regression: mmap for fd

[lunaix-os.git] / lunaix-os / kernel / fs / vfs.c

diff --git a/lunaix-os/kernel/fs/vfs.c b/lunaix-os/kernel/fs/vfs.c
index c13b8d3d9663cdddf88865c666b7744e5dc2ec28..cbe3c39ab507a8021d453b01ab38f07803858bdb 100644 (file)
--- a/lunaix-os/kernel/fs/vfs.c
+++ b/lunaix-os/kernel/fs/vfs.c
@@ -53,25 +53,10 @@
 #include <lunaix/process.h>
 #include <lunaix/spike.h>
 #include <lunaix/syscall.h>
 #include <lunaix/process.h>
 #include <lunaix/spike.h>
 #include <lunaix/syscall.h>

+#include <lunaix/syscall_utils.h>

 
 #include <lunaix/fs/twifs.h>
 
 
 #include <lunaix/fs/twifs.h>
 

-#define PATH_DELIM '/'
-
-#define unlock_inode(inode) mutex_unlock(&inode->lock)
-#define lock_inode(inode)                                                      \
-    ({                                                                         \
-        mutex_lock(&inode->lock);                                              \
-        lru_use_one(inode_lru, &inode->lru);                                   \
-    })
-
-#define unlock_dnode(dnode) mutex_unlock(&dnode->lock)
-#define lock_dnode(dnode)                                                      \
-    ({                                                                         \
-        mutex_lock(&dnode->lock);                                              \
-        lru_use_one(dnode_lru, &dnode->lru);                                   \
-    })
-

 static struct cake_pile* dnode_pile;
 static struct cake_pile* inode_pile;
 static struct cake_pile* file_pile;
 static struct cake_pile* dnode_pile;
 static struct cake_pile* inode_pile;
 static struct cake_pile* file_pile;


@@ -81,7 +66,7 @@ static struct cake_pile* fd_pile;
 struct v_dnode* vfs_sysroot;
 static struct hbucket* dnode_cache;
 
 struct v_dnode* vfs_sysroot;
 static struct hbucket* dnode_cache;
 

-static struct lru_zone *dnode_lru, *inode_lru;
+struct lru_zone *dnode_lru, *inode_lru;

 
 struct hstr vfs_ddot = HSTR("..", 2);
 struct hstr vfs_dot = HSTR(".", 1);
 
 struct hstr vfs_ddot = HSTR("..", 2);
 struct hstr vfs_dot = HSTR(".", 1);


@@ -120,17 +105,18 @@ vfs_init()
 
     // 创建一个根dnode。
     vfs_sysroot = vfs_d_alloc(NULL, &vfs_empty);
 
     // 创建一个根dnode。
     vfs_sysroot = vfs_d_alloc(NULL, &vfs_empty);

+    vfs_sysroot->parent = vfs_sysroot;

     atomic_fetch_add(&vfs_sysroot->ref_count, 1);
 }
 
 inline struct hbucket*
     atomic_fetch_add(&vfs_sysroot->ref_count, 1);
 }
 
 inline struct hbucket*

-__dcache_hash(struct v_dnode* parent, uint32_t* hash)
+__dcache_hash(struct v_dnode* parent, u32_t* hash)

 {
 {

-    uint32_t _hash = *hash;
-    // 与parent的指针值做加法，来减小碰撞的可能性。
-    _hash += (uint32_t)parent;
+    u32_t _hash = *hash;

     // 确保低位更加随机
     _hash = _hash ^ (_hash >> VFS_HASHBITS);
     // 确保低位更加随机
     _hash = _hash ^ (_hash >> VFS_HASHBITS);

+    // 与parent的指针值做加法，来减小碰撞的可能性。
+    _hash += (u32_t)parent;

     *hash = _hash;
     return &dnode_cache[_hash & VFS_HASH_MASK];
 }
     *hash = _hash;
     return &dnode_cache[_hash & VFS_HASH_MASK];
 }


@@ -142,10 +128,10 @@ vfs_dcache_lookup(struct v_dnode* parent, struct hstr* str)
         return parent;
 
     if (HSTR_EQ(str, &vfs_ddot)) {
         return parent;
 
     if (HSTR_EQ(str, &vfs_ddot)) {

-        return parent->parent ? parent->parent : parent;
+        return parent->parent;

     }
 
     }
 

-    uint32_t hash = str->hash;
+    u32_t hash = str->hash;

     struct hbucket* slot = __dcache_hash(parent, &hash);
 
     struct v_dnode *pos, *n;
     struct hbucket* slot = __dcache_hash(parent, &hash);
 
     struct v_dnode *pos, *n;


@@ -178,6 +164,7 @@ vfs_dcache_remove(struct v_dnode* dnode)
     assert(dnode->ref_count == 1);
 
     llist_delete(&dnode->siblings);
     assert(dnode->ref_count == 1);
 
     llist_delete(&dnode->siblings);

+    llist_delete(&dnode->aka_list);

     hlist_delete(&dnode->hash_list);
 
     dnode->parent = NULL;
     hlist_delete(&dnode->hash_list);
 
     dnode->parent = NULL;


@@ -194,176 +181,6 @@ vfs_dcache_rehash(struct v_dnode* new_parent, struct v_dnode* dnode)
     vfs_dcache_add(new_parent, dnode);
 }
 
     vfs_dcache_add(new_parent, dnode);
 }
 

-#define VFS_SYMLINK_DEPTH 16
-
-int
-__vfs_walk(struct v_dnode* start,
-           const char* path,
-           struct v_dnode** dentry,
-           struct hstr* component,
-           int walk_options,
-           size_t depth,
-           char* fname_buffer)
-{
-    int errno = 0;
-    int i = 0, j = 0;
-
-    if (depth >= VFS_SYMLINK_DEPTH) {
-        return ENAMETOOLONG;
-    }
-
-    if (path[0] == PATH_DELIM || !start) {
-        if ((walk_options & VFS_WALK_FSRELATIVE) && start) {
-            start = start->super_block->root;
-        } else {
-            start = vfs_sysroot;
-            if (!vfs_sysroot->mnt) {
-                panick("vfs: no root");
-            }
-        }
-        i++;
-    }
-
-    struct v_dnode* dnode;
-    struct v_inode* current_inode;
-    struct v_dnode* current_level = start;
-
-    struct hstr name = HSTR(fname_buffer, 0);
-
-    char current = path[i++], lookahead;
-    while (current && current_level) {
-        lookahead = path[i++];
-        if (current != PATH_DELIM) {
-            if (j >= VFS_NAME_MAXLEN - 1) {
-                return ENAMETOOLONG;
-            }
-            if (!VFS_VALID_CHAR(current)) {
-                return EINVAL;
-            }
-            fname_buffer[j++] = current;
-            if (lookahead) {
-                goto cont;
-            }
-        }
-
-        // handling cases like /^.*(\/+).*$/
-        if (lookahead == PATH_DELIM) {
-            goto cont;
-        }
-
-        fname_buffer[j] = 0;
-        name.len = j;
-        hstr_rehash(&name, HSTR_FULL_HASH);
-
-        if (!lookahead && (walk_options & VFS_WALK_PARENT)) {
-            if (component) {
-                component->hash = name.hash;
-                component->len = j;
-                strcpy(component->value, fname_buffer);
-            }
-            break;
-        }
-
-        current_inode = current_level->inode;
-
-        if ((current_inode->itype & VFS_IFSYMLINK)) {
-            const char* link;
-
-            lock_inode(current_inode);
-            if ((errno =
-                   current_inode->ops->read_symlink(current_inode, &link))) {
-                unlock_inode(current_inode);
-                goto error;
-            }
-            unlock_inode(current_inode);
-
-            errno = __vfs_walk(current_level->parent,
-                               link,
-                               &dnode,
-                               NULL,
-                               0,
-                               depth + 1,
-                               fname_buffer + name.len + 1);
-
-            if (errno) {
-                goto error;
-            }
-
-            // reposition the resolved subtree pointed by symlink
-            vfs_dcache_rehash(current_level->parent, dnode);
-            current_level = dnode;
-            current_inode = dnode->inode;
-        }
-
-        lock_dnode(current_level);
-
-        dnode = vfs_dcache_lookup(current_level, &name);
-
-        if (!dnode) {
-            dnode = vfs_d_alloc(current_level, &name);
-
-            if (!dnode) {
-                errno = ENOMEM;
-                goto error;
-            }
-
-            lock_inode(current_inode);
-
-            errno = current_inode->ops->dir_lookup(current_inode, dnode);
-
-            if (errno == ENOENT && (walk_options & VFS_WALK_MKPARENT)) {
-                if (!current_inode->ops->mkdir) {
-                    errno = ENOTSUP;
-                } else {
-                    errno = current_inode->ops->mkdir(current_inode, dnode);
-                }
-            }
-
-            vfs_dcache_add(current_level, dnode);
-            unlock_inode(current_inode);
-
-            if (errno) {
-                unlock_dnode(current_level);
-                goto cleanup;
-            }
-        }
-
-        unlock_dnode(current_level);
-
-        j = 0;
-        current_level = dnode;
-    cont:
-        current = lookahead;
-    };
-
-    *dentry = current_level;
-    return 0;
-
-cleanup:
-    vfs_d_free(dnode);
-error:
-    *dentry = NULL;
-    return errno;
-}
-
-int
-vfs_walk(struct v_dnode* start,
-         const char* path,
-         struct v_dnode** dentry,
-         struct hstr* component,
-         int options)
-{
-    // allocate a file name stack for path walking and recursion to resolve
-    // symlink
-    char* name_buffer = valloc(2048);
-
-    int errno =
-      __vfs_walk(start, path, dentry, component, options, 0, name_buffer);
-
-    vfree(name_buffer);
-    return errno;
-}
-

 int
 vfs_open(struct v_dnode* dnode, struct v_file** file)
 {
 int
 vfs_open(struct v_dnode* dnode, struct v_file** file)
 {


@@ -410,8 +227,10 @@ void
 vfs_assign_inode(struct v_dnode* assign_to, struct v_inode* inode)
 {
     if (assign_to->inode) {
 vfs_assign_inode(struct v_dnode* assign_to, struct v_inode* inode)
 {
     if (assign_to->inode) {

+        llist_delete(&assign_to->aka_list);

         assign_to->inode->link_count--;
     }
         assign_to->inode->link_count--;
     }

+    llist_append(&inode->aka_dnodes, &assign_to->aka_list);

     assign_to->inode = inode;
     inode->link_count++;
 }
     assign_to->inode = inode;
     inode->link_count++;
 }


@@ -421,6 +240,10 @@ vfs_link(struct v_dnode* to_link, struct v_dnode* name)
 {
     int errno;
 
 {
     int errno;
 

+    if ((errno = vfs_check_writable(to_link))) {
+        return errno;
+    }
+

     lock_inode(to_link->inode);
     if (to_link->super_block->root != name->super_block->root) {
         errno = EXDEV;
     lock_inode(to_link->inode);
     if (to_link->super_block->root != name->super_block->root) {
         errno = EXDEV;


@@ -435,12 +258,35 @@ vfs_link(struct v_dnode* to_link, struct v_dnode* name)
 }
 
 int
 }
 
 int

-vfs_close(struct v_file* file)
+vfs_pclose(struct v_file* file, pid_t pid)

 {
     int errno = 0;
 {
     int errno = 0;

-    if (!(errno = file->ops->close(file))) {
+    if (file->ref_count > 1) {
+        atomic_fetch_sub(&file->ref_count, 1);
+    } else if (!(errno = file->ops->close(file))) {

         atomic_fetch_sub(&file->dnode->ref_count, 1);
         file->inode->open_count--;
         atomic_fetch_sub(&file->dnode->ref_count, 1);
         file->inode->open_count--;

+
+        /*
+         * Prevent dead lock.
+         * This happened when process is terminated while blocking on read.
+         * In that case, the process is still holding the inode lock and it
+             will never get released.
+         * The unlocking should also include ownership check.
+         *
+         * To see why, consider two process both open the same file both with
+         * fd=x.
+         *      Process A: busy on reading x
+         *      Process B: do nothing with x
+         * Assuming that, after a very short time, process B get terminated
+         * while process A is still busy in it's reading business. By this
+         * design, the inode lock of this file x is get released by B rather
+         * than A. And this will cause a probable race condition on A if other
+         * process is writing to this file later after B exit.
+         */
+        if (mutex_on_hold(&file->inode->lock)) {
+            mutex_unlock_for(&file->inode->lock, pid);
+        }

         mnt_chillax(file->dnode->mnt);
 
         pcache_commit_all(file->inode);
         mnt_chillax(file->dnode->mnt);
 
         pcache_commit_all(file->inode);


@@ -449,14 +295,31 @@ vfs_close(struct v_file* file)
     return errno;
 }
 
     return errno;
 }
 

+int
+vfs_close(struct v_file* file)
+{
+    return vfs_pclose(file, __current->pid);
+}
+
+void
+vfs_free_fd(struct v_fd* fd)
+{
+    cake_release(fd_pile, fd);
+}
+

 int
 vfs_fsync(struct v_file* file)
 {
 int
 vfs_fsync(struct v_file* file)
 {

+    int errno;
+    if ((errno = vfs_check_writable(file->dnode))) {
+        return errno;
+    }
+

     lock_inode(file->inode);
 
     lock_inode(file->inode);
 

-    int errno = ENOTSUP;

     pcache_commit_all(file->inode);
 
     pcache_commit_all(file->inode);
 

+    errno = ENOTSUP;

     if (file->ops->sync) {
         errno = file->ops->sync(file);
     }
     if (file->ops->sync) {
         errno = file->ops->sync(file);
     }


@@ -534,6 +397,7 @@ vfs_d_alloc(struct v_dnode* parent, struct hstr* name)
     memset(dnode, 0, sizeof(*dnode));
     llist_init_head(&dnode->children);
     llist_init_head(&dnode->siblings);
     memset(dnode, 0, sizeof(*dnode));
     llist_init_head(&dnode->children);
     llist_init_head(&dnode->siblings);

+    llist_init_head(&dnode->aka_list);

     mutex_init(&dnode->lock);
 
     dnode->ref_count = ATOMIC_VAR_INIT(0);
     mutex_init(&dnode->lock);
 
     dnode->ref_count = ATOMIC_VAR_INIT(0);


@@ -543,6 +407,7 @@ vfs_d_alloc(struct v_dnode* parent, struct hstr* name)
 
     if (parent) {
         dnode->super_block = parent->super_block;
 
     if (parent) {
         dnode->super_block = parent->super_block;

+        dnode->mnt = parent->mnt;

     }
 
     lru_use_one(dnode_lru, &dnode->lru);
     }
 
     lru_use_one(dnode_lru, &dnode->lru);


@@ -575,7 +440,7 @@ vfs_d_free(struct v_dnode* dnode)
 }
 
 struct v_inode*
 }
 
 struct v_inode*

-vfs_i_find(struct v_superblock* sb, uint32_t i_id)
+vfs_i_find(struct v_superblock* sb, u32_t i_id)

 {
     struct hbucket* slot = &sb->i_cache[i_id & VFS_HASH_MASK];
     struct v_inode *pos, *n;
 {
     struct hbucket* slot = &sb->i_cache[i_id & VFS_HASH_MASK];
     struct v_inode *pos, *n;


@@ -614,6 +479,8 @@ vfs_i_alloc(struct v_superblock* sb)
 
     memset(inode, 0, sizeof(*inode));
     mutex_init(&inode->lock);
 
     memset(inode, 0, sizeof(*inode));
     mutex_init(&inode->lock);

+    llist_init_head(&inode->xattrs);
+    llist_init_head(&inode->aka_dnodes);

 
     sb->ops.init_inode(sb, inode);
 
 
     sb->ops.init_inode(sb, inode);
 


@@ -634,7 +501,12 @@ vfs_i_free(struct v_inode* inode)
         pcache_release(inode->pg_cache);
         vfree(inode->pg_cache);
     }
         pcache_release(inode->pg_cache);
         vfree(inode->pg_cache);
     }

-    inode->ops->sync(inode);
+    // we don't need to sync inode.
+    // If an inode can be free, then it must be properly closed.
+    // Hence it must be synced already!
+    if (inode->destruct) {
+        inode->destruct(inode);
+    }

     hlist_delete(&inode->hash_list);
     cake_release(inode_pile, inode);
 }
     hlist_delete(&inode->hash_list);
     cake_release(inode_pile, inode);
 }


@@ -644,7 +516,7 @@ vfs_i_free(struct v_inode* inode)
 #define FLOCATE_CREATE_EMPTY 1
 
 int
 #define FLOCATE_CREATE_EMPTY 1
 
 int

-__vfs_getfd(int fd, struct v_fd** fd_s)
+vfs_getfd(int fd, struct v_fd** fd_s)

 {
     if (TEST_FD(fd) && (*fd_s = __current->fdtable->fds[fd])) {
         return 0;
 {
     if (TEST_FD(fd) && (*fd_s = __current->fdtable->fds[fd])) {
         return 0;


@@ -661,8 +533,9 @@ __vfs_try_locate_file(const char* path,
     char name_str[VFS_NAME_MAXLEN];
     struct hstr name = HSTR(name_str, 0);
     int errno;
     char name_str[VFS_NAME_MAXLEN];
     struct hstr name = HSTR(name_str, 0);
     int errno;

-    if ((errno =
-           vfs_walk(__current->cwd, path, fdir, &name, VFS_WALK_PARENT))) {
+
+    name_str[0] = 0;
+    if ((errno = vfs_walk_proc(path, fdir, &name, VFS_WALK_PARENT))) {

         return errno;
     }
 
         return errno;
     }
 


@@ -697,19 +570,20 @@ vfs_do_open(const char* path, int options)
 {
     int errno, fd;
     struct v_dnode *dentry, *file;
 {
     int errno, fd;
     struct v_dnode *dentry, *file;

-    struct v_file* ofile = 0;
+    struct v_file* ofile = NULL;

 
     errno = __vfs_try_locate_file(
       path, &dentry, &file, (options & FO_CREATE) ? FLOCATE_CREATE_EMPTY : 0);
 
 
     errno = __vfs_try_locate_file(
       path, &dentry, &file, (options & FO_CREATE) ? FLOCATE_CREATE_EMPTY : 0);
 

-    if (errno || (errno = vfs_open(file, &ofile))) {
-        return errno;
-    }
+    if (!errno && !(errno = vfs_alloc_fdslot(&fd))) {

 
 

-    struct v_inode* o_inode = ofile->inode;
+        if (errno || (errno = vfs_open(file, &ofile))) {
+            return errno;
+        }
+
+        struct v_fd* fd_s = cake_grab(fd_pile);
+        memset(fd_s, 0, sizeof(*fd_s));

 
 

-    if (!errno && !(errno = vfs_alloc_fdslot(&fd))) {
-        struct v_fd* fd_s = vzalloc(sizeof(*fd_s));

         ofile->f_pos = ofile->inode->fsize & -((options & FO_APPEND) != 0);
         fd_s->file = ofile;
         fd_s->flags = options;
         ofile->f_pos = ofile->inode->fsize & -((options & FO_APPEND) != 0);
         fd_s->file = ofile;
         fd_s->flags = options;


@@ -730,17 +604,15 @@ __DEFINE_LXSYSCALL1(int, close, int, fd)
 {
     struct v_fd* fd_s;
     int errno = 0;
 {
     struct v_fd* fd_s;
     int errno = 0;

-    if ((errno = __vfs_getfd(fd, &fd_s))) {
+    if ((errno = vfs_getfd(fd, &fd_s))) {

         goto done_err;
     }
 
         goto done_err;
     }
 

-    if (fd_s->file->ref_count > 1) {
-        fd_s->file->ref_count--;
-    } else if ((errno = vfs_close(fd_s->file))) {
+    if ((errno = vfs_close(fd_s->file))) {

         goto done_err;
     }
 
         goto done_err;
     }
 

-    vfree(fd_s);
+    cake_release(fd_pile, fd_s);

     __current->fdtable->fds[fd] = 0;
 
 done_err:
     __current->fdtable->fds[fd] = 0;
 
 done_err:


@@ -759,12 +631,12 @@ __vfs_readdir_callback(struct dir_context* dctx,
     dent->d_type = dtype;
 }
 
     dent->d_type = dtype;
 }
 

-__DEFINE_LXSYSCALL2(int, readdir, int, fd, struct dirent*, dent)
+__DEFINE_LXSYSCALL2(int, sys_readdir, int, fd, struct dirent*, dent)

 {
     struct v_fd* fd_s;
     int errno;
 
 {
     struct v_fd* fd_s;
     int errno;
 

-    if ((errno = __vfs_getfd(fd, &fd_s))) {
+    if ((errno = vfs_getfd(fd, &fd_s))) {

         goto done;
     }
 
         goto done;
     }
 


@@ -782,9 +654,9 @@ __DEFINE_LXSYSCALL2(int, readdir, int, fd, struct dirent*, dent)
                                   __vfs_readdir_callback };
         errno = 1;
         if (dent->d_offset == 0) {
                                   __vfs_readdir_callback };
         errno = 1;
         if (dent->d_offset == 0) {

-            __vfs_readdir_callback(&dctx, vfs_dot.value, vfs_dot.len, 0);
+            __vfs_readdir_callback(&dctx, vfs_dot.value, vfs_dot.len, DT_DIR);

         } else if (dent->d_offset == 1) {
         } else if (dent->d_offset == 1) {

-            __vfs_readdir_callback(&dctx, vfs_ddot.value, vfs_ddot.len, 0);
+            __vfs_readdir_callback(&dctx, vfs_ddot.value, vfs_ddot.len, DT_DIR);

         } else {
             dctx.index -= 2;
             if ((errno = fd_s->file->ops->readdir(fd_s->file, &dctx)) != 1) {
         } else {
             dctx.index -= 2;
             if ((errno = fd_s->file->ops->readdir(fd_s->file, &dctx)) != 1) {


@@ -805,7 +677,7 @@ __DEFINE_LXSYSCALL3(int, read, int, fd, void*, buf, size_t, count)
 {
     int errno = 0;
     struct v_fd* fd_s;
 {
     int errno = 0;
     struct v_fd* fd_s;

-    if ((errno = __vfs_getfd(fd, &fd_s))) {
+    if ((errno = vfs_getfd(fd, &fd_s))) {

         goto done;
     }
 
         goto done;
     }
 


@@ -819,13 +691,11 @@ __DEFINE_LXSYSCALL3(int, read, int, fd, void*, buf, size_t, count)
 
     file->inode->atime = clock_unixtime();
 
 
     file->inode->atime = clock_unixtime();
 

-    __SYSCALL_INTERRUPTIBLE({
-        if ((file->inode->itype & VFS_IFSEQDEV) || (fd_s->flags & FO_DIRECT)) {
-            errno = file->ops->read(file->inode, buf, count, file->f_pos);
-        } else {
-            errno = pcache_read(file->inode, buf, count, file->f_pos);
-        }
-    })
+    if ((file->inode->itype & VFS_IFSEQDEV) || (fd_s->flags & FO_DIRECT)) {
+        errno = file->ops->read(file->inode, buf, count, file->f_pos);
+    } else {
+        errno = pcache_read(file->inode, buf, count, file->f_pos);
+    }

 
     if (errno > 0) {
         file->f_pos += errno;
 
     if (errno > 0) {
         file->f_pos += errno;


@@ -843,11 +713,16 @@ __DEFINE_LXSYSCALL3(int, write, int, fd, void*, buf, size_t, count)
 {
     int errno = 0;
     struct v_fd* fd_s;
 {
     int errno = 0;
     struct v_fd* fd_s;

-    if ((errno = __vfs_getfd(fd, &fd_s))) {
+    if ((errno = vfs_getfd(fd, &fd_s))) {

         goto done;
     }
 
     struct v_file* file = fd_s->file;
         goto done;
     }
 
     struct v_file* file = fd_s->file;

+
+    if ((errno = vfs_check_writable(file->dnode))) {
+        goto done;
+    }
+

     if ((file->inode->itype & VFS_IFDIR)) {
         errno = EISDIR;
         goto done;
     if ((file->inode->itype & VFS_IFDIR)) {
         errno = EISDIR;
         goto done;


@@ -857,13 +732,11 @@ __DEFINE_LXSYSCALL3(int, write, int, fd, void*, buf, size_t, count)
 
     file->inode->mtime = clock_unixtime();
 
 
     file->inode->mtime = clock_unixtime();
 

-    __SYSCALL_INTERRUPTIBLE({
-        if ((file->inode->itype & VFS_IFSEQDEV) || (fd_s->flags & FO_DIRECT)) {
-            errno = file->ops->write(file->inode, buf, count, file->f_pos);
-        } else {
-            errno = pcache_write(file->inode, buf, count, file->f_pos);
-        }
-    })
+    if ((file->inode->itype & VFS_IFSEQDEV) || (fd_s->flags & FO_DIRECT)) {
+        errno = file->ops->write(file->inode, buf, count, file->f_pos);
+    } else {
+        errno = pcache_write(file->inode, buf, count, file->f_pos);
+    }

 
     if (errno > 0) {
         file->f_pos += errno;
 
     if (errno > 0) {
         file->f_pos += errno;


@@ -881,27 +754,36 @@ __DEFINE_LXSYSCALL3(int, lseek, int, fd, int, offset, int, options)
 {
     int errno = 0;
     struct v_fd* fd_s;
 {
     int errno = 0;
     struct v_fd* fd_s;

-    if ((errno = __vfs_getfd(fd, &fd_s))) {
+    if ((errno = vfs_getfd(fd, &fd_s))) {

         goto done;
     }
 
     struct v_file* file = fd_s->file;
 
         goto done;
     }
 
     struct v_file* file = fd_s->file;
 

+    if (!file->ops->seek) {
+        errno = ENOTSUP;
+        goto done;
+    }
+

     lock_inode(file->inode);
 
     lock_inode(file->inode);
 

-    size_t fpos = file->f_pos;
+    int overflow = 0;
+    int fpos = file->f_pos;

     switch (options) {
         case FSEEK_CUR:
     switch (options) {
         case FSEEK_CUR:

-            fpos = (size_t)((int)file->f_pos + offset);
+            overflow = __builtin_sadd_overflow((int)file->f_pos, offset, &fpos);

             break;
         case FSEEK_END:
             break;
         case FSEEK_END:

-            fpos = (size_t)((int)file->inode->fsize + offset);
+            overflow =
+              __builtin_sadd_overflow((int)file->inode->fsize, offset, &fpos);

             break;
         case FSEEK_SET:
             fpos = offset;
             break;
     }
             break;
         case FSEEK_SET:
             fpos = offset;
             break;
     }

-    if (!(errno = file->ops->seek(file->inode, fpos))) {
+    if (overflow) {
+        errno = EOVERFLOW;
+    } else if (!(errno = file->ops->seek(file->inode, fpos))) {

         file->f_pos = fpos;
     }
 
         file->f_pos = fpos;
     }
 


@@ -919,23 +801,27 @@ vfs_get_path(struct v_dnode* dnode, char* buf, size_t size, int depth)
     }
 
     if (depth > 64) {
     }
 
     if (depth > 64) {

-        return ELOOP;
+        return ENAMETOOLONG;

     }
 
     }
 

-    size_t len = vfs_get_path(dnode->parent, buf, size, depth + 1);
+    size_t len = 0;
+
+    if (dnode->parent != dnode) {
+        len = vfs_get_path(dnode->parent, buf, size, depth + 1);
+    }

 
     if (len >= size) {
         return len;
     }
 
 
     if (len >= size) {
         return len;
     }
 

+    if (!len || buf[len - 1] != VFS_PATH_DELIM) {
+        buf[len++] = VFS_PATH_DELIM;
+    }
+

     size_t cpy_size = MIN(dnode->name.len, size - len);
     strncpy(buf + len, dnode->name.value, cpy_size);
     len += cpy_size;
 
     size_t cpy_size = MIN(dnode->name.len, size - len);
     strncpy(buf + len, dnode->name.value, cpy_size);
     len += cpy_size;
 

-    if (len < size) {
-        buf[len++] = PATH_DELIM;
-    }
-

     return len;
 }
 
     return len;
 }
 


@@ -956,11 +842,24 @@ vfs_readlink(struct v_dnode* dnode, char* buf, size_t size)
     return 0;
 }
 
     return 0;
 }
 

+int
+vfs_get_dtype(int itype)
+{
+    switch (itype) {
+        case VFS_IFDIR:
+            return DT_DIR;
+        case VFS_IFSYMLINK:
+            return DT_SYMLINK;
+        default:
+            return DT_PIPE;
+    }
+}
+

 __DEFINE_LXSYSCALL3(int, realpathat, int, fd, char*, buf, size_t, size)
 {
     int errno;
     struct v_fd* fd_s;
 __DEFINE_LXSYSCALL3(int, realpathat, int, fd, char*, buf, size_t, size)
 {
     int errno;
     struct v_fd* fd_s;

-    if ((errno = __vfs_getfd(fd, &fd_s))) {
+    if ((errno = vfs_getfd(fd, &fd_s))) {

         goto done;
     }
 
         goto done;
     }
 


@@ -979,8 +878,7 @@ __DEFINE_LXSYSCALL3(int, readlink, const char*, path, char*, buf, size_t, size)
 {
     int errno;
     struct v_dnode* dnode;
 {
     int errno;
     struct v_dnode* dnode;

-    if (!(errno =
-            vfs_walk(__current->cwd, path, &dnode, NULL, VFS_WALK_NOFOLLOW))) {
+    if (!(errno = vfs_walk_proc(path, &dnode, NULL, VFS_WALK_NOFOLLOW))) {

         errno = vfs_readlink(dnode, buf, size);
     }
 
         errno = vfs_readlink(dnode, buf, size);
     }
 


@@ -1004,7 +902,7 @@ __DEFINE_LXSYSCALL4(int,
 {
     int errno;
     struct v_fd* fd_s;
 {
     int errno;
     struct v_fd* fd_s;

-    if ((errno = __vfs_getfd(dirfd, &fd_s))) {
+    if ((errno = vfs_getfd(dirfd, &fd_s))) {

         goto done;
     }
 
         goto done;
     }
 


@@ -1034,12 +932,16 @@ __DEFINE_LXSYSCALL1(int, rmdir, const char*, pathname)
 {
     int errno;
     struct v_dnode* dnode;
 {
     int errno;
     struct v_dnode* dnode;

-    if ((errno = vfs_walk(__current->cwd, pathname, &dnode, NULL, 0))) {
+    if ((errno = vfs_walk_proc(pathname, &dnode, NULL, 0))) {

         return DO_STATUS(errno);
     }
 
     lock_dnode(dnode);
 
         return DO_STATUS(errno);
     }
 
     lock_dnode(dnode);
 

+    if ((errno = vfs_check_writable(dnode))) {
+        goto done;
+    }
+

     if ((dnode->super_block->fs->types & FSTYPE_ROFS)) {
         errno = EROFS;
         goto done;
     if ((dnode->super_block->fs->types & FSTYPE_ROFS)) {
         errno = EROFS;
         goto done;


@@ -1089,17 +991,18 @@ __DEFINE_LXSYSCALL1(int, mkdir, const char*, path)
     char name_value[VFS_NAME_MAXLEN];
     struct hstr name = HHSTR(name_value, 0, 0);
 
     char name_value[VFS_NAME_MAXLEN];
     struct hstr name = HHSTR(name_value, 0, 0);
 

-    if (!dir) {
-        errno = ENOMEM;
+    if ((errno = vfs_walk_proc(path, &parent, &name, VFS_WALK_PARENT))) {

         goto done;
     }
 
         goto done;
     }
 

-    if ((errno =
-           vfs_walk(__current->cwd, path, &parent, &name, VFS_WALK_PARENT))) {
+    if ((errno = vfs_check_writable(parent))) {

         goto done;
     }
 
         goto done;
     }
 

-    dir = vfs_d_alloc(parent, &name);
+    if (!(dir = vfs_d_alloc(parent, &name))) {
+        errno = ENOMEM;
+        goto done;
+    }

 
     lock_dnode(parent);
     lock_inode(parent->inode);
 
     lock_dnode(parent);
     lock_inode(parent->inode);


@@ -1127,15 +1030,19 @@ done:
 int
 __vfs_do_unlink(struct v_dnode* dnode)
 {
 int
 __vfs_do_unlink(struct v_dnode* dnode)
 {

+    int errno;

     struct v_inode* inode = dnode->inode;
 
     if (dnode->ref_count > 1) {
         return EBUSY;
     }
 
     struct v_inode* inode = dnode->inode;
 
     if (dnode->ref_count > 1) {
         return EBUSY;
     }
 

+    if ((errno = vfs_check_writable(dnode))) {
+        return errno;
+    }
+

     lock_inode(inode);
 
     lock_inode(inode);
 

-    int errno;

     if (inode->open_count) {
         errno = EBUSY;
     } else if (!(inode->itype & VFS_IFDIR)) {
     if (inode->open_count) {
         errno = EBUSY;
     } else if (!(inode->itype & VFS_IFDIR)) {


@@ -1158,11 +1065,7 @@ __DEFINE_LXSYSCALL1(int, unlink, const char*, pathname)
 {
     int errno;
     struct v_dnode* dnode;
 {
     int errno;
     struct v_dnode* dnode;

-    if ((errno = vfs_walk(__current->cwd, pathname, &dnode, NULL, 0))) {
-        goto done;
-    }
-    if ((dnode->super_block->fs->types & FSTYPE_ROFS)) {
-        errno = EROFS;
+    if ((errno = vfs_walk_proc(pathname, &dnode, NULL, 0))) {

         goto done;
     }
 
         goto done;
     }
 


@@ -1176,7 +1079,7 @@ __DEFINE_LXSYSCALL2(int, unlinkat, int, fd, const char*, pathname)
 {
     int errno;
     struct v_fd* fd_s;
 {
     int errno;
     struct v_fd* fd_s;

-    if ((errno = __vfs_getfd(fd, &fd_s))) {
+    if ((errno = vfs_getfd(fd, &fd_s))) {

         goto done;
     }
 
         goto done;
     }
 


@@ -1211,7 +1114,8 @@ __DEFINE_LXSYSCALL1(int, fsync, int, fildes)
 {
     int errno;
     struct v_fd* fd_s;
 {
     int errno;
     struct v_fd* fd_s;

-    if (!(errno = __vfs_getfd(fildes, &fd_s))) {
+
+    if (!(errno = vfs_getfd(fildes, &fd_s))) {

         errno = vfs_fsync(fd_s->file);
     }
 
         errno = vfs_fsync(fd_s->file);
     }
 


@@ -1242,7 +1146,7 @@ vfs_dup2(int oldfd, int newfd)
 
     int errno;
     struct v_fd *oldfd_s, *newfd_s;
 
     int errno;
     struct v_fd *oldfd_s, *newfd_s;

-    if ((errno = __vfs_getfd(oldfd, &oldfd_s))) {
+    if ((errno = vfs_getfd(oldfd, &oldfd_s))) {

         goto done;
     }
 
         goto done;
     }
 


@@ -1274,7 +1178,7 @@ __DEFINE_LXSYSCALL1(int, dup, int, oldfd)
 {
     int errno, newfd;
     struct v_fd *oldfd_s, *newfd_s;
 {
     int errno, newfd;
     struct v_fd *oldfd_s, *newfd_s;

-    if ((errno = __vfs_getfd(oldfd, &oldfd_s))) {
+    if ((errno = vfs_getfd(oldfd, &oldfd_s))) {

         goto done;
     }
 
         goto done;
     }
 


@@ -1297,13 +1201,14 @@ __DEFINE_LXSYSCALL2(int,
 {
     int errno;
     struct v_dnode* dnode;
 {
     int errno;
     struct v_dnode* dnode;

-    if ((errno = vfs_walk(__current->cwd, pathname, &dnode, NULL, 0))) {
+    if ((errno = vfs_walk_proc(pathname, &dnode, NULL, 0))) {

         goto done;
     }
         goto done;
     }

-    if ((dnode->super_block->fs->types & FSTYPE_ROFS)) {
-        errno = EROFS;
+
+    if (errno = vfs_check_writable(dnode)) {

         goto done;
     }
         goto done;
     }

+

     if (!dnode->inode->ops->set_symlink) {
         errno = ENOTSUP;
         goto done;
     if (!dnode->inode->ops->set_symlink) {
         errno = ENOTSUP;
         goto done;


@@ -1319,8 +1224,22 @@ done:
     return DO_STATUS(errno);
 }
 
     return DO_STATUS(errno);
 }
 

+void
+vfs_ref_dnode(struct v_dnode* dnode)
+{
+    atomic_fetch_add(&dnode->ref_count, 1);
+    mnt_mkbusy(dnode->mnt);
+}
+
+void
+vfs_unref_dnode(struct v_dnode* dnode)
+{
+    atomic_fetch_sub(&dnode->ref_count, 1);
+    mnt_chillax(dnode->mnt);
+}
+

 int
 int

-__vfs_do_chdir(struct v_dnode* dnode)
+vfs_do_chdir(struct proc_info* proc, struct v_dnode* dnode)

 {
     int errno = 0;
 
 {
     int errno = 0;
 


@@ -1331,14 +1250,12 @@ __vfs_do_chdir(struct v_dnode* dnode)
         goto done;
     }
 
         goto done;
     }
 

-    if (__current->cwd) {
-        atomic_fetch_sub(&__current->cwd->ref_count, 1);
-        mnt_chillax(__current->cwd->mnt);
+    if (proc->cwd) {
+        vfs_unref_dnode(proc->cwd);

     }
 
     }
 

-    atomic_fetch_add(&dnode->ref_count, 1);
-    mnt_mkbusy(dnode->mnt);
-    __current->cwd = dnode;
+    vfs_ref_dnode(dnode);
+    proc->cwd = dnode;

 
     unlock_dnode(dnode);
 
 
     unlock_dnode(dnode);
 


@@ -1351,11 +1268,11 @@ __DEFINE_LXSYSCALL1(int, chdir, const char*, path)
     struct v_dnode* dnode;
     int errno = 0;
 
     struct v_dnode* dnode;
     int errno = 0;
 

-    if ((errno = vfs_walk(__current->cwd, path, &dnode, NULL, 0))) {
+    if ((errno = vfs_walk_proc(path, &dnode, NULL, 0))) {

         goto done;
     }
 
         goto done;
     }
 

-    errno = __vfs_do_chdir(dnode);
+    errno = vfs_do_chdir(__current, dnode);

 
 done:
     return DO_STATUS(errno);
 
 done:
     return DO_STATUS(errno);


@@ -1366,11 +1283,11 @@ __DEFINE_LXSYSCALL1(int, fchdir, int, fd)
     struct v_fd* fd_s;
     int errno = 0;
 
     struct v_fd* fd_s;
     int errno = 0;
 

-    if ((errno = __vfs_getfd(fd, &fd_s))) {
+    if ((errno = vfs_getfd(fd, &fd_s))) {

         goto done;
     }
 
         goto done;
     }
 

-    errno = __vfs_do_chdir(fd_s->file->dnode);
+    errno = vfs_do_chdir(__current, fd_s->file->dnode);

 
 done:
     return DO_STATUS(errno);
 
 done:
     return DO_STATUS(errno);


@@ -1388,7 +1305,7 @@ __DEFINE_LXSYSCALL2(char*, getcwd, char*, buf, size_t, size)
     size_t len = 0;
 
     if (!__current->cwd) {
     size_t len = 0;
 
     if (!__current->cwd) {

-        *buf = PATH_DELIM;
+        *buf = VFS_PATH_DELIM;

         len = 1;
     } else {
         len = vfs_get_path(__current->cwd, buf, size, 0);
         len = 1;
     } else {
         len = vfs_get_path(__current->cwd, buf, size, 0);


@@ -1410,11 +1327,16 @@ done:
 int
 vfs_do_rename(struct v_dnode* current, struct v_dnode* target)
 {
 int
 vfs_do_rename(struct v_dnode* current, struct v_dnode* target)
 {

+    int errno = 0;

     if (current->inode->id == target->inode->id) {
         // hard link
         return 0;
     }
 
     if (current->inode->id == target->inode->id) {
         // hard link
         return 0;
     }
 

+    if (errno = vfs_check_writable(current)) {
+        return errno;
+    }
+

     if (current->ref_count > 1 || target->ref_count > 1) {
         return EBUSY;
     }
     if (current->ref_count > 1 || target->ref_count > 1) {
         return EBUSY;
     }


@@ -1423,8 +1345,6 @@ vfs_do_rename(struct v_dnode* current, struct v_dnode* target)
         return EXDEV;
     }
 
         return EXDEV;
     }
 

-    int errno = 0;
-

     struct v_dnode* oldparent = current->parent;
     struct v_dnode* newparent = target->parent;
 
     struct v_dnode* oldparent = current->parent;
     struct v_dnode* newparent = target->parent;
 


@@ -1472,7 +1392,7 @@ __DEFINE_LXSYSCALL2(int, rename, const char*, oldpath, const char*, newpath)
     struct hstr name = HSTR(valloc(VFS_NAME_MAXLEN), 0);
     int errno = 0;
 
     struct hstr name = HSTR(valloc(VFS_NAME_MAXLEN), 0);
     int errno = 0;
 

-    if ((errno = vfs_walk(__current->cwd, oldpath, &cur, NULL, 0))) {
+    if ((errno = vfs_walk_proc(oldpath, &cur, NULL, 0))) {

         goto done;
     }
 
         goto done;
     }