Skip to contentSkip to conversation

afs, rxrpc: Clean up refcounting on afs_cell and afs_server records

GIT PULL David Howells 2 participants 3 messages Mar 3, 2025 – Mar 6, 2025

Subsystems (4)
lore
Share & tools
lore

David Howells

author maintainer
#0 lore raw

Hi Christian,

Could you pull this into the VFS tree onto a stable branch, replacing the earlier pull? The patches were previously posted here as part of a longer series:

https://lore.kernel.org/r/[email protected]/

This fixes an occasional hang that's only really encountered when rmmod'ing the kafs module, one of the reasons why I'm proposing it for the next merge window rather than immediate upstreaming. The changes include:

(1) Remove the "-o autocell" mount option. This is obsolete with the dynamic root and removing it makes the next patch slightly easier.

(2) Change how the dynamic root mount is constructed. Currently, the root directory is (de)populated when it is (un)mounted if there are cells already configured and, further, pairs of automount points have to be created/removed each time a cell is added/deleted.

This is changed so that readdir on the root dir lists all the known cell automount pairs plus the @cell symlinks and the inodes and dentries are constructed by lookup on demand. This simplifies the cell management code.

(3) A few improvements to the afs_volume tracepoint.

(4) A few improvements to the afs_server tracepoint.

(5) Pass trace info into the afs_lookup_cell() function to allow the trace log to indicate the purpose of the lookup.

(6) Remove the 'net' parameter from afs_unuse_cell() as it's superfluous.

(7) In rxrpc, allow a kernel app (such as kafs) to store a word of information on rxrpc_peer records.

(8) Use the information stored on the rxrpc_peer record to point to the afs_server record. This allows the server address lookup to be done away with.

(9) Simplify the afs_server ref/activity accounting to make each one self-contained and not garbage collected from the cell management work item.

(10) Simplify the afs_cell ref/activity accounting to make each one of these also self-contained and not driven by a central management work item.

The current code was intended to make it such that a single timer for the namespace and one work item per cell could do all the work required to maintain these records. This, however, made for some sequencing problems when cleaning up these records. Further, the attempt to pass refs along with timers and work items made getting it right rather tricky when the timer or work item already had a ref attached and now a ref had to be got rid of.

Thanks,
David

Changes
=======
ver #3)
- Fix the fix for an error check of the form "unsigned value < 0".

ver #2)
- Fix an error check of the form "unsigned value < 0".

---
The following changes since commit 1e15510b71c99c6e49134d756df91069f7d18141:

Merge tag 'net-6.14-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net (2025-02-27 09:32:42 -0800)

are available in the Git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git tags/afs-next-20250303

for you to fetch changes up to 73f77882c18d849cc8a26e5f06af5e6116169aba:

afs: Simplify cell record handling (2025-03-03 09:44:36 +0000)

----------------------------------------------------------------
afs: Fix ref leak in rmmod

----------------------------------------------------------------
David Howells (10):
afs: Remove the "autocell" mount option
afs: Change dynroot to create contents on demand
afs: Improve afs_volume tracing to display a debug ID
afs: Improve server refcount/active count tracing
afs: Make afs_lookup_cell() take a trace note
afs: Drop the net parameter from afs_unuse_cell()
rxrpc: Allow the app to store private data on peer structs
afs: Use the per-peer app data provided by rxrpc
afs: Fix afs_server ref accounting
afs: Simplify cell record handling

 fs/afs/addr_list.c         |  50 ++++
 fs/afs/cell.c              | 437 ++++++++++++++------------------
 fs/afs/cmservice.c         |  82 +------
 fs/afs/dir.c               |   5 +-
 fs/afs/dynroot.c           | 484 +++++++++++++++---------------------
 fs/afs/fs_probe.c          |  32 ++-
 fs/afs/fsclient.c          |   4 +-
 fs/afs/internal.h          |  98 ++++----
 fs/afs/main.c              |  16 +-
 fs/afs/mntpt.c             |   5 +-
 fs/afs/proc.c              |  15 +-
 fs/afs/rxrpc.c             |   8 +-
 fs/afs/server.c            | 601 +++++++++++++++++++--------------------------
 fs/afs/server_list.c       |   6 +-
 fs/afs/super.c             |  25 +-
 fs/afs/vl_alias.c          |   7 +-
 fs/afs/vl_rotate.c         |   2 +-
 fs/afs/volume.c            |  15 +-
 include/net/af_rxrpc.h     |   2 +
 include/trace/events/afs.h |  83 ++++---
 net/rxrpc/ar-internal.h    |   1 +
 net/rxrpc/peer_object.c    |  30 ++-
 22 files changed, 904 insertions(+), 1104 deletions(-)
#1 lore raw
Quoted 11 lines · David Howells

Could you pull this into the VFS tree onto a stable branch, replacing the earlier pull? The patches were previously posted here as part of a longer series:

https://lore.kernel.org/r/[email protected]/

This fixes an occasional hang that's only really encountered when rmmod'ing the kafs module, one of the reasons why I'm proposing it for the next merge window rather than immediate upstreaming. The changes include:

[...]

Pulled into the vfs-6.15.shared.afs branch of the vfs/vfs.git tree. Patches in the vfs-6.15.shared.afs branch should appear in linux-next soon.

Please report any outstanding bugs that were missed during review in a new review to the original patch series or pull request allowing us to drop it.

It's encouraged to provide Acked-bys and Reviewed-bys even though the patch has now been applied. If possible patch trailers will be updated.

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs.git
branch: vfs-6.15.shared.afs

https://git.kernel.org/vfs/vfs/c/acf689e88306

author maintainer
#2 lore raw
In reply to David Howells

Al spotted another bug (fix below). Subject to his okaying of the patch, I'll fold it down and ask for a repull.

David
---
afs: Fix afs_atcell_get_link() to handle RCU pathwalk

The ->get_link() method may be entered under RCU pathwalk conditions (in which case, the dentry pointer is NULL). This is not taken account of by afs_atcell_get_link() and lockdep will complain when it tries to lock an rwsem.

Fix this by marking net->ws_cell as __rcu and using RCU access macros on it and by making afs_atcell_get_link() just return a pointer to the name in RCU pathwalk without taking net->cells_lock or a ref on the cell as RCU will protect the name storage (the cell is already freed via call_rcu()).

---
 fs/afs/cell.c     |   11 ++++++-----
 fs/afs/dynroot.c  |   21 +++++++++++++++++----
 fs/afs/internal.h |    2 +-
 fs/afs/proc.c     |    2 +-
 4 files changed, 25 insertions(+), 11 deletions(-)
fs/afs/cell.c +6 -5
index 4ca713d3862c..0168bbf53fe0 100644
@@ -57,7 +57,8 @@ static struct afs_cell *afs_find_cell_locked(struct afs_net *net,
 		return ERR_PTR(-ENAMETOOLONG);
 
 	if (!name) {
-		cell = net->ws_cell;
+		cell = rcu_dereference_protected(net->ws_cell,
+						 lockdep_is_held(&net->cells_lock));
 		if (!cell)
 			return ERR_PTR(-EDESTADDRREQ);
 		goto found;
@@ -394,8 +395,8 @@ int afs_cell_init(struct afs_net *net, const char *rootcell)
 
 	/* install the new cell */
 	down_write(&net->cells_lock);
-	old_root = net->ws_cell;
-	net->ws_cell = new_root;
+	old_root = rcu_replace_pointer(net->ws_cell, new_root,
+				       lockdep_is_held(&net->cells_lock));
 	up_write(&net->cells_lock);
 
 	afs_unuse_cell(old_root, afs_cell_trace_unuse_ws);
@@ -869,8 +870,8 @@ void afs_cell_purge(struct afs_net *net)
 	_enter("");
 
 	down_write(&net->cells_lock);
-	ws = net->ws_cell;
-	net->ws_cell = NULL;
+	ws = rcu_replace_pointer(net->ws_cell, NULL,
+				 lockdep_is_held(&net->cells_lock));
 	up_write(&net->cells_lock);
 	afs_unuse_cell(ws, afs_cell_trace_unuse_ws);
 
fs/afs/dynroot.c +17 -4
index 0b66865e3535..9732a1e17db3 100644
@@ -210,12 +210,23 @@ static const char *afs_atcell_get_link(struct dentry *dentry, struct inode *inod
 	const char *name;
 	bool dotted = vnode->fid.vnode == 3;
 
-	if (!net->ws_cell)
+	if (!dentry) {
+		/* We're in RCU-pathwalk. */
+		cell = rcu_dereference(net->ws_cell);
+		if (dotted)
+			name = cell->name - 1;
+		else
+			name = cell->name;
+		/* Shouldn't need to set a delayed call. */
+		return name;
+	}
+
+	if (!rcu_access_pointer(net->ws_cell))
 		return ERR_PTR(-ENOENT);
 
 	down_read(&net->cells_lock);
 
-	cell = net->ws_cell;
+	cell = rcu_dereference_protected(net->ws_cell, lockdep_is_held(&net->cells_lock));
 	if (dotted)
 		name = cell->name - 1;
 	else
@@ -324,12 +335,14 @@ static int afs_dynroot_readdir(struct file *file, struct dir_context *ctx)
 		return 0;
 
 	if (ctx->pos == 2) {
-		if (net->ws_cell && !dir_emit(ctx, "@cell", 5, 2, DT_LNK))
+		if (rcu_access_pointer(net->ws_cell) &&
+		    !dir_emit(ctx, "@cell", 5, 2, DT_LNK))
 			return 0;
 		ctx->pos = 3;
 	}
 	if (ctx->pos == 3) {
-		if (net->ws_cell && !dir_emit(ctx, ".@cell", 6, 3, DT_LNK))
+		if (rcu_access_pointer(net->ws_cell) &&
+		    !dir_emit(ctx, ".@cell", 6, 3, DT_LNK))
 			return 0;
 		ctx->pos = 4;
 	}
fs/afs/internal.h +1 -1
index addce2f03562..440b0e731093 100644
@@ -288,7 +288,7 @@ struct afs_net {
 	/* Cell database */
 	struct rb_root		cells;
 	struct idr		cells_dyn_ino;	/* cell->dynroot_ino mapping */
-	struct afs_cell		*ws_cell;
+	struct afs_cell __rcu	*ws_cell;
 	atomic_t		cells_outstanding;
 	struct rw_semaphore	cells_lock;
 	struct mutex		cells_alias_lock;
fs/afs/proc.c +1 -1
index 0af94c846504..9d4df9e4562b 100644
@@ -207,7 +207,7 @@ static int afs_proc_rootcell_show(struct seq_file *m, void *v)
 
 	net = afs_seq2net_single(m);
 	down_read(&net->cells_lock);
-	cell = net->ws_cell;
+	cell = rcu_dereference_protected(net->ws_cell, lockdep_is_held(&net->cells_lock));
 	if (cell)
 		seq_printf(m, "%s\n", cell->name);
 	up_read(&net->cells_lock);