First implementation of CommonStaticIncreasingLoop #2
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
A lot of NVVM intrinsics can use the default intrinsic attributes (e.g.,
nosync, nofree, ...) as well as `speculatable`. The latter is important
if we want to recompute intrinsics results instead of communicating them
via memory.
I did use default attributes for almost all `readnone` attributes but
speculatable only where I had reasonable confidence they cannot
experience UB. That said, someone should double check.
TODO: There seem to be various intrinsics marked `Commutative` which
should not, e.g., fma and div.
Differential Revision: https://reviews.llvm.org/D109987
The test case is the IR of:
```
void func(float * restrict a, float *b, int N) {
N = 199;
#pragma omp parallel for
for (int i = 1; i < N; i++)
a[i] = b[i] + 1.0;
}
```
Even if we look for `nocapture` we need to bail on escaping pointers. The crucial thing is that we might not look at a big enough scope when we derive the memory behavior. Thus, it might be `nocapture` in a larger context while it is "captured" in a smaller context.
A declare variant template is only compatible with a base when the number of template arguments is equal, otherwise our instantiations will produce nonsensical results. Exposes as part of D109344. Differential Revision: https://reviews.llvm.org/D109770
Inline assembly is scary but we need to support it for the OpenMP GPU device runtime. The new assumption expresses the fact that it may not have call semantics, that is, it will not call another function but simply perform an operation or side-effect. This is important for reachability in the presence of inline assembly. Differential Revision: https://reviews.llvm.org/D109986
We know a lot about the mapping that we can encode as assumptions for the optimizer. Our `mapping` vs `impl` layer makes this even easier. Assumptions are checked as assertions in debug mode and otherwise assumed (via `llvm.assume`). Differential Revision: https://reviews.llvm.org/D109985
In the last meeting nobody could tell what was problematic for some of the points so we mark them as done until we realize otherwise. Differential Revision: https://reviews.llvm.org/D109990
Since D104432 we can look through memory by analyzing all writes that might interfere with a load. This patch provides some logic to exclude writes that cannot interfere with a location, due to CFG reasoning. We make sure to avoid multi-thread write-read situations properly while we ignore writes that cannot reach a load or writes that will be overwritten before the load is reached. Differential Revision: https://reviews.llvm.org/D106397
AAPointerInfo, and thereby other places, can look already through internal global and stack memory. This patch enables them to look through heap memory returned by functions with a `noalias` return. In the future we can look through `noalias` arguments as well but that will require AAIsDead to learn that such memory can be inspected by the caller later on. We also need teach AAPointerInfo about dominance to actually deal with memory that might not be `null` or `undef` initialized. D106397 is a first step in that direction already. Differential Revision: https://reviews.llvm.org/D109170
While we skipped uses in stores if we can find all copies of the value when the memory is loaded, we did not correlate the use in the store with the use in the load. So far this lead to less precise results in the offset calculations which prevented deductions. With the new EquivalentUseCB callback argument the user of checkForAllUses can be informed of the correlation and act on it appropriately. Differential Revision: https://reviews.llvm.org/D109662
While the simple handling doesn't catch a corner case shown in the tests it is easier to reason about. As we extend PHI handling we will introduce proper recursive reasoning and this makes it simpler to do so.
Similar to loads, we want to be aggressive when it comes to store simplification. Not everything in LLVM handles dead stores well when address space casts are involved, we can simply ask the Attributor to do it for us though. Differential Revision: https://reviews.llvm.org/D109998
Before this patch we conflated the concept of a call instruction and the call return value in a bad way. Any instruction is dead if it has no side-effect and its result is not used. So far we only checked the latter for calls if we ended up creating an AAIsDead attribute for them. There was also no way to refer to the "call instruction" rather than the "call return value" through the IRPosition system. The new entry point `IRPosition::inst` makes this possible. In contrast to `IRPosition::value` it does not translate from call base to call site returned. This makes sense if you think of the "value" is the returned thing the instruction evaluates to and the "inst(ruction)" as the effect of the entire thing. To make this work we also had to remove the "context instruction" for functions, which was the first instruction in the entry block. It did not have an effect on its own and only became relevant as we properly distinguished between call instructions and call return values when we looked at liveness.
We have two attributes that can answer readnone queries. While there is a dependence between them, it seems best to not force the users to know what AA to ask. The helpers also allow to check for readonly nicely. Test changes show where we now deduce readnone but haven't before, mostly because we only asked AAMemoryBehavior and not AAMemoryLocation. AANoAlias has not been ported to the new API yet.
While we already ignored assumed dead internal functions as part of the initialization we did not remember them. Consequently we thought they might be live later on. This only became a real problem as we started to visit the uses of globals as they can be part of those functions. Before we always went through the call site to reach the function which makes it (assumed) live. For now we use AAIsDeadFunction to deal with dead internal functions which solves the problem. It also prevent updates of IR that is dead, see for example align.ll and depgraph.ll. The intended effect is in `read_global_only_written_in_dead_fn` which is now properly optimized.
We used a set with pointer hashes before which caused non-deterministic effects. The new implementation basically numbers accesses as we encounter them and therefore keeps the order deterministic. Helps also debugging a lot.
…Edges) With three reachability related AAs things got complicated. D106720 introduced features that did not work properly as we could add new queries after a fixpoint was reached and which could not be answered by the information gathered up to the fixpoint alone. This patch merges D106720 and AACallEdges to simplify the situation. It also moves the "backwards" reachability out of D106720 and into `AA::isPotentiallyReachable`. In addition to avoiding the fixpoint issue described above, the new scheme allows to track reachability in one place only. We do not have to redo work for backwards queries as the caller already caches reachability. Asking the caller explicitly is the right way to go here. To test inter-procedural reachability in a reasonable way this patch includes also an extension to `AAPointerInfo::forallInterferingWrites`. Basically, we can exclude writes if they cannot reach a load "during the lifetime" of the allocation. That is, we need to go up the call graph to determine reachability until we can determine the allocation would be dead in the caller. See `value-simplify-pointer-info-gpu.ll`. Note: The new code contains plenty debug output to determine how reachability queries are resolved. Differential Revision: https://reviews.llvm.org/D110078
We missed out on AANoRecurse in the module pass because we had no call graph. With AAFunctionReachability we can simply ask if the function may reach itself. Differential Revision: https://reviews.llvm.org/D110099
It is helpful to know that certain functions are entry points, thus that they will be called once and that we do not come back when they are done. The user is the reachability analysis in the Attributor which can now realize that kernels are basically independent.
This is useful to find errors when manually looking at the debug output.
As replacements will become more complex it is better to have a single AA responsible for replacing a use. Before this patch AAValueSimplify* and AAValueSimplifyReturned could both try to replace the returned value. The latter was marginally better for the old pass manager and when a function was already carrying a `returned` attribute. Now we explicitly create AAValueSimplify* AAs from the AAValueSimplifyReturned and avoid replacing returned value uses in the latter. Some local dominance logic for the old PM is included.
When simplify values we might end up with an instruction from a different scope or just one that does not dominate the use. If the instruction can be reproduced without side-effect (incl. UB) we can now do that. For now this is mostly used for speculatable (intrinsic) calls but as we learn to make things like arguments or memory available this will become more powerful.
OpenMP-Opt will use the fact that certain barriers are aligned to reason about potential synchronization. Similarly, we need to know that device (de)allocation functions could synchronize threads but in fact do not.
jdoerfert
pushed a commit
that referenced
this pull request
Dec 21, 2023
Internal builds of the unittests with msan flagged mempcpy_test.
==6862==WARNING: MemorySanitizer: use-of-uninitialized-value
#0 0x55e34d7d734a in length
llvm-project/libc/src/__support/CPP/string_view.h:41:11
#1 0x55e34d7d734a in string_view
llvm-project/libc/src/__support/CPP/string_view.h:71:24
#2 0x55e34d7d734a in
__llvm_libc_9999_0_0_git::testing::Test::testStrEq(char const*, char
const*, char const*, char const*,
__llvm_libc_9999_0_0_git::testing::internal::Location)
llvm-project/libc/test/UnitTest/LibcTest.cpp:284:13
#3 0x55e34d7d4e09 in LlvmLibcMempcpyTest_Simple::Run()
llvm-project/libc/test/src/string/mempcpy_test.cpp:20:3
#4 0x55e34d7d6dff in
__llvm_libc_9999_0_0_git::testing::Test::runTests(char const*)
llvm-project/libc/test/UnitTest/LibcTest.cpp:133:8
#5 0x55e34d7d86e0 in main
llvm-project/libc/test/UnitTest/LibcTestMain.cpp:21:10
SUMMARY: MemorySanitizer: use-of-uninitialized-value
llvm-project/libc/src/__support/CPP/string_view.h:41:11 in length
What's going on here is that mempcpy_test.cpp's Simple test is using
ASSERT_STREQ with a partially initialized char array. ASSERT_STREQ calls
Test::testStrEq which constructs a cpp:string_view. That constructor
calls the
private method cpp::string_view::length. When built with msan, the loop
is
transformed into multi-byte access, which then fails upon access.
I took a look at libc++'s __constexpr_strlen which just calls
__builtin_strlen(). Replacing the implementation of
cpp::string_view::length
with a call to __builtin_strlen() may still result in out of bounds
access when
the test is built with msan.
It's not safe to use ASSERT_STREQ with a partially initialized array.
Initialize the whole array so that the test passes.
jdoerfert
pushed a commit
that referenced
this pull request
Dec 21, 2023
We'd like a way to select the current thread by its thread ID (rather than its internal LLDB thread index). This PR adds a `-t` option (`--thread_id` long option) that tells the `thread select` command to interpret the `<thread-index>` argument as a thread ID. Here's an example of it working: ``` michristensen@devbig356 llvm/llvm-project (thread-select-tid) » ../Debug/bin/lldb ~/scratch/cpp/threading/a.out (lldb) target create "/home/michristensen/scratch/cpp/threading/a.out" Current executable set to '/home/michristensen/scratch/cpp/threading/a.out' (x86_64). (lldb) b 18 Breakpoint 1: where = a.out`main + 80 at main.cpp:18:12, address = 0x0000000000000850 (lldb) run Process 215715 launched: '/home/michristensen/scratch/cpp/threading/a.out' (x86_64) This is a thread, i=1 This is a thread, i=2 This is a thread, i=3 This is a thread, i=4 This is a thread, i=5 Process 215715 stopped * thread #1, name = 'a.out', stop reason = breakpoint 1.1 frame #0: 0x0000555555400850 a.out`main at main.cpp:18:12 15 for (int i = 0; i < 5; i++) { 16 pthread_create(&thread_ids[i], NULL, foo, NULL); 17 } -> 18 for (int i = 0; i < 5; i++) { 19 pthread_join(thread_ids[i], NULL); 20 } 21 return 0; (lldb) thread select 2 * thread #2, name = 'a.out' frame #0: 0x00007ffff68f9918 libc.so.6`__nanosleep + 72 libc.so.6`__nanosleep: -> 0x7ffff68f9918 <+72>: cmpq $-0x1000, %rax ; imm = 0xF000 0x7ffff68f991e <+78>: ja 0x7ffff68f9952 ; <+130> 0x7ffff68f9920 <+80>: movl %edx, %edi 0x7ffff68f9922 <+82>: movl %eax, 0xc(%rsp) (lldb) thread info thread #2: tid = 216047, 0x00007ffff68f9918 libc.so.6`__nanosleep + 72, name = 'a.out' (lldb) thread list Process 215715 stopped thread #1: tid = 215715, 0x0000555555400850 a.out`main at main.cpp:18:12, name = 'a.out', stop reason = breakpoint 1.1 * thread #2: tid = 216047, 0x00007ffff68f9918 libc.so.6`__nanosleep + 72, name = 'a.out' thread #3: tid = 216048, 0x00007ffff68f9918 libc.so.6`__nanosleep + 72, name = 'a.out' thread #4: tid = 216049, 0x00007ffff68f9918 libc.so.6`__nanosleep + 72, name = 'a.out' thread #5: tid = 216050, 0x00007ffff68f9918 libc.so.6`__nanosleep + 72, name = 'a.out' thread #6: tid = 216051, 0x00007ffff68f9918 libc.so.6`__nanosleep + 72, name = 'a.out' (lldb) thread select 215715 error: invalid thread #215715. (lldb) thread select -t 215715 * thread #1, name = 'a.out', stop reason = breakpoint 1.1 frame #0: 0x0000555555400850 a.out`main at main.cpp:18:12 15 for (int i = 0; i < 5; i++) { 16 pthread_create(&thread_ids[i], NULL, foo, NULL); 17 } -> 18 for (int i = 0; i < 5; i++) { 19 pthread_join(thread_ids[i], NULL); 20 } 21 return 0; (lldb) thread select -t 216051 * thread #6, name = 'a.out' frame #0: 0x00007ffff68f9918 libc.so.6`__nanosleep + 72 libc.so.6`__nanosleep: -> 0x7ffff68f9918 <+72>: cmpq $-0x1000, %rax ; imm = 0xF000 0x7ffff68f991e <+78>: ja 0x7ffff68f9952 ; <+130> 0x7ffff68f9920 <+80>: movl %edx, %edi 0x7ffff68f9922 <+82>: movl %eax, 0xc(%rsp) (lldb) thread select 3 * thread #3, name = 'a.out' frame #0: 0x00007ffff68f9918 libc.so.6`__nanosleep + 72 libc.so.6`__nanosleep: -> 0x7ffff68f9918 <+72>: cmpq $-0x1000, %rax ; imm = 0xF000 0x7ffff68f991e <+78>: ja 0x7ffff68f9952 ; <+130> 0x7ffff68f9920 <+80>: movl %edx, %edi 0x7ffff68f9922 <+82>: movl %eax, 0xc(%rsp) (lldb) thread select -t 216048 * thread #3, name = 'a.out' frame #0: 0x00007ffff68f9918 libc.so.6`__nanosleep + 72 libc.so.6`__nanosleep: -> 0x7ffff68f9918 <+72>: cmpq $-0x1000, %rax ; imm = 0xF000 0x7ffff68f991e <+78>: ja 0x7ffff68f9952 ; <+130> 0x7ffff68f9920 <+80>: movl %edx, %edi 0x7ffff68f9922 <+82>: movl %eax, 0xc(%rsp) (lldb) thread select --thread_id 216048 * thread #3, name = 'a.out' frame #0: 0x00007ffff68f9918 libc.so.6`__nanosleep + 72 libc.so.6`__nanosleep: -> 0x7ffff68f9918 <+72>: cmpq $-0x1000, %rax ; imm = 0xF000 0x7ffff68f991e <+78>: ja 0x7ffff68f9952 ; <+130> 0x7ffff68f9920 <+80>: movl %edx, %edi 0x7ffff68f9922 <+82>: movl %eax, 0xc(%rsp) (lldb) help thread select Change the currently selected thread. Syntax: thread select <cmd-options> <thread-index> Command Options Usage: thread select [-t] <thread-index> -t ( --thread_id ) Provide a thread ID instead of a thread index. This command takes options and free-form arguments. If your arguments resemble option specifiers (i.e., they start with a - or --), you must use ' -- ' between the end of the command options and the beginning of the arguments. (lldb) c Process 215715 resuming Process 215715 exited with status = 0 (0x00000000) ```
jdoerfert
pushed a commit
that referenced
this pull request
Dec 21, 2023
Linalg op fusion (`Linalg/Transforms/Fusion.cpp`) used to generate invalid fused producer ops: ``` error: 'linalg.conv_2d_nhwc_hwcf' op expected type of operand #2 ('tensor<1x8x16x4xf32>') to match type of corresponding result ('tensor<?x?x?x?xf32>') note: see current operation: %24 = "linalg.conv_2d_nhwc_hwcf"(%21, %22, %23) <{dilations = dense<1> : tensor<2xi64>, operandSegmentSizes = array<i32: 2, 1>, strides = dense<2> : tensor<2xi64>}> ({ ^bb0(%arg9: f32, %arg10: f32, %arg11: f32): %28 = "arith.mulf"(%arg9, %arg10) <{fastmath = #arith.fastmath<none>}> : (f32, f32) -> f32 %29 = "arith.addf"(%arg11, %28) <{fastmath = #arith.fastmath<none>}> : (f32, f32) -> f32 "linalg.yield"(%29) : (f32) -> () }) {linalg.memoized_indexing_maps = [affine_map<(d0, d1, d2, d3, d4, d5, d6) -> (d0, d1 * 2 + d4, d2 * 2 + d5, d6)>, affine_map<(d0, d1, d2, d3, d4, d5, d6) -> (d4, d5, d6, d3)>, affine_map<(d0, d1, d2, d3, d4, d5, d6) -> (d0, d1, d2, d3)>]} : (tensor<1x?x?x3xf32>, tensor<3x3x3x4xf32>, tensor<1x8x16x4xf32>) -> tensor<?x?x?x?xf32> ``` This is a problem because the input IR to greedy pattern rewriter during `-test-linalg-greedy-fusion` is invalid. This commit fixes tests such as `mlir/test/Dialect/Linalg/tile-and-fuse-tensors.mlir` when verifying the IR after each pattern application (llvm#74270).
jdoerfert
pushed a commit
that referenced
this pull request
Dec 21, 2023
This has been flaky for a while, for example https://lab.llvm.org/buildbot/#/builders/96/builds/50350 ``` Command Output (stdout): -- lldb version 18.0.0git (https://github.com/llvm/llvm-project.git revision 3974d89) clang revision 3974d89 llvm revision 3974d89 "can't evaluate expressions when the process is running." ``` ``` PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace. #0 0x0000ffffa46191a0 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x529a1a0) #1 0x0000ffffa4617144 llvm::sys::RunSignalHandlers() (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x5298144) #2 0x0000ffffa46198d0 SignalHandler(int) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x529a8d0) #3 0x0000ffffab25b7dc (linux-vdso.so.1+0x7dc) #4 0x0000ffffab13d050 /build/glibc-Q8DG8B/glibc-2.31/string/../sysdeps/aarch64/multiarch/memcpy_advsimd.S:92:0 #5 0x0000ffffa446f420 lldb_private::process_gdb_remote::GDBRemoteRegisterContext::PrivateSetRegisterValue(unsigned int, llvm::ArrayRef<unsigned char>) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x50f0420) #6 0x0000ffffa446f7b8 lldb_private::process_gdb_remote::GDBRemoteRegisterContext::GetPrimordialRegister(lldb_private::RegisterInfo const*, lldb_private::process_gdb_remote::GDBRemoteCommunicationClient&) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x50f07b8) #7 0x0000ffffa446f308 lldb_private::process_gdb_remote::GDBRemoteRegisterContext::ReadRegisterBytes(lldb_private::RegisterInfo const*) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x50f0308) #8 0x0000ffffa446ec1c lldb_private::process_gdb_remote::GDBRemoteRegisterContext::ReadRegister(lldb_private::RegisterInfo const*, lldb_private::RegisterValue&) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x50efc1c) #9 0x0000ffffa412eaa4 lldb_private::RegisterContext::ReadRegisterAsUnsigned(lldb_private::RegisterInfo const*, unsigned long) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x4dafaa4) #10 0x0000ffffa420861c ReadLinuxProcessAddressMask(std::shared_ptr<lldb_private::Process>, llvm::StringRef) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x4e8961c) #11 0x0000ffffa4208430 ABISysV_arm64::FixCodeAddress(unsigned long) (/home/tcwg-buildbot/worker/lldb-aarch64-ubuntu/build/lib/python3.8/site-packages/lldb/_lldb.cpython-38-aarch64-linux-gnu.so+0x4e89430) ``` Judging by the backtrace something is trying to read the pointer authentication address/code mask registers. This explains why I've not seen this issue locally, as the buildbot runs on Graviton 3 with has the pointer authentication extension. I will try to reproduce, fix and re-enable the test.
jdoerfert
pushed a commit
that referenced
this pull request
Mar 21, 2024
…lvm#85653) This reverts commit daebe5c. This commit causes the following asan issue: ``` <snip>/llvm-project/build/bin/mlir-opt <snip>/llvm-project/mlir/test/Dialect/XeGPU/XeGPUOps.mlir | <snip>/llvm-project/build/bin/FileCheck <snip>/llvm-project/mlir/test/Dialect/XeGPU/XeGPUOps.mlir # executed command: <snip>/llvm-project/build/bin/mlir-opt <snip>/llvm-project/mlir/test/Dialect/XeGPU/XeGPUOps.mlir # .---command stderr------------ # | ================================================================= # | ==2772558==ERROR: AddressSanitizer: stack-use-after-return on address 0x7fd2c2c42b90 at pc 0x55e406d54614 bp 0x7ffc810e4070 sp 0x7ffc810e4068 # | READ of size 8 at 0x7fd2c2c42b90 thread T0 # | #0 0x55e406d54613 in operator()<long int const*> /usr/include/c++/13/bits/predefined_ops.h:318 # | #1 0x55e406d54613 in __count_if<long int const*, __gnu_cxx::__ops::_Iter_pred<mlir::verifyListOfOperandsOrIntegers(Operation*, llvm::StringRef, unsigned int, llvm::ArrayRef<long int>, ValueRange)::<lambda(int64_t)> > > /usr/include/c++/13/bits/stl_algobase.h:2125 # | #2 0x55e406d54613 in count_if<long int const*, mlir::verifyListOfOperandsOrIntegers(Operation*, ... ```
jdoerfert
pushed a commit
that referenced
this pull request
Mar 21, 2024
…oint. (llvm#83821)" This reverts commit c2c1e6e. It creates a use after free. ==8342==ERROR: AddressSanitizer: heap-use-after-free on address 0x50f000001760 at pc 0x55b9fb84a8fb bp 0x7ffc18468a10 sp 0x7ffc18468a08 READ of size 1 at 0x50f000001760 thread T0 #0 0x55b9fb84a8fa in dropPoisonGeneratingFlags llvm/lib/Transforms/Vectorize/VPlan.h:1040:13 #1 0x55b9fb84a8fa in llvm::VPlanTransforms::dropPoisonGeneratingRecipes(llvm::VPlan&, llvm::function_ref<bool (llvm::BasicBlock*)>)::$_0::operator()(llvm::VPRecipeBase*) const llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp:1236:23 #2 0x55b9fb84a196 in llvm::VPlanTransforms::dropPoisonGeneratingRecipes(llvm::VPlan&, llvm::function_ref<bool (llvm::BasicBlock*)>) llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp Can be reproduced with asan on Transforms/LoopVectorize/AArch64/sve-interleaved-masked-accesses.ll Transforms/LoopVectorize/X86/pr81872.ll Transforms/LoopVectorize/X86/x86-interleaved-accesses-masked-group.ll
jdoerfert
pushed a commit
that referenced
this pull request
May 8, 2024
…e exception specification of a function (llvm#90760) [temp.deduct.general] p6 states: > At certain points in the template argument deduction process it is necessary to take a function type that makes use of template parameters and replace those template parameters with the corresponding template arguments. This is done at the beginning of template argument deduction when any explicitly specified template arguments are substituted into the function type, and again at the end of template argument deduction when any template arguments that were deduced or obtained from default arguments are substituted. [temp.deduct.general] p7 goes on to say: > The _deduction substitution loci_ are > - the function type outside of the _noexcept-specifier_, > - the explicit-specifier, > - the template parameter declarations, and > - the template argument list of a partial specialization > > The substitution occurs in all types and expressions that are used in the deduction substitution loci. [...] Consider the following: ```cpp struct A { static constexpr bool x = true; }; template<typename T, typename U> void f(T, U) noexcept(T::x); // #1 template<typename T, typename U> void f(T, U*) noexcept(T::y); // #2 template<> void f<A>(A, int*) noexcept; // clang currently accepts, GCC and EDG reject ``` Currently, `Sema::SubstituteExplicitTemplateArguments` will substitute into the _noexcept-specifier_ when deducing template arguments from a function declaration or when deducing template arguments for taking the address of a function template (and the substitution is treated as a SFINAE context). In the above example, `#1` is selected as the primary template because substitution of the explicit template arguments into the _noexcept-specifier_ of `#2` failed, which resulted in the candidate being ignored. This behavior is incorrect ([temp.deduct.general] note 4 says as much), and this patch corrects it by deferring all substitution into the _noexcept-specifier_ until it is instantiated. As part of the necessary changes to make this patch work, the instantiation of the exception specification of a function template specialization when taking the address of a function template is changed to only occur for the function selected by overload resolution per [except.spec] p13.1 (as opposed to being instantiated for every candidate).
jdoerfert
pushed a commit
that referenced
this pull request
May 8, 2024
…ined member functions & member function templates (llvm#88963) Consider the following snippet from the discussion of CWG2847 on the core reflector: ``` template<typename T> concept C = sizeof(T) <= sizeof(long); template<typename T> struct A { template<typename U> void f(U) requires C<U>; // #1, declares a function template void g() requires C<T>; // #2, declares a function template<> void f(char); // #3, an explicit specialization of a function template that declares a function }; template<> template<typename U> void A<short>::f(U) requires C<U>; // #4, an explicit specialization of a function template that declares a function template template<> template<> void A<int>::f(int); // #5, an explicit specialization of a function template that declares a function template<> void A<long>::g(); // #6, an explicit specialization of a function that declares a function ``` A number of problems exist: - Clang rejects `#4` because the trailing _requires-clause_ has `U` substituted with the wrong template parameter depth when `Sema::AreConstraintExpressionsEqual` is called to determine whether it matches the trailing _requires-clause_ of the implicitly instantiated function template. - Clang rejects `#5` because the function template specialization instantiated from `A<int>::f` has a trailing _requires-clause_, but `#5` does not (nor can it have one as it isn't a templated function). - Clang rejects `#6` for the same reasons it rejects `#5`. This patch resolves these issues by making the following changes: - To fix `#4`, `Sema::AreConstraintExpressionsEqual` is passed `FunctionTemplateDecl`s when comparing the trailing _requires-clauses_ of `#4` and the function template instantiated from `#1`. - To fix `#5` and `#6`, the trailing _requires-clauses_ are not compared for explicit specializations that declare functions. In addition to these changes, `CheckMemberSpecialization` now considers constraint satisfaction/constraint partial ordering when determining which member function is specialized by an explicit specialization of a member function for an implicit instantiation of a class template (we previously would select the first function that has the same type as the explicit specialization). With constraints taken under consideration, we match EDG's behavior for these declarations.
jdoerfert
pushed a commit
that referenced
this pull request
May 22, 2024
...which caused issues like > ==42==ERROR: AddressSanitizer failed to deallocate 0x32 (50) bytes at address 0x117e0000 (error code: 28) > ==42==Cannot dump memory map on emscriptenAddressSanitizer: CHECK failed: sanitizer_common.cpp:81 "((0 && "unable to unmmap")) != (0)" (0x0, 0x0) (tid=288045824) > #0 0x14f73b0c in __asan::CheckUnwind()+0x14f73b0c (this.program+0x14f73b0c) > #1 0x14f8a3c2 in __sanitizer::CheckFailed(char const*, int, char const*, unsigned long long, unsigned long long)+0x14f8a3c2 (this.program+0x14f8a3c2) > #2 0x14f7d6e1 in __sanitizer::ReportMunmapFailureAndDie(void*, unsigned long, int, bool)+0x14f7d6e1 (this.program+0x14f7d6e1) > #3 0x14f81fbd in __sanitizer::UnmapOrDie(void*, unsigned long)+0x14f81fbd (this.program+0x14f81fbd) > #4 0x14f875df in __sanitizer::SuppressionContext::ParseFromFile(char const*)+0x14f875df (this.program+0x14f875df) > #5 0x14f74eab in __asan::InitializeSuppressions()+0x14f74eab (this.program+0x14f74eab) > #6 0x14f73a1a in __asan::AsanInitInternal()+0x14f73a1a (this.program+0x14f73a1a) when trying to use an ASan suppressions file under Emscripten: Even though it would be considered OK by SUSv4, the Emscripten runtime states "We don't support partial munmapping" (see <emscripten-core/emscripten@f4115eb> "Implement MAP_ANONYMOUS on top of malloc in STANDALONE_WASM mode (llvm#16289)"). Co-authored-by: Stephan Bergmann <[email protected]>
jdoerfert
pushed a commit
that referenced
this pull request
May 22, 2024
…ication as used during partial ordering (llvm#91534) We do not deduce template arguments from the exception specification when determining the primary template of a function template specialization or when taking the address of a function template. Therefore, this patch changes `isAtLeastAsSpecializedAs` such that we do not mark template parameters in the exception specification as 'used' during partial ordering (per [temp.deduct.partial] p12) to prevent the following from being ambiguous: ``` template<typename T, typename U> void f(U) noexcept(noexcept(T())); // #1 template<typename T> void f(T*) noexcept; // #2 template<> void f<int>(int*) noexcept; // currently ambiguous, selects #2 with this patch applied ``` Although there is no corresponding wording in the standard (see core issue filed here cplusplus/CWG#537), this seems to be the intended behavior given the definition of _deduction substitution loci_ in [temp.deduct.general] p7 (and EDG does the same thing).
jdoerfert
pushed a commit
that referenced
this pull request
May 22, 2024
…erSize (llvm#67657)" This reverts commit f0b3654. This commit triggers UB by reading an uninitialized variable. `UP.PartialThreshold` is used uninitialized in `getUnrollingPreferences()` when it is called from `LoopVectorizationPlanner::executePlan()`. In this case the `UP` variable is created on the stack and its fields are not initialized. ``` ==8802==WARNING: MemorySanitizer: use-of-uninitialized-value #0 0x557c0b081b99 in llvm::BasicTTIImplBase<llvm::X86TTIImpl>::getUnrollingPreferences(llvm::Loop*, llvm::ScalarEvolution&, llvm::TargetTransformInfo::UnrollingPreferences&, llvm::OptimizationRemarkEmitter*) llvm-project/llvm/include/llvm/CodeGen/BasicTTIImpl.h #1 0x557c0b07a40c in llvm::TargetTransformInfo::Model<llvm::X86TTIImpl>::getUnrollingPreferences(llvm::Loop*, llvm::ScalarEvolution&, llvm::TargetTransformInfo::UnrollingPreferences&, llvm::OptimizationRemarkEmitter*) llvm-project/llvm/include/llvm/Analysis/TargetTransformInfo.h:2277:17 #2 0x557c0f5d69ee in llvm::TargetTransformInfo::getUnrollingPreferences(llvm::Loop*, llvm::ScalarEvolution&, llvm::TargetTransformInfo::UnrollingPreferences&, llvm::OptimizationRemarkEmitter*) const llvm-project/llvm/lib/Analysis/TargetTransformInfo.cpp:387:19 #3 0x557c0e6b96a0 in llvm::LoopVectorizationPlanner::executePlan(llvm::ElementCount, unsigned int, llvm::VPlan&, llvm::InnerLoopVectorizer&, llvm::DominatorTree*, bool, llvm::DenseMap<llvm::SCEV const*, llvm::Value*, llvm::DenseMapInfo<llvm::SCEV const*, void>, llvm::detail::DenseMapPair<llvm::SCEV const*, llvm::Value*>> const*) llvm-project/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:7624:7 #4 0x557c0e6e4b63 in llvm::LoopVectorizePass::processLoop(llvm::Loop*) llvm-project/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:10253:13 #5 0x557c0e6f2429 in llvm::LoopVectorizePass::runImpl(llvm::Function&, llvm::ScalarEvolution&, llvm::LoopInfo&, llvm::TargetTransformInfo&, llvm::DominatorTree&, llvm::BlockFrequencyInfo*, llvm::TargetLibraryInfo*, llvm::DemandedBits&, llvm::AssumptionCache&, llvm::LoopAccessInfoManager&, llvm::OptimizationRemarkEmitter&, llvm::ProfileSummaryInfo*) llvm-project/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:10344:30 #6 0x557c0e6f2f97 in llvm::LoopVectorizePass::run(llvm::Function&, llvm::AnalysisManager<llvm::Function>&) llvm-project/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:10383:9 [...] Uninitialized value was created by an allocation of 'UP' in the stack frame #0 0x557c0e6b961e in llvm::LoopVectorizationPlanner::executePlan(llvm::ElementCount, unsigned int, llvm::VPlan&, llvm::InnerLoopVectorizer&, llvm::DominatorTree*, bool, llvm::DenseMap<llvm::SCEV const*, llvm::Value*, llvm::DenseMapInfo<llvm::SCEV const*, void>, llvm::detail::DenseMapPair<llvm::SCEV const*, llvm::Value*>> const*) llvm-project/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:7623:3 ```
jdoerfert
pushed a commit
that referenced
this pull request
May 22, 2024
…vm#90820) This solves some ambuguity introduced in P0522 regarding how template template parameters are partially ordered, and should reduce the negative impact of enabling `-frelaxed-template-template-args` by default. When performing template argument deduction, a template template parameter containing no packs should be more specialized than one that does. Given the following example: ```C++ template<class T2> struct A; template<template<class ...T3s> class TT1, class T4> struct A<TT1<T4>>; // #1 template<template<class T5 > class TT2, class T6> struct A<TT2<T6>>; // #2 template<class T1> struct B; template struct A<B<char>>; ``` Prior to P0522, candidate `#2` would be more specialized. After P0522, neither is more specialized, so this becomes ambiguous. With this change, `#2` becomes more specialized again, maintaining compatibility with pre-P0522 implementations. The problem is that in P0522, candidates are at least as specialized when matching packs to fixed-size lists both ways, whereas before, a fixed-size list is more specialized. This patch keeps the original behavior when checking template arguments outside deduction, but restores this aspect of pre-P0522 matching during deduction. --- Since this changes provisional implementation of CWG2398 which has not been released yet, and already contains a changelog entry, we don't provide a changelog entry here.
jdoerfert
pushed a commit
that referenced
this pull request
May 22, 2024
'reduction' has a few restrictions over normal 'var-list' clauses: 1- On parallel, a num_gangs can only have 1 argument when combined with reduction. These two aren't able to be combined on any other of the compute constructs however. 2- The vars all must be 'numerical data types' types of some sort, or a 'composite of numerical data types'. A list of types is given in the standard as a minimum, so we choose 'isScalar', which covers all of these types and keeps types that are actually numeric. Other compilers don't seem to implement the 'composite of numerical data types', though we do. 3- Because of the above restrictions, member-of-composite is not allowed, so any access via a memberexpr is disallowed. Array-element and sub-arrays (aka array sections) are both permitted, so long as they meet the requirements of #2. This patch implements all of these for compute constructs.
jdoerfert
pushed a commit
that referenced
this pull request
May 22, 2024
…llvm#92855) This solves some ambuguity introduced in P0522 regarding how template template parameters are partially ordered, and should reduce the negative impact of enabling `-frelaxed-template-template-args` by default. When performing template argument deduction, we extend the provisional wording introduced in llvm#89807 so it also covers deduction of class templates. Given the following example: ```C++ template <class T1, class T2 = float> struct A; template <class T3> struct B; template <template <class T4> class TT1, class T5> struct B<TT1<T5>>; // #1 template <class T6, class T7> struct B<A<T6, T7>>; // #2 template struct B<A<int>>; ``` Prior to P0522, `#2` was picked. Afterwards, this became ambiguous. This patch restores the pre-P0522 behavior, `#2` is picked again. This has the beneficial side effect of making the following code valid: ```C++ template<class T, class U> struct A {}; A<int, float> v; template<template<class> class TT> void f(TT<int>); // OK: TT picks 'float' as the default argument for the second parameter. void g() { f(v); } ``` --- Since this changes provisional implementation of CWG2398 which has not been released yet, and already contains a changelog entry, we don't provide a changelog entry here.
shiltian
pushed a commit
that referenced
this pull request
Jul 24, 2024
…linux (llvm#99613) Examples of the output: ARM: ``` # ./a.out AddressSanitizer:DEADLYSIGNAL ================================================================= ==122==ERROR: AddressSanitizer: SEGV on unknown address 0x0000007a (pc 0x76e13ac0 bp 0x7eb7fd00 sp 0x7eb7fcc8 T0) ==122==The signal is caused by a READ memory access. ==122==Hint: address points to the zero page. #0 0x76e13ac0 (/lib/libc.so.6+0x7cac0) #1 0x76dce680 in gsignal (/lib/libc.so.6+0x37680) #2 0x005c2250 (/root/a.out+0x145250) #3 0x76db982c (/lib/libc.so.6+0x2282c) #4 0x76db9918 in __libc_start_main (/lib/libc.so.6+0x22918) ==122==Register values: r0 = 0x00000000 r1 = 0x0000007a r2 = 0x0000000b r3 = 0x76d95020 r4 = 0x0000007a r5 = 0x00000001 r6 = 0x005dcc5c r7 = 0x0000010c r8 = 0x0000000b r9 = 0x76f9ece0 r10 = 0x00000000 r11 = 0x7eb7fd00 r12 = 0x76dce670 sp = 0x7eb7fcc8 lr = 0x76e13ab4 pc = 0x76e13ac0 AddressSanitizer can not provide additional info. SUMMARY: AddressSanitizer: SEGV (/lib/libc.so.6+0x7cac0) ==122==ABORTING ``` AArch64: ``` # ./a.out UndefinedBehaviorSanitizer:DEADLYSIGNAL ==99==ERROR: UndefinedBehaviorSanitizer: SEGV on unknown address 0x000000000063 (pc 0x007fbbbc5860 bp 0x007fcfdcb700 sp 0x007fcfdcb700 T99) ==99==The signal is caused by a UNKNOWN memory access. ==99==Hint: address points to the zero page. #0 0x007fbbbc5860 (/lib64/libc.so.6+0x82860) #1 0x007fbbb81578 (/lib64/libc.so.6+0x3e578) #2 0x00556051152c (/root/a.out+0x3152c) #3 0x007fbbb6e268 (/lib64/libc.so.6+0x2b268) #4 0x007fbbb6e344 (/lib64/libc.so.6+0x2b344) #5 0x0055604e45ec (/root/a.out+0x45ec) ==99==Register values: x0 = 0x0000000000000000 x1 = 0x0000000000000063 x2 = 0x000000000000000b x3 = 0x0000007fbbb41440 x4 = 0x0000007fbbb41580 x5 = 0x3669288942d44cce x6 = 0x0000000000000000 x7 = 0x00000055605110b0 x8 = 0x0000000000000083 x9 = 0x0000000000000000 x10 = 0x0000000000000000 x11 = 0x0000000000000000 x12 = 0x0000007fbbdb3360 x13 = 0x0000000000010000 x14 = 0x0000000000000039 x15 = 0x00000000004113a0 x16 = 0x0000007fbbb81560 x17 = 0x0000005560540138 x18 = 0x000000006474e552 x19 = 0x0000000000000063 x20 = 0x0000000000000001 x21 = 0x000000000000000b x22 = 0x0000005560511510 x23 = 0x0000007fcfdcb918 x24 = 0x0000007fbbdb1b50 x25 = 0x0000000000000000 x26 = 0x0000007fbbdb2000 x27 = 0x000000556053f858 x28 = 0x0000000000000000 fp = 0x0000007fcfdcb700 lr = 0x0000007fbbbc584c sp = 0x0000007fcfdcb700 UndefinedBehaviorSanitizer can not provide additional info. SUMMARY: UndefinedBehaviorSanitizer: SEGV (/lib64/libc.so.6+0x82860) ==99==ABORTING ```
jdoerfert
pushed a commit
that referenced
this pull request
Aug 4, 2024
```
UBSan-Standalone-sparc :: TestCases/Misc/Linux/diag-stacktrace.cpp
```
`FAIL`s on 32 and 64-bit Linux/sparc64 (and on Solaris/sparcv9, too: the
test isn't Linux-specific at all). With
`UBSAN_OPTIONS=fast_unwind_on_fatal=1`, the stack trace shows a
duplicate innermost frame:
```
compiler-rt/test/ubsan/TestCases/Misc/Linux/diag-stacktrace.cpp:14:31: runtime error: execution reached the end of a value-returning function without returning a value
#0 0x7003a708 in f() compiler-rt/test/ubsan/TestCases/Misc/Linux/diag-stacktrace.cpp:14:35
#1 0x7003a708 in f() compiler-rt/test/ubsan/TestCases/Misc/Linux/diag-stacktrace.cpp:14:35
#2 0x7003a714 in g() compiler-rt/test/ubsan/TestCases/Misc/Linux/diag-stacktrace.cpp:17:38
```
which isn't seen with `fast_unwind_on_fatal=0`.
This turns out to be another fallout from fixing
`__builtin_return_address`/`__builtin_extract_return_addr` on SPARC. In
`sanitizer_stacktrace_sparc.cpp` (`BufferedStackTrace::UnwindFast`) the
`pc` arg is the return address, while `pc1` from the stack frame
(`fr_savpc`) is the address of the `call` insn, leading to a double
entry for the innermost frame in `trace_buffer[]`.
This patch fixes this by moving the adjustment before all uses.
Tested on `sparc64-unknown-linux-gnu` and `sparcv9-sun-solaris2.11`
(with the `ubsan/TestCases/Misc/Linux` tests enabled).
jdoerfert
pushed a commit
that referenced
this pull request
Aug 30, 2024
…lvm#104148) `hasOperands` does not always execute matchers in the order they are written. This can cause issue in code using bindings when one operand matcher is relying on a binding set by the other. With this change, the first matcher present in the code is always executed first and any binding it sets are available to the second matcher. Simple example with current version (1 match) and new version (2 matches): ```bash > cat tmp.cpp int a = 13; int b = ((int) a) - a; int c = a - ((int) a); > clang-query tmp.cpp clang-query> set traversal IgnoreUnlessSpelledInSource clang-query> m binaryOperator(hasOperands(cStyleCastExpr(has(declRefExpr(hasDeclaration(valueDecl().bind("d"))))), declRefExpr(hasDeclaration(valueDecl(equalsBoundNode("d")))))) Match #1: tmp.cpp:1:1: note: "d" binds here int a = 13; ^~~~~~~~~~ tmp.cpp:2:9: note: "root" binds here int b = ((int)a) - a; ^~~~~~~~~~~~ 1 match. > ./build/bin/clang-query tmp.cpp clang-query> set traversal IgnoreUnlessSpelledInSource clang-query> m binaryOperator(hasOperands(cStyleCastExpr(has(declRefExpr(hasDeclaration(valueDecl().bind("d"))))), declRefExpr(hasDeclaration(valueDecl(equalsBoundNode("d")))))) Match #1: tmp.cpp:1:1: note: "d" binds here 1 | int a = 13; | ^~~~~~~~~~ tmp.cpp:2:9: note: "root" binds here 2 | int b = ((int)a) - a; | ^~~~~~~~~~~~ Match #2: tmp.cpp:1:1: note: "d" binds here 1 | int a = 13; | ^~~~~~~~~~ tmp.cpp:3:9: note: "root" binds here 3 | int c = a - ((int)a); | ^~~~~~~~~~~~ 2 matches. ``` If this should be documented or regression tested anywhere please let me know where.
jdoerfert
pushed a commit
that referenced
this pull request
Aug 30, 2024
…104523) Compilers and language runtimes often use helper functions that are fundamentally uninteresting when debugging anything but the compiler/runtime itself. This patch introduces a user-extensible mechanism that allows for these frames to be hidden from backtraces and automatically skipped over when navigating the stack with `up` and `down`. This does not affect the numbering of frames, so `f <N>` will still provide access to the hidden frames. The `bt` output will also print a hint that frames have been hidden. My primary motivation for this feature is to hide thunks in the Swift programming language, but I'm including an example recognizer for `std::function::operator()` that I wished for myself many times while debugging LLDB. rdar://126629381 Example output. (Yes, my proof-of-concept recognizer could hide even more frames if we had a method that returned the function name without the return type or I used something that isn't based off regex, but it's really only meant as an example). before: ``` (lldb) thread backtrace --filtered=false * thread #1, queue = 'com.apple.main-thread', stop reason = breakpoint 1.1 * frame #0: 0x0000000100001f04 a.out`foo(x=1, y=1) at main.cpp:4:10 frame #1: 0x0000000100003a00 a.out`decltype(std::declval<int (*&)(int, int)>()(std::declval<int>(), std::declval<int>())) std::__1::__invoke[abi:se200000]<int (*&)(int, int), int, int>(__f=0x000000016fdff280, __args=0x000000016fdff224, __args=0x000000016fdff220) at invoke.h:149:25 frame #2: 0x000000010000399c a.out`int std::__1::__invoke_void_return_wrapper<int, false>::__call[abi:se200000]<int (*&)(int, int), int, int>(__args=0x000000016fdff280, __args=0x000000016fdff224, __args=0x000000016fdff220) at invoke.h:216:12 frame #3: 0x0000000100003968 a.out`std::__1::__function::__alloc_func<int (*)(int, int), std::__1::allocator<int (*)(int, int)>, int (int, int)>::operator()[abi:se200000](this=0x000000016fdff280, __arg=0x000000016fdff224, __arg=0x000000016fdff220) at function.h:171:12 frame #4: 0x00000001000026bc a.out`std::__1::__function::__func<int (*)(int, int), std::__1::allocator<int (*)(int, int)>, int (int, int)>::operator()(this=0x000000016fdff278, __arg=0x000000016fdff224, __arg=0x000000016fdff220) at function.h:313:10 frame #5: 0x0000000100003c38 a.out`std::__1::__function::__value_func<int (int, int)>::operator()[abi:se200000](this=0x000000016fdff278, __args=0x000000016fdff224, __args=0x000000016fdff220) const at function.h:430:12 frame #6: 0x0000000100002038 a.out`std::__1::function<int (int, int)>::operator()(this= Function = foo(int, int) , __arg=1, __arg=1) const at function.h:989:10 frame #7: 0x0000000100001f64 a.out`main(argc=1, argv=0x000000016fdff4f8) at main.cpp:9:10 frame #8: 0x0000000183cdf154 dyld`start + 2476 (lldb) ``` after ``` (lldb) bt * thread #1, queue = 'com.apple.main-thread', stop reason = breakpoint 1.1 * frame #0: 0x0000000100001f04 a.out`foo(x=1, y=1) at main.cpp:4:10 frame #1: 0x0000000100003a00 a.out`decltype(std::declval<int (*&)(int, int)>()(std::declval<int>(), std::declval<int>())) std::__1::__invoke[abi:se200000]<int (*&)(int, int), int, int>(__f=0x000000016fdff280, __args=0x000000016fdff224, __args=0x000000016fdff220) at invoke.h:149:25 frame #2: 0x000000010000399c a.out`int std::__1::__invoke_void_return_wrapper<int, false>::__call[abi:se200000]<int (*&)(int, int), int, int>(__args=0x000000016fdff280, __args=0x000000016fdff224, __args=0x000000016fdff220) at invoke.h:216:12 frame #6: 0x0000000100002038 a.out`std::__1::function<int (int, int)>::operator()(this= Function = foo(int, int) , __arg=1, __arg=1) const at function.h:989:10 frame #7: 0x0000000100001f64 a.out`main(argc=1, argv=0x000000016fdff4f8) at main.cpp:9:10 frame #8: 0x0000000183cdf154 dyld`start + 2476 Note: Some frames were hidden by frame recognizers ```
jdoerfert
pushed a commit
that referenced
this pull request
Aug 30, 2024
) Currently, process of replacing bitwise operations consisting of `LSR`/`LSL` with `And` is performed by `DAGCombiner`. However, in certain cases, the `AND` generated by this process can be removed. Consider following case: ``` lsr x8, x8, llvm#56 and x8, x8, #0xfc ldr w0, [x2, x8] ret ``` In this case, we can remove the `AND` by changing the target of `LDR` to `[X2, X8, LSL #2]` and right-shifting amount change to 56 to 58. after changed: ``` lsr x8, x8, llvm#58 ldr w0, [x2, x8, lsl #2] ret ``` This patch checks to see if the `SHIFTING` + `AND` operation on load target can be optimized and optimizes it if it can.
jdoerfert
pushed a commit
that referenced
this pull request
Aug 30, 2024
`JITDylibSearchOrderResolver` local variable can be destroyed before
completion of all callbacks. Capture it together with `Deps` in
`OnEmitted` callback.
Original error:
```
==2035==ERROR: AddressSanitizer: stack-use-after-return on address 0x7bebfa155b70 at pc 0x7ff2a9a88b4a bp 0x7bec08d51980 sp 0x7bec08d51978
READ of size 8 at 0x7bebfa155b70 thread T87 (tf_xla-cpu-llvm)
#0 0x7ff2a9a88b49 in operator() llvm/lib/ExecutionEngine/Orc/RTDyldObjectLinkingLayer.cpp:55:58
#1 0x7ff2a9a88b49 in __invoke<(lambda at llvm/lib/ExecutionEngine/Orc/RTDyldObjectLinkingLayer.cpp:55:9) &, const llvm::DenseMap<llvm::orc::JITDylib *, llvm::DenseSet<llvm::orc::SymbolStringPtr, llvm::DenseMapInfo<llvm::orc::SymbolStringPtr, void> >, llvm::DenseMapInfo<llvm::orc::JITDylib *, void>, llvm::detail::DenseMapPair<llvm::orc::JITDylib *, llvm::DenseSet<llvm::orc::SymbolStringPtr, llvm::DenseMapInfo<llvm::orc::SymbolStringPtr, void> > > > &> libcxx/include/__type_traits/invoke.h:149:25
#2 0x7ff2a9a88b49 in __call<(lambda at llvm/lib/ExecutionEngine/Orc/RTDyldObjectLinkingLayer.cpp:55:9) &, const llvm::DenseMap<llvm::orc::JITDylib *, llvm::DenseSet<llvm::orc::SymbolStringPtr, llvm::DenseMapInfo<llvm::orc::SymbolStringPtr, void> >, llvm::DenseMapInfo<llvm::orc::JITDylib *, void>, llvm::detail::DenseMapPair<llvm::orc::JITDylib *, llvm::DenseSet<llvm::orc::SymbolStringPtr, llvm::DenseMapInfo<llvm::orc::SymbolStringPtr, void> > > > &> libcxx/include/__type_traits/invoke.h:224:5
#3 0x7ff2a9a88b49 in operator() libcxx/include/__functional/function.h:210:12
#4 0x7ff2a9a88b49 in void std::__u::__function::__policy_invoker<void (llvm::DenseMap<llvm::orc::JITDylib*, llvm::DenseSet<llvm::orc::SymbolStringPtr,
```
jdoerfert
pushed a commit
that referenced
this pull request
Sep 5, 2024
Static destructor can race with calls to notify and trigger tsan
warning.
```
WARNING: ThreadSanitizer: data race (pid=5787)
Write of size 1 at 0x55bec9df8de8 by thread T23:
#0 pthread_mutex_destroy [third_party/llvm/llvm-project/compiler-rt/lib/tsan/rtl/tsan_interceptors_posix.cpp:1344](third_party/llvm/llvm-project/compiler-rt/lib/tsan/rtl/tsan_interceptors_posix.cpp?l=1344&cl=669089572):3 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x1b12affb) (BuildId: ff25ace8b17d9863348bb1759c47246c)
#1 __libcpp_recursive_mutex_destroy [third_party/crosstool/v18/stable/src/libcxx/include/__thread/support/pthread.h:91](third_party/crosstool/v18/stable/src/libcxx/include/__thread/support/pthread.h?l=91&cl=669089572):10 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x4523d4e9) (BuildId: ff25ace8b17d9863348bb1759c47246c)
#2 std::__tsan::recursive_mutex::~recursive_mutex() [third_party/crosstool/v18/stable/src/libcxx/src/mutex.cpp:52](third_party/crosstool/v18/stable/src/libcxx/src/mutex.cpp?l=52&cl=669089572):11 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x4523d4e9)
#3 ~SmartMutex [third_party/llvm/llvm-project/llvm/include/llvm/Support/Mutex.h:28](third_party/llvm/llvm-project/llvm/include/llvm/Support/Mutex.h?l=28&cl=669089572):11 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x2bcaedfe) (BuildId: ff25ace8b17d9863348bb1759c47246c)
#4 (anonymous namespace)::PerfJITEventListener::~PerfJITEventListener() [third_party/llvm/llvm-project/llvm/lib/ExecutionEngine/PerfJITEvents/PerfJITEventListener.cpp:65](third_party/llvm/llvm-project/llvm/lib/ExecutionEngine/PerfJITEvents/PerfJITEventListener.cpp?l=65&cl=669089572):3 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x2bcaedfe)
#5 cxa_at_exit_callback_installed_at(void*) [third_party/llvm/llvm-project/compiler-rt/lib/tsan/rtl/tsan_interceptors_posix.cpp:437](third_party/llvm/llvm-project/compiler-rt/lib/tsan/rtl/tsan_interceptors_posix.cpp?l=437&cl=669089572):3 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x1b172cb9) (BuildId: ff25ace8b17d9863348bb1759c47246c)
#6 llvm::JITEventListener::createPerfJITEventListener() [third_party/llvm/llvm-project/llvm/lib/ExecutionEngine/PerfJITEvents/PerfJITEventListener.cpp:496](third_party/llvm/llvm-project/llvm/lib/ExecutionEngine/PerfJITEvents/PerfJITEventListener.cpp?l=496&cl=669089572):3 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x2bcad8f5) (BuildId: ff25ace8b17d9863348bb1759c47246c)
```
```
Previous atomic read of size 1 at 0x55bec9df8de8 by thread T192 (mutexes: write M0, write M1):
#0 pthread_mutex_unlock [third_party/llvm/llvm-project/compiler-rt/lib/tsan/rtl/tsan_interceptors_posix.cpp:1387](third_party/llvm/llvm-project/compiler-rt/lib/tsan/rtl/tsan_interceptors_posix.cpp?l=1387&cl=669089572):3 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x1b12b6bb) (BuildId: ff25ace8b17d9863348bb1759c47246c)
#1 __libcpp_recursive_mutex_unlock [third_party/crosstool/v18/stable/src/libcxx/include/__thread/support/pthread.h:87](third_party/crosstool/v18/stable/src/libcxx/include/__thread/support/pthread.h?l=87&cl=669089572):10 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x4523d589) (BuildId: ff25ace8b17d9863348bb1759c47246c)
#2 std::__tsan::recursive_mutex::unlock() [third_party/crosstool/v18/stable/src/libcxx/src/mutex.cpp:64](third_party/crosstool/v18/stable/src/libcxx/src/mutex.cpp?l=64&cl=669089572):11 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x4523d589)
#3 unlock [third_party/llvm/llvm-project/llvm/include/llvm/Support/Mutex.h:47](third_party/llvm/llvm-project/llvm/include/llvm/Support/Mutex.h?l=47&cl=669089572):16 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x2bcaf968) (BuildId: ff25ace8b17d9863348bb1759c47246c)
#4 ~lock_guard [third_party/crosstool/v18/stable/src/libcxx/include/__mutex/lock_guard.h:39](third_party/crosstool/v18/stable/src/libcxx/include/__mutex/lock_guard.h?l=39&cl=669089572):101 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x2bcaf968)
#5 (anonymous namespace)::PerfJITEventListener::notifyObjectLoaded(unsigned long, llvm::object::ObjectFile const&, llvm::RuntimeDyld::LoadedObjectInfo const&) [third_party/llvm/llvm-project/llvm/lib/ExecutionEngine/PerfJITEvents/PerfJITEventListener.cpp:290](https://cs.corp.google.com/piper///depot/google3/third_party/llvm/llvm-project/llvm/lib/ExecutionEngine/PerfJITEvents/PerfJITEventListener.cpp?l=290&cl=669089572):1 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x2bcaf968)
#6 llvm::orc::RTDyldObjectLinkingLayer::onObjEmit(llvm::orc::MaterializationResponsibility&, llvm::object::OwningBinary<llvm::object::ObjectFile>, std::__tsan::unique_ptr<llvm::RuntimeDyld::MemoryManager, std::__tsan::default_delete<llvm::RuntimeDyld::MemoryManager>>, std::__tsan::unique_ptr<llvm::RuntimeDyld::LoadedObjectInfo, std::__tsan::default_delete<llvm::RuntimeDyld::LoadedObjectInfo>>, std::__tsan::unique_ptr<llvm::DenseMap<llvm::orc::JITDylib*, llvm::DenseSet<llvm::orc::SymbolStringPtr, llvm::DenseMapInfo<llvm::orc::SymbolStringPtr, void>>, llvm::DenseMapInfo<llvm::orc::JITDylib*, void>, llvm::detail::DenseMapPair<llvm::orc::JITDylib*, llvm::DenseSet<llvm::orc::SymbolStringPtr, llvm::DenseMapInfo<llvm::orc::SymbolStringPtr, void>>>>, std::__tsan::default_delete<llvm::DenseMap<llvm::orc::JITDylib*, llvm::DenseSet<llvm::orc::SymbolStringPtr, llvm::DenseMapInfo<llvm::orc::SymbolStringPtr, void>>, llvm::DenseMapInfo<llvm::orc::JITDylib*, void>, llvm::detail::DenseMapPair<llvm::orc::JITDylib*, llvm::DenseSet<llvm::orc::SymbolStringPtr, llvm::DenseMapInfo<llvm::orc::SymbolStringPtr, void>>>>>>, llvm::Error) [third_party/llvm/llvm-project/llvm/lib/ExecutionEngine/Orc/RTDyldObjectLinkingLayer.cpp:386](https://cs.corp.google.com/piper///depot/google3/third_party/llvm/llvm-project/llvm/lib/ExecutionEngine/Orc/RTDyldObjectLinkingLayer.cpp?l=386&cl=669089572):10 (be1eb158bb70fc9cf7be2db70407e512890e5c6e20720cd88c69d7d9c26ea531_0200d5f71908+0x2bc404a8) (BuildId: ff25ace8b17d9863348bb1759c47246c)
```
jdoerfert
pushed a commit
that referenced
this pull request
Sep 20, 2024
…llvm#94981) This extends default argument deduction to cover class templates as well, applying only to partial ordering, adding to the provisional wording introduced in llvm#89807. This solves some ambuguity introduced in P0522 regarding how template template parameters are partially ordered, and should reduce the negative impact of enabling `-frelaxed-template-template-args` by default. Given the following example: ```C++ template <class T1, class T2 = float> struct A; template <class T3> struct B; template <template <class T4> class TT1, class T5> struct B<TT1<T5>>; // #1 template <class T6, class T7> struct B<A<T6, T7>>; // #2 template struct B<A<int>>; ``` Prior to P0522, `#2` was picked. Afterwards, this became ambiguous. This patch restores the pre-P0522 behavior, `#2` is picked again.
jdoerfert
pushed a commit
that referenced
this pull request
Sep 20, 2024
When SPARC Asan testing is enabled by PR llvm#107405, many Linux/sparc64 tests just hang like ``` #0 0xf7ae8e90 in syscall () from /usr/lib32/libc.so.6 #1 0x701065e8 in __sanitizer::FutexWait(__sanitizer::atomic_uint32_t*, unsigned int) () at compiler-rt/lib/sanitizer_common/sanitizer_linux.cpp:766 #2 0x70107c90 in Wait () at compiler-rt/lib/sanitizer_common/sanitizer_mutex.cpp:35 #3 0x700f7cac in Lock () at compiler-rt/lib/asan/../sanitizer_common/sanitizer_mutex.h:196 #4 Lock () at compiler-rt/lib/asan/../sanitizer_common/sanitizer_thread_registry.h:98 #5 LockThreads () at compiler-rt/lib/asan/asan_thread.cpp:489 #6 0x700e9c8c in __asan::BeforeFork() () at compiler-rt/lib/asan/asan_posix.cpp:157 #7 0xf7ac83f4 in ?? () from /usr/lib32/libc.so.6 Backtrace stopped: previous frame identical to this frame (corrupt stack?) ``` It turns out that this happens in tests using `internal_fork` (e.g. invoking `llvm-symbolizer`): unlike most other Linux targets, which use `clone`, Linux/sparc64 has to use `__fork` instead. While `clone` doesn't trigger `pthread_atfork` handlers, `__fork` obviously does, causing the hang. To avoid this, this patch disables `InstallAtForkHandler` and lets the ASan tests run to completion. Tested on `sparc64-unknown-linux-gnu`.
jdoerfert
pushed a commit
that referenced
this pull request
Sep 20, 2024
…ap (llvm#108825) This attempts to improve user-experience when LLDB stops on a verbose_trap. Currently if a `__builtin_verbose_trap` triggers, we display the first frame above the call to the verbose_trap. So in the newly added test case, we would've previously stopped here: ``` (lldb) run Process 28095 launched: '/Users/michaelbuch/a.out' (arm64) Process 28095 stopped * thread #1, queue = 'com.apple.main-thread', stop reason = Bounds error: out-of-bounds access frame #1: 0x0000000100003f5c a.out`std::__1::vector<int>::operator[](this=0x000000016fdfebef size=0, (null)=10) at verbose_trap.cpp:6:9 3 template <typename T> 4 struct vector { 5 void operator[](unsigned) { -> 6 __builtin_verbose_trap("Bounds error", "out-of-bounds access"); 7 } 8 }; ``` After this patch, we would stop in the first non-`std` frame: ``` (lldb) run Process 27843 launched: '/Users/michaelbuch/a.out' (arm64) Process 27843 stopped * thread #1, queue = 'com.apple.main-thread', stop reason = Bounds error: out-of-bounds access frame #2: 0x0000000100003f44 a.out`g() at verbose_trap.cpp:14:5 11 12 void g() { 13 std::vector<int> v; -> 14 v[10]; 15 } 16 ``` rdar://134490328
jdoerfert
pushed a commit
that referenced
this pull request
Oct 2, 2024
…ext is not fully initialized (llvm#110481) As this comment around target initialization implies: ``` // This can be NULL if we don't know anything about the architecture or if // the target for an architecture isn't enabled in the llvm/clang that we // built ``` There are cases where we might fail to call `InitBuiltinTypes` when creating the backing `ASTContext` for a `TypeSystemClang`. If that happens, the builtins `QualType`s, e.g., `VoidPtrTy`/`IntTy`/etc., are not initialized and dereferencing them as we do in `GetBuiltinTypeForEncodingAndBitSize` (and other places) will lead to nullptr-dereferences. Example backtrace: ``` (lldb) run Assertion failed: (!isNull() && "Cannot retrieve a NULL type pointer"), function getCommonPtr, file Type.h, line 958. Process 2680 stopped * thread #15, name = '<lldb.process.internal-state(pid=2712)>', stop reason = hit program assert frame #4: 0x000000010cdf3cdc liblldb.20.0.0git.dylib`DWARFASTParserClang::ExtractIntFromFormValue(lldb_private::CompilerType const&, lldb_private::plugin::dwarf::DWARFFormValue const&) const (.cold.1) + liblldb.20.0.0git.dylib`DWARFASTParserClang::ParseObjCMethod(lldb_private::ObjCLanguage::MethodName const&, lldb_private::plugin::dwarf::DWARFDIE const&, lldb_private::CompilerType, ParsedDWARFTypeAttributes , bool) (.cold.1): -> 0x10cdf3cdc <+0>: stp x29, x30, [sp, #-0x10]! 0x10cdf3ce0 <+4>: mov x29, sp 0x10cdf3ce4 <+8>: adrp x0, 545 0x10cdf3ce8 <+12>: add x0, x0, #0xa25 ; "ParseObjCMethod" Target 0: (lldb) stopped. (lldb) bt * thread #15, name = '<lldb.process.internal-state(pid=2712)>', stop reason = hit program assert frame #0: 0x0000000180d08600 libsystem_kernel.dylib`__pthread_kill + 8 frame #1: 0x0000000180d40f50 libsystem_pthread.dylib`pthread_kill + 288 frame #2: 0x0000000180c4d908 libsystem_c.dylib`abort + 128 frame #3: 0x0000000180c4cc1c libsystem_c.dylib`__assert_rtn + 284 * frame #4: 0x000000010cdf3cdc liblldb.20.0.0git.dylib`DWARFASTParserClang::ExtractIntFromFormValue(lldb_private::CompilerType const&, lldb_private::plugin::dwarf::DWARFFormValue const&) const (.cold.1) + frame #5: 0x0000000109d30acc liblldb.20.0.0git.dylib`lldb_private::TypeSystemClang::GetBuiltinTypeForEncodingAndBitSize(lldb::Encoding, unsigned long) + 1188 frame #6: 0x0000000109aaaed4 liblldb.20.0.0git.dylib`DynamicLoaderMacOS::NotifyBreakpointHit(void*, lldb_private::StoppointCallbackContext*, unsigned long long, unsigned long long) + 384 ``` This patch adds a one-time user-visible warning for when we fail to initialize the AST to indicate that initialization went wrong for the given target. Additionally, we add checks for whether one of the `ASTContext` `QualType`s is invalid before dereferencing any builtin types. The warning would look as follows: ``` (lldb) target create "a.out" Current executable set to 'a.out' (arm64). (lldb) b main warning: Failed to initialize builtin ASTContext types for target 'some-unknown-triple'. Printing variables may behave unexpectedly. Breakpoint 1: where = a.out`main + 8 at stepping.cpp:5:14, address = 0x0000000100003f90 ``` rdar://134869779
jdoerfert
pushed a commit
that referenced
this pull request
Oct 9, 2024
Fixes llvm#102703. https://godbolt.org/z/nfj8xsb1Y The following pattern: ``` %2 = and i32 %0, 254 %3 = icmp eq i32 %2, 0 ``` is optimised by instcombine into: ```%3 = icmp ult i32 %0, 2``` However, post instcombine leads to worse aarch64 than the unoptimised version. Pre instcombine: ``` tst w0, #0xfe cset w0, eq ret ``` Post instcombine: ``` and w8, w0, #0xff cmp w8, #2 cset w0, lo ret ``` In the unoptimised version, SelectionDAG converts `SETCC (AND X 254) 0 EQ` into `CSEL 0 1 1 (ANDS X 254)`, which gets emitted as a `tst`. In the optimised version, SelectionDAG converts `SETCC (AND X 255) 2 ULT` into `CSEL 0 1 2 (SUBS (AND X 255) 2)`, which gets emitted as an `and`/`cmp`. This PR adds an optimisation to `AArch64ISelLowering`, converting `SETCC (AND X Y) Z ULT` into `SETCC (AND X (Y & ~(Z - 1))) 0 EQ` when `Z` is a power of two. This makes SelectionDAG/Codegen produce the same optimised code for both examples.
jdoerfert
pushed a commit
that referenced
this pull request
Jan 7, 2025
llvm#118923) …d reentry. These utilities provide new, more generic and easier to use support for lazy compilation in ORC. LazyReexportsManager is an alternative to LazyCallThroughManager. It takes requests for lazy re-entry points in the form of an alias map: lazy-reexports = { ( <entry point symbol #1>, <implementation symbol #1> ), ( <entry point symbol #2>, <implementation symbol #2> ), ... ( <entry point symbol #n>, <implementation symbol #n> ) } LazyReexportsManager then: 1. binds the entry points to the implementation names in an internal table. 2. creates a JIT re-entry trampoline for each entry point. 3. creates a redirectable symbol for each of the entry point name and binds redirectable symbol to the corresponding reentry trampoline. When an entry point symbol is first called at runtime (which may be on any thread of the JIT'd program) it will re-enter the JIT via the trampoline and trigger a lookup for the implementation symbol stored in LazyReexportsManager's internal table. When the lookup completes the entry point symbol will be updated (via the RedirectableSymbolManager) to point at the implementation symbol, and execution will proceed to the implementation symbol. Actual construction of the re-entry trampolines and redirectable symbols is delegated to an EmitTrampolines functor and the RedirectableSymbolsManager respectively. JITLinkReentryTrampolines.h provides a JITLink-based implementation of the EmitTrampolines functor. (AArch64 only in this patch, but other architectures will be added in the near future). Register state save and reentry functionality is added to the ORC runtime in the __orc_rt_sysv_resolve and __orc_rt_resolve_implementation functions (the latter is generic, the former will need custom implementations for each ABI and architecture to be supported, however this should be much less effort than the existing OrcABISupport approach, since the ORC runtime allows this code to be written as native assembly). The resulting system: 1. Works equally well for in-process and out-of-process JIT'd code. 2. Requires less boilerplate to set up. Given an ObjectLinkingLayer and PlatformJD (JITDylib containing the ORC runtime), setup is just: ```c++ auto RSMgr = JITLinkRedirectableSymbolManager::Create(OLL); if (!RSMgr) return RSMgr.takeError(); auto LRMgr = createJITLinkLazyReexportsManager(OLL, **RSMgr, PlatformJD); if (!LRMgr) return LRMgr.takeError(); ``` after which lazy reexports can be introduced with: ```c++ JD.define(lazyReexports(LRMgr, <alias map>)); ``` LazyObectLinkingLayer is updated to use this new method, but the LLVM-IR level CompileOnDemandLayer will continue to use LazyCallThroughManager and OrcABISupport until the new system supports a wider range of architectures and ABIs. The llvm-jitlink utility's -lazy option now uses the new scheme. Since it depends on the ORC runtime, the lazy-link.ll testcase and associated helpers are moved to the ORC runtime.
jdoerfert
pushed a commit
that referenced
this pull request
Jan 7, 2025
The Clang binary (and any binary linking Clang as a library), when built
using PIE, ends up with a pretty shocking number of dynamic relocations
to apply to the executable image: roughly 400k.
Each of these takes up binary space in the executable, and perhaps most
interestingly takes start-up time to apply the relocations.
The largest pattern I identified were the strings used to describe
target builtins. The addresses of these string literals were stored into
huge arrays, each one requiring a dynamic relocation. The way to avoid
this is to design the target builtins to use a single large table of
strings and offsets within the table for the individual strings. This
switches the builtin management to such a scheme.
This saves over 100k dynamic relocations by my measurement, an over 25%
reduction. Just looking at byte size improvements, using the `bloaty`
tool to compare a newly built `clang` binary to an old one:
```
FILE SIZE VM SIZE
-------------- --------------
+1.4% +653Ki +1.4% +653Ki .rodata
+0.0% +960 +0.0% +960 .text
+0.0% +197 +0.0% +197 .dynstr
+0.0% +184 +0.0% +184 .eh_frame
+0.0% +96 +0.0% +96 .dynsym
+0.0% +40 +0.0% +40 .eh_frame_hdr
+114% +32 [ = ] 0 [Unmapped]
+0.0% +20 +0.0% +20 .gnu.hash
+0.0% +8 +0.0% +8 .gnu.version
+0.9% +7 +0.9% +7 [LOAD #2 [R]]
[ = ] 0 -75.4% -3.00Ki .relro_padding
-16.1% -802Ki -16.1% -802Ki .data.rel.ro
-27.3% -2.52Mi -27.3% -2.52Mi .rela.dyn
-1.6% -2.66Mi -1.6% -2.66Mi TOTAL
```
We get a 16% reduction in the `.data.rel.ro` section, and nearly 30%
reduction in `.rela.dyn` where those reloctaions are stored.
This is also visible in my benchmarking of binary start-up overhead at
least:
```
Benchmark 1: ./old_clang --version
Time (mean ± σ): 17.6 ms ± 1.5 ms [User: 4.1 ms, System: 13.3 ms]
Range (min … max): 14.2 ms … 22.8 ms 162 runs
Benchmark 2: ./new_clang --version
Time (mean ± σ): 15.5 ms ± 1.4 ms [User: 3.6 ms, System: 11.8 ms]
Range (min … max): 12.4 ms … 20.3 ms 216 runs
Summary
'./new_clang --version' ran
1.13 ± 0.14 times faster than './old_clang --version'
```
We get about 2ms faster `--version` runs. While there is a lot of noise
in binary execution time, this delta is pretty consistent, and
represents over 10% improvement. This is particularly interesting to me
because for very short source files, repeatedly starting the `clang`
binary is actually the dominant cost. For example, `configure` scripts
running against the `clang` compiler are slow in large part because of
binary start up time, not the time to process the actual inputs to the
compiler.
----
This PR implements the string tables using `constexpr` code and the
existing macro system. I understand that the builtins are moving towards
a TableGen model, and if complete that would provide more options for
modeling this. Unfortunately, that migration isn't complete, and even
the parts that are migrated still rely on the ability to break out of
the TableGen model and directly expand an X-macro style `BUILTIN(...)`
textually. I looked at trying to complete the move to TableGen, but it
would both require the difficult migration of the remaining targets, and
solving some tricky problems with how to move away from any macro-based
expansion.
I was also able to find a reasonably clean and effective way of doing
this with the existing macros and some `constexpr` code that I think is
clean enough to be a pretty good intermediate state, and maybe give a
good target for the eventual TableGen solution. I was also able to
factor the macros into set of consistent patterns that avoids a
significant regression in overall boilerplate.
jdoerfert
pushed a commit
that referenced
this pull request
May 9, 2025
The mcmodel=tiny memory model is only valid on ARM targets. While trying this on X86 compiler throws an internal error along with stack dump. llvm#125641 This patch resolves the issue. Reduced test case: ``` #include <stdio.h> int main( void ) { printf( "Hello, World!\n" ); return 0; } ``` ``` 0. Program arguments: /opt/compiler-explorer/clang-trunk/bin/clang++ -gdwarf-4 -g -o /app/output.s -fno-verbose-asm -S --gcc-toolchain=/opt/compiler-explorer/gcc-snapshot -fcolor-diagnostics -fno-crash-diagnostics -mcmodel=tiny <source> 1. <eof> parser at end of file #0 0x0000000003b10218 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3b10218) #1 0x0000000003b0e35c llvm::sys::CleanupOnSignal(unsigned long) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3b0e35c) #2 0x0000000003a5dbc3 llvm::CrashRecoveryContext::HandleExit(int) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3a5dbc3) #3 0x0000000003b05cfe llvm::sys::Process::Exit(int, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3b05cfe) #4 0x0000000000d4e3eb LLVMErrorHandler(void*, char const*, bool) cc1_main.cpp:0:0 #5 0x0000000003a67c93 llvm::report_fatal_error(llvm::Twine const&, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3a67c93) #6 0x0000000003a67df8 (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3a67df8) #7 0x0000000002549148 llvm::X86TargetMachine::X86TargetMachine(llvm::Target const&, llvm::Triple const&, llvm::StringRef, llvm::StringRef, llvm::TargetOptions const&, std::optional<llvm::Reloc::Model>, std::optional<llvm::CodeModel::Model>, llvm::CodeGenOptLevel, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x2549148) #8 0x00000000025491fc llvm::RegisterTargetMachine<llvm::X86TargetMachine>::Allocator(llvm::Target const&, llvm::Triple const&, llvm::StringRef, llvm::StringRef, llvm::TargetOptions const&, std::optional<llvm::Reloc::Model>, std::optional<llvm::CodeModel::Model>, llvm::CodeGenOptLevel, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x25491fc) #9 0x0000000003db74cc clang::emitBackendOutput(clang::CompilerInstance&, clang::CodeGenOptions&, llvm::StringRef, llvm::Module*, clang::BackendAction, llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem>, std::unique_ptr<llvm::raw_pwrite_stream, std::default_delete<llvm::raw_pwrite_stream>>, clang::BackendConsumer*) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3db74cc) #10 0x0000000004460d95 clang::BackendConsumer::HandleTranslationUnit(clang::ASTContext&) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x4460d95) #11 0x00000000060005ec clang::ParseAST(clang::Sema&, bool, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x60005ec) #12 0x00000000044614b5 clang::CodeGenAction::ExecuteAction() (/opt/compiler-explorer/clang-trunk/bin/clang+++0x44614b5) #13 0x0000000004737121 clang::FrontendAction::Execute() (/opt/compiler-explorer/clang-trunk/bin/clang+++0x4737121) #14 0x00000000046b777b clang::CompilerInstance::ExecuteAction(clang::FrontendAction&) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x46b777b) #15 0x00000000048229e3 clang::ExecuteCompilerInvocation(clang::CompilerInstance*) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x48229e3) #16 0x0000000000d50621 cc1_main(llvm::ArrayRef<char const*>, char const*, void*) (/opt/compiler-explorer/clang-trunk/bin/clang+++0xd50621) #17 0x0000000000d48e2d ExecuteCC1Tool(llvm::SmallVectorImpl<char const*>&, llvm::ToolContext const&) driver.cpp:0:0 #18 0x00000000044acc99 void llvm::function_ref<void ()>::callback_fn<clang::driver::CC1Command::Execute(llvm::ArrayRef<std::optional<llvm::StringRef>>, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char>>*, bool*) const::'lambda'()>(long) Job.cpp:0:0 #19 0x0000000003a5dac3 llvm::CrashRecoveryContext::RunSafely(llvm::function_ref<void ()>) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3a5dac3) #20 0x00000000044aceb9 clang::driver::CC1Command::Execute(llvm::ArrayRef<std::optional<llvm::StringRef>>, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char>>*, bool*) const (.part.0) Job.cpp:0:0 #21 0x00000000044710dd clang::driver::Compilation::ExecuteCommand(clang::driver::Command const&, clang::driver::Command const*&, bool) const (/opt/compiler-explorer/clang-trunk/bin/clang+++0x44710dd) #22 0x0000000004472071 clang::driver::Compilation::ExecuteJobs(clang::driver::JobList const&, llvm::SmallVectorImpl<std::pair<int, clang::driver::Command const*>>&, bool) const (/opt/compiler-explorer/clang-trunk/bin/clang+++0x4472071) #23 0x000000000447c3fc clang::driver::Driver::ExecuteCompilation(clang::driver::Compilation&, llvm::SmallVectorImpl<std::pair<int, clang::driver::Command const*>>&) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x447c3fc) llvm#24 0x0000000000d4d2b1 clang_main(int, char**, llvm::ToolContext const&) (/opt/compiler-explorer/clang-trunk/bin/clang+++0xd4d2b1) llvm#25 0x0000000000c12464 main (/opt/compiler-explorer/clang-trunk/bin/clang+++0xc12464) llvm#26 0x00007ae43b029d90 (/lib/x86_64-linux-gnu/libc.so.6+0x29d90) llvm#27 0x00007ae43b029e40 __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x29e40) llvm#28 0x0000000000d488c5 _start (/opt/compiler-explorer/clang-trunk/bin/clang+++0xd488c5) ``` --------- Co-authored-by: Shashwathi N <[email protected]>
jdoerfert
pushed a commit
that referenced
this pull request
May 19, 2025
… `getForwardSlice` matchers (llvm#115670) Improve mlir-query tool by implementing `getBackwardSlice` and `getForwardSlice` matchers. As an addition `SetQuery` also needed to be added to enable custom configuration for each query. e.g: `inclusive`, `omitUsesFromAbove`, `omitBlockArguments`. Note: backwardSlice and forwardSlice algoritms are the same as the ones in `mlir/lib/Analysis/SliceAnalysis.cpp` Example of current matcher. The query was made to the file: `mlir/test/mlir-query/complex-test.mlir` ```mlir ./mlir-query /home/dbudii/personal/llvm-project/mlir/test/mlir-query/complex-test.mlir -c "match getDefinitions(hasOpName(\"arith.add f\"),2)" Match #1: /home/dbudii/personal/llvm-project/mlir/test/mlir-query/complex-test.mlir:5:8: %0 = linalg.generic {indexing_maps = [#map, #map], iterator_types = ["parallel", "parallel"]} ins(%arg0 : tensor<5x5xf32>) outs(%arg1 : tensor<5x5xf32>) { ^ /home/dbudii/personal/llvm-project/mlir/test/mlir-query/complex-test.mlir:7:10: note: "root" binds here %2 = arith.addf %in, %in : f32 ^ Match #2: /home/dbudii/personal/llvm-project/mlir/test/mlir-query/complex-test.mlir:10:16: %collapsed = tensor.collapse_shape %0 [[0, 1]] : tensor<5x5xf32> into tensor<25xf32> ^ /home/dbudii/personal/llvm-project/mlir/test/mlir-query/complex-test.mlir:13:11: %c2 = arith.constant 2 : index ^ /home/dbudii/personal/llvm-project/mlir/test/mlir-query/complex-test.mlir:14:18: %extracted = tensor.extract %collapsed[%c2] : tensor<25xf32> ^ /home/dbudii/personal/llvm-project/mlir/test/mlir-query/complex-test.mlir:15:10: note: "root" binds here %2 = arith.addf %extracted, %extracted : f32 ^ 2 matches. ```
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.