With a bit guidance from Alexander Viro, here's a (tested) update of
romfs for the 2.3.7+ (in fact some newer 2.2.x -- need to backport them)
VFS changes, highlights (as an overview for what needs to be done for
other fs'es, but just as a beginning):
- romfs_lookup() returns a negative dentry for non-existent filenames;
instead of a "real" error. It seems to me, this is the most correct
way to handle the traditional ENOENT case; although it should only
matter for writable filesystems
- romfs_readpage() updated to the new semantics, it's mostly
straightforward, hopefully the lock_kernel()/unlock_kernel() is ok
too. (AFAICS, readpage is at a much lower level that "normal"
filesystems don't need to care about, but I do). The page->owner
hack possibly can go away once the debugging is resolved?
- 2.3.7+ inode_ops reorganization
- romfs_read_inode() is a bit streamlined, and now uses the all new
init_special_inode to create fifo/socket/device inodes, as per hints
of Alexander. (Also the possibly finally correct way to generate a
device number; maybe not, but it's working as one would expect).
The careful observer might note a few hundred bytes obesity in romfs
resulting from this patch (although none of the changes should warrant
bloat), and in general with 2.2/2.3 itself too. Perusal of the
generated assembly however showed that the fat is coming only from the
few new debugging hooks/printks, and the strings used by them, related
to wait queues, and (page) locking. Once they get removed, we can talk
about fat again :)
"Sadly", most of the page/buffer cache issues won't matter in this fs,
so I can't provide another example of doing it :) BTW, it's mostly
because of the "no page boundary" policy; which is not necessarily true,
a future optimization of the user-space tool will lay out data
differently, I guess I'll still need an special readpage to cope with
the partial pages; the question is, how crazy is a scheme which allows
the tail end of the file "float" to blocks reserved for slack, and even
move when extending the file :)
BTW, modular loop.c needs __mark_buffer_dirty exported in (possibly)
kernel/ksyms.c; and there are a lot of compile errors caused by locks
hidden behind __SMP__ around the masquerading code, but used
nevertheless..
diff -urpN linux-2.3.8-3-pre/fs/romfs/inode.c linux/fs/romfs/inode.c
--- linux-2.3.8-3-pre/fs/romfs/inode.c Thu Jun 10 07:01:55 1999
+++ linux/fs/romfs/inode.c Wed Jun 23 06:24:25 1999
@@ -1,7 +1,7 @@
/*
* ROMFS file system, Linux implementation
*
- * Copyright (C) 1997 Janos Farkas <chexum@shadow.banki.hu>
+ * Copyright (C) 1997-1999 Janos Farkas <chexum@shadow.banki.hu>
*
* Using parts of the minix filesystem
* Copyright (C) 1991, 1992 Linus Torvalds
@@ -57,6 +57,7 @@
#include <linux/fs.h>
#include <linux/locks.h>
#include <linux/init.h>
+#include <linux/smp_lock.h>
#include <asm/uaccess.h>
@@ -320,7 +321,7 @@ romfs_lookup(struct inode *dir, struct d
const char *name; /* got from dentry */
int len;
- res = 0; /* instead of ENOENT */
+ res = -EACCES; /* placeholder for "no data here" */
offset = dir->i_ino & ROMFH_MASK;
if (romfs_copyfrom(dir, &ri, offset, ROMFH_SIZE) <= 0)
goto out;
@@ -335,8 +336,9 @@ romfs_lookup(struct inode *dir, struct d
len = dentry->d_name.len;
for(;;) {
- if (!offset || offset >= maxoff
- || romfs_copyfrom(dir, &ri, offset, ROMFH_SIZE) <= 0)
+ if (!offset || offset >= maxoff)
+ goto out0;
+ if (romfs_copyfrom(dir, &ri, offset, ROMFH_SIZE) <= 0)
goto out;
/* try to match the first 16 bytes of name */
@@ -365,19 +367,28 @@ romfs_lookup(struct inode *dir, struct d
if ((ntohl(ri.next) & ROMFH_TYPE) == ROMFH_HRD)
offset = ntohl(ri.spec) & ROMFH_MASK;
- if ((inode = iget(dir->i_sb, offset))==NULL) {
- res = -EACCES;
- } else {
- d_add(dentry, inode);
- }
+ if ((inode = iget(dir->i_sb, offset)))
+ goto outi;
-out:
- return ERR_PTR(res);
+ /*
+ * it's a bit funky, _lookup needs to return an error code
+ * (negative) or a NULL, both as a dentry. ENOENT should not
+ * be returned, instead we need to create a negative dentry by
+ * d_add(dentry, NULL); and return 0 as no error.
+ * (Although as I see, it only matters on writable file
+ * systems).
+ */
+
+out0: inode = NULL;
+outi: res = 0;
+ d_add (dentry, inode);
+
+out: return ERR_PTR(res);
}
/*
* Ok, we do readpage, to be able to execute programs. Unfortunately,
- * we can't use bmap, since we have looser alignments.
+ * we can't use bmap, since we may have looser alignments.
*/
static int
@@ -389,12 +400,13 @@ romfs_readpage(struct file * file, struc
unsigned long offset, avail, readlen;
int result = -EIO;
- atomic_inc(&page->count);
- set_bit(PG_locked, &page->flags);
-
+ lock_kernel();
+ get_page(page);
buf = page_address(page);
- clear_bit(PG_uptodate, &page->flags);
- clear_bit(PG_error, &page->flags);
+
+ /* hack? */
+ page->owner = (int)current;
+
offset = page->offset;
if (offset < inode->i_size) {
avail = inode->i_size-offset;
@@ -403,18 +415,19 @@ romfs_readpage(struct file * file, struc
if (readlen < PAGE_SIZE) {
memset((void *)(buf+readlen),0,PAGE_SIZE-readlen);
}
- set_bit(PG_uptodate, &page->flags);
+ SetPageUptodate(page);
result = 0;
}
}
if (result) {
- set_bit(PG_error, &page->flags);
memset((void *)buf, 0, PAGE_SIZE);
+ SetPageError(page);
}
- clear_bit(PG_locked, &page->flags);
- wake_up(&page->wait);
+ UnlockPage(page);
+
free_page(buf);
+ unlock_kernel();
return result;
}
@@ -508,12 +521,16 @@ static struct inode_operations romfs_fil
NULL, /* rename */
NULL, /* readlink */
NULL, /* follow_link */
+
+ NULL, /* bmap -- not really */
romfs_readpage, /* readpage */
NULL, /* writepage */
- NULL, /* bmap -- not really */
+ NULL, /* flushpage */
+
NULL, /* truncate */
NULL, /* permission */
NULL, /* smap */
+ NULL, /* revalidate */
};
static struct file_operations romfs_dir_operations = {
@@ -550,12 +567,16 @@ static struct inode_operations romfs_dir
NULL, /* rename */
NULL, /* readlink */
NULL, /* follow_link */
+
+ NULL, /* bmap */
NULL, /* readpage */
NULL, /* writepage */
- NULL, /* bmap */
+ NULL, /* flushpage */
+
NULL, /* truncate */
NULL, /* permission */
NULL, /* smap */
+ NULL, /* revalidate */
};
static struct inode_operations romfs_link_inode_operations = {
@@ -571,18 +592,22 @@ static struct inode_operations romfs_lin
NULL, /* rename */
romfs_readlink, /* readlink */
romfs_follow_link, /* follow_link */
+
+ NULL, /* bmap */
NULL, /* readpage */
NULL, /* writepage */
- NULL, /* bmap */
+ NULL, /* flushpage */
+
NULL, /* truncate */
NULL, /* permission */
NULL, /* smap */
+ NULL, /* revalidate */
};
static mode_t romfs_modemap[] =
{
- 0, S_IFDIR, S_IFREG, S_IFLNK+0777,
- S_IFBLK, S_IFCHR, S_IFSOCK, S_IFIFO
+ 0, S_IFDIR+0644, S_IFREG+0644, S_IFLNK+0777,
+ S_IFBLK+0600, S_IFCHR+0600, S_IFSOCK+0644, S_IFIFO+0644
};
static struct inode_operations *romfs_inoops[] =
@@ -591,10 +616,10 @@ static struct inode_operations *romfs_in
&romfs_dir_inode_operations,
&romfs_file_inode_operations,
&romfs_link_inode_operations,
- &blkdev_inode_operations, /* standard handlers */
- &chrdev_inode_operations,
- NULL, /* socket */
- NULL, /* fifo */
+ NULL, /* device/fifo/socket nodes, */
+ NULL, /* set by init_special_inode */
+ NULL,
+ NULL,
};
static void
@@ -627,34 +652,30 @@ romfs_read_inode(struct inode *i)
i->i_mtime = i->i_atime = i->i_ctime = 0;
i->i_uid = i->i_gid = 0;
- i->i_op = romfs_inoops[nextfh & ROMFH_TYPE];
-
- /* Precalculate the data offset */
- ino = romfs_strnlen(i, ino+ROMFH_SIZE, ROMFS_MAXFN);
- if (ino >= 0)
- ino = ((ROMFH_SIZE+ino+1+ROMFH_PAD)&ROMFH_MASK);
- else
- ino = 0;
-
- i->u.romfs_i.i_metasize = ino;
- i->u.romfs_i.i_dataoffset = ino+(i->i_ino&ROMFH_MASK);
-
- /* Compute permissions */
- ino = S_IRUGO|S_IWUSR;
- ino |= romfs_modemap[nextfh & ROMFH_TYPE];
- if (nextfh & ROMFH_EXEC) {
- ino |= S_IXUGO;
- }
- i->i_mode = ino;
-
- if (S_ISFIFO(ino))
- init_fifo(i);
- else if (S_ISDIR(ino))
- i->i_size = i->u.romfs_i.i_metasize;
- else if (S_ISBLK(ino) || S_ISCHR(ino)) {
- i->i_mode &= ~(S_IRWXG|S_IRWXO);
- ino = ntohl(ri.spec);
- i->i_rdev = MKDEV(ino>>16,ino&0xffff);
+ /* Precalculate the data offset */
+ ino = romfs_strnlen(i, ino+ROMFH_SIZE, ROMFS_MAXFN);
+ if (ino >= 0)
+ ino = ((ROMFH_SIZE+ino+1+ROMFH_PAD)&ROMFH_MASK);
+ else
+ ino = 0;
+
+ i->u.romfs_i.i_metasize = ino;
+ i->u.romfs_i.i_dataoffset = ino+(i->i_ino&ROMFH_MASK);
+
+ /* Compute permissions */
+ ino = romfs_modemap[nextfh & ROMFH_TYPE];
+ /* only "normal" files have ops */
+ if ((i->i_op = romfs_inoops[nextfh & ROMFH_TYPE])) {
+ if (nextfh & ROMFH_EXEC)
+ ino |= S_IXUGO;
+ i->i_mode = ino;
+ if (S_ISDIR(ino))
+ i->i_size = i->u.romfs_i.i_metasize;
+ } else {
+ /* depending on MBZ for sock/fifos */
+ nextfh = ntohl(ri.spec);
+ nextfh = kdev_t_to_nr(MKDEV(nextfh>>16,nextfh&0xffff));
+ init_special_inode(i, ino, nextfh);
}
}
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/