diff options
| author | Mark Rowe <mrowe@bdash.net.nz> | 2025-01-30 09:48:06 -0800 |
|---|---|---|
| committer | Mason Reed <mason@vector35.com> | 2025-02-12 18:43:02 -0500 |
| commit | 1ffdd4da176f1990cdf0ff4d0079a3ccb4e51206 (patch) | |
| tree | c71085230d083aadea24639466da297fda4165ba /view/sharedcache/core/VM.h | |
| parent | c9bbe933dd53fff5d54c2e4f218e4a89f6c58fcd (diff) | |
[SharedCache] Rework how file accessors are handled
Previously, `MMappedFileAccessor::Open` attempted to impose a fixed limit
on the number of file accessors that were live at one time. This was
done because the default file descriptor limit on some platforms is
relatively low (256 on macOS). Given that recent iOS shared caches can
contain 60+ files it is easy to tie up a large percentage of this limit
by opening one or two shared caches.
This is problematic as if the limit imposed by `MMappedFileAccessor::Open`
is reached, the attempt to access the file will block waiting for
another file accessor to be closed. This can lead to a deadlock.
This commit makes three changes to this strategy:
1. It attempts to raise the file descriptor limit to 1024. Unix systems
support both soft and hard file descriptor limits. The soft file
descriptor limit is what is enforced, but a process can explicitly
raise the limit to any value below the hard limit if it wishes.
2. The fixed limit on open files accessors is changed to a soft limit.
`FileAccessorCache` is introduced to manage the caching of file
accessors. It provides a basic LRU cache. Whenever a new file is
opened, the cache of open accessors is pruned to stay below the
target limit (50% of the soft file descriptor limit). This limiting
is primarily done to allow files containing dirty pages to be
unmapped if they're no longer being used.
LRU isn't the optimal strategy for this, but it is simple to
implement and understand. Ideally there'd be some access time
component to the cache so files that haven't been accessed can be
released.
3. File accessors are cached even for sessions corresponding to views
that have been closed. The "Open Selection with Options..." context
menu hits this case as a view is created and destroyed as part of
populating the options dialog, and the real view is then created in
the same session.
The most significant benefit of this change is that the logic around
opening / closing file accessors is simpler and can no longer deadlock.
While working on this I noticed that `SharedCache` opened some files via
`MMappedFileAccessor::Open` without specifying a post-open operation to
apply slide information. Instead, it would explicitly apply the slide
information after opening the file. This works fine so long as the file
accessor limit is never hit. If it is hit and the accessor is closed,
the next time the file is opened it will not have any slide information
applied. This would lead to very confusing bugs.
Diffstat (limited to 'view/sharedcache/core/VM.h')
| -rw-r--r-- | view/sharedcache/core/VM.h | 110 |
1 files changed, 50 insertions, 60 deletions
diff --git a/view/sharedcache/core/VM.h b/view/sharedcache/core/VM.h index dd4f3842..402df196 100644 --- a/view/sharedcache/core/VM.h +++ b/view/sharedcache/core/VM.h @@ -4,62 +4,26 @@ #ifndef SHAREDCACHE_VM_H #define SHAREDCACHE_VM_H + #include <binaryninjaapi.h> -#include <condition_variable> + +#include <list> +#include <mutex> +#include <unordered_map> void VMShutdown(); std::string ResolveFilePath(BinaryNinja::Ref<BinaryNinja::BinaryView> dscView, const std::string& path); -class counting_semaphore { -public: - explicit counting_semaphore(int count = 0) : count_(count) {} - - void release(int update = 1) { - std::unique_lock<std::mutex> lock(mutex_); - count_ += update; - cv_.notify_all(); - } - - void acquire() { - std::unique_lock<std::mutex> lock(mutex_); - cv_.wait(lock, [this]() { return count_ > 0; }); - --count_; - } - - bool try_acquire() { - std::unique_lock<std::mutex> lock(mutex_); - if (count_ > 0) { - --count_; - return true; - } - return false; - } - - void set_count(int new_count) { - std::unique_lock<std::mutex> lock(mutex_); - count_ = new_count; - cv_.notify_all(); - } - -private: - std::mutex mutex_; - std::condition_variable cv_; - int count_; -}; - - template <typename T> class SelfAllocatingWeakPtr { public: - SelfAllocatingWeakPtr(std::function<std::shared_ptr<T>()> allocator, std::function<void(std::shared_ptr<T>)> postAlloc) - : allocator(allocator), postAlloc(postAlloc) {} + SelfAllocatingWeakPtr(std::function<std::shared_ptr<T>()> allocator) : allocator(allocator) {} std::shared_ptr<T> lock() { std::shared_ptr<T> sharedPtr = weakPtr.lock(); if (!sharedPtr) { sharedPtr = allocator(); - postAlloc(sharedPtr); weakPtr = sharedPtr; } return sharedPtr; @@ -71,8 +35,7 @@ public: private: std::weak_ptr<T> weakPtr; // Weak reference to the object - std::function<std::shared_ptr<T>()> allocator; // Function to recreate the object - std::function<void(std::shared_ptr<T>)> postAlloc; // Function to call after the object is allocated + std::function<std::shared_ptr<T>()> allocator; // Function to recreate the object }; @@ -106,26 +69,20 @@ class MMAP { class LazyMappedFileAccessor : public SelfAllocatingWeakPtr<MMappedFileAccessor> { public: - LazyMappedFileAccessor(std::string filePath, std::function<std::shared_ptr<MMappedFileAccessor>()> allocator, - std::function<void(std::shared_ptr<MMappedFileAccessor>)> postAlloc) - : SelfAllocatingWeakPtr(std::move(allocator), std::move(postAlloc)), m_filePath(std::move(filePath)) { - } + LazyMappedFileAccessor( + std::string filePath, std::function<std::shared_ptr<MMappedFileAccessor>(const std::string&)> allocator) : + SelfAllocatingWeakPtr([this, allocator = std::move(allocator)] { return allocator(m_filePath); }), + m_filePath(std::move(filePath)) + {} + ~LazyMappedFileAccessor() = default; - std::string_view filePath() const { return m_filePath; } + std::string_view filePath() const { return m_filePath; } private: std::string m_filePath; }; -static uint64_t maxFPLimit; -static std::mutex fileAccessorDequeMutex; -static std::unordered_map<uint64_t, std::deque<std::shared_ptr<MMappedFileAccessor>>> fileAccessorReferenceHolder; -static std::set<uint64_t> blockedSessionIDs; -static std::mutex fileAccessorsMutex; -static std::unordered_map<std::string, std::shared_ptr<LazyMappedFileAccessor>> fileAccessors; -static counting_semaphore fileAccessorSemaphore(0); - -static std::atomic<uint64_t> mmapCount = 0; +uint64_t MMapCount(); class MMappedFileAccessor { std::string m_path; @@ -142,9 +99,9 @@ public: static void InitialVMSetup(); - std::string Path() const { return m_path; }; + const std::string& Path() const { return m_path; }; - size_t Length() const { return m_mmap.len; }; + size_t Length() const { return m_mmap.len; }; void *Data() const { return m_mmap._mmap; }; @@ -201,6 +158,39 @@ public: T Read(size_t address); }; +class FileAccessorCache +{ +public: + static FileAccessorCache& Shared(); + + std::shared_ptr<LazyMappedFileAccessor> OpenLazily(BinaryNinja::Ref<BinaryNinja::BinaryView> dscView, + const uint64_t sessionID, const std::string& path, + std::function<void(std::shared_ptr<MMappedFileAccessor>)> postAllocationRoutine); + + void SetCacheSize(uint64_t size); + +private: + FileAccessorCache(); + + std::shared_ptr<MMappedFileAccessor> Open( + BinaryNinja::Ref<BinaryNinja::BinaryView> dscView, const uint64_t sessionID, const std::string& path); + + void Close(MMappedFileAccessor* accessor); + + void RecordAccess(const std::string& path); + void EvictFromCacheIfNeeded(); + + std::mutex m_mutex; + std::unordered_map<std::string, std::shared_ptr<LazyMappedFileAccessor>> m_lazyAccessors; + + // Ordered from least recently opened (front) to most recently opened (end). + std::list<std::string> m_leastRecentlyOpened; + std::unordered_map<std::string, std::pair<std::shared_ptr<MMappedFileAccessor>, std::list<std::string>::iterator>> + m_accessors; + + uint64_t m_cacheSize = 8; +}; + struct PageMapping { std::shared_ptr<LazyMappedFileAccessor> fileAccessor; size_t fileOffset; |
