dynarmic: Inline exclusive memory accesses
Inlines implementation of exclusive instructions into JITted code, improving performance of applications relying heavily on these instructions. We also fastmem these instructions for additional speed, with support for appropriate recompilation on fastmem failure. An unsafe optimization to disable the intercore global_monitor is also provided, should one wish to rely solely on cmpxchg semantics for safety. See also: merryhime/dynarmic#664
This commit is contained in:
parent
96d90be59f
commit
16784e5bb3
17 changed files with 114 additions and 8 deletions
|
@ -137,6 +137,8 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
|
|||
config.page_table_pointer_mask_bits = Common::PageTable::ATTRIBUTE_BITS;
|
||||
config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
|
||||
config.only_detect_misalignment_via_page_table_on_page_boundary = true;
|
||||
config.fastmem_exclusive_access = true;
|
||||
config.recompile_on_exclusive_fastmem_failure = true;
|
||||
|
||||
// Multi-process state
|
||||
config.processor_id = core_index;
|
||||
|
@ -178,6 +180,12 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
|
|||
if (!Settings::values.cpuopt_fastmem) {
|
||||
config.fastmem_pointer = nullptr;
|
||||
}
|
||||
if (!Settings::values.cpuopt_fastmem_exclusives) {
|
||||
config.fastmem_exclusive_access = false;
|
||||
}
|
||||
if (!Settings::values.cpuopt_recompile_exclusives) {
|
||||
config.recompile_on_exclusive_fastmem_failure = false;
|
||||
}
|
||||
}
|
||||
|
||||
// Unsafe optimizations
|
||||
|
@ -195,6 +203,9 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
|
|||
if (Settings::values.cpuopt_unsafe_inaccurate_nan) {
|
||||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_InaccurateNaN;
|
||||
}
|
||||
if (Settings::values.cpuopt_unsafe_ignore_global_monitor) {
|
||||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreGlobalMonitor;
|
||||
}
|
||||
}
|
||||
|
||||
// Curated optimizations
|
||||
|
@ -203,6 +214,7 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
|
|||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_UnfuseFMA;
|
||||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreStandardFPCRValue;
|
||||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_InaccurateNaN;
|
||||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreGlobalMonitor;
|
||||
}
|
||||
|
||||
return std::make_unique<Dynarmic::A32::Jit>(config);
|
||||
|
|
|
@ -185,6 +185,9 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
|
|||
config.fastmem_pointer = page_table->fastmem_arena;
|
||||
config.fastmem_address_space_bits = address_space_bits;
|
||||
config.silently_mirror_fastmem = false;
|
||||
|
||||
config.fastmem_exclusive_access = true;
|
||||
config.recompile_on_exclusive_fastmem_failure = true;
|
||||
}
|
||||
|
||||
// Multi-process state
|
||||
|
@ -237,6 +240,12 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
|
|||
if (!Settings::values.cpuopt_fastmem) {
|
||||
config.fastmem_pointer = nullptr;
|
||||
}
|
||||
if (!Settings::values.cpuopt_fastmem_exclusives) {
|
||||
config.fastmem_exclusive_access = false;
|
||||
}
|
||||
if (!Settings::values.cpuopt_recompile_exclusives) {
|
||||
config.recompile_on_exclusive_fastmem_failure = false;
|
||||
}
|
||||
}
|
||||
|
||||
// Unsafe optimizations
|
||||
|
@ -254,6 +263,9 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
|
|||
if (Settings::values.cpuopt_unsafe_fastmem_check) {
|
||||
config.fastmem_address_space_bits = 64;
|
||||
}
|
||||
if (Settings::values.cpuopt_unsafe_ignore_global_monitor) {
|
||||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreGlobalMonitor;
|
||||
}
|
||||
}
|
||||
|
||||
// Curated optimizations
|
||||
|
@ -262,6 +274,7 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
|
|||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_UnfuseFMA;
|
||||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_InaccurateNaN;
|
||||
config.fastmem_address_space_bits = 64;
|
||||
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreGlobalMonitor;
|
||||
}
|
||||
|
||||
return std::make_shared<Dynarmic::A64::Jit>(config);
|
||||
|
|
|
@ -37,8 +37,8 @@ u128 DynarmicExclusiveMonitor::ExclusiveRead128(std::size_t core_index, VAddr ad
|
|||
});
|
||||
}
|
||||
|
||||
void DynarmicExclusiveMonitor::ClearExclusive() {
|
||||
monitor.Clear();
|
||||
void DynarmicExclusiveMonitor::ClearExclusive(std::size_t core_index) {
|
||||
monitor.ClearProcessor(core_index);
|
||||
}
|
||||
|
||||
bool DynarmicExclusiveMonitor::ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) {
|
||||
|
|
|
@ -29,7 +29,7 @@ public:
|
|||
u32 ExclusiveRead32(std::size_t core_index, VAddr addr) override;
|
||||
u64 ExclusiveRead64(std::size_t core_index, VAddr addr) override;
|
||||
u128 ExclusiveRead128(std::size_t core_index, VAddr addr) override;
|
||||
void ClearExclusive() override;
|
||||
void ClearExclusive(std::size_t core_index) override;
|
||||
|
||||
bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) override;
|
||||
bool ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) override;
|
||||
|
|
|
@ -23,7 +23,7 @@ public:
|
|||
virtual u32 ExclusiveRead32(std::size_t core_index, VAddr addr) = 0;
|
||||
virtual u64 ExclusiveRead64(std::size_t core_index, VAddr addr) = 0;
|
||||
virtual u128 ExclusiveRead128(std::size_t core_index, VAddr addr) = 0;
|
||||
virtual void ClearExclusive() = 0;
|
||||
virtual void ClearExclusive(std::size_t core_index) = 0;
|
||||
|
||||
virtual bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) = 0;
|
||||
virtual bool ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) = 0;
|
||||
|
|
|
@ -49,7 +49,7 @@ bool DecrementIfLessThan(Core::System& system, s32* out, VAddr address, s32 valu
|
|||
}
|
||||
} else {
|
||||
// Otherwise, clear our exclusive hold and finish
|
||||
monitor.ClearExclusive();
|
||||
monitor.ClearExclusive(current_core);
|
||||
}
|
||||
|
||||
// We're done.
|
||||
|
@ -78,7 +78,7 @@ bool UpdateIfEqual(Core::System& system, s32* out, VAddr address, s32 value, s32
|
|||
}
|
||||
} else {
|
||||
// Otherwise, clear our exclusive hold and finish.
|
||||
monitor.ClearExclusive();
|
||||
monitor.ClearExclusive(current_core);
|
||||
}
|
||||
|
||||
// We're done.
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue