kernel: Add initial support for tmpfs

Introduce the initial support for tmpfs, a readable and writable in-memory
filesystem.

Signed-off-by: Ian Moffett <ian@osmora.org>

5 files changed, 482 insertions, 1 deletions

diff --git a/share/misc/tmpfs b/share/misc/tmpfs
new file mode 100644
index 0000000..38eac22
--- /dev/null
+++ b/share/misc/tmpfs
@@ -0,0 +1,15 @@
+----------------------------------------------------
+| Readable + writable in-memory filesystem (tmpfs) |
++--------------------------------------------------+
+| Author: Ian Marco Moffett                        |
++--------------------------------------------------+
+
+------------------------------------------------------------------------------
+
+    [d] /tmp/ -> [next] -> foo.txt -> [next] -> bar.txt
+          \
+           +> /tmp/noises/ -> [next] -> mrow.txt -> [next] -> squeak.txt
+            \
+             +> /tmp/noises1/ -> [next] -> bark.bin -> [next] -> wrff.txt
+
+------------------------------------------------------------------------------
diff --git a/sys/fs/tmpfs.c b/sys/fs/tmpfs.c
new file mode 100644
index 0000000..9dce89a
--- /dev/null
+++ b/sys/fs/tmpfs.c
@@ -0,0 +1,390 @@
+/*
+ * Copyright (c) 2023-2025 Ian Marco Moffett and the Osmora Team.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of Hyra nor the names of its
+ *    contributors may be used to endorse or promote products derived from
+ *    this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <sys/mount.h>
+#include <sys/errno.h>
+#include <sys/syslog.h>
+#include <sys/types.h>
+#include <sys/cdefs.h>
+#include <sys/param.h>
+#include <sys/panic.h>
+#include <sys/vnode.h>
+#include <vm/dynalloc.h>
+#include <vm/vm_obj.h>
+#include <vm/vm_page.h>
+#include <vm/vm_pager.h>
+#include <fs/tmpfs.h>
+#include <string.h>
+
+#define ROOT_RPATH "/tmp"
+#define TMPFS_BSIZE DEFAULT_PAGESIZE
+
+#define pr_trace(fmt, ...) kprintf("tmpfs: " fmt, ##__VA_ARGS__)
+#define pr_error(...) pr_trace(__VA_ARGS__)
+
+static TAILQ_HEAD(, tmpfs_node) root;
+
+/*
+ * Generate a vnode for a specific tmpfs
+ * node.
+ */
+static int
+tmpfs_ref(struct tmpfs_node *np)
+{
+    struct vnode *vp = NULL;
+    int retval = 0;
+
+    if (np->vp == NULL) {
+        spinlock_acquire(&np->lock);
+        retval = vfs_alloc_vnode(&vp, np->type);
+        np->vp = vp;
+        spinlock_release(&np->lock);
+    }
+
+    if (vp != NULL) {
+        vp->data = np;
+        vp->vops = &g_tmpfs_vops;
+    }
+
+    return retval;
+}
+
+/*
+ * Perform lookup within the tmpfs namespace
+ *
+ * XXX: This operations is serialized
+ * TODO: Support multiple directories (only fs root now)
+ *
+ * @rpath: /tmp/ relative path to lookup
+ * @res: The result is written here (must NOT be NULL)
+ */
+static int
+tmpfs_do_lookup(const char *rpath, struct tmpfs_node **res)
+{
+    struct tmpfs_node *cnp;
+    struct tmpfs_node *dirent;
+    int error = 0;
+
+    /*
+     * If the directory is the node that we are
+     * looking for, return it. But if it is not
+     * and it is empty then there is nothing
+     * we can do.
+     */
+    cnp = TAILQ_FIRST(&root);
+    if (strcmp(cnp->rpath, rpath) == 0) {
+        *res = cnp;
+        return 0;
+    }
+    if (TAILQ_NELEM(&cnp->dirents) == 0) {
+        return -ENOENT;
+    }
+
+    /*
+     * Go through each tmpfs dirent to see if we can
+     * find the file we are looking for.
+     */
+    spinlock_acquire(&cnp->lock);
+    dirent = TAILQ_FIRST(&cnp->dirents);
+    while (dirent != NULL) {
+        if (strcmp(dirent->rpath, rpath) == 0) {
+            break;
+        }
+
+        dirent = TAILQ_NEXT(dirent, link);
+    }
+
+    spinlock_release(&cnp->lock);
+    if (res == NULL) {
+        return -ENOENT;
+    }
+
+    if ((error = tmpfs_ref(dirent)) != 0) {
+        return error;
+    }
+
+    *res = dirent;
+    return 0;
+}
+
+/*
+ * TMPFS lookup callback for the VFS
+ *
+ * Takes some arguments and returns a vnode
+ * in args->vpp
+ */
+static int
+tmpfs_lookup(struct vop_lookup_args *args)
+{
+    struct tmpfs_node *np;
+    int error;
+
+    if (args == NULL) {
+        return -EINVAL;
+    }
+    if (args->name == NULL) {
+        return -EINVAL;
+    }
+
+    /*
+     * Attempt to find the node we want, if it already
+     * has a vnode attached to it then that's something we
+     * want. However we should allocate a new vnode if we
+     * need to.
+     */
+    error = tmpfs_do_lookup(args->name, &np);
+    if (error != 0) {
+        return error;
+    }
+
+    *args->vpp = np->vp;
+    return 0;
+}
+
+/*
+ * TMPFS create callback for the VFS
+ *
+ * Creates a new TMPFS node
+ */
+static int
+tmpfs_create(struct vop_create_args *args)
+{
+    const char *pcp = args->path;   /* Stay away from boat, kids */
+    struct vnode *dirvp;
+    struct tmpfs_node *np;
+    struct tmpfs_node *root_np;
+    int error;
+
+    /* Validate inputs */
+    if (args == NULL)
+        return -EINVAL;
+    if (pcp == NULL)
+        return -EIO;
+    if ((dirvp = args->dirvp) == NULL)
+        return -EIO;
+
+    /* Remove the leading "/tmp/" */
+    pcp += sizeof(ROOT_RPATH);
+    if (*pcp == '\0') {
+        return -ENOENT;
+    }
+
+    np = dynalloc(sizeof(*np));
+    if (np == NULL) {
+        return -ENOMEM;
+    }
+
+    memset(np, 0, sizeof(*np));
+
+    /*
+     * TODO: Support multiple directories.
+     *
+     * XXX: We currently only create a TMPFS_REG node as
+     *      to keep things initially simple.
+     */
+    root_np = TAILQ_FIRST(&root);
+    np->dirvp = dirvp;
+    np->type = TMPFS_REG;
+    memcpy(np->rpath, pcp, strlen(pcp) + 1);
+    TAILQ_INSERT_TAIL(&root_np->dirents, np, link);
+
+    if ((error = tmpfs_ref(np)) != 0) {
+        return error;
+    }
+
+    *args->vpp = np->vp;
+    return 0;
+}
+
+/*
+ * TMPFS write callback for VFS
+ *
+ * Node buffers are orthogonally managed. That is, each
+ * node has their own respective data buffers. When
+ * writing to a node, we need to take into account of the
+ * length of the buffer. This value may need to expanded as
+ * well as more pages allocated if the amount of bytes to
+ * be written exceeds it.
+ */
+static int
+tmpfs_write(struct vnode *vp, struct sio_txn *sio)
+{
+    struct tmpfs_node *np;
+    uint8_t *buf;
+
+    if (sio->buf == NULL || sio->len == 0) {
+        return -EINVAL;
+    }
+
+    /* This should not happen but you never know */
+    if ((np = vp->data) == NULL) {
+        return -EIO;
+    }
+
+    /* Is this even a regular file? */
+    if (np->type != VREG) {
+        return -EISDIR;
+    }
+
+    spinlock_acquire(&np->lock);
+
+    /*
+     * If the residual byte count is zero, we need to
+     * allocate a new page to be used. However if this
+     * fails we'll throw back an -ENOMEM.
+     */
+    if (np->len == 0) {
+        np->data = dynalloc(TMPFS_BSIZE);
+        if (np->data == NULL) {
+            spinlock_release(&np->lock);
+            return -ENOMEM;
+        }
+        np->len += TMPFS_BSIZE;
+    }
+
+    /*
+     * If the length to be written exceeds the residual byte
+     * count. We will try to expand the buffer by the page
+     * size. However, if this fails, we will split the write
+     * into a suitable size that does not overflow what we
+     * have left.
+     */
+    if ((sio->offset + sio->len) > np->len) {
+        np->data = dynrealloc(np->data, (sio->offset + sio->len));
+        if (np->data == NULL) {
+            sio->len = np->len;
+        } else {
+            np->len = sio->offset + sio->len;
+        }
+    }
+
+    buf = np->data;
+    memcpy(&buf[sio->offset], sio->buf, sio->len);
+    spinlock_release(&np->lock);
+    kprintf("%d\n", sio->len);
+    return sio->len;
+}
+
+/*
+ * TMPFS read callback for VFS
+ */
+static int
+tmpfs_read(struct vnode *vp, struct sio_txn *sio)
+{
+    struct tmpfs_node *np;
+    uint8_t *buf;
+
+    if (sio->buf == NULL || sio->len == 0) {
+        return -EINVAL;
+    }
+
+    /* This should not happen but you never know */
+    if ((np = vp->data) == NULL) {
+        return -EIO;
+    }
+
+    /* Is this even a regular file? */
+    if (np->type != VREG) {
+        return -EISDIR;
+    }
+
+    spinlock_acquire(&np->lock);
+
+    if (sio->offset > np->len - 1) {
+        return -EINVAL;
+    }
+    if ((sio->offset + sio->len) > np->len) {
+        sio->len = np->len;
+    }
+
+    buf = np->data;
+    memcpy(sio->buf, &buf[sio->offset], sio->len);
+    spinlock_release(&np->lock);
+    return sio->len;
+}
+
+static int
+tmpfs_reclaim(struct vnode *vp)
+{
+    struct tmpfs_node *np;
+
+    if ((np = vp->data) == NULL) {
+        return 0;
+    }
+
+    np->vp = NULL;
+    vp->data = NULL;
+    return 0;
+}
+
+static int
+tmpfs_init(struct fs_info *fip)
+{
+    struct tmpfs_node *np;
+    struct vnode *vp;
+    struct mount *mp;
+    int error;
+
+    /* Grab ourselves a new vnode for /tmp */
+    if ((error = vfs_alloc_vnode(&vp, VDIR)) != 0) {
+        return error;
+    }
+
+    vp->vops = &g_tmpfs_vops;
+    mp = vfs_alloc_mount(vp, fip);
+    vfs_name_mount(mp, "tmp");
+    TAILQ_INSERT_TAIL(&g_mountlist, mp, mnt_list);
+
+    /* Pre-allocate the first entry */
+    if ((np = dynalloc(sizeof(*np))) == NULL) {
+        return -ENOMEM;
+    }
+
+    TAILQ_INIT(&root);
+    memset(np, 0, sizeof(*np));
+
+    memcpy(np->rpath, ROOT_RPATH, sizeof(ROOT_RPATH));
+    np->type = TMPFS_DIR;
+    TAILQ_INIT(&np->dirents);
+    TAILQ_INSERT_TAIL(&root, np, link);
+    return 0;
+}
+
+const struct vops g_tmpfs_vops = {
+    .lookup = tmpfs_lookup,
+    .getattr = NULL,
+    .read = tmpfs_read,
+    .write = tmpfs_write,
+    .reclaim = tmpfs_reclaim,
+    .create = tmpfs_create,
+};
+
+const struct vfsops g_tmpfs_vfsops = {
+    .init = tmpfs_init
+};
diff --git a/sys/include/fs/tmpfs.h b/sys/include/fs/tmpfs.h
new file mode 100644
index 0000000..b2a5bbe
--- /dev/null
+++ b/sys/include/fs/tmpfs.h
@@ -0,0 +1,73 @@
+/*
+ * Copyright (c) 2023-2025 Ian Marco Moffett and the Osmora Team.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of Hyra nor the names of its
+ *    contributors may be used to endorse or promote products derived from
+ *    this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef _FS_TMPFS_H_
+#define _FS_TMPFS_H_
+
+#include <sys/types.h>
+#include <sys/limits.h>
+#include <sys/vnode.h>
+#include <sys/queue.h>
+#include <sys/spinlock.h>
+#include <vm/vm_obj.h>
+
+extern const struct vops g_tmpfs_vops;
+
+/* Tmpfs node types */
+#define TMPFS_NONE  (VNON)        /* No type */
+#define TMPFS_REG   (VREG)        /* Regular file [f] */
+#define TMPFS_DIR   (VDIR)        /* Directory    [d] */
+
+struct tmpfs_node;
+
+/*
+ * A tmpfs node represents an object within the
+ * tmpfs namespace such as a file, directory, etc.
+ *
+ * @rpath: /tmp/ relative path (for lookups)
+ * @type: The tmpfs node type [one-to-one to vtype]
+ * @len: Length of buffer
+ * @data: The backing file data
+ * @dirvp: Vnode of the parent node
+ * @vp: Vnode of the current node
+ * @lock: Lock protecting this node
+ */
+struct tmpfs_node {
+    char rpath[PATH_MAX];
+    uint8_t type;
+    size_t len;
+    void *data;
+    struct vnode *dirvp;
+    struct vnode *vp;
+    struct spinlock lock;
+    TAILQ_HEAD(, tmpfs_node) dirents;
+    TAILQ_ENTRY(tmpfs_node) link;
+};
+
+#endif  /* !_FS_TMPFS_H_ */
diff --git a/sys/include/sys/mount.h b/sys/include/sys/mount.h
index 1fcdbfa..636c7bf 100644
--- a/sys/include/sys/mount.h
+++ b/sys/include/sys/mount.h
@@ -47,6 +47,7 @@
 #define MOUNT_RAMFS "initramfs"
 #define MOUNT_DEVFS "devfs"
 #define MOUNT_CTLFS "ctlfs"
+#define MOUNT_TMPFS "tmpfs"
 
 struct vfsops;
 struct mount;
@@ -59,6 +60,7 @@ extern mountlist_t g_mountlist;
 extern const struct vfsops g_initramfs_vfsops;
 extern const struct vfsops g_devfs_vfsops;
 extern const struct vfsops g_ctlfs_vfsops;
+extern const struct vfsops g_tmpfs_vfsops;
 
 struct mount {
     char *name;
diff --git a/sys/kern/vfs_init.c b/sys/kern/vfs_init.c
index 8c1bc74..bc7f8b0 100644
--- a/sys/kern/vfs_init.c
+++ b/sys/kern/vfs_init.c
@@ -37,7 +37,8 @@ struct vnode *g_root_vnode = NULL;
 static struct fs_info fs_list[] = {
     {MOUNT_RAMFS, &g_initramfs_vfsops, 0, 0},
     {MOUNT_DEVFS, &g_devfs_vfsops, 0, 0},
-    {MOUNT_CTLFS, &g_ctlfs_vfsops, 0, 0}
+    {MOUNT_CTLFS, &g_ctlfs_vfsops, 0, 0},
+    {MOUNT_TMPFS, &g_tmpfs_vfsops, 0, 0}
 };
 
 void