yuzu-android/src/video_core/rasterizer_accelerated.cpp
comex d7c532d889 Fixes and workarounds to make UBSan happier on macOS
There are still some other issues not addressed here, but it's a start.

Workarounds for false-positive reports:

- `RasterizerAccelerated`: Put a gigantic array behind a `unique_ptr`,
  because UBSan has a [hardcoded limit](https://stackoverflow.com/questions/64531383/c-runtime-error-using-fsanitize-undefined-object-has-a-possibly-invalid-vp)
  of how big it thinks objects can be, specifically when dealing with
  offset-to-top values used with multiple inheritance.  Hopefully this
  doesn't have a performance impact.

- `QueryCacheBase::QueryCacheBase`: Avoid an operation that UBSan thinks
  is UB even though it at least arguably isn't.  See the link in the
  comment for more information.

Fixes for correct reports:

- `PageTable`, `Memory`: Use `uintptr_t` values instead of pointers to
  avoid UB from pointer overflow (when pointer arithmetic wraps around
  the address space).

- `KScheduler::Reload`: `thread->GetOwnerProcess()` can be `nullptr`;
  avoid calling methods on it in this case.  (The existing code returns
  a garbage reference to a field, which is then passed into
  `LoadWatchpointArray`, and apparently it's never used, so it's
  harmless in practice but still triggers UBSan.)

- `KAutoObject::Close`: This function calls `this->Destroy()`, which
  overwrites the beginning of the object with junk (specifically a free
  list pointer).  Then it calls `this->UnregisterWithKernel()`.  UBSan
  complains about a type mismatch because the vtable has been
  overwritten, and I believe this is indeed UB.  `UnregisterWithKernel`
  also loads `m_kernel` from the 'freed' object, which seems to be
  technically safe (the overwriting doesn't extend as far as that
  field), but seems dubious.  Switch to a `static` method and load
  `m_kernel` in advance.
2023-07-15 12:00:28 -07:00

73 lines
2.6 KiB
C++

// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#include <atomic>
#include "common/assert.h"
#include "common/common_types.h"
#include "common/div_ceil.h"
#include "core/memory.h"
#include "video_core/rasterizer_accelerated.h"
namespace VideoCore {
using namespace Core::Memory;
RasterizerAccelerated::RasterizerAccelerated(Memory& cpu_memory_)
: cached_pages(std::make_unique<CachedPages>()), cpu_memory{cpu_memory_} {}
RasterizerAccelerated::~RasterizerAccelerated() = default;
void RasterizerAccelerated::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
u64 uncache_begin = 0;
u64 cache_begin = 0;
u64 uncache_bytes = 0;
u64 cache_bytes = 0;
std::atomic_thread_fence(std::memory_order_acquire);
const u64 page_end = Common::DivCeil(addr + size, YUZU_PAGESIZE);
for (u64 page = addr >> YUZU_PAGEBITS; page != page_end; ++page) {
std::atomic_uint16_t& count = cached_pages->at(page >> 2).Count(page);
if (delta > 0) {
ASSERT_MSG(count.load(std::memory_order::relaxed) < UINT16_MAX, "Count may overflow!");
} else if (delta < 0) {
ASSERT_MSG(count.load(std::memory_order::relaxed) > 0, "Count may underflow!");
} else {
ASSERT_MSG(false, "Delta must be non-zero!");
}
// Adds or subtracts 1, as count is a unsigned 8-bit value
count.fetch_add(static_cast<u16>(delta), std::memory_order_release);
// Assume delta is either -1 or 1
if (count.load(std::memory_order::relaxed) == 0) {
if (uncache_bytes == 0) {
uncache_begin = page;
}
uncache_bytes += YUZU_PAGESIZE;
} else if (uncache_bytes > 0) {
cpu_memory.RasterizerMarkRegionCached(uncache_begin << YUZU_PAGEBITS, uncache_bytes,
false);
uncache_bytes = 0;
}
if (count.load(std::memory_order::relaxed) == 1 && delta > 0) {
if (cache_bytes == 0) {
cache_begin = page;
}
cache_bytes += YUZU_PAGESIZE;
} else if (cache_bytes > 0) {
cpu_memory.RasterizerMarkRegionCached(cache_begin << YUZU_PAGEBITS, cache_bytes, true);
cache_bytes = 0;
}
}
if (uncache_bytes > 0) {
cpu_memory.RasterizerMarkRegionCached(uncache_begin << YUZU_PAGEBITS, uncache_bytes, false);
}
if (cache_bytes > 0) {
cpu_memory.RasterizerMarkRegionCached(cache_begin << YUZU_PAGEBITS, cache_bytes, true);
}
}
} // namespace VideoCore