Make NFS userspace and server directories cookies independant [patch, fc3, 2.6.10-1.766_FC3]

From: Olivier Galibert
Date: Mon Mar 21 2005 - 11:23:33 EST


Following Trond's advice, I made a patch to make the userspace nfs
directory cookies independant of the nfs server ones. It nows behaves
pretty much identically to tmpfs and friends, with f_pos being the
entry number and a 1-entry cache to ensure that you can read part of
the directory, delete files and keep reading.

With that patch and the one added to bk not so long ago (10-15 days)
about NLM packets authentication I'm a happy camper in an heterogenous
environment including old SGIs used as fileservers.

Patch is against a fedora core 3 2.6.10, it requires minor
modifications (context change, not contents) in the nfs_do_filldir par
of the dir.c change to account for the dirent addition to apply to
2.6.11. I can update it if useful.

OG.

Signed-off-by: Olivier Galibert <galibert@xxxxxxxxx>

fs/nfs/dir.c | 114 ++++++++++++++++++++++++++++++++++++++-----------
fs/nfs/inode.c | 2
include/linux/nfs_fs.h | 3 +
3 files changed, 95 insertions(+), 24 deletions(-)

--- ../save/kernel-2.6.10/linux-2.6.10/fs/nfs/dir.c 2005-03-16 14:41:24.000000000 +0100
+++ linux-2.6.10/fs/nfs/dir.c 2005-03-21 16:57:42.495392504 +0100
@@ -118,7 +118,9 @@ typedef struct {
struct page *page;
unsigned long page_index;
u32 *ptr;
- u64 target;
+ u64 target_cookie;
+ int target_index;
+ int current_index;
struct nfs_entry *entry;
decode_dirent_t decode;
int plus;
@@ -204,14 +206,14 @@ void dir_page_release(nfs_readdir_descri

/*
* Given a pointer to a buffer that has already been filled by a call
- * to readdir, find the next entry.
+ * to readdir, find the next entry with cookie 'desc->target_cookie'.
*
* If the end of the buffer has been reached, return -EAGAIN, if not,
* return the offset within the buffer of the next entry to be
* read.
*/
static inline
-int find_dirent(nfs_readdir_descriptor_t *desc, struct page *page)
+int find_dirent(nfs_readdir_descriptor_t *desc)
{
struct nfs_entry *entry = desc->entry;
int loop_count = 0,
@@ -219,7 +221,7 @@ int find_dirent(nfs_readdir_descriptor_t

while((status = dir_decode(desc)) == 0) {
dfprintk(VFS, "NFS: found cookie %Lu\n", (long long)entry->cookie);
- if (entry->prev_cookie == desc->target)
+ if (entry->prev_cookie == desc->target_cookie)
break;
if (loop_count++ > 200) {
loop_count = 0;
@@ -231,8 +233,44 @@ int find_dirent(nfs_readdir_descriptor_t
}

/*
- * Find the given page, and call find_dirent() in order to try to
- * return the next entry.
+ * Given a pointer to a buffer that has already been filled by a call
+ * to readdir, find the entry at offset 'desc->target_index'.
+ *
+ * If the end of the buffer has been reached, return -EAGAIN, if not,
+ * return the offset within the buffer of the next entry to be
+ * read.
+ */
+static inline
+int find_dirent_index(nfs_readdir_descriptor_t *desc)
+{
+ struct nfs_entry *entry = desc->entry;
+ int loop_count = 0,
+ status;
+
+ for(;;) {
+ status = dir_decode(desc);
+ if (status)
+ break;
+
+ dfprintk(VFS, "NFS: found cookie %Lu at index %d\n", (long long)entry->cookie, desc->current_index);
+
+ if (desc->target_index == desc->current_index) {
+ desc->target_cookie = entry->cookie;
+ break;
+ }
+ desc->current_index++;
+ if (loop_count++ > 200) {
+ loop_count = 0;
+ schedule();
+ }
+ }
+ dfprintk(VFS, "NFS: find_dirent_index() returns %d\n", status);
+ return status;
+}
+
+/*
+ * Find the given page, and call find_dirent() or find_dirent_index in
+ * order to try to return the next entry.
*/
static inline
int find_dirent_page(nfs_readdir_descriptor_t *desc)
@@ -255,7 +293,10 @@ int find_dirent_page(nfs_readdir_descrip
/* NOTE: Someone else may have changed the READDIRPLUS flag */
desc->page = page;
desc->ptr = kmap(page); /* matching kunmap in nfs_do_filldir */
- status = find_dirent(desc, page);
+ if (desc->target_cookie)
+ status = find_dirent(desc);
+ else
+ status = find_dirent_index(desc);
if (status < 0)
dir_page_release(desc);
out:
@@ -270,7 +311,8 @@ int find_dirent_page(nfs_readdir_descrip
* Recurse through the page cache pages, and return a
* filled nfs_entry structure of the next directory entry if possible.
*
- * The target for the search is 'desc->target'.
+ * The target for the search is 'desc->target_cookie' if non-0,
+ * 'desc->target_index' otherwise
*/
static inline
int readdir_search_pagecache(nfs_readdir_descriptor_t *desc)
@@ -278,7 +320,19 @@ int readdir_search_pagecache(nfs_readdir
int loop_count = 0;
int res;

- dfprintk(VFS, "NFS: readdir_search_pagecache() searching for cookie %Lu\n", (long long)desc->target);
+ if (desc->target_cookie)
+ dfprintk(VFS, "NFS: readdir_search_pagecache() searching for cookie %Lu\n", (long long)desc->target_cookie);
+ else
+ dfprintk(VFS, "NFS: readdir_search_pagecache() searching for cookie number %d\n", desc->target_index);
+
+ /* Always search-by-index from the beginning of the cache */
+ if (!(desc->target_cookie)) {
+ desc->page_index = 0;
+ desc->entry->cookie = desc->entry->prev_cookie = 0;
+ desc->entry->eof = 0;
+ desc->current_index = 0;
+ }
+
for (;;) {
res = find_dirent_page(desc);
if (res != -EAGAIN)
@@ -321,11 +375,12 @@ int nfs_do_filldir(nfs_readdir_descripto
{
struct file *file = desc->file;
struct nfs_entry *entry = desc->entry;
+ struct nfs_open_context *ctx = file->private_data;
unsigned long fileid;
int loop_count = 0,
res;

- dfprintk(VFS, "NFS: nfs_do_filldir() filling starting @ cookie %Lu\n", (long long)desc->target);
+ dfprintk(VFS, "NFS: nfs_do_filldir() filling starting @ cookie %Lu\n", (long long)entry->cookie);

for(;;) {
unsigned d_type = DT_UNKNOWN;
@@ -338,10 +393,10 @@ int nfs_do_filldir(nfs_readdir_descripto
d_type = nfs_type_to_d_type(entry->fattr->type);

res = filldir(dirent, entry->name, entry->len,
- entry->prev_cookie, fileid, d_type);
+ file->f_pos, fileid, d_type);
if (res < 0)
break;
- file->f_pos = desc->target = entry->cookie;
+ file->f_pos++;
if (dir_decode(desc) != 0) {
desc->page_index ++;
break;
@@ -351,9 +406,12 @@ int nfs_do_filldir(nfs_readdir_descripto
schedule();
}
}
+ ctx->dir_pos = file->f_pos;
+ ctx->dir_cookie = entry->cookie;
+ desc->target_cookie = entry->cookie;
dir_page_release(desc);

- dfprintk(VFS, "NFS: nfs_do_filldir() filling ended @ cookie %Lu; returning = %d\n", (long long)desc->target, res);
+ dfprintk(VFS, "NFS: nfs_do_filldir() filling ended @ cookie %Lu; returning = %d\n", (long long)ctx->dir_cookie, res);
return res;
}

@@ -379,14 +437,14 @@ int uncached_readdir(nfs_readdir_descrip
struct page *page = NULL;
int status;

- dfprintk(VFS, "NFS: uncached_readdir() searching for cookie %Lu\n", (long long)desc->target);
+ dfprintk(VFS, "NFS: uncached_readdir() searching for cookie %Lu\n", (long long)desc->target_cookie);

page = alloc_page(GFP_HIGHUSER);
if (!page) {
status = -ENOMEM;
goto out;
}
- desc->error = NFS_PROTO(inode)->readdir(file->f_dentry, cred, desc->target,
+ desc->error = NFS_PROTO(inode)->readdir(file->f_dentry, cred, desc->target_cookie,
page,
NFS_SERVER(inode)->dtsize,
desc->plus);
@@ -395,7 +453,7 @@ int uncached_readdir(nfs_readdir_descrip
desc->ptr = kmap(page); /* matching kunmap in nfs_do_filldir */
if (desc->error >= 0) {
if ((status = dir_decode(desc)) == 0)
- desc->entry->prev_cookie = desc->target;
+ desc->entry->prev_cookie = desc->target_cookie;
} else
status = -EIO;
if (status < 0)
@@ -416,13 +474,15 @@ int uncached_readdir(nfs_readdir_descrip
goto out;
}

-/* The file offset position is now represented as a true offset into the
- * page cache as is the case in most of the other filesystems.
+/* The file offset position represents the dirent entry number. A
+ last cookie cache takes care of the common case of reading the
+ whole directory.
*/
static int nfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
{
struct dentry *dentry = filp->f_dentry;
struct inode *inode = dentry->d_inode;
+ struct nfs_open_context *ctx = filp->private_data;
nfs_readdir_descriptor_t my_desc,
*desc = &my_desc;
struct nfs_entry my_entry;
@@ -439,17 +499,22 @@ static int nfs_readdir(struct file *filp
}

/*
- * filp->f_pos points to the file offset in the page cache.
- * but if the cache has meanwhile been zapped, we need to
- * read from the last dirent to revalidate f_pos
- * itself.
+ * filp->f_pos points to the dirent entry number.
+ * ctx->dir_pos has the number of the cached cookie. We have
+ * to either find the entry with the appropriate number or
+ * revalidate the cookie.
*/
memset(desc, 0, sizeof(*desc));

desc->file = filp;
- desc->target = filp->f_pos;
desc->decode = NFS_PROTO(inode)->decode_dirent;
desc->plus = NFS_USE_READDIRPLUS(inode);
+ desc->target_index = filp->f_pos;
+
+ if (filp->f_pos == ctx->dir_pos)
+ desc->target_cookie = ctx->dir_cookie;
+ else
+ desc->target_cookie = 0;

my_entry.cookie = my_entry.prev_cookie = 0;
my_entry.eof = 0;
@@ -459,9 +524,10 @@ static int nfs_readdir(struct file *filp

while(!desc->entry->eof) {
res = readdir_search_pagecache(desc);
+
if (res == -EBADCOOKIE) {
/* This means either end of directory */
- if (desc->entry->cookie != desc->target) {
+ if (desc->target_cookie && desc->entry->cookie != desc->target_cookie) {
/* Or that the server has 'lost' a cookie */
res = uncached_readdir(desc, dirent, filldir);
if (res >= 0)
--- ../save/kernel-2.6.10/linux-2.6.10/fs/nfs/inode.c 2004-12-24 22:35:24.000000000 +0100
+++ linux-2.6.10/fs/nfs/inode.c 2005-03-16 18:02:03.000000000 +0100
@@ -831,6 +831,8 @@ struct nfs_open_context *alloc_nfs_open_
ctx->state = NULL;
ctx->lockowner = current->files;
ctx->error = 0;
+ ctx->dir_pos = 0;
+ ctx->dir_cookie = 0;
init_waitqueue_head(&ctx->waitq);
}
return ctx;
--- ../save/kernel-2.6.10/linux-2.6.10/include/linux/nfs_fs.h 2005-03-16 14:41:24.000000000 +0100
+++ linux-2.6.10/include/linux/nfs_fs.h 2005-03-16 14:41:53.000000000 +0100
@@ -96,6 +96,9 @@ struct nfs_open_context {

struct list_head list;
wait_queue_head_t waitq;
+
+ int dir_pos; /* Directory cookie cache */
+ __u64 dir_cookie;
};

/*
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/