arm64: arm_mpu: Add userspace
Add dynamic_areas_init. It will mark a mpu region as a dynamic region area. The dynamic region areas is designed to be the background regions, so that the system could re-program the thread regions on the backgroud regions. Add configure_dynamic_mpu_regions to re-program the thread regions on the backgroud regions. The configure_dynamic_mpu_regions function is the core function of implementing the userspace for the MPU. This function is used in thread creation and context switch. During context switch, the pre thread's regions should be disabled, and the new thread's regions will be re-programed. Since the thread's stack region will also be switched, there will be a problem before new thread's region being re-programed which is the new thread's stack usage. To avoid the exception generated by stack usage caused by unprogramed new thread's stack region, I disable mpu first before flush_dynamic_regions_to_mpu and then enable it. Signed-off-by: Jaxson Han <jaxson.han@arm.com>
This commit is contained in:
Jaxson Han
Christopher Friedt
parent
fa494e78b2
commit
27ed237f6d
1 changed files with 311 additions and 0 deletions
|
|
@ -15,6 +15,21 @@
|
||||||
|
|
||||||
LOG_MODULE_REGISTER(mpu, CONFIG_MPU_LOG_LEVEL);
|
LOG_MODULE_REGISTER(mpu, CONFIG_MPU_LOG_LEVEL);
|
||||||
|
|
||||||
|
#define MPU_DYNAMIC_REGION_AREAS_NUM 1
|
||||||
|
|
||||||
|
#define _MAX_DYNAMIC_MPU_REGIONS_NUM \
|
||||||
|
((IS_ENABLED(CONFIG_USERSPACE) ? (CONFIG_MAX_DOMAIN_PARTITIONS + 1) : 0) + \
|
||||||
|
(IS_ENABLED(CONFIG_MPU_STACK_GUARD) ? 1 : 0))
|
||||||
|
|
||||||
|
#ifdef CONFIG_USERSPACE
|
||||||
|
static int dynamic_areas_init(uintptr_t start, size_t size);
|
||||||
|
#define MPU_DYNAMIC_REGIONS_AREA_START ((uintptr_t)&_app_smem_start)
|
||||||
|
#else
|
||||||
|
#define MPU_DYNAMIC_REGIONS_AREA_START ((uintptr_t)&__kernel_ram_start)
|
||||||
|
#endif
|
||||||
|
#define MPU_DYNAMIC_REGIONS_AREA_SIZE \
|
||||||
|
((size_t)((uintptr_t)&__kernel_ram_end - MPU_DYNAMIC_REGIONS_AREA_START))
|
||||||
|
|
||||||
/*
|
/*
|
||||||
* AArch64 Memory Model Feature Register 0
|
* AArch64 Memory Model Feature Register 0
|
||||||
* Provides information about the implemented memory model and memory
|
* Provides information about the implemented memory model and memory
|
||||||
|
|
@ -182,4 +197,300 @@ void z_arm64_mm_init(bool is_primary_core)
|
||||||
static_regions_num = mpu_config.num_regions;
|
static_regions_num = mpu_config.num_regions;
|
||||||
|
|
||||||
arm_core_mpu_enable();
|
arm_core_mpu_enable();
|
||||||
|
|
||||||
|
#ifdef CONFIG_USERSPACE
|
||||||
|
int rc = dynamic_areas_init(MPU_DYNAMIC_REGIONS_AREA_START,
|
||||||
|
MPU_DYNAMIC_REGIONS_AREA_SIZE);
|
||||||
|
if (rc <= 0) {
|
||||||
|
__ASSERT(0, "Dynamic areas init fail");
|
||||||
|
return;
|
||||||
|
}
|
||||||
|
#endif
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|
||||||
|
#ifdef CONFIG_USERSPACE
|
||||||
|
|
||||||
|
struct dynamic_region_info {
|
||||||
|
int index;
|
||||||
|
struct arm_mpu_region region_conf;
|
||||||
|
};
|
||||||
|
|
||||||
|
static struct dynamic_region_info sys_dyn_regions[MPU_DYNAMIC_REGION_AREAS_NUM];
|
||||||
|
static int sys_dyn_regions_num;
|
||||||
|
|
||||||
|
static int dynamic_areas_init(uintptr_t start, size_t size)
|
||||||
|
{
|
||||||
|
const struct arm_mpu_region *region;
|
||||||
|
struct dynamic_region_info *tmp_info;
|
||||||
|
|
||||||
|
uint64_t base = start;
|
||||||
|
uint64_t limit = base + size;
|
||||||
|
|
||||||
|
if (sys_dyn_regions_num + 1 > MPU_DYNAMIC_REGION_AREAS_NUM) {
|
||||||
|
return -1;
|
||||||
|
}
|
||||||
|
|
||||||
|
for (size_t i = 0; i < mpu_config.num_regions; i++) {
|
||||||
|
region = &mpu_config.mpu_regions[i];
|
||||||
|
tmp_info = &sys_dyn_regions[sys_dyn_regions_num];
|
||||||
|
if (base >= region->base && limit <= region->limit) {
|
||||||
|
tmp_info->index = i;
|
||||||
|
tmp_info->region_conf = *region;
|
||||||
|
return ++sys_dyn_regions_num;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
return -1;
|
||||||
|
}
|
||||||
|
|
||||||
|
static int dup_dynamic_regions(struct dynamic_region_info *dst, int len)
|
||||||
|
{
|
||||||
|
size_t i;
|
||||||
|
|
||||||
|
__ASSERT(sys_dyn_regions_num < len, "system dynamic region nums too large.");
|
||||||
|
|
||||||
|
for (i = 0; i < sys_dyn_regions_num; i++) {
|
||||||
|
dst[i] = sys_dyn_regions[i];
|
||||||
|
}
|
||||||
|
for (; i < len; i++) {
|
||||||
|
dst[i].index = -1;
|
||||||
|
}
|
||||||
|
|
||||||
|
return sys_dyn_regions_num;
|
||||||
|
}
|
||||||
|
|
||||||
|
static void set_region(struct arm_mpu_region *region,
|
||||||
|
uint64_t base, uint64_t limit,
|
||||||
|
struct arm_mpu_region_attr *attr)
|
||||||
|
{
|
||||||
|
region->base = base;
|
||||||
|
region->limit = limit;
|
||||||
|
region->attr = *attr;
|
||||||
|
}
|
||||||
|
|
||||||
|
static int get_underlying_region_idx(struct dynamic_region_info *dyn_regions,
|
||||||
|
uint8_t region_num, uint64_t base,
|
||||||
|
uint64_t limit)
|
||||||
|
{
|
||||||
|
for (size_t idx = 0; idx < region_num; idx++) {
|
||||||
|
struct arm_mpu_region *region = &(dyn_regions[idx].region_conf);
|
||||||
|
|
||||||
|
if (base >= region->base && limit <= region->limit) {
|
||||||
|
return idx;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
return -1;
|
||||||
|
}
|
||||||
|
|
||||||
|
static uint8_t insert_region(struct dynamic_region_info *dyn_regions,
|
||||||
|
uint8_t region_idx, uint8_t region_num,
|
||||||
|
uintptr_t start, size_t size,
|
||||||
|
struct arm_mpu_region_attr *attr)
|
||||||
|
{
|
||||||
|
|
||||||
|
/* base: inclusive, limit: exclusive */
|
||||||
|
uint64_t base = (uint64_t)start;
|
||||||
|
uint64_t limit = base + size;
|
||||||
|
int u_idx;
|
||||||
|
struct arm_mpu_region *u_region;
|
||||||
|
uint64_t u_base;
|
||||||
|
uint64_t u_limit;
|
||||||
|
struct arm_mpu_region_attr *u_attr;
|
||||||
|
|
||||||
|
|
||||||
|
__ASSERT(region_idx < region_num,
|
||||||
|
"Out-of-bounds error for dynamic region map. region idx: %d, region num: %d",
|
||||||
|
region_idx, region_num);
|
||||||
|
|
||||||
|
u_idx = get_underlying_region_idx(dyn_regions, region_idx, base, limit);
|
||||||
|
|
||||||
|
__ASSERT(u_idx >= 0, "Invalid underlying region index");
|
||||||
|
|
||||||
|
/* Get underlying region range and attr */
|
||||||
|
u_region = &(dyn_regions[u_idx].region_conf);
|
||||||
|
u_base = u_region->base;
|
||||||
|
u_limit = u_region->limit;
|
||||||
|
u_attr = &u_region->attr;
|
||||||
|
|
||||||
|
/* Temporally holding new region available to be configured */
|
||||||
|
struct arm_mpu_region *curr_region = &(dyn_regions[region_idx].region_conf);
|
||||||
|
|
||||||
|
if (base == u_base && limit == u_limit) {
|
||||||
|
/*
|
||||||
|
* The new region overlaps entirely with the
|
||||||
|
* underlying region. Simply update the attr.
|
||||||
|
*/
|
||||||
|
set_region(u_region, base, limit, attr);
|
||||||
|
} else if (base == u_base) {
|
||||||
|
set_region(curr_region, base, limit, attr);
|
||||||
|
set_region(u_region, limit, u_limit, u_attr);
|
||||||
|
region_idx++;
|
||||||
|
} else if (limit == u_limit) {
|
||||||
|
set_region(u_region, u_base, base, u_attr);
|
||||||
|
set_region(curr_region, base, limit, attr);
|
||||||
|
region_idx++;
|
||||||
|
} else {
|
||||||
|
set_region(u_region, u_base, base, u_attr);
|
||||||
|
set_region(curr_region, base, limit, attr);
|
||||||
|
region_idx++;
|
||||||
|
curr_region = &(dyn_regions[region_idx].region_conf);
|
||||||
|
set_region(curr_region, limit, u_limit, u_attr);
|
||||||
|
region_idx++;
|
||||||
|
}
|
||||||
|
|
||||||
|
return region_idx;
|
||||||
|
}
|
||||||
|
|
||||||
|
static int flush_dynamic_regions_to_mpu(struct dynamic_region_info *dyn_regions,
|
||||||
|
uint8_t region_num)
|
||||||
|
{
|
||||||
|
int reg_avail_idx = static_regions_num;
|
||||||
|
/*
|
||||||
|
* Clean the dynamic regions
|
||||||
|
*/
|
||||||
|
for (size_t i = reg_avail_idx; i < get_num_regions(); i++) {
|
||||||
|
mpu_set_region(i, 0, 0);
|
||||||
|
}
|
||||||
|
|
||||||
|
/*
|
||||||
|
* flush the dyn_regions to MPU
|
||||||
|
*/
|
||||||
|
for (size_t i = 0; i < region_num; i++) {
|
||||||
|
int region_idx = dyn_regions[i].index;
|
||||||
|
/*
|
||||||
|
* dyn_regions has two types of regions:
|
||||||
|
* 1) The fixed dyn background region which has a real index.
|
||||||
|
* 2) The normal region whose index will accumulate from
|
||||||
|
* static_regions_num.
|
||||||
|
*
|
||||||
|
* Region_idx < 0 means not the fixed dyn background region.
|
||||||
|
* In this case, region_idx should be the reg_avail_idx which
|
||||||
|
* is accumulated from static_regions_num.
|
||||||
|
*/
|
||||||
|
if (region_idx < 0) {
|
||||||
|
region_idx = reg_avail_idx++;
|
||||||
|
}
|
||||||
|
__ASSERT(region_idx < get_num_regions(),
|
||||||
|
"Out-of-bounds error for mpu regions. region idx: %d, total mpu regions: %d",
|
||||||
|
region_idx, get_num_regions());
|
||||||
|
region_init(region_idx, &(dyn_regions[i].region_conf));
|
||||||
|
}
|
||||||
|
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
|
|
||||||
|
static void configure_dynamic_mpu_regions(struct k_thread *thread)
|
||||||
|
{
|
||||||
|
/*
|
||||||
|
* Allocate double space for dyn_regions. Because when split
|
||||||
|
* the background dynamic regions, it will cause double regions numbers
|
||||||
|
* generated.
|
||||||
|
*/
|
||||||
|
struct dynamic_region_info dyn_regions[_MAX_DYNAMIC_MPU_REGIONS_NUM * 2];
|
||||||
|
const uint8_t max_region_num = ARRAY_SIZE(dyn_regions);
|
||||||
|
uint8_t region_num;
|
||||||
|
|
||||||
|
region_num = dup_dynamic_regions(dyn_regions, max_region_num);
|
||||||
|
|
||||||
|
struct k_mem_domain *mem_domain = thread->mem_domain_info.mem_domain;
|
||||||
|
|
||||||
|
if (mem_domain) {
|
||||||
|
LOG_DBG("configure domain: %p", mem_domain);
|
||||||
|
|
||||||
|
uint32_t num_parts = mem_domain->num_partitions;
|
||||||
|
uint32_t max_parts = CONFIG_MAX_DOMAIN_PARTITIONS;
|
||||||
|
struct k_mem_partition *partition;
|
||||||
|
|
||||||
|
for (size_t i = 0; i < max_parts && num_parts > 0; i++, num_parts--) {
|
||||||
|
partition = &mem_domain->partitions[i];
|
||||||
|
if (partition->size == 0) {
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
LOG_DBG("set region 0x%lx 0x%lx",
|
||||||
|
partition->start, partition->size);
|
||||||
|
region_num = insert_region(dyn_regions,
|
||||||
|
region_num,
|
||||||
|
max_region_num,
|
||||||
|
partition->start,
|
||||||
|
partition->size,
|
||||||
|
&partition->attr);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
LOG_DBG("configure user thread %p's context", thread);
|
||||||
|
if ((thread->base.user_options & K_USER) != 0) {
|
||||||
|
/* K_USER thread stack needs a region */
|
||||||
|
region_num = insert_region(dyn_regions,
|
||||||
|
region_num,
|
||||||
|
max_region_num,
|
||||||
|
thread->stack_info.start,
|
||||||
|
thread->stack_info.size,
|
||||||
|
&K_MEM_PARTITION_P_RW_U_RW);
|
||||||
|
}
|
||||||
|
|
||||||
|
arm_core_mpu_disable();
|
||||||
|
flush_dynamic_regions_to_mpu(dyn_regions, region_num);
|
||||||
|
arm_core_mpu_enable();
|
||||||
|
}
|
||||||
|
|
||||||
|
int arch_mem_domain_max_partitions_get(void)
|
||||||
|
{
|
||||||
|
int max_parts = get_num_regions() - static_regions_num;
|
||||||
|
|
||||||
|
if (max_parts > CONFIG_MAX_DOMAIN_PARTITIONS) {
|
||||||
|
max_parts = CONFIG_MAX_DOMAIN_PARTITIONS;
|
||||||
|
}
|
||||||
|
|
||||||
|
return max_parts;
|
||||||
|
}
|
||||||
|
|
||||||
|
void arch_mem_domain_partition_add(struct k_mem_domain *domain, uint32_t partition_id)
|
||||||
|
{
|
||||||
|
ARG_UNUSED(domain);
|
||||||
|
ARG_UNUSED(partition_id);
|
||||||
|
}
|
||||||
|
|
||||||
|
void arch_mem_domain_partition_remove(struct k_mem_domain *domain, uint32_t partition_id)
|
||||||
|
{
|
||||||
|
ARG_UNUSED(domain);
|
||||||
|
ARG_UNUSED(partition_id);
|
||||||
|
}
|
||||||
|
|
||||||
|
void arch_mem_domain_thread_add(struct k_thread *thread)
|
||||||
|
{
|
||||||
|
if (thread == _current) {
|
||||||
|
configure_dynamic_mpu_regions(thread);
|
||||||
|
}
|
||||||
|
#ifdef CONFIG_SMP
|
||||||
|
else {
|
||||||
|
/* the thread could be running on another CPU right now */
|
||||||
|
z_arm64_mem_cfg_ipi();
|
||||||
|
}
|
||||||
|
#endif
|
||||||
|
}
|
||||||
|
|
||||||
|
void arch_mem_domain_thread_remove(struct k_thread *thread)
|
||||||
|
{
|
||||||
|
if (thread == _current) {
|
||||||
|
configure_dynamic_mpu_regions(thread);
|
||||||
|
}
|
||||||
|
#ifdef CONFIG_SMP
|
||||||
|
else {
|
||||||
|
/* the thread could be running on another CPU right now */
|
||||||
|
z_arm64_mem_cfg_ipi();
|
||||||
|
}
|
||||||
|
#endif
|
||||||
|
}
|
||||||
|
|
||||||
|
void z_arm64_thread_mem_domains_init(struct k_thread *thread)
|
||||||
|
{
|
||||||
|
configure_dynamic_mpu_regions(thread);
|
||||||
|
}
|
||||||
|
|
||||||
|
void z_arm64_swap_mem_domains(struct k_thread *thread)
|
||||||
|
{
|
||||||
|
configure_dynamic_mpu_regions(thread);
|
||||||
|
}
|
||||||
|
|
||||||
|
#endif /* CONFIG_USERSPACE */
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue