Skip to content

Re-org address unalignment access for fast-interp (#597) #249

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Apr 1, 2021
Merged

Re-org address unalignment access for fast-interp (#597) #249

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 8 additions & 6 deletions core/config.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,29 +131,31 @@
#define WASM_ENABLE_LOG 1
#endif

#if defined(BUILD_TARGET_X86_32) || defined(BUILD_TARGET_X86_64)
#define WASM_CPU_SUPPORTS_UNALIGNED_64BIT_ACCESS 1
#ifndef WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS
#if defined(BUILD_TARGET_X86_32) || defined(BUILD_TARGET_X86_64) \
|| defined(BUILD_TARGET_AARCH64)
#define WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS 1
#else
#define WASM_CPU_SUPPORTS_UNALIGNED_64BIT_ACCESS 0
#define WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS 0
#endif
#endif

/* WASM Interpreter labels-as-values feature */
#ifndef WASM_ENABLE_LABELS_AS_VALUES
#ifdef __GNUC__
#define WASM_ENABLE_LABELS_AS_VALUES 1
#else
#define WASM_ENABLE_LABELS_AS_VALUES 0
#endif
#endif

/* Enable fast interpreter or not */
#ifndef WASM_ENABLE_FAST_INTERP
#define WASM_ENABLE_FAST_INTERP 0
#endif

#if WASM_ENABLE_FAST_INTERP != 0
#define WASM_ENABLE_ABS_LABEL_ADDR 1
#define WASM_DEBUG_PREPROCESSOR 0
#else
#define WASM_ENABLE_ABS_LABEL_ADDR 0
#endif

/* Enable opcode counter or not */
Expand Down
15 changes: 0 additions & 15 deletions core/iwasm/aot/aot_runtime.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1001,21 +1001,6 @@ aot_lookup_function(const AOTModuleInstance *module_inst,
return NULL;
}

#define PUT_I64_TO_ADDR(addr, value) do { \
union { int64 val; uint32 parts[2]; } u; \
u.val = (value); \
(addr)[0] = u.parts[0]; \
(addr)[1] = u.parts[1]; \
} while (0)

#define PUT_F64_TO_ADDR(addr, value) do { \
union { float64 val; uint32 parts[2]; } u; \
u.val = (value); \
(addr)[0] = u.parts[0]; \
(addr)[1] = u.parts[1]; \
} while (0)


#ifdef OS_ENABLE_HW_BOUND_CHECK

#define STACK_OVERFLOW_CHECK_GUARD_PAGE_COUNT 3
Expand Down
14 changes: 0 additions & 14 deletions core/iwasm/common/wasm_runtime_common.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2825,20 +2825,6 @@ wasm_runtime_invoke_native_raw(WASMExecEnv *exec_env, void *func_ptr,
* Implementation of wasm_runtime_invoke_native()
*/

#define PUT_I64_TO_ADDR(addr, value) do { \
union { int64 val; uint32 parts[2]; } u; \
u.val = (value); \
(addr)[0] = u.parts[0]; \
(addr)[1] = u.parts[1]; \
} while (0)

#define PUT_F64_TO_ADDR(addr, value) do { \
union { float64 val; uint32 parts[2]; } u; \
u.val = (value); \
(addr)[0] = u.parts[0]; \
(addr)[1] = u.parts[1]; \
} while (0)

/* The invoke native implementation on ARM platform with VFP co-processor */
#if defined(BUILD_TARGET_ARM_VFP) \
|| defined(BUILD_TARGET_THUMB_VFP) \
Expand Down
224 changes: 224 additions & 0 deletions core/iwasm/common/wasm_runtime_common.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,230 @@
extern "C" {
#endif

#if WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS != 0

#define PUT_I64_TO_ADDR(addr, value) do { \
*(int64*)(addr) = (int64)(value); \
} while (0)
#define PUT_F64_TO_ADDR(addr, value) do { \
*(float64*)(addr) = (float64)(value); \
} while (0)

#define GET_I64_FROM_ADDR(addr) (*(int64*)(addr))
#define GET_F64_FROM_ADDR(addr) (*(float64*)(addr))

/* For STORE opcodes */
#define STORE_I64 PUT_I64_TO_ADDR
#define STORE_U32(addr, value) do { \
*(uint32*)(addr) = (uint32)(value); \
} while (0)
#define STORE_U16(addr, value) do { \
*(uint16*)(addr) = (uint16)(value); \
} while (0)

/* For LOAD opcodes */
#define LOAD_I64(addr) (*(int64*)(addr))
#define LOAD_F64(addr) (*(float64*)(addr))
#define LOAD_I32(addr) (*(int32*)(addr))
#define LOAD_U32(addr) (*(uint32*)(addr))
#define LOAD_I16(addr) (*(int16*)(addr))
#define LOAD_U16(addr) (*(uint16*)(addr))

#define STORE_PTR(addr, ptr) do { \
*(void**)addr = (void*)ptr; \
} while (0)
#define LOAD_PTR(addr) (*(void**)(addr))

#else /* WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS != 0 */

#define PUT_I64_TO_ADDR(addr, value) do { \
uint32 *addr_u32 = (uint32*)(addr); \
union { int64 val; uint32 parts[2]; } u; \
u.val = (int64)(value); \
addr_u32[0] = u.parts[0]; \
addr_u32[1] = u.parts[1]; \
} while (0)
#define PUT_F64_TO_ADDR(addr, value) do { \
uint32 *addr_u32 = (uint32*)(addr); \
union { float64 val; uint32 parts[2]; } u; \
u.val = (value); \
addr_u32[0] = u.parts[0]; \
addr_u32[1] = u.parts[1]; \
} while (0)

static inline int64
GET_I64_FROM_ADDR(uint32 *addr)
{
union { int64 val; uint32 parts[2]; } u;
u.parts[0] = addr[0];
u.parts[1] = addr[1];
return u.val;
}

static inline float64
GET_F64_FROM_ADDR (uint32 *addr)
{
union { float64 val; uint32 parts[2]; } u;
u.parts[0] = addr[0];
u.parts[1] = addr[1];
return u.val;
}

/* For STORE opcodes */
#define STORE_I64(addr, value) do { \
uintptr_t addr1 = (uintptr_t)(addr); \
union { int64 val; uint32 u32[2]; \
uint16 u16[4]; uint8 u8[8]; } u; \
if ((addr1 & (uintptr_t)7) == 0) \
*(int64*)(addr) = (int64)(value); \
else { \
u.val = (int64)(value); \
if ((addr1 & (uintptr_t)3) == 0) { \
((uint32*)(addr))[0] = u.u32[0]; \
((uint32*)(addr))[1] = u.u32[1]; \
} \
else if ((addr1 & (uintptr_t)1) == 0) { \
((uint16*)(addr))[0] = u.u16[0]; \
((uint16*)(addr))[1] = u.u16[1]; \
((uint16*)(addr))[2] = u.u16[2]; \
((uint16*)(addr))[3] = u.u16[3]; \
} \
else { \
int32 t; \
for (t = 0; t < 8; t++) \
((uint8*)(addr))[t] = u.u8[t]; \
} \
} \
} while (0)

#define STORE_U32(addr, value) do { \
uintptr_t addr1 = (uintptr_t)(addr); \
union { uint32 val; \
uint16 u16[2]; uint8 u8[4]; } u; \
if ((addr1 & (uintptr_t)3) == 0) \
*(uint32*)(addr) = (uint32)(value); \
else { \
u.val = (uint32)(value); \
if ((addr1 & (uintptr_t)1) == 0) { \
((uint16*)(addr))[0] = u.u16[0]; \
((uint16*)(addr))[1] = u.u16[1]; \
} \
else { \
((uint8*)(addr))[0] = u.u8[0]; \
((uint8*)(addr))[1] = u.u8[1]; \
((uint8*)(addr))[2] = u.u8[2]; \
((uint8*)(addr))[3] = u.u8[3]; \
} \
} \
} while (0)

#define STORE_U16(addr, value) do { \
union { uint16 val; uint8 u8[2]; } u; \
u.val = (uint16)(value); \
((uint8*)(addr))[0] = u.u8[0]; \
((uint8*)(addr))[1] = u.u8[1]; \
} while (0)

/* For LOAD opcodes */
static inline int64
LOAD_I64(void *addr)
{
uintptr_t addr1 = (uintptr_t)addr;
union { int64 val; uint32 u32[2];
uint16 u16[4]; uint8 u8[8]; } u;
if ((addr1 & (uintptr_t)7) == 0)
return *(int64*)addr;

if ((addr1 & (uintptr_t)3) == 0) {
u.u32[0] = ((uint32*)addr)[0];
u.u32[1] = ((uint32*)addr)[1];
}
else if ((addr1 & (uintptr_t)1) == 0) {
u.u16[0] = ((uint16*)addr)[0];
u.u16[1] = ((uint16*)addr)[1];
u.u16[2] = ((uint16*)addr)[2];
u.u16[3] = ((uint16*)addr)[3];
}
else {
int32 t;
for (t = 0; t < 8; t++)
u.u8[t] = ((uint8*)addr)[t];
}
return u.val;
}

static inline float64
LOAD_F64(void *addr)
{
uintptr_t addr1 = (uintptr_t)addr;
union { float64 val; uint32 u32[2];
uint16 u16[4]; uint8 u8[8]; } u;
if ((addr1 & (uintptr_t)7) == 0)
return *(float64*)addr;

if ((addr1 & (uintptr_t)3) == 0) {
u.u32[0] = ((uint32*)addr)[0];
u.u32[1] = ((uint32*)addr)[1];
}
else if ((addr1 & (uintptr_t)1) == 0) {
u.u16[0] = ((uint16*)addr)[0];
u.u16[1] = ((uint16*)addr)[1];
u.u16[2] = ((uint16*)addr)[2];
u.u16[3] = ((uint16*)addr)[3];
}
else {
int32 t;
for (t = 0; t < 8; t++)
u.u8[t] = ((uint8*)addr)[t];
}
return u.val;
}

static inline int32
LOAD_I32(void *addr)
{
uintptr_t addr1 = (uintptr_t)addr;
union { int32 val; uint16 u16[2]; uint8 u8[4]; } u;
if ((addr1 & (uintptr_t)3) == 0)
return *(int32*)addr;

if ((addr1 & (uintptr_t)1) == 0) {
u.u16[0] = ((uint16*)addr)[0];
u.u16[1] = ((uint16*)addr)[1];
}
else {
u.u8[0] = ((uint8*)addr)[0];
u.u8[1] = ((uint8*)addr)[1];
u.u8[2] = ((uint8*)addr)[2];
u.u8[3] = ((uint8*)addr)[3];
}
return u.val;
}

static inline int16
LOAD_I16(void *addr)
{
uintptr_t addr1 = (uintptr_t)addr;
union { int16 val; uint8 u8[2]; } u;
if ((addr1 & (uintptr_t)1)) {
u.u8[0] = ((uint8*)addr)[0];
u.u8[1] = ((uint8*)addr)[1];
return u.val;
}
return *(int16*)addr;
}

#define LOAD_U32(addr) ((uint32)LOAD_I32(addr))
#define LOAD_U16(addr) ((uint16)LOAD_I16(addr))

#if UINTPTR_MAX == UINT32_MAX
#define STORE_PTR(addr, ptr) STORE_U32(addr, (uintptr_t)ptr)
#elif UINTPTR_MAX == UINT64_MAX
#define STORE_PTR(addr, ptr) STORE_I64(addr, (uintptr_t)ptr)
#endif

#endif /* WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS != 0 */

typedef struct WASMModuleCommon {
/* Module type, for module loaded from WASM bytecode binary,
this field is Wasm_Module_Bytecode, and this structure should
Expand Down
Loading