="+bulk-memory,+nontrapping-fptoint,+sign-ext" }
attributes #2 = { argmemonly nofree nosync nounwind willreturn }
attributes #3 = { nounwind }