linux/fs/afs/dir.c

1202 lines
28 KiB
C
Raw Normal View History

/* dir.c: AFS filesystem directory handling
*
* Copyright (C) 2002 Red Hat, Inc. All Rights Reserved.
* Written by David Howells (dhowells@redhat.com)
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/namei.h>
#include <linux/pagemap.h>
#include <linux/ctype.h>
#include <linux/sched.h>
#include "internal.h"
static struct dentry *afs_lookup(struct inode *dir, struct dentry *dentry,
unsigned int flags);
static int afs_dir_open(struct inode *inode, struct file *file);
static int afs_readdir(struct file *file, struct dir_context *ctx);
static int afs_d_revalidate(struct dentry *dentry, unsigned int flags);
static int afs_d_delete(const struct dentry *dentry);
static void afs_d_release(struct dentry *dentry);
static int afs_lookup_filldir(struct dir_context *ctx, const char *name, int nlen,
[PATCH] VFS: Make filldir_t and struct kstat deal in 64-bit inode numbers These patches make the kernel pass 64-bit inode numbers internally when communicating to userspace, even on a 32-bit system. They are required because some filesystems have intrinsic 64-bit inode numbers: NFS3+ and XFS for example. The 64-bit inode numbers are then propagated to userspace automatically where the arch supports it. Problems have been seen with userspace (eg: ld.so) using the 64-bit inode number returned by stat64() or getdents64() to differentiate files, and failing because the 64-bit inode number space was compressed to 32-bits, and so overlaps occur. This patch: Make filldir_t take a 64-bit inode number and struct kstat carry a 64-bit inode number so that 64-bit inode numbers can be passed back to userspace. The stat functions then returns the full 64-bit inode number where available and where possible. If it is not possible to represent the inode number supplied by the filesystem in the field provided by userspace, then error EOVERFLOW will be issued. Similarly, the getdents/readdir functions now pass the full 64-bit inode number to userspace where possible, returning EOVERFLOW instead when a directory entry is encountered that can't be properly represented. Note that this means that some inodes will not be stat'able on a 32-bit system with old libraries where they were before - but it does mean that there will be no ambiguity over what a 32-bit inode number refers to. Note similarly that directory scans may be cut short with an error on a 32-bit system with old libraries where the scan would work before for the same reasons. It is judged unlikely that this situation will occur because modern glibc uses 64-bit capable versions of stat and getdents class functions exclusively, and that older systems are unlikely to encounter unrepresentable inode numbers anyway. [akpm: alpha build fix] Signed-off-by: David Howells <dhowells@redhat.com> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-10-03 08:13:46 +00:00
loff_t fpos, u64 ino, unsigned dtype);
static int afs_create(struct inode *dir, struct dentry *dentry, umode_t mode,
bool excl);
static int afs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode);
static int afs_rmdir(struct inode *dir, struct dentry *dentry);
static int afs_unlink(struct inode *dir, struct dentry *dentry);
static int afs_link(struct dentry *from, struct inode *dir,
struct dentry *dentry);
static int afs_symlink(struct inode *dir, struct dentry *dentry,
const char *content);
static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
struct inode *new_dir, struct dentry *new_dentry,
unsigned int flags);
const struct file_operations afs_dir_file_operations = {
.open = afs_dir_open,
.release = afs_release,
.iterate_shared = afs_readdir,
.lock = afs_lock,
.llseek = generic_file_llseek,
};
const struct inode_operations afs_dir_inode_operations = {
.create = afs_create,
.lookup = afs_lookup,
.link = afs_link,
.unlink = afs_unlink,
.symlink = afs_symlink,
.mkdir = afs_mkdir,
.rmdir = afs_rmdir,
.rename = afs_rename,
.permission = afs_permission,
.getattr = afs_getattr,
.setattr = afs_setattr,
.listxattr = afs_listxattr,
};
const struct dentry_operations afs_fs_dentry_operations = {
.d_revalidate = afs_d_revalidate,
.d_delete = afs_d_delete,
.d_release = afs_d_release,
.d_automount = afs_d_automount,
};
#define AFS_DIR_HASHTBL_SIZE 128
#define AFS_DIR_DIRENT_SIZE 32
#define AFS_DIRENT_PER_BLOCK 64
union afs_dirent {
struct {
uint8_t valid;
uint8_t unused[1];
__be16 hash_next;
__be32 vnode;
__be32 unique;
uint8_t name[16];
uint8_t overflow[4]; /* if any char of the name (inc
* NUL) reaches here, consume
* the next dirent too */
} u;
uint8_t extended_name[32];
};
/* AFS directory page header (one at the beginning of every 2048-byte chunk) */
struct afs_dir_pagehdr {
__be16 npages;
__be16 magic;
#define AFS_DIR_MAGIC htons(1234)
uint8_t nentries;
uint8_t bitmap[8];
uint8_t pad[19];
};
/* directory block layout */
union afs_dir_block {
struct afs_dir_pagehdr pagehdr;
struct {
struct afs_dir_pagehdr pagehdr;
uint8_t alloc_ctrs[128];
/* dir hash table */
uint16_t hashtable[AFS_DIR_HASHTBL_SIZE];
} hdr;
union afs_dirent dirents[AFS_DIRENT_PER_BLOCK];
};
/* layout on a linux VM page */
struct afs_dir_page {
union afs_dir_block blocks[PAGE_SIZE / sizeof(union afs_dir_block)];
};
struct afs_lookup_cookie {
struct dir_context ctx;
struct afs_fid fid;
struct qstr name;
int found;
};
/*
* check that a directory page is valid
*/
static inline bool afs_dir_check_page(struct inode *dir, struct page *page)
{
struct afs_dir_page *dbuf;
loff_t latter;
int tmp, qty;
#if 0
/* check the page count */
qty = desc.size / sizeof(dbuf->blocks[0]);
if (qty == 0)
goto error;
if (page->index == 0 && qty != ntohs(dbuf->blocks[0].pagehdr.npages)) {
printk("kAFS: %s(%lu): wrong number of dir blocks %d!=%hu\n",
__func__, dir->i_ino, qty,
ntohs(dbuf->blocks[0].pagehdr.npages));
goto error;
}
#endif
/* determine how many magic numbers there should be in this page */
[PATCH] fix possible PAGE_CACHE_SHIFT overflows We've had two instances recently of overflows when doing 64_bit_value = (32_bit_value << PAGE_CACHE_SHIFT) I did a tree-wide grep of `<<.*PAGE_CACHE_SHIFT' and this is the result. - afs_rxfs_fetch_descriptor.offset is of type off_t, which seems broken. - jfs and jffs are limited to 4GB anyway. - reiserfs map_block_for_writepage() takes an unsigned long for the block - it should take sector_t. (It'll fail for huge filesystems with blocksize<PAGE_CACHE_SIZE) - cramfs_read() needs to use sector_t (I think cramsfs is busted on large filesystems anyway) - affs is limited in file size anyway. - I generally didn't fix 32-bit overflows in directory operations. - arm's __flush_dcache_page() is peculiar. What if the page lies beyond 4G? - gss_wrap_req_priv() needs checking (snd_buf->page_base) Cc: Oleg Drokin <green@linuxhacker.ru> Cc: David Howells <dhowells@redhat.com> Cc: David Woodhouse <dwmw2@infradead.org> Cc: <reiserfs-dev@namesys.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Anton Altaparmakov <aia21@cantab.net> Cc: Jeff Dike <jdike@addtoit.com> Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Roman Zippel <zippel@linux-m68k.org> Cc: <linux-fsdevel@vger.kernel.org> Cc: Miklos Szeredi <miklos@szeredi.hu> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Trond Myklebust <trond.myklebust@fys.uio.no> Cc: Neil Brown <neilb@cse.unsw.edu.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 09:03:05 +00:00
latter = dir->i_size - page_offset(page);
if (latter >= PAGE_SIZE)
qty = PAGE_SIZE;
else
qty = latter;
qty /= sizeof(union afs_dir_block);
/* check them */
dbuf = page_address(page);
for (tmp = 0; tmp < qty; tmp++) {
if (dbuf->blocks[tmp].pagehdr.magic != AFS_DIR_MAGIC) {
printk("kAFS: %s(%lu): bad magic %d/%d is %04hx\n",
__func__, dir->i_ino, tmp, qty,
ntohs(dbuf->blocks[tmp].pagehdr.magic));
goto error;
}
}
SetPageChecked(page);
return true;
error:
SetPageError(page);
return false;
}
/*
* discard a page cached in the pagecache
*/
static inline void afs_dir_put_page(struct page *page)
{
kunmap(page);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(page);
}
/*
* get a page into the pagecache
*/
static struct page *afs_dir_get_page(struct inode *dir, unsigned long index,
struct key *key)
{
struct page *page;
_enter("{%lu},%lu", dir->i_ino, index);
page = read_cache_page(dir->i_mapping, index, afs_page_filler, key);
if (!IS_ERR(page)) {
kmap(page);
if (unlikely(!PageChecked(page))) {
if (PageError(page) || !afs_dir_check_page(dir, page))
goto fail;
}
}
return page;
fail:
afs_dir_put_page(page);
_leave(" = -EIO");
return ERR_PTR(-EIO);
}
/*
* open an AFS directory file
*/
static int afs_dir_open(struct inode *inode, struct file *file)
{
_enter("{%lu}", inode->i_ino);
BUILD_BUG_ON(sizeof(union afs_dir_block) != 2048);
BUILD_BUG_ON(sizeof(union afs_dirent) != 32);
if (test_bit(AFS_VNODE_DELETED, &AFS_FS_I(inode)->flags))
return -ENOENT;
return afs_open(inode, file);
}
/*
* deal with one block in an AFS directory
*/
static int afs_dir_iterate_block(struct dir_context *ctx,
union afs_dir_block *block,
unsigned blkoff)
{
union afs_dirent *dire;
unsigned offset, next, curr;
size_t nlen;
int tmp;
_enter("%u,%x,%p,,",(unsigned)ctx->pos,blkoff,block);
curr = (ctx->pos - blkoff) / sizeof(union afs_dirent);
/* walk through the block, an entry at a time */
for (offset = AFS_DIRENT_PER_BLOCK - block->pagehdr.nentries;
offset < AFS_DIRENT_PER_BLOCK;
offset = next
) {
next = offset + 1;
/* skip entries marked unused in the bitmap */
if (!(block->pagehdr.bitmap[offset / 8] &
(1 << (offset % 8)))) {
_debug("ENT[%zu.%u]: unused",
blkoff / sizeof(union afs_dir_block), offset);
if (offset >= curr)
ctx->pos = blkoff +
next * sizeof(union afs_dirent);
continue;
}
/* got a valid entry */
dire = &block->dirents[offset];
nlen = strnlen(dire->u.name,
sizeof(*block) -
offset * sizeof(union afs_dirent));
_debug("ENT[%zu.%u]: %s %zu \"%s\"",
blkoff / sizeof(union afs_dir_block), offset,
(offset < curr ? "skip" : "fill"),
nlen, dire->u.name);
/* work out where the next possible entry is */
for (tmp = nlen; tmp > 15; tmp -= sizeof(union afs_dirent)) {
if (next >= AFS_DIRENT_PER_BLOCK) {
_debug("ENT[%zu.%u]:"
" %u travelled beyond end dir block"
" (len %u/%zu)",
blkoff / sizeof(union afs_dir_block),
offset, next, tmp, nlen);
return -EIO;
}
if (!(block->pagehdr.bitmap[next / 8] &
(1 << (next % 8)))) {
_debug("ENT[%zu.%u]:"
" %u unmarked extension (len %u/%zu)",
blkoff / sizeof(union afs_dir_block),
offset, next, tmp, nlen);
return -EIO;
}
_debug("ENT[%zu.%u]: ext %u/%zu",
blkoff / sizeof(union afs_dir_block),
next, tmp, nlen);
next++;
}
/* skip if starts before the current position */
if (offset < curr)
continue;
/* found the next entry */
if (!dir_emit(ctx, dire->u.name, nlen,
ntohl(dire->u.vnode),
ctx->actor == afs_lookup_filldir ?
ntohl(dire->u.unique) : DT_UNKNOWN)) {
_leave(" = 0 [full]");
return 0;
}
ctx->pos = blkoff + next * sizeof(union afs_dirent);
}
_leave(" = 1 [more]");
return 1;
}
/*
* iterate through the data blob that lists the contents of an AFS directory
*/
static int afs_dir_iterate(struct inode *dir, struct dir_context *ctx,
struct key *key)
{
union afs_dir_block *dblock;
struct afs_dir_page *dbuf;
struct page *page;
unsigned blkoff, limit;
int ret;
_enter("{%lu},%u,,", dir->i_ino, (unsigned)ctx->pos);
if (test_bit(AFS_VNODE_DELETED, &AFS_FS_I(dir)->flags)) {
_leave(" = -ESTALE");
return -ESTALE;
}
/* round the file position up to the next entry boundary */
ctx->pos += sizeof(union afs_dirent) - 1;
ctx->pos &= ~(sizeof(union afs_dirent) - 1);
/* walk through the blocks in sequence */
ret = 0;
while (ctx->pos < dir->i_size) {
blkoff = ctx->pos & ~(sizeof(union afs_dir_block) - 1);
/* fetch the appropriate page from the directory */
page = afs_dir_get_page(dir, blkoff / PAGE_SIZE, key);
if (IS_ERR(page)) {
ret = PTR_ERR(page);
break;
}
limit = blkoff & ~(PAGE_SIZE - 1);
dbuf = page_address(page);
/* deal with the individual blocks stashed on this page */
do {
dblock = &dbuf->blocks[(blkoff % PAGE_SIZE) /
sizeof(union afs_dir_block)];
ret = afs_dir_iterate_block(ctx, dblock, blkoff);
if (ret != 1) {
afs_dir_put_page(page);
goto out;
}
blkoff += sizeof(union afs_dir_block);
} while (ctx->pos < dir->i_size && blkoff < limit);
afs_dir_put_page(page);
ret = 0;
}
out:
_leave(" = %d", ret);
return ret;
}
/*
* read an AFS directory
*/
static int afs_readdir(struct file *file, struct dir_context *ctx)
{
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
return afs_dir_iterate(file_inode(file),
ctx, file->private_data);
}
/*
* search the directory for a name
* - if afs_dir_iterate_block() spots this function, it'll pass the FID
* uniquifier through dtype
*/
static int afs_lookup_filldir(struct dir_context *ctx, const char *name,
int nlen, loff_t fpos, u64 ino, unsigned dtype)
{
struct afs_lookup_cookie *cookie =
container_of(ctx, struct afs_lookup_cookie, ctx);
_enter("{%s,%u},%s,%u,,%llu,%u",
cookie->name.name, cookie->name.len, name, nlen,
(unsigned long long) ino, dtype);
/* insanity checks first */
BUILD_BUG_ON(sizeof(union afs_dir_block) != 2048);
BUILD_BUG_ON(sizeof(union afs_dirent) != 32);
if (cookie->name.len != nlen ||
memcmp(cookie->name.name, name, nlen) != 0) {
_leave(" = 0 [no]");
return 0;
}
cookie->fid.vnode = ino;
cookie->fid.unique = dtype;
cookie->found = 1;
_leave(" = -1 [found]");
return -1;
}
/*
* do a lookup in a directory
* - just returns the FID the dentry name maps to if found
*/
static int afs_do_lookup(struct inode *dir, struct dentry *dentry,
struct afs_fid *fid, struct key *key)
{
struct afs_super_info *as = dir->i_sb->s_fs_info;
struct afs_lookup_cookie cookie = {
.ctx.actor = afs_lookup_filldir,
.name = dentry->d_name,
.fid.vid = as->volume->vid
};
int ret;
_enter("{%lu},%p{%pd},", dir->i_ino, dentry, dentry);
/* search the directory */
ret = afs_dir_iterate(dir, &cookie.ctx, key);
if (ret < 0) {
_leave(" = %d [iter]", ret);
return ret;
}
ret = -ENOENT;
if (!cookie.found) {
_leave(" = -ENOENT [not found]");
return -ENOENT;
}
*fid = cookie.fid;
_leave(" = 0 { vn=%u u=%u }", fid->vnode, fid->unique);
return 0;
}
/*
* Try to auto mount the mountpoint with pseudo directory, if the autocell
* operation is setted.
*/
static struct inode *afs_try_auto_mntpt(
int ret, struct dentry *dentry, struct inode *dir, struct key *key,
struct afs_fid *fid)
{
const char *devname = dentry->d_name.name;
struct afs_vnode *vnode = AFS_FS_I(dir);
struct inode *inode;
_enter("%d, %p{%pd}, {%x:%u}, %p",
ret, dentry, dentry, vnode->fid.vid, vnode->fid.vnode, key);
if (ret != -ENOENT ||
!test_bit(AFS_VNODE_AUTOCELL, &vnode->flags))
goto out;
inode = afs_iget_autocell(dir, devname, strlen(devname), key);
if (IS_ERR(inode)) {
ret = PTR_ERR(inode);
goto out;
}
*fid = AFS_FS_I(inode)->fid;
_leave("= %p", inode);
return inode;
out:
_leave("= %d", ret);
return ERR_PTR(ret);
}
/*
* look up an entry in a directory
*/
static struct dentry *afs_lookup(struct inode *dir, struct dentry *dentry,
unsigned int flags)
{
struct afs_vnode *vnode;
struct afs_fid fid;
struct inode *inode;
struct key *key;
int ret;
vnode = AFS_FS_I(dir);
_enter("{%x:%u},%p{%pd},",
vnode->fid.vid, vnode->fid.vnode, dentry, dentry);
ASSERTCMP(d_inode(dentry), ==, NULL);
if (dentry->d_name.len >= AFSNAMEMAX) {
_leave(" = -ENAMETOOLONG");
return ERR_PTR(-ENAMETOOLONG);
}
if (test_bit(AFS_VNODE_DELETED, &vnode->flags)) {
_leave(" = -ESTALE");
return ERR_PTR(-ESTALE);
}
key = afs_request_key(vnode->volume->cell);
if (IS_ERR(key)) {
_leave(" = %ld [key]", PTR_ERR(key));
return ERR_CAST(key);
}
ret = afs_validate(vnode, key);
if (ret < 0) {
key_put(key);
_leave(" = %d [val]", ret);
return ERR_PTR(ret);
}
ret = afs_do_lookup(dir, dentry, &fid, key);
if (ret < 0) {
inode = afs_try_auto_mntpt(ret, dentry, dir, key, &fid);
if (!IS_ERR(inode)) {
key_put(key);
goto success;
}
ret = PTR_ERR(inode);
key_put(key);
if (ret == -ENOENT) {
d_add(dentry, NULL);
_leave(" = NULL [negative]");
return NULL;
}
_leave(" = %d [do]", ret);
return ERR_PTR(ret);
}
dentry->d_fsdata = (void *)(unsigned long) vnode->status.data_version;
/* instantiate the dentry */
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
inode = afs_iget(dir->i_sb, key, &fid, NULL, NULL, NULL);
key_put(key);
if (IS_ERR(inode)) {
_leave(" = %ld", PTR_ERR(inode));
return ERR_CAST(inode);
}
success:
d_add(dentry, inode);
_leave(" = 0 { vn=%u u=%u } -> { ino=%lu v=%u }",
fid.vnode,
fid.unique,
d_inode(dentry)->i_ino,
d_inode(dentry)->i_generation);
return NULL;
}
/*
* check that a dentry lookup hit has found a valid entry
* - NOTE! the hit can be a negative hit too, so we can't assume we have an
* inode
*/
static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
{
struct afs_vnode *vnode, *dir;
struct afs_fid uninitialized_var(fid);
struct dentry *parent;
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
struct inode *inode;
struct key *key;
void *dir_version;
int ret;
if (flags & LOOKUP_RCU)
return -ECHILD;
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
if (d_really_is_positive(dentry)) {
vnode = AFS_FS_I(d_inode(dentry));
_enter("{v={%x:%u} n=%pd fl=%lx},",
vnode->fid.vid, vnode->fid.vnode, dentry,
vnode->flags);
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
} else {
_enter("{neg n=%pd}", dentry);
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
}
key = afs_request_key(AFS_FS_S(dentry->d_sb)->volume->cell);
if (IS_ERR(key))
key = NULL;
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
if (d_really_is_positive(dentry)) {
inode = d_inode(dentry);
if (inode) {
vnode = AFS_FS_I(inode);
afs_validate(vnode, key);
if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
goto out_bad;
}
}
/* lock down the parent dentry so we can peer at it */
parent = dget_parent(dentry);
dir = AFS_FS_I(d_inode(parent));
/* validate the parent directory */
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
afs_validate(dir, key);
if (test_bit(AFS_VNODE_DELETED, &dir->flags)) {
_debug("%pd: parent dir deleted", dentry);
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
goto out_bad_parent;
}
dir_version = (void *) (unsigned long) dir->status.data_version;
if (dentry->d_fsdata == dir_version)
goto out_valid; /* the dir contents are unchanged */
_debug("dir modified");
/* search the directory for this vnode */
ret = afs_do_lookup(&dir->vfs_inode, dentry, &fid, key);
switch (ret) {
case 0:
/* the filename maps to something */
if (d_really_is_negative(dentry))
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
goto out_bad_parent;
inode = d_inode(dentry);
if (is_bad_inode(inode)) {
printk("kAFS: afs_d_revalidate: %pd2 has bad inode\n",
dentry);
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
goto out_bad_parent;
}
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
vnode = AFS_FS_I(inode);
/* if the vnode ID has changed, then the dirent points to a
* different file */
if (fid.vnode != vnode->fid.vnode) {
_debug("%pd: dirent changed [%u != %u]",
dentry, fid.vnode,
vnode->fid.vnode);
goto not_found;
}
/* if the vnode ID uniqifier has changed, then the file has
* been deleted and replaced, and the original vnode ID has
* been reused */
if (fid.unique != vnode->fid.unique) {
_debug("%pd: file deleted (uq %u -> %u I:%u)",
dentry, fid.unique,
vnode->fid.unique,
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
vnode->vfs_inode.i_generation);
write_seqlock(&vnode->cb_lock);
set_bit(AFS_VNODE_DELETED, &vnode->flags);
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
write_sequnlock(&vnode->cb_lock);
goto not_found;
}
goto out_valid;
case -ENOENT:
/* the filename is unknown */
_debug("%pd: dirent not found", dentry);
if (d_really_is_positive(dentry))
goto not_found;
goto out_valid;
default:
_debug("failed to iterate dir %pd: %d",
parent, ret);
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
goto out_bad_parent;
}
out_valid:
dentry->d_fsdata = dir_version;
dput(parent);
key_put(key);
_leave(" = 1 [valid]");
return 1;
/* the dirent, if it exists, now points to a different vnode */
not_found:
spin_lock(&dentry->d_lock);
dentry->d_flags |= DCACHE_NFSFS_RENAMED;
spin_unlock(&dentry->d_lock);
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
out_bad_parent:
_debug("dropping dentry %pd2", dentry);
dput(parent);
afs: Overhaul the callback handling Overhaul the AFS callback handling by the following means: (1) Don't give up callback promises on vnodes that we are no longer using, rather let them just expire on the server or let the server break them. This is actually more efficient for the server as the callback lookup is expensive if there are lots of extant callbacks. (2) Only give up the callback promises we have from a server when the server record is destroyed. Then we can just give up *all* the callback promises on it in one go. (3) Servers can end up being shared between cells if cells are aliased, so don't add all the vnodes being backed by a particular server into a big FID-indexed tree on that server as there may be duplicates. Instead have each volume instance (~= superblock) register an interest in a server as it starts to make use of it and use this to allow the processor for callbacks from the server to find the superblock and thence the inode corresponding to the FID being broken by means of ilookup_nowait(). (4) Rather than iterating over the entire callback list when a mass-break comes in from the server, maintain a counter of mass-breaks in afs_server (cb_seq) and make afs_validate() check it against the copy in afs_vnode. It would be nice not to have to take a read_lock whilst doing this, but that's tricky without using RCU. (5) Save a ref on the fileserver we're using for a call in the afs_call struct so that we can access its cb_s_break during call decoding. (6) Write-lock around callback and status storage in a vnode and read-lock around getattr so that we don't see the status mid-update. This has the following consequences: (1) Data invalidation isn't seen until someone calls afs_validate() on a vnode. Unfortunately, we need to use a key to query the server, but getting one from a background thread is tricky without caching loads of keys all over the place. (2) Mass invalidation isn't seen until someone calls afs_validate(). (3) Callback breaking is going to hit the inode_hash_lock quite a bit. Could this be replaced with rcu_read_lock() since inodes are destroyed under RCU conditions. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:49 +00:00
out_bad:
key_put(key);
_leave(" = 0 [bad]");
return 0;
}
/*
* allow the VFS to enquire as to whether a dentry should be unhashed (mustn't
* sleep)
* - called from dput() when d_count is going to 0.
* - return 1 to request dentry be unhashed, 0 otherwise
*/
static int afs_d_delete(const struct dentry *dentry)
{
_enter("%pd", dentry);
if (dentry->d_flags & DCACHE_NFSFS_RENAMED)
goto zap;
if (d_really_is_positive(dentry) &&
(test_bit(AFS_VNODE_DELETED, &AFS_FS_I(d_inode(dentry))->flags) ||
test_bit(AFS_VNODE_PSEUDODIR, &AFS_FS_I(d_inode(dentry))->flags)))
goto zap;
_leave(" = 0 [keep]");
return 0;
zap:
_leave(" = 1 [zap]");
return 1;
}
/*
* handle dentry release
*/
static void afs_d_release(struct dentry *dentry)
{
_enter("%pd", dentry);
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
/*
* Create a new inode for create/mkdir/symlink
*/
static void afs_vnode_new_inode(struct afs_fs_cursor *fc,
struct dentry *new_dentry,
struct afs_fid *newfid,
struct afs_file_status *newstatus,
struct afs_callback *newcb)
{
struct inode *inode;
if (fc->ac.error < 0)
return;
inode = afs_iget(fc->vnode->vfs_inode.i_sb, fc->key,
newfid, newstatus, newcb, fc->cbi);
if (IS_ERR(inode)) {
/* ENOMEM or EINTR at a really inconvenient time - just abandon
* the new directory on the server.
*/
fc->ac.error = PTR_ERR(inode);
return;
}
d_instantiate(new_dentry, inode);
if (d_unhashed(new_dentry))
d_rehash(new_dentry);
}
/*
* create a directory on an AFS filesystem
*/
static int afs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
{
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
struct afs_file_status newstatus;
struct afs_fs_cursor fc;
struct afs_callback newcb;
struct afs_vnode *dvnode = AFS_FS_I(dir);
struct afs_fid newfid;
struct key *key;
int ret;
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
mode |= S_IFDIR;
_enter("{%x:%u},{%pd},%ho",
dvnode->fid.vid, dvnode->fid.vnode, dentry, mode);
key = afs_request_key(dvnode->volume->cell);
if (IS_ERR(key)) {
ret = PTR_ERR(key);
goto error;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ERESTARTSYS;
if (afs_begin_vnode_operation(&fc, dvnode, key)) {
while (afs_select_fileserver(&fc)) {
fc.cb_break = dvnode->cb_break + dvnode->cb_s_break;
afs_fs_create(&fc, dentry->d_name.name, mode,
&newfid, &newstatus, &newcb);
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
afs_check_for_remote_deletion(&fc, fc.vnode);
afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
afs_vnode_new_inode(&fc, dentry, &newfid, &newstatus, &newcb);
ret = afs_end_vnode_operation(&fc);
if (ret < 0)
goto error_key;
}
key_put(key);
_leave(" = 0");
return 0;
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
error_key:
key_put(key);
error:
d_drop(dentry);
_leave(" = %d", ret);
return ret;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
/*
* Remove a subdir from a directory.
*/
static void afs_dir_remove_subdir(struct dentry *dentry)
{
if (d_really_is_positive(dentry)) {
struct afs_vnode *vnode = AFS_FS_I(d_inode(dentry));
clear_nlink(&vnode->vfs_inode);
set_bit(AFS_VNODE_DELETED, &vnode->flags);
clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
}
}
/*
* remove a directory from an AFS filesystem
*/
static int afs_rmdir(struct inode *dir, struct dentry *dentry)
{
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
struct afs_fs_cursor fc;
struct afs_vnode *dvnode = AFS_FS_I(dir);
struct key *key;
int ret;
_enter("{%x:%u},{%pd}",
dvnode->fid.vid, dvnode->fid.vnode, dentry);
key = afs_request_key(dvnode->volume->cell);
if (IS_ERR(key)) {
ret = PTR_ERR(key);
goto error;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ERESTARTSYS;
if (afs_begin_vnode_operation(&fc, dvnode, key)) {
while (afs_select_fileserver(&fc)) {
fc.cb_break = dvnode->cb_break + dvnode->cb_s_break;
afs_fs_remove(&fc, dentry->d_name.name, true);
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
ret = afs_end_vnode_operation(&fc);
if (ret == 0)
afs_dir_remove_subdir(dentry);
}
key_put(key);
error:
return ret;
}
/*
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
* Remove a link to a file or symlink from a directory.
*
* If the file was not deleted due to excess hard links, the fileserver will
* break the callback promise on the file - if it had one - before it returns
* to us, and if it was deleted, it won't
*
* However, if we didn't have a callback promise outstanding, or it was
* outstanding on a different server, then it won't break it either...
*/
static int afs_dir_remove_link(struct dentry *dentry, struct key *key)
{
int ret = 0;
if (d_really_is_positive(dentry)) {
struct afs_vnode *vnode = AFS_FS_I(d_inode(dentry));
if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
kdebug("AFS_VNODE_DELETED");
clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
ret = afs_validate(vnode, key);
if (ret == -ESTALE)
ret = 0;
_debug("nlink %d [val %d]", vnode->vfs_inode.i_nlink, ret);
}
return ret;
}
/*
* Remove a file or symlink from an AFS filesystem.
*/
static int afs_unlink(struct inode *dir, struct dentry *dentry)
{
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
struct afs_fs_cursor fc;
struct afs_vnode *dvnode = AFS_FS_I(dir), *vnode;
struct key *key;
int ret;
_enter("{%x:%u},{%pd}",
dvnode->fid.vid, dvnode->fid.vnode, dentry);
if (dentry->d_name.len >= AFSNAMEMAX)
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
return -ENAMETOOLONG;
key = afs_request_key(dvnode->volume->cell);
if (IS_ERR(key)) {
ret = PTR_ERR(key);
goto error;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
/* Try to make sure we have a callback promise on the victim. */
if (d_really_is_positive(dentry)) {
vnode = AFS_FS_I(d_inode(dentry));
ret = afs_validate(vnode, key);
if (ret < 0)
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
goto error_key;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ERESTARTSYS;
if (afs_begin_vnode_operation(&fc, dvnode, key)) {
while (afs_select_fileserver(&fc)) {
fc.cb_break = dvnode->cb_break + dvnode->cb_s_break;
afs_fs_remove(&fc, dentry->d_name.name, false);
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
ret = afs_end_vnode_operation(&fc);
if (ret == 0)
ret = afs_dir_remove_link(dentry, key);
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
error_key:
key_put(key);
error:
_leave(" = %d", ret);
return ret;
}
/*
* create a regular file on an AFS filesystem
*/
static int afs_create(struct inode *dir, struct dentry *dentry, umode_t mode,
bool excl)
{
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
struct afs_fs_cursor fc;
struct afs_file_status newstatus;
struct afs_callback newcb;
struct afs_vnode *dvnode = dvnode = AFS_FS_I(dir);
struct afs_fid newfid;
struct key *key;
int ret;
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
mode |= S_IFREG;
_enter("{%x:%u},{%pd},%ho,",
dvnode->fid.vid, dvnode->fid.vnode, dentry, mode);
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ENAMETOOLONG;
if (dentry->d_name.len >= AFSNAMEMAX)
goto error;
key = afs_request_key(dvnode->volume->cell);
if (IS_ERR(key)) {
ret = PTR_ERR(key);
goto error;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ERESTARTSYS;
if (afs_begin_vnode_operation(&fc, dvnode, key)) {
while (afs_select_fileserver(&fc)) {
fc.cb_break = dvnode->cb_break + dvnode->cb_s_break;
afs_fs_create(&fc, dentry->d_name.name, mode,
&newfid, &newstatus, &newcb);
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
afs_check_for_remote_deletion(&fc, fc.vnode);
afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
afs_vnode_new_inode(&fc, dentry, &newfid, &newstatus, &newcb);
ret = afs_end_vnode_operation(&fc);
if (ret < 0)
goto error_key;
}
key_put(key);
_leave(" = 0");
return 0;
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
error_key:
key_put(key);
error:
d_drop(dentry);
_leave(" = %d", ret);
return ret;
}
/*
* create a hard link between files in an AFS filesystem
*/
static int afs_link(struct dentry *from, struct inode *dir,
struct dentry *dentry)
{
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
struct afs_fs_cursor fc;
struct afs_vnode *dvnode, *vnode;
struct key *key;
int ret;
vnode = AFS_FS_I(d_inode(from));
dvnode = AFS_FS_I(dir);
_enter("{%x:%u},{%x:%u},{%pd}",
vnode->fid.vid, vnode->fid.vnode,
dvnode->fid.vid, dvnode->fid.vnode,
dentry);
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ENAMETOOLONG;
if (dentry->d_name.len >= AFSNAMEMAX)
goto error;
key = afs_request_key(dvnode->volume->cell);
if (IS_ERR(key)) {
ret = PTR_ERR(key);
goto error;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ERESTARTSYS;
if (afs_begin_vnode_operation(&fc, dvnode, key)) {
if (mutex_lock_interruptible_nested(&vnode->io_lock, 1) < 0) {
afs_end_vnode_operation(&fc);
return -ERESTARTSYS;
}
while (afs_select_fileserver(&fc)) {
fc.cb_break = dvnode->cb_break + dvnode->cb_s_break;
fc.cb_break_2 = vnode->cb_break + vnode->cb_s_break;
afs_fs_link(&fc, vnode, dentry->d_name.name);
}
afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
afs_vnode_commit_status(&fc, vnode, fc.cb_break_2);
ihold(&vnode->vfs_inode);
d_instantiate(dentry, &vnode->vfs_inode);
mutex_unlock(&vnode->io_lock);
ret = afs_end_vnode_operation(&fc);
if (ret < 0)
goto error_key;
}
key_put(key);
_leave(" = 0");
return 0;
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
error_key:
key_put(key);
error:
d_drop(dentry);
_leave(" = %d", ret);
return ret;
}
/*
* create a symlink in an AFS filesystem
*/
static int afs_symlink(struct inode *dir, struct dentry *dentry,
const char *content)
{
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
struct afs_fs_cursor fc;
struct afs_file_status newstatus;
struct afs_vnode *dvnode = AFS_FS_I(dir);
struct afs_fid newfid;
struct key *key;
int ret;
_enter("{%x:%u},{%pd},%s",
dvnode->fid.vid, dvnode->fid.vnode, dentry,
content);
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ENAMETOOLONG;
if (dentry->d_name.len >= AFSNAMEMAX)
goto error;
ret = -EINVAL;
if (strlen(content) >= AFSPATHMAX)
goto error;
key = afs_request_key(dvnode->volume->cell);
if (IS_ERR(key)) {
ret = PTR_ERR(key);
goto error;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ERESTARTSYS;
if (afs_begin_vnode_operation(&fc, dvnode, key)) {
while (afs_select_fileserver(&fc)) {
fc.cb_break = dvnode->cb_break + dvnode->cb_s_break;
afs_fs_symlink(&fc, dentry->d_name.name, content,
&newfid, &newstatus);
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
afs_check_for_remote_deletion(&fc, fc.vnode);
afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
afs_vnode_new_inode(&fc, dentry, &newfid, &newstatus, NULL);
ret = afs_end_vnode_operation(&fc);
if (ret < 0)
goto error_key;
}
key_put(key);
_leave(" = 0");
return 0;
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
error_key:
key_put(key);
error:
d_drop(dentry);
_leave(" = %d", ret);
return ret;
}
/*
* rename a file in an AFS filesystem and/or move it between directories
*/
static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
struct inode *new_dir, struct dentry *new_dentry,
unsigned int flags)
{
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
struct afs_fs_cursor fc;
struct afs_vnode *orig_dvnode, *new_dvnode, *vnode;
struct key *key;
int ret;
if (flags)
return -EINVAL;
vnode = AFS_FS_I(d_inode(old_dentry));
orig_dvnode = AFS_FS_I(old_dir);
new_dvnode = AFS_FS_I(new_dir);
_enter("{%x:%u},{%x:%u},{%x:%u},{%pd}",
orig_dvnode->fid.vid, orig_dvnode->fid.vnode,
vnode->fid.vid, vnode->fid.vnode,
new_dvnode->fid.vid, new_dvnode->fid.vnode,
new_dentry);
key = afs_request_key(orig_dvnode->volume->cell);
if (IS_ERR(key)) {
ret = PTR_ERR(key);
goto error;
}
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
ret = -ERESTARTSYS;
if (afs_begin_vnode_operation(&fc, orig_dvnode, key)) {
if (orig_dvnode != new_dvnode) {
if (mutex_lock_interruptible_nested(&new_dvnode->io_lock, 1) < 0) {
afs_end_vnode_operation(&fc);
return -ERESTARTSYS;
}
}
while (afs_select_fileserver(&fc)) {
fc.cb_break = orig_dvnode->cb_break + orig_dvnode->cb_s_break;
fc.cb_break_2 = new_dvnode->cb_break + new_dvnode->cb_s_break;
afs_fs_rename(&fc, old_dentry->d_name.name,
new_dvnode, new_dentry->d_name.name);
}
afs_vnode_commit_status(&fc, orig_dvnode, fc.cb_break);
afs_vnode_commit_status(&fc, new_dvnode, fc.cb_break_2);
if (orig_dvnode != new_dvnode)
mutex_unlock(&new_dvnode->io_lock);
ret = afs_end_vnode_operation(&fc);
if (ret < 0)
goto error_key;
}
key_put(key);
_leave(" = 0");
return 0;
afs: Overhaul volume and server record caching and fileserver rotation The current code assumes that volumes and servers are per-cell and are never shared, but this is not enforced, and, indeed, public cells do exist that are aliases of each other. Further, an organisation can, say, set up a public cell and a private cell with overlapping, but not identical, sets of servers. The difference is purely in the database attached to the VL servers. The current code will malfunction if it sees a server in two cells as it assumes global address -> server record mappings and that each server is in just one cell. Further, each server may have multiple addresses - and may have addresses of different families (IPv4 and IPv6, say). To this end, the following structural changes are made: (1) Server record management is overhauled: (a) Server records are made independent of cell. The namespace keeps track of them, volume records have lists of them and each vnode has a server on which its callback interest currently resides. (b) The cell record no longer keeps a list of servers known to be in that cell. (c) The server records are now kept in a flat list because there's no single address to sort on. (d) Server records are now keyed by their UUID within the namespace. (e) The addresses for a server are obtained with the VL.GetAddrsU rather than with VL.GetEntryByName, using the server's UUID as a parameter. (f) Cached server records are garbage collected after a period of non-use and are counted out of existence before purging is allowed to complete. This protects the work functions against rmmod. (g) The servers list is now in /proc/fs/afs/servers. (2) Volume record management is overhauled: (a) An RCU-replaceable server list is introduced. This tracks both servers and their coresponding callback interests. (b) The superblock is now keyed on cell record and numeric volume ID. (c) The volume record is now tied to the superblock which mounts it, and is activated when mounted and deactivated when unmounted. This makes it easier to handle the cache cookie without causing a double-use in fscache. (d) The volume record is loaded from the VLDB using VL.GetEntryByNameU to get the server UUID list. (e) The volume name is updated if it is seen to have changed when the volume is updated (the update is keyed on the volume ID). (3) The vlocation record is got rid of and VLDB records are no longer cached. Sufficient information is stored in the volume record, though an update to a volume record is now no longer shared between related volumes (volumes come in bundles of three: R/W, R/O and backup). and the following procedural changes are made: (1) The fileserver cursor introduced previously is now fleshed out and used to iterate over fileservers and their addresses. (2) Volume status is checked during iteration, and the server list is replaced if a change is detected. (3) Server status is checked during iteration, and the address list is replaced if a change is detected. (4) The abort code is saved into the address list cursor and -ECONNABORTED returned in afs_make_call() if a remote abort happened rather than translating the abort into an error message. This allows actions to be taken depending on the abort code more easily. (a) If a VMOVED abort is seen then this is handled by rechecking the volume and restarting the iteration. (b) If a VBUSY, VRESTARTING or VSALVAGING abort is seen then this is handled by sleeping for a short period and retrying and/or trying other servers that might serve that volume. A message is also displayed once until the condition has cleared. (c) If a VOFFLINE abort is seen, then this is handled as VBUSY for the moment. (d) If a VNOVOL abort is seen, the volume is rechecked in the VLDB to see if it has been deleted; if not, the fileserver is probably indicating that the volume couldn't be attached and needs salvaging. (e) If statfs() sees one of these aborts, it does not sleep, but rather returns an error, so as not to block the umount program. (5) The fileserver iteration functions in vnode.c are now merged into their callers and more heavily macroised around the cursor. vnode.c is removed. (6) Operations on a particular vnode are serialised on that vnode because the server will lock that vnode whilst it operates on it, so a second op sent will just have to wait. (7) Fileservers are probed with FS.GetCapabilities before being used. This is where service upgrade will be done. (8) A callback interest on a fileserver is set up before an FS operation is performed and passed through to afs_make_call() so that it can be set on the vnode if the operation returns a callback. The callback interest is passed through to afs_iget() also so that it can be set there too. In general, record updating is done on an as-needed basis when we try to access servers, volumes or vnodes rather than offloading it to work items and special threads. Notes: (1) Pre AFS-3.4 servers are no longer supported, though this can be added back if necessary (AFS-3.4 was released in 1998). (2) VBUSY is retried forever for the moment at intervals of 1s. (3) /proc/fs/afs/<cell>/servers no longer exists. Signed-off-by: David Howells <dhowells@redhat.com>
2017-11-02 15:27:50 +00:00
error_key:
key_put(key);
error:
d_drop(new_dentry);
_leave(" = %d", ret);
return ret;
}