linux/fs/afs/callback.c

229 lines
5.9 KiB
C
Raw Normal View History

/*
* Copyright (c) 2002, 2007 Red Hat, Inc. All rights reserved.
*
* This software may be freely redistributed under the terms of the
* GNU General Public License.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*
* Authors: David Woodhouse <dwmw2@infradead.org>
* David Howells <dhowells@redhat.com>
*
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/circ_buf.h>
#include <linux/sched.h>
#include "internal.h"
afs: Fix mmap coherency vs 3rd-party changes Fix the coherency management of mmap'd data such that 3rd-party changes become visible as soon as possible after the callback notification is delivered by the fileserver. This is done by the following means: (1) When we break a callback on a vnode specified by the CB.CallBack call from the server, we queue a work item (vnode->cb_work) to go and clobber all the PTEs mapping to that inode. This causes the CPU to trip through the ->map_pages() and ->page_mkwrite() handlers if userspace attempts to access the page(s) again. (Ideally, this would be done in the service handler for CB.CallBack, but the server is waiting for our reply before considering, and we have a list of vnodes, all of which need breaking - and the process of getting the mmap_lock and stripping the PTEs on all CPUs could be quite slow.) (2) Call afs_validate() from the ->map_pages() handler to check to see if the file has changed and to get a new callback promise from the server. Also handle the fileserver telling us that it's dropping all callbacks, possibly after it's been restarted by sending us a CB.InitCallBackState* call by the following means: (3) Maintain a per-cell list of afs files that are currently mmap'd (cell->fs_open_mmaps). (4) Add a work item to each server that is invoked if there are any open mmaps when CB.InitCallBackState happens. This work item goes through the aforementioned list and invokes the vnode->cb_work work item for each one that is currently using this server. This causes the PTEs to be cleared, causing ->map_pages() or ->page_mkwrite() to be called again, thereby calling afs_validate() again. I've chosen to simply strip the PTEs at the point of notification reception rather than invalidate all the pages as well because (a) it's faster, (b) we may get a notification for other reasons than the data being altered (in which case we don't want to clobber the pagecache) and (c) we need to ask the server to find out - and I don't want to wait for the reply before holding up userspace. This was tested using the attached test program: #include <stdbool.h> #include <stdio.h> #include <stdlib.h> #include <unistd.h> #include <fcntl.h> #include <sys/mman.h> int main(int argc, char *argv[]) { size_t size = getpagesize(); unsigned char *p; bool mod = (argc == 3); int fd; if (argc != 2 && argc != 3) { fprintf(stderr, "Format: %s <file> [mod]\n", argv[0]); exit(2); } fd = open(argv[1], mod ? O_RDWR : O_RDONLY); if (fd < 0) { perror(argv[1]); exit(1); } p = mmap(NULL, size, mod ? PROT_READ|PROT_WRITE : PROT_READ, MAP_SHARED, fd, 0); if (p == MAP_FAILED) { perror("mmap"); exit(1); } for (;;) { if (mod) { p[0]++; msync(p, size, MS_ASYNC); fsync(fd); } printf("%02x", p[0]); fflush(stdout); sleep(1); } } It runs in two modes: in one mode, it mmaps a file, then sits in a loop reading the first byte, printing it and sleeping for a second; in the second mode it mmaps a file, then sits in a loop incrementing the first byte and flushing, then printing and sleeping. Two instances of this program can be run on different machines, one doing the reading and one doing the writing. The reader should see the changes made by the writer, but without this patch, they aren't because validity checking is being done lazily - only on entry to the filesystem. Testing the InitCallBackState change is more complicated. The server has to be taken offline, the saved callback state file removed and then the server restarted whilst the reading-mode program continues to run. The client machine then has to poke the server to trigger the InitCallBackState call. Signed-off-by: David Howells <dhowells@redhat.com> Tested-by: Markus Suvanto <markus.suvanto@gmail.com> cc: linux-afs@lists.infradead.org Link: https://lore.kernel.org/r/163111668833.283156.382633263709075739.stgit@warthog.procyon.org.uk/
2021-09-02 23:43:10 +08:00
/*
* Handle invalidation of an mmap'd file. We invalidate all the PTEs referring
* to the pages in this file's pagecache, forcing the kernel to go through
* ->fault() or ->page_mkwrite() - at which point we can handle invalidation
* more fully.
*/
void afs_invalidate_mmap_work(struct work_struct *work)
{
struct afs_vnode *vnode = container_of(work, struct afs_vnode, cb_work);
unmap_mapping_pages(vnode->vfs_inode.i_mapping, 0, 0, false);
}
void afs_server_init_callback_work(struct work_struct *work)
{
struct afs_server *server = container_of(work, struct afs_server, initcb_work);
struct afs_vnode *vnode;
struct afs_cell *cell = server->cell;
down_read(&cell->fs_open_mmaps_lock);
list_for_each_entry(vnode, &cell->fs_open_mmaps, cb_mmap_link) {
if (vnode->cb_server == server) {
clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
queue_work(system_unbound_wq, &vnode->cb_work);
}
}
up_read(&cell->fs_open_mmaps_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 23:27:49 +08:00
/*
* Allow the fileserver to request callback state (re-)initialisation.
* Unfortunately, UUIDs are not guaranteed 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 23:27:49 +08:00
*/
void afs_init_callback_state(struct afs_server *server)
{
rcu_read_lock();
do {
server->cb_s_break++;
afs: Try to avoid taking RCU read lock when checking vnode validity Try to avoid taking the RCU read lock when checking the validity of a vnode's callback state. The only thing it's needed for is to pin the parent volume's server list whilst we search it to find the record of the server we're currently using to see if it has been reinitialised (ie. it sent us a CB.InitCallBackState* RPC). Do this by the following means: (1) Keep an additional per-cell counter (fs_s_break) that's incremented each time any of the fileservers in the cell reinitialises. Since the new counter can be accessed without RCU from the vnode, we can check that first - and only if it differs, get the RCU read lock and check the volume's server list. (2) Replace afs_get_s_break_rcu() with afs_check_server_good() which now indicates whether the callback promise is still expected to be present on the server. This does the checks as described in (1). (3) Restructure afs_check_validity() to take account of the change in (2). We can also get rid of the valid variable and just use the need_clear variable with the addition of the afs_cb_break_no_promise reason. (4) afs_check_validity() probably shouldn't be altering vnode->cb_v_break and vnode->cb_s_break when it doesn't have cb_lock exclusively locked. Move the change to vnode->cb_v_break to __afs_break_callback(). Delegate the change to vnode->cb_s_break to afs_select_fileserver() and set vnode->cb_fs_s_break there also. (5) afs_validate() no longer needs to get the RCU read lock around its call to afs_check_validity() - and can skip the call entirely if we don't have a promise. Signed-off-by: David Howells <dhowells@redhat.com> Tested-by: Markus Suvanto <markus.suvanto@gmail.com> cc: linux-afs@lists.infradead.org Link: https://lore.kernel.org/r/163111669583.283156.1397603105683094563.stgit@warthog.procyon.org.uk/
2021-09-03 04:51:01 +08:00
atomic_inc(&server->cell->fs_s_break);
afs: Fix mmap coherency vs 3rd-party changes Fix the coherency management of mmap'd data such that 3rd-party changes become visible as soon as possible after the callback notification is delivered by the fileserver. This is done by the following means: (1) When we break a callback on a vnode specified by the CB.CallBack call from the server, we queue a work item (vnode->cb_work) to go and clobber all the PTEs mapping to that inode. This causes the CPU to trip through the ->map_pages() and ->page_mkwrite() handlers if userspace attempts to access the page(s) again. (Ideally, this would be done in the service handler for CB.CallBack, but the server is waiting for our reply before considering, and we have a list of vnodes, all of which need breaking - and the process of getting the mmap_lock and stripping the PTEs on all CPUs could be quite slow.) (2) Call afs_validate() from the ->map_pages() handler to check to see if the file has changed and to get a new callback promise from the server. Also handle the fileserver telling us that it's dropping all callbacks, possibly after it's been restarted by sending us a CB.InitCallBackState* call by the following means: (3) Maintain a per-cell list of afs files that are currently mmap'd (cell->fs_open_mmaps). (4) Add a work item to each server that is invoked if there are any open mmaps when CB.InitCallBackState happens. This work item goes through the aforementioned list and invokes the vnode->cb_work work item for each one that is currently using this server. This causes the PTEs to be cleared, causing ->map_pages() or ->page_mkwrite() to be called again, thereby calling afs_validate() again. I've chosen to simply strip the PTEs at the point of notification reception rather than invalidate all the pages as well because (a) it's faster, (b) we may get a notification for other reasons than the data being altered (in which case we don't want to clobber the pagecache) and (c) we need to ask the server to find out - and I don't want to wait for the reply before holding up userspace. This was tested using the attached test program: #include <stdbool.h> #include <stdio.h> #include <stdlib.h> #include <unistd.h> #include <fcntl.h> #include <sys/mman.h> int main(int argc, char *argv[]) { size_t size = getpagesize(); unsigned char *p; bool mod = (argc == 3); int fd; if (argc != 2 && argc != 3) { fprintf(stderr, "Format: %s <file> [mod]\n", argv[0]); exit(2); } fd = open(argv[1], mod ? O_RDWR : O_RDONLY); if (fd < 0) { perror(argv[1]); exit(1); } p = mmap(NULL, size, mod ? PROT_READ|PROT_WRITE : PROT_READ, MAP_SHARED, fd, 0); if (p == MAP_FAILED) { perror("mmap"); exit(1); } for (;;) { if (mod) { p[0]++; msync(p, size, MS_ASYNC); fsync(fd); } printf("%02x", p[0]); fflush(stdout); sleep(1); } } It runs in two modes: in one mode, it mmaps a file, then sits in a loop reading the first byte, printing it and sleeping for a second; in the second mode it mmaps a file, then sits in a loop incrementing the first byte and flushing, then printing and sleeping. Two instances of this program can be run on different machines, one doing the reading and one doing the writing. The reader should see the changes made by the writer, but without this patch, they aren't because validity checking is being done lazily - only on entry to the filesystem. Testing the InitCallBackState change is more complicated. The server has to be taken offline, the saved callback state file removed and then the server restarted whilst the reading-mode program continues to run. The client machine then has to poke the server to trigger the InitCallBackState call. Signed-off-by: David Howells <dhowells@redhat.com> Tested-by: Markus Suvanto <markus.suvanto@gmail.com> cc: linux-afs@lists.infradead.org Link: https://lore.kernel.org/r/163111668833.283156.382633263709075739.stgit@warthog.procyon.org.uk/
2021-09-02 23:43:10 +08:00
if (!list_empty(&server->cell->fs_open_mmaps))
queue_work(system_unbound_wq, &server->initcb_work);
} while ((server = rcu_dereference(server->uuid_next)));
rcu_read_unlock();
}
/*
* actually break a callback
*/
void __afs_break_callback(struct afs_vnode *vnode, enum afs_cb_break_reason reason)
{
_enter("");
clear_bit(AFS_VNODE_NEW_CONTENT, &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 23:27:49 +08:00
if (test_and_clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
vnode->cb_break++;
afs: Try to avoid taking RCU read lock when checking vnode validity Try to avoid taking the RCU read lock when checking the validity of a vnode's callback state. The only thing it's needed for is to pin the parent volume's server list whilst we search it to find the record of the server we're currently using to see if it has been reinitialised (ie. it sent us a CB.InitCallBackState* RPC). Do this by the following means: (1) Keep an additional per-cell counter (fs_s_break) that's incremented each time any of the fileservers in the cell reinitialises. Since the new counter can be accessed without RCU from the vnode, we can check that first - and only if it differs, get the RCU read lock and check the volume's server list. (2) Replace afs_get_s_break_rcu() with afs_check_server_good() which now indicates whether the callback promise is still expected to be present on the server. This does the checks as described in (1). (3) Restructure afs_check_validity() to take account of the change in (2). We can also get rid of the valid variable and just use the need_clear variable with the addition of the afs_cb_break_no_promise reason. (4) afs_check_validity() probably shouldn't be altering vnode->cb_v_break and vnode->cb_s_break when it doesn't have cb_lock exclusively locked. Move the change to vnode->cb_v_break to __afs_break_callback(). Delegate the change to vnode->cb_s_break to afs_select_fileserver() and set vnode->cb_fs_s_break there also. (5) afs_validate() no longer needs to get the RCU read lock around its call to afs_check_validity() - and can skip the call entirely if we don't have a promise. Signed-off-by: David Howells <dhowells@redhat.com> Tested-by: Markus Suvanto <markus.suvanto@gmail.com> cc: linux-afs@lists.infradead.org Link: https://lore.kernel.org/r/163111669583.283156.1397603105683094563.stgit@warthog.procyon.org.uk/
2021-09-03 04:51:01 +08:00
vnode->cb_v_break = vnode->volume->cb_v_break;
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 23:27:49 +08:00
afs_clear_permits(vnode);
if (vnode->lock_state == AFS_VNODE_LOCK_WAITING_FOR_CB)
afs_lock_may_be_available(vnode);
afs: Fix mmap coherency vs 3rd-party changes Fix the coherency management of mmap'd data such that 3rd-party changes become visible as soon as possible after the callback notification is delivered by the fileserver. This is done by the following means: (1) When we break a callback on a vnode specified by the CB.CallBack call from the server, we queue a work item (vnode->cb_work) to go and clobber all the PTEs mapping to that inode. This causes the CPU to trip through the ->map_pages() and ->page_mkwrite() handlers if userspace attempts to access the page(s) again. (Ideally, this would be done in the service handler for CB.CallBack, but the server is waiting for our reply before considering, and we have a list of vnodes, all of which need breaking - and the process of getting the mmap_lock and stripping the PTEs on all CPUs could be quite slow.) (2) Call afs_validate() from the ->map_pages() handler to check to see if the file has changed and to get a new callback promise from the server. Also handle the fileserver telling us that it's dropping all callbacks, possibly after it's been restarted by sending us a CB.InitCallBackState* call by the following means: (3) Maintain a per-cell list of afs files that are currently mmap'd (cell->fs_open_mmaps). (4) Add a work item to each server that is invoked if there are any open mmaps when CB.InitCallBackState happens. This work item goes through the aforementioned list and invokes the vnode->cb_work work item for each one that is currently using this server. This causes the PTEs to be cleared, causing ->map_pages() or ->page_mkwrite() to be called again, thereby calling afs_validate() again. I've chosen to simply strip the PTEs at the point of notification reception rather than invalidate all the pages as well because (a) it's faster, (b) we may get a notification for other reasons than the data being altered (in which case we don't want to clobber the pagecache) and (c) we need to ask the server to find out - and I don't want to wait for the reply before holding up userspace. This was tested using the attached test program: #include <stdbool.h> #include <stdio.h> #include <stdlib.h> #include <unistd.h> #include <fcntl.h> #include <sys/mman.h> int main(int argc, char *argv[]) { size_t size = getpagesize(); unsigned char *p; bool mod = (argc == 3); int fd; if (argc != 2 && argc != 3) { fprintf(stderr, "Format: %s <file> [mod]\n", argv[0]); exit(2); } fd = open(argv[1], mod ? O_RDWR : O_RDONLY); if (fd < 0) { perror(argv[1]); exit(1); } p = mmap(NULL, size, mod ? PROT_READ|PROT_WRITE : PROT_READ, MAP_SHARED, fd, 0); if (p == MAP_FAILED) { perror("mmap"); exit(1); } for (;;) { if (mod) { p[0]++; msync(p, size, MS_ASYNC); fsync(fd); } printf("%02x", p[0]); fflush(stdout); sleep(1); } } It runs in two modes: in one mode, it mmaps a file, then sits in a loop reading the first byte, printing it and sleeping for a second; in the second mode it mmaps a file, then sits in a loop incrementing the first byte and flushing, then printing and sleeping. Two instances of this program can be run on different machines, one doing the reading and one doing the writing. The reader should see the changes made by the writer, but without this patch, they aren't because validity checking is being done lazily - only on entry to the filesystem. Testing the InitCallBackState change is more complicated. The server has to be taken offline, the saved callback state file removed and then the server restarted whilst the reading-mode program continues to run. The client machine then has to poke the server to trigger the InitCallBackState call. Signed-off-by: David Howells <dhowells@redhat.com> Tested-by: Markus Suvanto <markus.suvanto@gmail.com> cc: linux-afs@lists.infradead.org Link: https://lore.kernel.org/r/163111668833.283156.382633263709075739.stgit@warthog.procyon.org.uk/
2021-09-02 23:43:10 +08:00
if (reason != afs_cb_break_for_deleted &&
vnode->status.type == AFS_FTYPE_FILE &&
atomic_read(&vnode->cb_nr_mmap))
queue_work(system_unbound_wq, &vnode->cb_work);
trace_afs_cb_break(&vnode->fid, vnode->cb_break, reason, true);
} else {
trace_afs_cb_break(&vnode->fid, vnode->cb_break, reason, false);
}
}
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 23:27:49 +08:00
void afs_break_callback(struct afs_vnode *vnode, enum afs_cb_break_reason reason)
{
write_seqlock(&vnode->cb_lock);
__afs_break_callback(vnode, reason);
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 23:27:49 +08:00
write_sequnlock(&vnode->cb_lock);
}
/*
* Look up a volume by volume ID under RCU conditions.
*/
static struct afs_volume *afs_lookup_volume_rcu(struct afs_cell *cell,
afs_volid_t vid)
{
struct afs_volume *volume = NULL;
struct rb_node *p;
int seq = 0;
do {
/* Unfortunately, rbtree walking doesn't give reliable results
* under just the RCU read lock, so we have to check for
* changes.
*/
read_seqbegin_or_lock(&cell->volume_lock, &seq);
p = rcu_dereference_raw(cell->volumes.rb_node);
while (p) {
volume = rb_entry(p, struct afs_volume, cell_node);
if (volume->vid < vid)
p = rcu_dereference_raw(p->rb_left);
else if (volume->vid > vid)
p = rcu_dereference_raw(p->rb_right);
else
break;
volume = NULL;
}
} while (need_seqretry(&cell->volume_lock, seq));
done_seqretry(&cell->volume_lock, seq);
return volume;
}
/*
* allow the fileserver to explicitly break one callback
* - happens when
* - the backing file is changed
* - a lock is released
*/
static void afs_break_one_callback(struct afs_volume *volume,
struct afs_fid *fid)
{
struct super_block *sb;
struct afs_vnode *vnode;
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 23:27:49 +08:00
struct inode *inode;
if (fid->vnode == 0 && fid->unique == 0) {
/* The callback break applies to an entire volume. */
write_lock(&volume->cb_v_break_lock);
volume->cb_v_break++;
trace_afs_cb_break(fid, volume->cb_v_break,
afs_cb_break_for_volume_callback, false);
write_unlock(&volume->cb_v_break_lock);
return;
}
/* See if we can find a matching inode - even an I_NEW inode needs to
* be marked as it can have its callback broken before we finish
* setting up the local inode.
*/
sb = rcu_dereference(volume->sb);
if (!sb)
return;
inode = find_inode_rcu(sb, fid->vnode, afs_ilookup5_test_by_fid, fid);
if (inode) {
vnode = AFS_FS_I(inode);
afs_break_callback(vnode, afs_cb_break_for_callback);
} else {
trace_afs_cb_miss(fid, afs_cb_break_for_callback);
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 23:27:49 +08:00
}
}
static void afs_break_some_callbacks(struct afs_server *server,
struct afs_callback_break *cbb,
size_t *_count)
{
struct afs_callback_break *residue = cbb;
struct afs_volume *volume;
afs_volid_t vid = cbb->fid.vid;
size_t i;
volume = afs_lookup_volume_rcu(server->cell, vid);
/* TODO: Find all matching volumes if we couldn't match the server and
* break them anyway.
*/
for (i = *_count; i > 0; cbb++, i--) {
if (cbb->fid.vid == vid) {
_debug("- Fid { vl=%08llx n=%llu u=%u }",
cbb->fid.vid,
cbb->fid.vnode,
cbb->fid.unique);
--*_count;
if (volume)
afs_break_one_callback(volume, &cbb->fid);
} else {
*residue++ = *cbb;
}
}
}
/*
* allow the fileserver to break callback promises
*/
void afs_break_callbacks(struct afs_server *server, size_t count,
struct afs_callback_break *callbacks)
{
_enter("%p,%zu,", server, count);
ASSERT(server != NULL);
rcu_read_lock();
while (count > 0)
afs_break_some_callbacks(server, callbacks, &count);
rcu_read_unlock();
return;
}