mirror of
https://github.com/llvm/llvm-project.git
synced 2025-04-17 23:36:40 +00:00
b2dea4fd22
This is an optional mechanism that automatically detects roots. It's a best-effort mechanism, and its main goal is to *avoid* pointing at the message pump function as a root. This is the function that polls message queue(s) in an infinite loop, and is thus a bad root (it never exits). High-level, when collection is requested - which should happen when a server has already been set up and handing requests - we spend a bit of time sampling all the server's threads. Each sample is a stack which we insert in a `PerThreadCallsiteTrie`. After a while, we run for each `PerThreadCallsiteTrie` the root detection logic. We then traverse all the `FunctionData`, find the ones matching the detected roots, and allocate a `ContextRoot` for them. From here, we special case `FunctionData` objects, in `__llvm_ctx_profile_get_context, that have a `CtxRoot` and route them to `__llvm_ctx_profile_start_context`. For this to work, on the llvm side, we need to have all functions call `__llvm_ctx_profile_release_context` because they _might_ be roots. This comes at a slight (percentages) penalty during collection - which we can afford since the overall technique is ~5x faster than normal instrumentation. We can later explore conditionally enabling autoroot detection and avoiding this penalty, if desired. Note that functions that `musttail call` can't have their return instrumented this way, and a subsequent patch will harden the mechanism against this case. The mechanism could be used in combination with explicit root specification, too.
101 lines
3.4 KiB
C++
101 lines
3.4 KiB
C++
/*===- RootAutodetector.h- auto-detect roots for ctxprof -----------------===*\
|
|
|*
|
|
|* Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
|* See https://llvm.org/LICENSE.txt for license information.
|
|
|* SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
|
|*
|
|
\*===----------------------------------------------------------------------===*/
|
|
|
|
#ifndef CTX_PROFILE_ROOTAUTODETECTOR_H_
|
|
#define CTX_PROFILE_ROOTAUTODETECTOR_H_
|
|
|
|
#include "sanitizer_common/sanitizer_dense_map.h"
|
|
#include "sanitizer_common/sanitizer_internal_defs.h"
|
|
#include "sanitizer_common/sanitizer_stacktrace.h"
|
|
#include "sanitizer_common/sanitizer_vector.h"
|
|
#include <pthread.h>
|
|
#include <sanitizer/common_interface_defs.h>
|
|
|
|
using namespace __asan;
|
|
using namespace __sanitizer;
|
|
|
|
namespace __ctx_profile {
|
|
|
|
/// Capture all the stack traces observed for a specific thread. The "for a
|
|
/// specific thread" part is not enforced, but assumed in determineRoots.
|
|
class PerThreadCallsiteTrie {
|
|
protected:
|
|
/// A trie. A node is the address of a callsite in a function activation. A
|
|
/// child is a callsite in the activation made from the callsite
|
|
/// corresponding to the parent.
|
|
struct Trie final {
|
|
const uptr CallsiteAddress;
|
|
uint64_t Count = 0;
|
|
DenseMap<uptr, Trie> Children;
|
|
|
|
Trie(uptr CallsiteAddress = 0) : CallsiteAddress(CallsiteAddress) {}
|
|
};
|
|
Trie TheTrie;
|
|
|
|
/// Return the runtime start address of the function that contains the call at
|
|
/// the runtime address CallsiteAddress. May be overriden for easy testing.
|
|
virtual uptr getFctStartAddr(uptr CallsiteAddress) const;
|
|
|
|
public:
|
|
PerThreadCallsiteTrie(const PerThreadCallsiteTrie &) = delete;
|
|
PerThreadCallsiteTrie(PerThreadCallsiteTrie &&) = default;
|
|
PerThreadCallsiteTrie() = default;
|
|
|
|
virtual ~PerThreadCallsiteTrie() = default;
|
|
|
|
void insertStack(const StackTrace &ST);
|
|
|
|
/// Return the runtime address of root functions, as determined for this
|
|
/// thread, together with the number of samples that included them.
|
|
DenseMap<uptr, uint64_t> determineRoots() const;
|
|
};
|
|
|
|
class RootAutoDetector final {
|
|
// A prime number. We may want to make this configurable at collection start.
|
|
static const uint64_t SampleRate = 6113;
|
|
const unsigned WaitSeconds;
|
|
pthread_t WorkerThread;
|
|
|
|
struct PerThreadSamples {
|
|
PerThreadSamples(RootAutoDetector &Parent);
|
|
|
|
PerThreadCallsiteTrie TrieRoot;
|
|
SpinMutex M;
|
|
};
|
|
SpinMutex AllSamplesMutex;
|
|
SANITIZER_GUARDED_BY(AllSamplesMutex)
|
|
Vector<PerThreadSamples *> AllSamples;
|
|
atomic_uintptr_t &FunctionDataListHead;
|
|
atomic_uintptr_t &Self;
|
|
void collectStack();
|
|
|
|
public:
|
|
RootAutoDetector(atomic_uintptr_t &FunctionDataListHead,
|
|
atomic_uintptr_t &Self, unsigned WaitSeconds)
|
|
: WaitSeconds(WaitSeconds), FunctionDataListHead(FunctionDataListHead),
|
|
Self(Self) {}
|
|
|
|
// Samples the stack at `SampleRate` (rate observed independently on each
|
|
// thread) in thread local `PerThreadCallsiteTrie`s.
|
|
void sample();
|
|
|
|
// Start a thread waiting `WaitSeconds`, after which it uses the
|
|
// `PerThreadCallsiteTrie` data observed so far over all threads to determine
|
|
// roots. Marks those roots by traversing the linked list of FunctionData that
|
|
// starts at `FunctionDataListHead`, and assigning their `CtxRoot`. Finally,
|
|
// resets the `Self` atomic, so that other threads don't continue calling
|
|
// `sample`.
|
|
void start();
|
|
|
|
// join the waiting thread.
|
|
void join();
|
|
};
|
|
|
|
} // namespace __ctx_profile
|
|
#endif
|