nv-tegra-mirror/linux-nvgpu
1
0
Fork 0
mirror of git://nv-tegra.nvidia.com/linux-nvgpu.git synced 2025-12-22 17:36:20 +03:00
Code Issues Packages Projects Releases Wiki Activity

gpu: nvgpu: unit: atomic: add non-atomic functions

Browse Source 
In order to show that atomicity tests prove atomicity, we need to show
non-atomic operations fail. This patch is the first step toward that,
which is to add support for non-atomic operations.

Since the common functions for the different operations must now support
3 options (32-bit, 64-bit, and non-atomic), change the macros to
inline functions for decreased complexity and better readability.

JIRA NVGPU-2251

Change-Id: I721373af1eff02d82ade45d1668ad6781bd6d7e2
Signed-off-by: Philip Elcan <pelcan@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2079376
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This commit is contained in:
Philip Elcan
2019-03-21 13:30:25 -04:00
committed by mobile promotions
parent 46121225a3
commit 2c88cd50a4
1 changed files with 374 additions and 147 deletions
Download Patch File Download Diff File Expand all files Collapse all files

521
userspace/units/interface/atomic/atomic.c
View File

@@ -29,12 +29,14 @@
#include <nvgpu/bug.h> #include <nvgpu/bug.h>
struct atomic_struct { struct atomic_struct {
long not_atomic;
nvgpu_atomic_t atomic; nvgpu_atomic_t atomic;
nvgpu_atomic64_t atomic64; nvgpu_atomic64_t atomic64;
}; };
enum atomic_width { enum atomic_type {
WIDTH_32, ATOMIC_32,
WIDTH_64, ATOMIC_64,
NOT_ATOMIC,
}; };
enum atomic_op { enum atomic_op {
op_inc, op_inc,
@@ -48,7 +50,7 @@ enum atomic_op {
}; };
struct atomic_test_args { struct atomic_test_args {
enum atomic_op op; enum atomic_op op;
enum atomic_width width; enum atomic_type type;
long start_val; long start_val;
long loop_count; long loop_count;
long value; /* for add/sub ops */ long value; /* for add/sub ops */
@@ -66,88 +68,313 @@ struct atomic_thread_info {
static pthread_barrier_t thread_barrier; static pthread_barrier_t thread_barrier;
/* /*
* Define macros for atomic ops that have 32b and 64b versions so we can * Define functions for atomic ops that handle all types so we can
* keep the code cleaner. * keep the code cleaner.
*/ */
#define ATOMIC_SET(width, ref, i) \ static inline void func_set(enum atomic_type type, struct atomic_struct *ref,
((width == WIDTH_32) ? \ long val)
nvgpu_atomic_set(&((ref)->atomic), i) : \ {
nvgpu_atomic64_set(&((ref)->atomic64), i)) switch (type) {
case NOT_ATOMIC:
ref->not_atomic = val;
break;
case ATOMIC_32:
nvgpu_atomic_set(&(ref->atomic), val);
break;
case ATOMIC_64:
nvgpu_atomic64_set(&(ref->atomic64), val);
break;
}
}
#define ATOMIC_READ(width, ref) \ static inline long func_read(enum atomic_type type, struct atomic_struct *ref)
((width == WIDTH_32) ? \ {
nvgpu_atomic_read(&((ref)->atomic)) : \ long ret = 0;
nvgpu_atomic64_read(&((ref)->atomic64)))
#define ATOMIC_INC(width, ref) \ switch (type) {
((width == WIDTH_32) ? \ case NOT_ATOMIC:
nvgpu_atomic_inc(&((ref)->atomic)) : \ ret = ref->not_atomic;
nvgpu_atomic64_inc(&((ref)->atomic64))) break;
case ATOMIC_32:
ret = nvgpu_atomic_read(&(ref->atomic));
break;
case ATOMIC_64:
ret = nvgpu_atomic64_read(&(ref->atomic64));
break;
}
return ret;
}
#define ATOMIC_INC_RETURN(width, ref) \ static inline void func_inc(enum atomic_type type, struct atomic_struct *ref)
((width == WIDTH_32) ? \ {
nvgpu_atomic_inc_return(&((ref)->atomic)) : \ switch (type) {
nvgpu_atomic64_inc_return(&((ref)->atomic64))) case NOT_ATOMIC:
++ref->not_atomic;
break;
case ATOMIC_32:
nvgpu_atomic_inc(&(ref->atomic));
break;
case ATOMIC_64:
nvgpu_atomic64_inc(&(ref->atomic64));
break;
}
}
#define ATOMIC_INC_AND_TEST(width, ref) \ static inline long func_inc_return(enum atomic_type type,
((width == WIDTH_32) ? \ struct atomic_struct *ref)
nvgpu_atomic_inc_and_test(&((ref)->atomic)) : \ {
nvgpu_atomic64_inc_and_test(&((ref)->atomic64))) long ret = 0;
#define ATOMIC_DEC(width, ref) \ switch (type) {
((width == WIDTH_32) ? \ case NOT_ATOMIC:
nvgpu_atomic_dec(&((ref)->atomic)) : \ ++ref->not_atomic;
nvgpu_atomic64_dec(&((ref)->atomic64))) ret = ref->not_atomic;
break;
case ATOMIC_32:
ret = nvgpu_atomic_inc_return(&(ref->atomic));
break;
case ATOMIC_64:
ret = nvgpu_atomic64_inc_return(&(ref->atomic64));
break;
}
return ret;
}
#define ATOMIC_DEC_RETURN(width, ref) \ static inline bool func_inc_and_test(enum atomic_type type,
((width == WIDTH_32) ? \ struct atomic_struct *ref)
nvgpu_atomic_dec_return(&((ref)->atomic)) : \ {
nvgpu_atomic64_dec_return(&((ref)->atomic64))) bool ret = false;
#define ATOMIC_DEC_AND_TEST(width, ref) \ switch (type) {
((width == WIDTH_32) ? \ case NOT_ATOMIC:
nvgpu_atomic_dec_and_test(&((ref)->atomic)) : \ ++ref->not_atomic;
nvgpu_atomic64_dec_and_test(&((ref)->atomic64))) ret = (ref->not_atomic == 0);
break;
case ATOMIC_32:
ret = nvgpu_atomic_inc_and_test(&(ref->atomic));
break;
case ATOMIC_64:
ret = nvgpu_atomic64_inc_and_test(&(ref->atomic64));
break;
}
return ret;
}
#define ATOMIC_ADD(width, x, ref) \ static inline void func_dec(enum atomic_type type, struct atomic_struct *ref)
((width == WIDTH_32) ? \ {
nvgpu_atomic_add(x, &((ref)->atomic)) : \ switch (type) {
nvgpu_atomic64_add(x, &((ref)->atomic64))) case NOT_ATOMIC:
--ref->not_atomic;
break;
case ATOMIC_32:
nvgpu_atomic_dec(&(ref->atomic));
break;
case ATOMIC_64:
nvgpu_atomic64_dec(&(ref->atomic64));
break;
}
}
#define ATOMIC_ADD_RETURN(width, x, ref) \ static inline long func_dec_return(enum atomic_type type,
((width == WIDTH_32) ? \ struct atomic_struct *ref)
nvgpu_atomic_add_return(x, &((ref)->atomic)) : \ {
nvgpu_atomic64_add_return(x, &((ref)->atomic64))) long ret = 0;
#define ATOMIC_ADD_UNLESS(width, ref, a, u) \ switch (type) {
(((width == WIDTH_32) ? \ case NOT_ATOMIC:
nvgpu_atomic_add_unless(&((ref)->atomic), a, u) : \ --ref->not_atomic;
nvgpu_atomic64_add_unless(&((ref)->atomic64), a, u))) ret = ref->not_atomic;
break;
case ATOMIC_32:
ret = nvgpu_atomic_dec_return(&(ref->atomic));
break;
case ATOMIC_64:
ret = nvgpu_atomic64_dec_return(&(ref->atomic64));
break;
}
return ret;
}
#define ATOMIC_SUB(width, x, ref) \ static inline bool func_dec_and_test(enum atomic_type type,
((width == WIDTH_32) ? \ struct atomic_struct *ref)
nvgpu_atomic_sub(x, &((ref)->atomic)) : \ {
nvgpu_atomic64_sub(x, &((ref)->atomic64))) bool ret = false;
#define ATOMIC_SUB_RETURN(width, x, ref) \ switch (type) {
((width == WIDTH_32) ? \ case NOT_ATOMIC:
nvgpu_atomic_sub_return(x, &((ref)->atomic)) : \ --ref->not_atomic;
nvgpu_atomic64_sub_return(x, &((ref)->atomic64))) ret = (ref->not_atomic == 0);
break;
case ATOMIC_32:
ret = nvgpu_atomic_dec_and_test(&(ref->atomic));
break;
case ATOMIC_64:
ret = nvgpu_atomic64_dec_and_test(&(ref->atomic64));
break;
}
return ret;
}
#define ATOMIC_SUB_AND_TEST(width, x, ref) \ static inline void func_add(enum atomic_type type, long val,
((width == WIDTH_32) ? \ struct atomic_struct *ref)
nvgpu_atomic_sub_and_test(x, &((ref)->atomic)) : \ {
nvgpu_atomic64_sub_and_test(x, &((ref)->atomic64))) switch (type) {
case NOT_ATOMIC:
ref->not_atomic += val;
break;
case ATOMIC_32:
nvgpu_atomic_add(val, &(ref->atomic));
break;
case ATOMIC_64:
nvgpu_atomic64_add(val, &(ref->atomic64));
break;
}
}
#define ATOMIC_XCHG(width, ref, new) \ static inline long func_add_return(enum atomic_type type, long val,
((width == WIDTH_32) ? \ struct atomic_struct *ref)
nvgpu_atomic_xchg(&((ref)->atomic), new) : \ {
nvgpu_atomic64_xchg(&((ref)->atomic64), new)) long ret = 0;
#define ATOMIC_CMPXCHG(width, ref, old, new) \ switch (type) {
((width == WIDTH_32) ? \ case NOT_ATOMIC:
nvgpu_atomic_cmpxchg(&((ref)->atomic), old, new) : \ ref->not_atomic += val;
nvgpu_atomic64_cmpxchg(&((ref)->atomic64), old, new)) ret = ref->not_atomic;
break;
case ATOMIC_32:
ret = nvgpu_atomic_add_return(val, &(ref->atomic));
break;
case ATOMIC_64:
ret = nvgpu_atomic64_add_return(val, &(ref->atomic64));
break;
}
return ret;
}
static inline long func_add_unless(enum atomic_type type,
struct atomic_struct *ref, long val,
long unless)
{
long ret = 0;
switch (type) {
case NOT_ATOMIC:
ret = ref->not_atomic;
if (ret != unless) {
ref->not_atomic += val;
}
break;
case ATOMIC_32:
ret = nvgpu_atomic_add_unless(&(ref->atomic), val,
unless);
break;
case ATOMIC_64:
ret = nvgpu_atomic64_add_unless(&(ref->atomic64), val,
unless);
break;
}
return ret;
}
static inline void func_sub(enum atomic_type type, long val,
struct atomic_struct *ref)
{
switch (type) {
case NOT_ATOMIC:
ref->not_atomic -= val;
break;
case ATOMIC_32:
nvgpu_atomic_sub(val, &(ref->atomic));
break;
case ATOMIC_64:
nvgpu_atomic64_sub(val, &(ref->atomic64));
break;
}
}
static inline long func_sub_return(enum atomic_type type, long val,
struct atomic_struct *ref)
{
long ret = 0;
switch (type) {
case NOT_ATOMIC:
ref->not_atomic -= val;
ret = ref->not_atomic;
break;
case ATOMIC_32:
ret = nvgpu_atomic_sub_return(val, &(ref->atomic));
break;
case ATOMIC_64:
ret = nvgpu_atomic64_sub_return(val, &(ref->atomic64));
break;
}
return ret;
}
static inline bool func_sub_and_test(enum atomic_type type, long val,
struct atomic_struct *ref)
{
bool ret = 0;
switch (type) {
case NOT_ATOMIC:
ref->not_atomic -= val;
ret = (ref->not_atomic == 0);
break;
case ATOMIC_32:
ret = nvgpu_atomic_sub_and_test(val, &(ref->atomic));
break;
case ATOMIC_64:
ret = nvgpu_atomic64_sub_and_test(val,
&(ref->atomic64));
break;
}
return ret;
}
static inline long func_xchg(enum atomic_type type, struct atomic_struct *ref,
long new)
{
long ret = 0;
switch (type) {
case NOT_ATOMIC:
ret = ref->not_atomic;
ref->not_atomic = new;
break;
case ATOMIC_32:
ret = nvgpu_atomic_xchg(&(ref->atomic), new);
break;
case ATOMIC_64:
ret = nvgpu_atomic64_xchg(&(ref->atomic64), new);
break;
}
return ret;
}
static inline long func_cmpxchg(enum atomic_type type,
struct atomic_struct *ref, long old, long new)
{
long ret = 0;
switch (type) {
case NOT_ATOMIC:
ret = ref->not_atomic;
if (ret == old) {
ref->not_atomic = new;
}
break;
case ATOMIC_32:
ret = nvgpu_atomic_cmpxchg(&(ref->atomic), old, new);
break;
case ATOMIC_64:
ret = nvgpu_atomic64_cmpxchg(&(ref->atomic64), old,
new);
break;
}
return ret;
}
/* /*
* Helper macro that takes an atomic op from the enum and returns +1/-1 * Helper macro that takes an atomic op from the enum and returns +1/-1
@@ -172,17 +399,17 @@ static pthread_barrier_t thread_barrier;
/* Support function to do an atomic set and read verification */ /* Support function to do an atomic set and read verification */
static int single_set_and_read(struct unit_module *m, static int single_set_and_read(struct unit_module *m,
struct atomic_struct *atomic, struct atomic_struct *atomic,
enum atomic_width width, const long set_val) enum atomic_type type, const long set_val)
{ {
long read_val; long read_val;
if ((width == WIDTH_32) && if ((type == ATOMIC_32) &&
((set_val < INT_MIN) || (set_val > INT_MAX))) { ((set_val < INT_MIN) || (set_val > INT_MAX))) {
unit_return_fail(m, "Invalid value for 32 op\n"); unit_return_fail(m, "Invalid value for 32 op\n");
} }
ATOMIC_SET(width, atomic, set_val); func_set(type, atomic, set_val);
read_val = ATOMIC_READ(width, atomic); read_val = func_read(type, atomic);
if (read_val != set_val) { if (read_val != set_val) {
unit_err(m, "Atomic returned wrong value. Expected: %ld " unit_err(m, "Atomic returned wrong value. Expected: %ld "
"Received: %ld\n", (long)set_val, (long)read_val); "Received: %ld\n", (long)set_val, (long)read_val);
@@ -201,21 +428,21 @@ static int test_atomic_set_and_read(struct unit_module *m,
struct gk20a *g, void *__args) struct gk20a *g, void *__args)
{ {
struct atomic_test_args *args = (struct atomic_test_args *)__args; struct atomic_test_args *args = (struct atomic_test_args *)__args;
const int loop_limit = args->width == WIDTH_32 ? (sizeof(int) * 8) : const int loop_limit = args->type == ATOMIC_32 ? (sizeof(int) * 8) :
(sizeof(long) * 8); (sizeof(long) * 8);
const long min_value = args->width == WIDTH_32 ? INT_MIN : const long min_value = args->type == ATOMIC_32 ? INT_MIN :
LONG_MIN; LONG_MIN;
const long max_value = args->width == WIDTH_32 ? INT_MAX : const long max_value = args->type == ATOMIC_32 ? INT_MAX :
LONG_MAX; LONG_MAX;
struct atomic_struct atomic; struct atomic_struct atomic = {0};
int i; int i;
single_set_and_read(m, &atomic, args->width, min_value); single_set_and_read(m, &atomic, args->type, min_value);
single_set_and_read(m, &atomic, args->width, max_value); single_set_and_read(m, &atomic, args->type, max_value);
single_set_and_read(m, &atomic, args->width, 0); single_set_and_read(m, &atomic, args->type, 0);
for (i = 0; i < loop_limit; i++) { for (i = 0; i < loop_limit; i++) {
if (single_set_and_read(m, &atomic, args->width, (1 << i)) if (single_set_and_read(m, &atomic, args->type, (1 << i))
!= UNIT_SUCCESS) { != UNIT_SUCCESS) {
return UNIT_FAIL; return UNIT_FAIL;
} }
@@ -238,7 +465,7 @@ static int test_atomic_arithmetic(struct unit_module *m,
struct gk20a *g, void *__args) struct gk20a *g, void *__args)
{ {
struct atomic_test_args *args = (struct atomic_test_args *)__args; struct atomic_test_args *args = (struct atomic_test_args *)__args;
struct atomic_struct atomic; struct atomic_struct atomic = {0};
int i; int i;
long delta_magnitude; long delta_magnitude;
long read_val; long read_val;
@@ -246,7 +473,7 @@ static int test_atomic_arithmetic(struct unit_module *m,
bool result_bool; bool result_bool;
bool check_result_bool = false; bool check_result_bool = false;
if (single_set_and_read(m, &atomic, args->width, args->start_val) if (single_set_and_read(m, &atomic, args->type, args->start_val)
!= UNIT_SUCCESS) { != UNIT_SUCCESS) {
return UNIT_FAIL; return UNIT_FAIL;
} }
@@ -255,39 +482,39 @@ static int test_atomic_arithmetic(struct unit_module *m,
if (args->op == op_inc) { if (args->op == op_inc) {
/* use 2 since we test both inc and inc_return */ /* use 2 since we test both inc and inc_return */
delta_magnitude = 2; delta_magnitude = 2;
ATOMIC_INC(args->width, &atomic); func_inc(args->type, &atomic);
read_val = ATOMIC_INC_RETURN(args->width, &atomic); read_val = func_inc_return(args->type, &atomic);
} else if (args->op == op_inc_and_test) { } else if (args->op == op_inc_and_test) {
delta_magnitude = 1; delta_magnitude = 1;
check_result_bool = true; check_result_bool = true;
result_bool = ATOMIC_INC_AND_TEST(args->width, &atomic); result_bool = func_inc_and_test(args->type, &atomic);
read_val = ATOMIC_READ(args->width, &atomic); read_val = func_read(args->type, &atomic);
} else if (args->op == op_dec) { } else if (args->op == op_dec) {
/* use 2 since we test both dec and dec_return */ /* use 2 since we test both dec and dec_return */
delta_magnitude = 2; delta_magnitude = 2;
ATOMIC_DEC(args->width, &atomic); func_dec(args->type, &atomic);
read_val = ATOMIC_DEC_RETURN(args->width, &atomic); read_val = func_dec_return(args->type, &atomic);
} else if (args->op == op_dec_and_test) { } else if (args->op == op_dec_and_test) {
delta_magnitude = 1; delta_magnitude = 1;
check_result_bool = true; check_result_bool = true;
result_bool = ATOMIC_DEC_AND_TEST(args->width, &atomic); result_bool = func_dec_and_test(args->type, &atomic);
read_val = ATOMIC_READ(args->width, &atomic); read_val = func_read(args->type, &atomic);
} else if (args->op == op_add) { } else if (args->op == op_add) {
delta_magnitude = args->value * 2; delta_magnitude = args->value * 2;
ATOMIC_ADD(args->width, args->value, &atomic); func_add(args->type, args->value, &atomic);
read_val = ATOMIC_ADD_RETURN(args->width, args->value, read_val = func_add_return(args->type, args->value,
&atomic); &atomic);
} else if (args->op == op_sub) { } else if (args->op == op_sub) {
delta_magnitude = args->value * 2; delta_magnitude = args->value * 2;
ATOMIC_SUB(args->width, args->value, &atomic); func_sub(args->type, args->value, &atomic);
read_val = ATOMIC_SUB_RETURN(args->width, args->value, read_val = func_sub_return(args->type, args->value,
&atomic); &atomic);
} else if (args->op == op_sub_and_test) { } else if (args->op == op_sub_and_test) {
delta_magnitude = args->value; delta_magnitude = args->value;
check_result_bool = true; check_result_bool = true;
result_bool = ATOMIC_SUB_AND_TEST(args->width, result_bool = func_sub_and_test(args->type,
args->value, &atomic); args->value, &atomic);
read_val = ATOMIC_READ(args->width, &atomic); read_val = func_read(args->type, &atomic);
} else { } else {
unit_return_fail(m, "Test error: invalid op in %s\n", unit_return_fail(m, "Test error: invalid op in %s\n",
__func__); __func__);
@@ -297,7 +524,7 @@ static int test_atomic_arithmetic(struct unit_module *m,
(i * delta_magnitude * ATOMIC_OP_SIGN(args->op)); (i * delta_magnitude * ATOMIC_OP_SIGN(args->op));
/* sanity check */ /* sanity check */
if ((args->width == WIDTH_32) && if ((args->type == ATOMIC_32) &&
((expected_val > INT_MAX) || (expected_val < INT_MIN))) { ((expected_val > INT_MAX) || (expected_val < INT_MIN))) {
unit_return_fail(m, "Test error: invalid value in %s\n", unit_return_fail(m, "Test error: invalid value in %s\n",
__func__); __func__);
@@ -334,26 +561,26 @@ static void *arithmetic_thread(void *__args)
for (i = 0; i < targs->margs->loop_count; i++) { for (i = 0; i < targs->margs->loop_count; i++) {
if (targs->margs->op == op_inc) { if (targs->margs->op == op_inc) {
ATOMIC_INC(targs->margs->width, targs->atomic); func_inc(targs->margs->type, targs->atomic);
} else if (targs->margs->op == op_dec) { } else if (targs->margs->op == op_dec) {
ATOMIC_DEC(targs->margs->width, targs->atomic); func_dec(targs->margs->type, targs->atomic);
} else if (targs->margs->op == op_add) { } else if (targs->margs->op == op_add) {
/* /*
* Save the last value to sanity that threads aren't * Save the last value to sanity that threads aren't
* running sequentially * running sequentially
*/ */
targs->final_val = ATOMIC_ADD_RETURN( targs->final_val = func_add_return(
targs->margs->width, targs->margs->type,
targs->margs->value, targs->margs->value,
targs->atomic); targs->atomic);
} else if (targs->margs->op == op_add) { } else if (targs->margs->op == op_add) {
ATOMIC_ADD(targs->margs->width, targs->margs->value, func_add(targs->margs->type, targs->margs->value,
targs->atomic); targs->atomic);
} else if (targs->margs->op == op_sub) { } else if (targs->margs->op == op_sub) {
ATOMIC_SUB(targs->margs->width, targs->margs->value, func_sub(targs->margs->type, targs->margs->value,
targs->atomic); targs->atomic);
} else if (targs->margs->op == op_inc_and_test) { } else if (targs->margs->op == op_inc_and_test) {
if (ATOMIC_INC_AND_TEST(targs->margs->width, if (func_inc_and_test(targs->margs->type,
targs->atomic)) { targs->atomic)) {
/* /*
* Only increment if atomic op returns true * Only increment if atomic op returns true
@@ -362,7 +589,7 @@ static void *arithmetic_thread(void *__args)
targs->iterations++; targs->iterations++;
} }
} else if (targs->margs->op == op_dec_and_test) { } else if (targs->margs->op == op_dec_and_test) {
if (ATOMIC_DEC_AND_TEST(targs->margs->width, if (func_dec_and_test(targs->margs->type,
targs->atomic)) { targs->atomic)) {
/* /*
* Only increment if atomic op returns true * Only increment if atomic op returns true
@@ -371,7 +598,7 @@ static void *arithmetic_thread(void *__args)
targs->iterations++; targs->iterations++;
} }
} else if (targs->margs->op == op_sub_and_test) { } else if (targs->margs->op == op_sub_and_test) {
if (ATOMIC_SUB_AND_TEST(targs->margs->width, if (func_sub_and_test(targs->margs->type,
targs->margs->value, targs->margs->value,
targs->atomic)) { targs->atomic)) {
/* /*
@@ -381,7 +608,7 @@ static void *arithmetic_thread(void *__args)
targs->iterations++; targs->iterations++;
} }
} else if (targs->margs->op == op_add_unless) { } else if (targs->margs->op == op_add_unless) {
if (ATOMIC_ADD_UNLESS(targs->margs->width, if (func_add_unless(targs->margs->type,
targs->atomic, targs->margs->value, targs->atomic, targs->margs->value,
targs->unless) != targs->unless) { targs->unless) != targs->unless) {
/* /*
@@ -442,14 +669,14 @@ static int test_atomic_arithmetic_threaded(struct unit_module *m,
struct gk20a *g, void *__args) struct gk20a *g, void *__args)
{ {
struct atomic_test_args *args = (struct atomic_test_args *)__args; struct atomic_test_args *args = (struct atomic_test_args *)__args;
struct atomic_struct atomic; struct atomic_struct atomic = {0};
const int num_threads = 100; const int num_threads = 100;
struct atomic_thread_info threads[num_threads]; struct atomic_thread_info threads[num_threads];
int i; int i;
long expected_val, val, expected_iterations; long expected_val, val, expected_iterations;
int ret = UNIT_SUCCESS; int ret = UNIT_SUCCESS;
if (single_set_and_read(m, &atomic, args->width, args->start_val) if (single_set_and_read(m, &atomic, args->type, args->start_val)
!= UNIT_SUCCESS) { != UNIT_SUCCESS) {
return UNIT_FAIL; return UNIT_FAIL;
} }
@@ -474,7 +701,7 @@ static int test_atomic_arithmetic_threaded(struct unit_module *m,
pthread_join(threads[i].thread, NULL); pthread_join(threads[i].thread, NULL);
} }
val = ATOMIC_READ(args->width, &atomic); val = func_read(args->type, &atomic);
switch (args->op) { switch (args->op) {
case op_add_unless: case op_add_unless:
@@ -525,7 +752,7 @@ static int test_atomic_arithmetic_threaded(struct unit_module *m,
} }
/* sanity check */ /* sanity check */
if ((args->width == WIDTH_32) && if ((args->type == ATOMIC_32) &&
((expected_val > INT_MAX) || (expected_val < INT_MIN))) { ((expected_val > INT_MAX) || (expected_val < INT_MIN))) {
unit_err(m, "Test error: invalid value in %s\n", __func__); unit_err(m, "Test error: invalid value in %s\n", __func__);
ret = UNIT_FAIL; ret = UNIT_FAIL;
@@ -573,11 +800,11 @@ static int test_atomic_xchg(struct unit_module *m,
struct gk20a *g, void *__args) struct gk20a *g, void *__args)
{ {
struct atomic_test_args *args = (struct atomic_test_args *)__args; struct atomic_test_args *args = (struct atomic_test_args *)__args;
struct atomic_struct atomic; struct atomic_struct atomic = {0};
int i; int i;
long new_val, old_val, ret_val; long new_val, old_val, ret_val;
if (single_set_and_read(m, &atomic, args->width, args->start_val) if (single_set_and_read(m, &atomic, args->type, args->start_val)
!= UNIT_SUCCESS) { != UNIT_SUCCESS) {
return UNIT_FAIL; return UNIT_FAIL;
} }
@@ -590,7 +817,7 @@ static int test_atomic_xchg(struct unit_module *m,
*/ */
new_val = (i % 2 ? 1 : -1) * (args->start_val + i); new_val = (i % 2 ? 1 : -1) * (args->start_val + i);
/* only a 32bit xchg op */ /* only a 32bit xchg op */
ret_val = ATOMIC_XCHG(args->width, &atomic, new_val); ret_val = func_xchg(args->type, &atomic, new_val);
if (ret_val != old_val) { if (ret_val != old_val) {
unit_return_fail(m, "xchg returned bad old val " unit_return_fail(m, "xchg returned bad old val "
"Expected: %ld, Received: %ld\n", "Expected: %ld, Received: %ld\n",
@@ -612,13 +839,13 @@ static int test_atomic_cmpxchg(struct unit_module *m,
struct gk20a *g, void *__args) struct gk20a *g, void *__args)
{ {
struct atomic_test_args *args = (struct atomic_test_args *)__args; struct atomic_test_args *args = (struct atomic_test_args *)__args;
struct atomic_struct atomic; struct atomic_struct atomic = {0};
const int switch_interval = 5; const int switch_interval = 5;
int i; int i;
long new_val, old_val, ret_val; long new_val, old_val, ret_val;
bool should_match = true; bool should_match = true;
if (single_set_and_read(m, &atomic, args->width, args->start_val) if (single_set_and_read(m, &atomic, args->type, args->start_val)
!= UNIT_SUCCESS) { != UNIT_SUCCESS) {
return UNIT_FAIL; return UNIT_FAIL;
} }
@@ -635,7 +862,7 @@ static int test_atomic_cmpxchg(struct unit_module *m,
new_val = args->start_val + i; new_val = args->start_val + i;
if (should_match) { if (should_match) {
ret_val = ATOMIC_CMPXCHG(args->width, &atomic, ret_val = func_cmpxchg(args->type, &atomic,
old_val, new_val); old_val, new_val);
if (ret_val != old_val) { if (ret_val != old_val) {
unit_return_fail(m, unit_return_fail(m,
@@ -643,7 +870,7 @@ static int test_atomic_cmpxchg(struct unit_module *m,
"Expected: %ld, Received: %ld\n", "Expected: %ld, Received: %ld\n",
old_val, ret_val); old_val, ret_val);
} }
ret_val = ATOMIC_READ(args->width, &atomic); ret_val = func_read(args->type, &atomic);
if (ret_val != new_val) { if (ret_val != new_val) {
unit_return_fail(m, unit_return_fail(m,
"cmpxchg did not update " "cmpxchg did not update "
@@ -652,7 +879,7 @@ static int test_atomic_cmpxchg(struct unit_module *m,
} }
old_val = new_val; old_val = new_val;
} else { } else {
ret_val = ATOMIC_CMPXCHG(args->width, &atomic, ret_val = func_cmpxchg(args->type, &atomic,
-1 * old_val, new_val); -1 * old_val, new_val);
if (ret_val != old_val) { if (ret_val != old_val) {
unit_return_fail(m, unit_return_fail(m,
@@ -660,7 +887,7 @@ static int test_atomic_cmpxchg(struct unit_module *m,
"Expected: %ld, Received: %ld\n", "Expected: %ld, Received: %ld\n",
old_val, ret_val); old_val, ret_val);
} }
ret_val = ATOMIC_READ(args->width, &atomic); ret_val = func_read(args->type, &atomic);
if (ret_val != old_val) { if (ret_val != old_val) {
unit_return_fail(m, unit_return_fail(m,
"cmpxchg should not have updated " "cmpxchg should not have updated "
@@ -686,13 +913,13 @@ static int test_atomic_add_unless(struct unit_module *m,
struct gk20a *g, void *__args) struct gk20a *g, void *__args)
{ {
struct atomic_test_args *args = (struct atomic_test_args *)__args; struct atomic_test_args *args = (struct atomic_test_args *)__args;
struct atomic_struct atomic; struct atomic_struct atomic = {0};
const int switch_interval = 5; const int switch_interval = 5;
int i; int i;
int new_val, old_val, ret_val; int new_val, old_val, ret_val;
bool should_update = true; bool should_update = true;
if (single_set_and_read(m, &atomic, args->width, args->start_val) if (single_set_and_read(m, &atomic, args->type, args->start_val)
!= UNIT_SUCCESS) { != UNIT_SUCCESS) {
return UNIT_FAIL; return UNIT_FAIL;
} }
@@ -705,7 +932,7 @@ static int test_atomic_add_unless(struct unit_module *m,
if (should_update) { if (should_update) {
/* This will fail to match and do the add */ /* This will fail to match and do the add */
ret_val = ATOMIC_ADD_UNLESS(args->width, &atomic, ret_val = func_add_unless(args->type, &atomic,
args->value, old_val - 1); args->value, old_val - 1);
if (ret_val != old_val) { if (ret_val != old_val) {
unit_return_fail(m, unit_return_fail(m,
@@ -714,7 +941,7 @@ static int test_atomic_add_unless(struct unit_module *m,
old_val, ret_val); old_val, ret_val);
} }
new_val = old_val + args->value; new_val = old_val + args->value;
ret_val = ATOMIC_READ(args->width, &atomic); ret_val = func_read(args->type, &atomic);
if (ret_val != new_val) { if (ret_val != new_val) {
unit_return_fail(m, "add_unless did not " unit_return_fail(m, "add_unless did not "
"update Expected: %d, " "update Expected: %d, "
@@ -724,7 +951,7 @@ static int test_atomic_add_unless(struct unit_module *m,
old_val = ret_val; old_val = ret_val;
} else { } else {
/* This will match the old value and won't add */ /* This will match the old value and won't add */
ret_val = ATOMIC_ADD_UNLESS(args->width, &atomic, ret_val = func_add_unless(args->type, &atomic,
args->value, old_val); args->value, old_val);
if (ret_val != old_val) { if (ret_val != old_val) {
unit_return_fail(m, unit_return_fail(m,
@@ -732,7 +959,7 @@ static int test_atomic_add_unless(struct unit_module *m,
"Expected: %d, Received: %d\n", "Expected: %d, Received: %d\n",
old_val, ret_val); old_val, ret_val);
} }
ret_val = ATOMIC_READ(args->width, &atomic); ret_val = func_read(args->type, &atomic);
if (ret_val != old_val) { if (ret_val != old_val) {
unit_return_fail(m, "add_unless should not " unit_return_fail(m, "add_unless should not "
"have updated Expected: %d, " "have updated Expected: %d, "
@@ -746,14 +973,14 @@ static int test_atomic_add_unless(struct unit_module *m,
} }
static struct atomic_test_args set_and_read_32_arg = { static struct atomic_test_args set_and_read_32_arg = {
.width = WIDTH_32, .type = ATOMIC_32,
}; };
static struct atomic_test_args set_and_read_64_arg = { static struct atomic_test_args set_and_read_64_arg = {
.width = WIDTH_64, .type = ATOMIC_64,
}; };
static struct atomic_test_args inc_32_arg = { static struct atomic_test_args inc_32_arg = {
.op = op_inc, .op = op_inc,
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = -500, .start_val = -500,
.loop_count = 10000, .loop_count = 10000,
.value = 1, .value = 1,
@@ -761,7 +988,7 @@ static struct atomic_test_args inc_32_arg = {
static struct atomic_test_args inc_and_test_32_arg = { static struct atomic_test_args inc_and_test_32_arg = {
/* must cross 0 */ /* must cross 0 */
.op = op_inc_and_test, .op = op_inc_and_test,
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = -500, .start_val = -500,
.loop_count = 10000, .loop_count = 10000,
.value = 1, .value = 1,
@@ -769,21 +996,21 @@ static struct atomic_test_args inc_and_test_32_arg = {
static struct atomic_test_args inc_and_test_64_arg = { static struct atomic_test_args inc_and_test_64_arg = {
/* must cross 0 */ /* must cross 0 */
.op = op_inc_and_test, .op = op_inc_and_test,
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = -500, .start_val = -500,
.loop_count = 10000, .loop_count = 10000,
.value = 1, .value = 1,
}; };
static struct atomic_test_args inc_64_arg = { static struct atomic_test_args inc_64_arg = {
.op = op_inc, .op = op_inc,
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = INT_MAX - 500, .start_val = INT_MAX - 500,
.loop_count = 10000, .loop_count = 10000,
.value = 1, .value = 1,
}; };
static struct atomic_test_args dec_32_arg = { static struct atomic_test_args dec_32_arg = {
.op = op_dec, .op = op_dec,
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = 500, .start_val = 500,
.loop_count = 10000, .loop_count = 10000,
.value = 1, .value = 1,
@@ -791,7 +1018,7 @@ static struct atomic_test_args dec_32_arg = {
static struct atomic_test_args dec_and_test_32_arg = { static struct atomic_test_args dec_and_test_32_arg = {
/* must cross 0 */ /* must cross 0 */
.op = op_dec_and_test, .op = op_dec_and_test,
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = 500, .start_val = 500,
.loop_count = 10000, .loop_count = 10000,
.value = 1, .value = 1,
@@ -799,42 +1026,42 @@ static struct atomic_test_args dec_and_test_32_arg = {
static struct atomic_test_args dec_and_test_64_arg = { static struct atomic_test_args dec_and_test_64_arg = {
/* must cross 0 */ /* must cross 0 */
.op = op_dec_and_test, .op = op_dec_and_test,
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = 500, .start_val = 500,
.loop_count = 10000, .loop_count = 10000,
.value = 1, .value = 1,
}; };
static struct atomic_test_args dec_64_arg = { static struct atomic_test_args dec_64_arg = {
.op = op_dec, .op = op_dec,
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = INT_MIN + 500, .start_val = INT_MIN + 500,
.loop_count = 10000, .loop_count = 10000,
.value = 1, .value = 1,
}; };
static struct atomic_test_args add_32_arg = { static struct atomic_test_args add_32_arg = {
.op = op_add, .op = op_add,
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = -500, .start_val = -500,
.loop_count = 10000, .loop_count = 10000,
.value = 7, .value = 7,
}; };
static struct atomic_test_args add_64_arg = { static struct atomic_test_args add_64_arg = {
.op = op_add, .op = op_add,
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = INT_MAX - 500, .start_val = INT_MAX - 500,
.loop_count = 10000, .loop_count = 10000,
.value = 7, .value = 7,
}; };
struct atomic_test_args sub_32_arg = { struct atomic_test_args sub_32_arg = {
.op = op_sub, .op = op_sub,
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = 500, .start_val = 500,
.loop_count = 10000, .loop_count = 10000,
.value = 7, .value = 7,
}; };
static struct atomic_test_args sub_64_arg = { static struct atomic_test_args sub_64_arg = {
.op = op_sub, .op = op_sub,
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = INT_MIN + 500, .start_val = INT_MIN + 500,
.loop_count = 10000, .loop_count = 10000,
.value = 7, .value = 7,
@@ -842,7 +1069,7 @@ static struct atomic_test_args sub_64_arg = {
static struct atomic_test_args sub_and_test_32_arg = { static struct atomic_test_args sub_and_test_32_arg = {
/* must cross 0 */ /* must cross 0 */
.op = op_sub_and_test, .op = op_sub_and_test,
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = 500, .start_val = 500,
.loop_count = 10000, .loop_count = 10000,
.value = 5, .value = 5,
@@ -850,25 +1077,25 @@ static struct atomic_test_args sub_and_test_32_arg = {
static struct atomic_test_args sub_and_test_64_arg = { static struct atomic_test_args sub_and_test_64_arg = {
/* must cross 0 */ /* must cross 0 */
.op = op_sub_and_test, .op = op_sub_and_test,
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = 500, .start_val = 500,
.loop_count = 10000, .loop_count = 10000,
.value = 5, .value = 5,
}; };
struct atomic_test_args xchg_32_arg = { struct atomic_test_args xchg_32_arg = {
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = 1, .start_val = 1,
.loop_count = 10000, .loop_count = 10000,
}; };
struct atomic_test_args xchg_64_arg = { struct atomic_test_args xchg_64_arg = {
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = INT_MAX, .start_val = INT_MAX,
.loop_count = 10000, .loop_count = 10000,
}; };
static struct atomic_test_args add_unless_32_arg = { static struct atomic_test_args add_unless_32_arg = {
/* must loop at least 10 times */ /* must loop at least 10 times */
.op = op_add_unless, .op = op_add_unless,
.width = WIDTH_32, .type = ATOMIC_32,
.start_val = -500, .start_val = -500,
.loop_count = 10000, .loop_count = 10000,
.value = 5, .value = 5,
@@ -876,7 +1103,7 @@ static struct atomic_test_args add_unless_32_arg = {
static struct atomic_test_args add_unless_64_arg = { static struct atomic_test_args add_unless_64_arg = {
/* must loop at least 10 times */ /* must loop at least 10 times */
.op = op_add_unless, .op = op_add_unless,
.width = WIDTH_64, .type = ATOMIC_64,
.start_val = -500, .start_val = -500,
.loop_count = 10000, .loop_count = 10000,
.value = 5, .value = 5,