A Total Overhaul on the Lunaix's Virtual Memory Model (#26)

[lunaix-os.git] / lunaix-os / kernel / mm / mmap.c

diff --git a/lunaix-os/kernel/mm/mmap.c b/lunaix-os/kernel/mm/mmap.c
index f700fd533179d9528ff84191e6f14402d0f5eca2..3f273c08935bfc492bc08f6fd30f22f1352e9bd9 100644 (file)
--- a/lunaix-os/kernel/mm/mmap.c
+++ b/lunaix-os/kernel/mm/mmap.c
@@ -3,12 +3,15 @@
 #include <lunaix/mm/valloc.h>
 #include <lunaix/mm/vmm.h>
 #include <lunaix/spike.h>
 #include <lunaix/mm/valloc.h>
 #include <lunaix/mm/vmm.h>
 #include <lunaix/spike.h>

-

 #include <lunaix/syscall.h>
 #include <lunaix/syscall_utils.h>
 
 #include <lunaix/syscall.h>
 #include <lunaix/syscall_utils.h>
 

+#include <sys/mm/mm_defs.h>
+
+#include <usr/lunaix/mann_flags.h>
+

 // any size beyond this is bullshit
 // any size beyond this is bullshit

-#define BS_SIZE (KERNEL_MM_BASE - UMMAP_START)
+#define BS_SIZE (KERNEL_RESIDENT - USR_MMAP)

 
 int
 mem_has_overlap(vm_regions_t* regions, ptr_t start, ptr_t end)
 
 int
 mem_has_overlap(vm_regions_t* regions, ptr_t start, ptr_t end)


@@ -55,16 +58,117 @@ mem_adjust_inplace(vm_regions_t* regions,
     return 0;
 }
 
     return 0;
 }
 

+int 
+mmap_user(void** addr_out,
+        struct mm_region** created,
+        ptr_t addr,
+        struct v_file* file,
+        struct mmap_param* param) 
+{
+    param->range_end = KERNEL_RESIDENT;
+    param->range_start = USR_EXEC;
+
+    return mem_map(addr_out, created, addr, file, param);
+}
+
+static ptr_t
+__mem_find_slot_backward(struct mm_region* lead, struct mmap_param* param, struct mm_region* anchor)
+{
+    ptr_t size = param->mlen;
+    struct mm_region *pos = anchor, 
+                     *n = next_region(pos);
+    while (pos != lead)
+    {
+        if (pos == lead) {
+            break;
+        }
+
+        ptr_t end = n->start;
+        if (n == lead) {
+            end = param->range_end;
+        }
+
+        if (end - pos->end >= size) {
+            return pos->end;
+        }
+
+        pos = n;
+        n = next_region(pos);
+    }
+    
+    return 0;
+}
+
+static ptr_t
+__mem_find_slot_forward(struct mm_region* lead, struct mmap_param* param, struct mm_region* anchor)
+{
+    ptr_t size = param->mlen;
+    struct mm_region *pos = anchor, 
+                     *prev = prev_region(pos);
+    while (lead != pos)
+    {
+        ptr_t end = prev->end;
+        if (prev == lead) {
+            end = param->range_start;
+        }
+
+        if (pos->start - end >= size) {
+            return pos->start - size;
+        }
+
+        pos = prev;
+        prev = prev_region(pos);
+    }
+
+    return 0;
+}
+
+static ptr_t
+__mem_find_slot(vm_regions_t* lead, struct mmap_param* param, struct mm_region* anchor)
+{
+    ptr_t result = 0;
+    struct mm_region* _lead = get_region(lead);
+    if ((result = __mem_find_slot_backward(_lead, param, anchor))) {
+        return result;
+    }
+
+    return __mem_find_slot_forward(_lead, param, anchor);
+}
+
+static struct mm_region*
+__mem_find_nearest(vm_regions_t* lead, ptr_t addr)
+{   
+    ptr_t min_dist = (ptr_t)-1;
+    struct mm_region *pos, *n, *min = NULL;
+    llist_for_each(pos, n, lead, head) {
+        if (region_contains(pos, addr)) {
+            return pos;
+        }
+
+        ptr_t dist = addr - pos->end;
+        if (addr < pos->start) {
+            dist = pos->start - addr;
+        }
+
+        if (dist < min_dist) {
+            min_dist = dist;
+            min = pos;
+        }
+    }
+
+    return min;
+}
+

 int
 mem_map(void** addr_out,
         struct mm_region** created,
 int
 mem_map(void** addr_out,
         struct mm_region** created,

-        void* addr,
+        ptr_t addr,

         struct v_file* file,
         struct mmap_param* param)
 {
     assert_msg(addr, "addr can not be NULL");
 
         struct v_file* file,
         struct mmap_param* param)
 {
     assert_msg(addr, "addr can not be NULL");
 

-    ptr_t last_end = USER_START, found_loc = (ptr_t)addr;
+    ptr_t last_end = USR_EXEC, found_loc = va_align(addr);

     struct mm_region *pos, *n;
 
     vm_regions_t* vm_regions = &param->pvms->regions;
     struct mm_region *pos, *n;
 
     vm_regions_t* vm_regions = &param->pvms->regions;


@@ -85,23 +189,19 @@ mem_map(void** addr_out,
         goto found;
     }
 
         goto found;
     }
 

-    llist_for_each(pos, n, vm_regions, head)
-    {
-        if (last_end < found_loc) {
-            size_t avail_space = pos->start - found_loc;
-            if (pos->start > found_loc && avail_space > param->mlen) {
-                goto found;
-            }
-            found_loc = pos->end + PG_SIZE;
-        }
+    if (llist_empty(vm_regions)) {
+        goto found;
+    }

 
 

-        last_end = pos->end;
+    struct mm_region* anchor = __mem_find_nearest(vm_regions, found_loc);
+    if ((found_loc = __mem_find_slot(vm_regions, param, anchor))) {
+        goto found;

     }
 
     return ENOMEM;
 
 found:
     }
 
     return ENOMEM;
 
 found:

-    if (found_loc >= KERNEL_MM_BASE || found_loc < USER_START) {
+    if (found_loc >= param->range_end || found_loc < param->range_start) {

         return ENOMEM;
     }
 
         return ENOMEM;
     }
 


@@ -112,26 +212,16 @@ found:
 
     region->mfile = file;
     region->foff = param->offset;
 
     region->mfile = file;
     region->foff = param->offset;

-    region->flen = param->flen;

     region->proc_vms = param->pvms;
 
     region_add(vm_regions, region);
     region->proc_vms = param->pvms;
 
     region_add(vm_regions, region);

-
-    u32_t attr = PG_ALLOW_USER;
-    if ((param->proct & REGION_WRITE)) {
-        attr |= PG_WRITE;
-    }
-
-    for (u32_t i = 0; i < param->mlen; i += PG_SIZE) {
-        vmm_set_mapping(param->vms_mnt, found_loc + i, 0, attr, 0);
-    }
-
+    

     if (file) {
         vfs_ref_file(file);
     }
 
     if (addr_out) {
     if (file) {
         vfs_ref_file(file);
     }
 
     if (addr_out) {

-        *addr_out = found_loc;
+        *addr_out = (void*)found_loc;

     }
     if (created) {
         *created = region;
     }
     if (created) {
         *created = region;


@@ -139,6 +229,18 @@ found:
     return 0;
 }
 
     return 0;
 }
 

+int
+mem_remap(void** addr_out,
+          struct mm_region** remapped,
+          void* addr,
+          struct v_file* file,
+          struct mmap_param* param)
+{
+    // TODO
+
+    return EINVAL;
+}
+

 void
 mem_sync_pages(ptr_t mnt,
                struct mm_region* region,
 void
 mem_sync_pages(ptr_t mnt,
                struct mm_region* region,


@@ -149,19 +251,24 @@ mem_sync_pages(ptr_t mnt,
     if (!region->mfile || !(region->attr & REGION_WSHARED)) {
         return;
     }
     if (!region->mfile || !(region->attr & REGION_WSHARED)) {
         return;
     }

+    
+    pte_t* ptep = mkptep_va(mnt, start);
+    ptr_t va    = va_align(start);

 
 

-    v_mapping mapping;
-    for (size_t i = 0; i < length; i += PG_SIZE) {
-        if (!vmm_lookupat(mnt, start + i, &mapping)) {
+    for (; va < start + length; va += PAGE_SIZE, ptep++) {
+        pte_t pte = vmm_tryptep(ptep, LFT_SIZE);
+        if (pte_isnull(pte)) {

             continue;
         }
 
             continue;
         }
 

-        if (PG_IS_DIRTY(*mapping.pte)) {
-            size_t offset = mapping.va - region->start + region->foff;
+        if (pte_dirty(pte)) {
+            size_t offset = va - region->start + region->foff;

             struct v_inode* inode = region->mfile->inode;
             struct v_inode* inode = region->mfile->inode;

-            region->mfile->ops->write_page(inode, mapping.va, PG_SIZE, offset);
-            *mapping.pte &= ~PG_DIRTY;
-            cpu_invplg(mapping.pte);
+
+            region->mfile->ops->write_page(inode, (void*)va, offset);
+
+            set_pte(ptep, pte_mkclean(pte));
+            cpu_flush_page(va);

         } else if ((options & MS_INVALIDATE)) {
             goto invalidate;
         }
         } else if ((options & MS_INVALIDATE)) {
             goto invalidate;
         }


@@ -173,9 +280,9 @@ mem_sync_pages(ptr_t mnt,
         continue;
 
     invalidate:
         continue;
 
     invalidate:

-        *mapping.pte &= ~PG_PRESENT;
-        pmm_free_page(KERNEL_PID, mapping.pa);
-        cpu_invplg(mapping.pte);
+        set_pte(ptep, null_pte);
+        pmm_free_page(pte_paddr(pte));
+        cpu_flush_page(va);

     }
 }
 
     }
 }
 


@@ -208,64 +315,138 @@ mem_msync(ptr_t mnt,
 void
 mem_unmap_region(ptr_t mnt, struct mm_region* region)
 {
 void
 mem_unmap_region(ptr_t mnt, struct mm_region* region)
 {

-    size_t len = ROUNDUP(region->end - region->start, PG_SIZE);
-    mem_sync_pages(mnt, region, region->start, len, 0);
-
-    for (size_t i = region->start; i <= region->end; i += PG_SIZE) {
-        ptr_t pa = vmm_del_mapping(mnt, i);
-        if (pa) {
-            pmm_free_page(__current->pid, pa);
+    if (!region) {
+        return;
+    }
+    
+    valloc_ensure_valid(region);
+    
+    pfn_t pglen = leaf_count(region->end - region->start);
+    mem_sync_pages(mnt, region, region->start, pglen * PAGE_SIZE, 0);
+
+    pte_t* ptep = mkptep_va(mnt, region->start);
+    for (size_t i = 0; i < pglen; i++, ptep++) {
+        pte_t pte = pte_at(ptep);
+        ptr_t pa  = pte_paddr(pte);
+
+        set_pte(ptep, null_pte);
+        if (pte_isloaded(pte)) {
+            pmm_free_page(pte_paddr(pte));

         }
     }
         }
     }

+    

     llist_delete(&region->head);
     region_release(region);
 }
 
     llist_delete(&region->head);
     region_release(region);
 }
 

-int
-mem_unmap(ptr_t mnt, vm_regions_t* regions, void* addr, size_t length)
+// Case: head inseted, tail inseted
+#define CASE_HITI(vmr, addr, len)                                              \
+    ((vmr)->start <= (addr) && ((addr) + (len)) <= (vmr)->end)
+
+// Case: head inseted, tail extruded
+#define CASE_HITE(vmr, addr, len)                                              \
+    ((vmr)->start <= (addr) && ((addr) + (len)) > (vmr)->end)
+
+// Case: head extruded, tail inseted
+#define CASE_HETI(vmr, addr, len)                                              \
+    ((vmr)->start > (addr) && ((addr) + (len)) <= (vmr)->end)
+
+// Case: head extruded, tail extruded
+#define CASE_HETE(vmr, addr, len)                                              \
+    ((vmr)->start > (addr) && ((addr) + (len)) > (vmr)->end)
+
+static void
+__unmap_overlapped_cases(ptr_t mnt,
+                         struct mm_region* vmr,
+                         ptr_t* addr,
+                         size_t* length)

 {
 {

-    length = ROUNDUP(length, PG_SIZE);
-    ptr_t cur_addr = PG_ALIGN(addr);
-    struct mm_region *pos, *n;
+    // seg start, umapped segement start
+    ptr_t seg_start = *addr, umps_start = 0;

 
 

-    llist_for_each(pos, n, regions, head)
-    {
-        if (pos->start <= cur_addr && pos->end >= cur_addr) {
-            break;
+    // seg len, umapped segement len
+    size_t seg_len = *length, umps_len = 0;
+
+    size_t displ = 0, shrink = 0;
+
+    if (CASE_HITI(vmr, seg_start, seg_len)) {
+        size_t new_start = seg_start + seg_len;
+
+        // Require a split
+        if (new_start < vmr->end) {
+            struct mm_region* region = region_dup(vmr);
+            if (region->mfile) {
+                size_t f_shifted = new_start - region->start;
+                region->foff += f_shifted;
+            }
+            region->start = new_start;
+            llist_insert_after(&vmr->head, &region->head);

         }
         }

+
+        shrink = vmr->end - seg_start;
+        umps_len = shrink;
+        umps_start = seg_start;
+    } 
+    else if (CASE_HITE(vmr, seg_start, seg_len)) {
+        shrink = vmr->end - seg_start;
+        umps_len = shrink;
+        umps_start = seg_start;
+    } 
+    else if (CASE_HETI(vmr, seg_start, seg_len)) {
+        displ = seg_len - (vmr->start - seg_start);
+        umps_len = displ;
+        umps_start = vmr->start;
+    } 
+    else if (CASE_HETE(vmr, seg_start, seg_len)) {
+        shrink = vmr->end - vmr->start;
+        umps_len = shrink;
+        umps_start = vmr->start;

     }
 
     }
 

-    while (&pos->head != regions && cur_addr >= pos->start) {
-        u32_t l = pos->end - cur_addr;
-        pos->end = cur_addr;
-
-        if (l > length) {
-            // unmap cause discontinunity in a memory region -  do split
-            struct mm_region* region = valloc(sizeof(struct mm_region));
-            *region = *pos;
-            region->start = cur_addr + length;
-            llist_insert_after(&pos->head, &region->head);
-            l = length;
+    mem_sync_pages(mnt, vmr, vmr->start, umps_len, 0);
+    for (size_t i = 0; i < umps_len; i += PAGE_SIZE) {
+        ptr_t pa = vmm_del_mapping(mnt, vmr->start + i);
+        if (pa) {
+            pmm_free_page(pa);

         }
         }

+    }

 
 

-        mem_sync_pages(mnt, pos, cur_addr, l, 0);
+    vmr->start += displ;
+    vmr->end -= shrink;

 
 

-        for (size_t i = 0; i < l; i += PG_SIZE) {
-            ptr_t pa = vmm_del_mapping(mnt, cur_addr + i);
-            if (pa) {
-                pmm_free_page(pos->proc_vms->pid, pa);
-            }
+    if (vmr->start >= vmr->end) {
+        llist_delete(&vmr->head);
+        region_release(vmr);
+    } else if (vmr->mfile) {
+        vmr->foff += displ;
+    }
+
+    *addr = umps_start + umps_len;
+
+    size_t ump_len = *addr - seg_start;
+    *length = MAX(seg_len, ump_len) - ump_len;
+}
+
+int
+mem_unmap(ptr_t mnt, vm_regions_t* regions, ptr_t addr, size_t length)
+{
+    length = ROUNDUP(length, PAGE_SIZE);
+    ptr_t cur_addr = va_align(addr);
+    struct mm_region *pos, *n;
+
+    llist_for_each(pos, n, regions, head)
+    {
+        u32_t l = pos->start - cur_addr;
+        if ((pos->start <= cur_addr && cur_addr < pos->end) || l <= length) {
+            break;

         }
         }

+    }

 
 

+    while (&pos->head != regions && length) {

         n = container_of(pos->head.next, typeof(*pos), head);
         n = container_of(pos->head.next, typeof(*pos), head);

-        if (pos->end == pos->start) {
-            llist_delete(&pos->head);
-            region_release(pos);
-        }
+        __unmap_overlapped_cases(mnt, pos, &cur_addr, &length);

 
         pos = n;
 
         pos = n;

-        length -= l;
-        cur_addr += length;

     }
 
     return 0;
     }
 
     return 0;


@@ -280,62 +461,69 @@ __DEFINE_LXSYSCALL3(void*, sys_mmap, void*, addr, size_t, length, va_list, lst)
     int errno = 0;
     void* result = (void*)-1;
 
     int errno = 0;
     void* result = (void*)-1;
 

-    if (!length || length > BS_SIZE || !PG_ALIGNED(addr)) {
-        errno = EINVAL;
-        goto done;
-    }
+    ptr_t addr_ptr = (ptr_t)addr;

 
 

-    if (!addr) {
-        addr = UMMAP_START;
-    } else if (addr < UMMAP_START || addr + length >= UMMAP_END) {
-        errno = ENOMEM;
+    if (!length || length > BS_SIZE || va_offset(addr_ptr)) {
+        errno = EINVAL;

         goto done;
     }
 
         goto done;
     }
 

-    struct v_fd* vfd;
-    if ((errno = vfs_getfd(fd, &vfd))) {
-        goto done;
+    if (!addr_ptr) {
+        addr_ptr = USR_MMAP;
+    } else if (addr_ptr < USR_MMAP || addr_ptr + length >= USR_MMAP_END) {
+        if (!(options & (MAP_FIXED | MAP_FIXED_NOREPLACE))) {
+            errno = ENOMEM;
+            goto done;
+        }

     }
 
     }
 

-    struct v_file* file = vfd->file;
+    struct v_file* file = NULL;

 
     if (!(options & MAP_ANON)) {
 
     if (!(options & MAP_ANON)) {

+        struct v_fd* vfd;
+        if ((errno = vfs_getfd(fd, &vfd))) {
+            goto done;
+        }
+
+        file = vfd->file;

         if (!file->ops->read_page) {
             errno = ENODEV;
             goto done;
         }
         if (!file->ops->read_page) {
             errno = ENODEV;
             goto done;
         }

-    } else {
-        file = NULL;

     }
 
     struct mmap_param param = { .flags = options,
     }
 
     struct mmap_param param = { .flags = options,

-                                .mlen = ROUNDUP(length, PG_SIZE),
+                                .mlen = ROUNDUP(length, PAGE_SIZE),

                                 .offset = offset,
                                 .type = REGION_TYPE_GENERAL,
                                 .proct = proct,
                                 .offset = offset,
                                 .type = REGION_TYPE_GENERAL,
                                 .proct = proct,

-                                .pvms = &__current->mm,
+                                .pvms = vmspace(__current),

                                 .vms_mnt = VMS_SELF };
 
                                 .vms_mnt = VMS_SELF };
 

-    errno = mem_map(&result, NULL, addr, file, &param);
+    errno = mmap_user(&result, NULL, addr_ptr, file, &param);

 
 done:
 
 done:

-    __current->k_status = errno;
+    syscall_result(errno);

     return result;
 }
 
     return result;
 }
 

-__DEFINE_LXSYSCALL2(void, munmap, void*, addr, size_t, length)
+__DEFINE_LXSYSCALL2(int, munmap, void*, addr, size_t, length)

 {
 {

-    return mem_unmap(VMS_SELF, &__current->mm.regions, addr, length);
+    return mem_unmap(
+      VMS_SELF, vmregions(__current), (ptr_t)addr, length);

 }
 
 __DEFINE_LXSYSCALL3(int, msync, void*, addr, size_t, length, int, flags)
 {
 }
 
 __DEFINE_LXSYSCALL3(int, msync, void*, addr, size_t, length, int, flags)
 {

-    if (!PG_ALIGNED(addr) || ((flags & MS_ASYNC) && (flags & MS_SYNC))) {
+    if (va_offset((ptr_t)addr) || ((flags & MS_ASYNC) && (flags & MS_SYNC))) {

         return DO_STATUS(EINVAL);
     }
 
         return DO_STATUS(EINVAL);
     }
 

-    int status =
-      mem_msync(VMS_SELF, &__current->mm.regions, addr, length, flags);
+    int status = mem_msync(VMS_SELF,
+                           vmregions(__current),
+                           (ptr_t)addr,
+                           length,
+                           flags);

 
     return DO_STATUS(status);
 }
\ No newline at end of file
 
     return DO_STATUS(status);
 }
\ No newline at end of file