feat: support ioctl() syscall for direct control to devices

[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 a69803ef6a3be08d68e9ea2ad6d784b44c10b3b5..476b00bbc7e28c4426f9711464cbfeee2052cf31 100644 (file)
--- a/lunaix-os/kernel/fs/vfs.c
+++ b/lunaix-os/kernel/fs/vfs.c
@@ -104,6 +104,7 @@ 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);
 }
 
     atomic_fetch_add(&vfs_sysroot->ref_count, 1);
 }
 


@@ -126,7 +127,7 @@ 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;
     }
 
     uint32_t hash = str->hash;


@@ -235,6 +236,10 @@ 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;


@@ -252,9 +257,33 @@ int
 vfs_close(struct v_file* file)
 {
     int errno = 0;
 vfs_close(struct v_file* file)
 {
     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--;

+
+        // Remove 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.
+        // FIXME is this a good solution?
+        /*
+         * Consider two process both open the same file both with fd=x.
+         *      Process A: busy on reading x
+         *      Process B: do nothing with x
+         * Assume 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.
+         *
+         * A possible solution is to add a owner identification in the lock
+         * context, so only the lock holder can do the release.
+         */
+        if (mutex_on_hold(&file->inode->lock)) {
+            unlock_inode(file->inode);
+        }

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


@@ -266,11 +295,16 @@ vfs_close(struct v_file* file)
 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);
     }


@@ -357,6 +391,7 @@ 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);


@@ -476,8 +511,9 @@ __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;
     }
 


@@ -549,9 +585,7 @@ __DEFINE_LXSYSCALL1(int, close, int, fd)
         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;
     }
 


@@ -597,9 +631,9 @@ __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) {


@@ -634,13 +668,11 @@ __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;


@@ -663,6 +695,11 @@ __DEFINE_LXSYSCALL3(int, write, int, fd, void*, buf, size_t, count)
     }
 
     struct v_file* file = fd_s->file;
     }
 
     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;


@@ -672,13 +709,11 @@ __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;


@@ -702,21 +737,30 @@ __DEFINE_LXSYSCALL3(int, lseek, int, fd, int, offset, int, options)
 
     struct v_file* file = fd_s->file;
 
 
     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;
     }
 


@@ -729,12 +773,12 @@ done:
 int
 vfs_get_path(struct v_dnode* dnode, char* buf, size_t size, int depth)
 {
 int
 vfs_get_path(struct v_dnode* dnode, char* buf, size_t size, int depth)
 {

-    if (!dnode) {
+    if (!dnode || dnode->parent == dnode) {

         return 0;
     }
 
     if (depth > 64) {
         return 0;
     }
 
     if (depth > 64) {

-        return ELOOP;
+        return ENAMETOOLONG;

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


@@ -743,14 +787,12 @@ vfs_get_path(struct v_dnode* dnode, char* buf, size_t size, int depth)
         return len;
     }
 
         return len;
     }
 

+    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++] = VFS_PATH_DELIM;
-    }
-

     return len;
 }
 
     return len;
 }
 


@@ -771,6 +813,19 @@ 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;
 __DEFINE_LXSYSCALL3(int, realpathat, int, fd, char*, buf, size_t, size)
 {
     int errno;


@@ -794,8 +849,7 @@ __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);
     }
 


@@ -849,12 +903,16 @@ __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;


@@ -904,17 +962,18 @@ __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);


@@ -942,15 +1001,19 @@ 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)) {


@@ -973,11 +1036,7 @@ __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;
     }
 


@@ -1026,6 +1085,7 @@ __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))) {
         errno = vfs_fsync(fd_s->file);
     }
     if (!(errno = vfs_getfd(fildes, &fd_s))) {
         errno = vfs_fsync(fd_s->file);
     }


@@ -1112,13 +1172,14 @@ __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;


@@ -1134,8 +1195,22 @@ 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;
 


@@ -1146,14 +1221,12 @@ __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);
 


@@ -1166,11 +1239,11 @@ __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);


@@ -1185,7 +1258,7 @@ __DEFINE_LXSYSCALL1(int, fchdir, int, fd)
         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);


@@ -1225,11 +1298,16 @@ 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;
     }


@@ -1238,8 +1316,6 @@ 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;
 


@@ -1287,7 +1363,7 @@ __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;
     }