From: David Howells <dhowells@redhat.com>
To: Marc Dionne <marc.dionne@auristor.com>
Cc: David Howells <dhowells@redhat.com>,
Christian Brauner <christian@brauner.io>,
linux-afs@lists.infradead.org, linux-fsdevel@vger.kernel.org,
linux-kernel@vger.kernel.org,
Alexander Viro <viro@zeniv.linux.org.uk>
Subject: [PATCH v4 01/11] afs: Fix afs_atcell_get_link() to handle RCU pathwalk
Date: Mon, 10 Mar 2025 09:41:54 +0000 [thread overview]
Message-ID: <20250310094206.801057-2-dhowells@redhat.com> (raw)
In-Reply-To: <20250310094206.801057-1-dhowells@redhat.com>
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()).
Fixes: 30bca65bbbae ("afs: Make /afs/@cell and /afs/.@cell symlinks")
Reported-by: Alexander Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org
cc: linux-fsdevel@vger.kernel.org
---
fs/afs/cell.c | 11 ++++++-----
fs/afs/dynroot.c | 15 +++++++++++++--
fs/afs/internal.h | 2 +-
fs/afs/proc.c | 4 ++--
4 files changed, 22 insertions(+), 10 deletions(-)
diff --git a/fs/afs/cell.c b/fs/afs/cell.c
index cee42646736c..96a6781f3653 100644
--- a/fs/afs/cell.c
+++ b/fs/afs/cell.c
@@ -64,7 +64,8 @@ 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;
@@ -388,8 +389,8 @@ int afs_cell_init(struct afs_net *net, const char *rootcell)
/* install the new cell */
down_write(&net->cells_lock);
afs_see_cell(new_root, afs_cell_trace_see_ws);
- 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(net, old_root, afs_cell_trace_unuse_ws);
@@ -945,8 +946,8 @@ 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(net, ws, afs_cell_trace_unuse_ws);
diff --git a/fs/afs/dynroot.c b/fs/afs/dynroot.c
index d8bf52f77d93..008698d706ca 100644
--- a/fs/afs/dynroot.c
+++ b/fs/afs/dynroot.c
@@ -314,12 +314,23 @@ 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
diff --git a/fs/afs/internal.h b/fs/afs/internal.h
index 90f407774a9a..df30bd62da79 100644
--- a/fs/afs/internal.h
+++ b/fs/afs/internal.h
@@ -287,7 +287,7 @@ struct afs_net {
/* Cell database */
struct rb_root cells;
- struct afs_cell *ws_cell;
+ struct afs_cell __rcu *ws_cell;
struct work_struct cells_manager;
struct timer_list cells_timer;
atomic_t cells_outstanding;
diff --git a/fs/afs/proc.c b/fs/afs/proc.c
index e7614f4f30c2..12c88d8be3fe 100644
--- a/fs/afs/proc.c
+++ b/fs/afs/proc.c
@@ -206,7 +206,7 @@ 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);
@@ -242,7 +242,7 @@ static int afs_proc_rootcell_write(struct file *file, char *buf, size_t size)
ret = -EEXIST;
inode_lock(file_inode(file));
- if (!net->ws_cell)
+ if (!rcu_access_pointer(net->ws_cell))
ret = afs_cell_init(net, buf);
else
printk("busy\n");
next prev parent reply other threads:[~2025-03-10 9:42 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-03-10 9:41 [PATCH v4 00/11] afs, rxrpc: Clean up refcounting on afs_cell and afs_server records David Howells
2025-03-10 9:41 ` David Howells [this message]
2025-03-10 9:41 ` [PATCH v4 02/11] afs: Remove the "autocell" mount option David Howells
2025-03-10 9:41 ` [PATCH v4 03/11] afs: Change dynroot to create contents on demand David Howells
2025-03-10 9:41 ` [PATCH v4 04/11] afs: Improve afs_volume tracing to display a debug ID David Howells
2025-03-10 9:41 ` [PATCH v4 05/11] afs: Improve server refcount/active count tracing David Howells
2025-03-10 9:41 ` [PATCH v4 06/11] afs: Make afs_lookup_cell() take a trace note David Howells
2025-03-10 9:42 ` [PATCH v4 07/11] afs: Drop the net parameter from afs_unuse_cell() David Howells
2025-03-10 9:42 ` [PATCH v4 08/11] rxrpc: Allow the app to store private data on peer structs David Howells
2025-03-10 9:42 ` [PATCH v4 09/11] afs: Use the per-peer app data provided by rxrpc David Howells
2025-03-10 9:42 ` [PATCH v4 10/11] afs: Fix afs_server ref accounting David Howells
2025-03-10 9:42 ` [PATCH v4 11/11] afs: Simplify cell record handling David Howells
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20250310094206.801057-2-dhowells@redhat.com \
--to=dhowells@redhat.com \
--cc=christian@brauner.io \
--cc=linux-afs@lists.infradead.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=marc.dionne@auristor.com \
--cc=viro@zeniv.linux.org.uk \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).