The patch titled sysctl: don't take the use count of multiple heads at a time. has been removed from the -mm tree. Its filename was sysctl-dont-take-the-use-count-of-multiple-heads-at-a-time.patch This patch was dropped because an updated version will be merged The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/ ------------------------------------------------------ Subject: sysctl: don't take the use count of multiple heads at a time. From: Eric Biederman <ebiederm@xxxxxxxxxxxxxxxxxx> The problem: There is a class of deadlocks that we currently have in the kernel that lockdep does not recognize. In particular with the network stack we can have: rtnl_lock(); use_table(); unregister_netdevice(); rtnl_lock(); unregister_sysctl_table(); .... wait_for_completion(); rtnl_lock(); .... unuse_table() rtnl_unlock(); complete(); Where we never make it to the lines labled .... My patch following patches treats the sysctl use count as a read/writer lock for purposes of lockdep. The code works but I'm not certain I have plugged into lockdep quite correctly. This patch: The current code works fine, and is actually not buggy but it does prevent enabling the use of lockdep to check refcounting vs lock holding ordering problems. So since we can ensure that we are only hold a single sysctl_head at a time. Allowing lockdep to complain. Signed-off-by: Eric Biederman <ebiederm@xxxxxxxxxxxxxxxxxx> Cc: Alexey Dobriyan <adobriyan@xxxxxxxxx> Cc: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- fs/proc/proc_sysctl.c | 24 ++++++++++-------------- 1 file changed, 10 insertions(+), 14 deletions(-) diff -puN fs/proc/proc_sysctl.c~sysctl-dont-take-the-use-count-of-multiple-heads-at-a-time fs/proc/proc_sysctl.c --- a/fs/proc/proc_sysctl.c~sysctl-dont-take-the-use-count-of-multiple-heads-at-a-time +++ a/fs/proc/proc_sysctl.c @@ -79,7 +79,6 @@ static struct dentry *proc_sys_lookup(st { struct ctl_table_header *head = grab_header(dir); struct ctl_table *table = PROC_I(dir)->sysctl_entry; - struct ctl_table_header *h = NULL; struct qstr *name = &dentry->d_name; struct ctl_table *p; struct inode *inode; @@ -97,10 +96,11 @@ static struct dentry *proc_sys_lookup(st p = find_in_table(table, name); if (!p) { - for (h = sysctl_head_next(NULL); h; h = sysctl_head_next(h)) { - if (h->attached_to != table) + sysctl_head_finish(head); + for (head = sysctl_head_next(NULL); head; head = sysctl_head_next(head)) { + if (head->attached_to != table) continue; - p = find_in_table(h->attached_by, name); + p = find_in_table(head->attached_by, name); if (p) break; } @@ -110,9 +110,7 @@ static struct dentry *proc_sys_lookup(st goto out; err = ERR_PTR(-ENOMEM); - inode = proc_sys_make_inode(dir->i_sb, h ? h : head, p); - if (h) - sysctl_head_finish(h); + inode = proc_sys_make_inode(dir->i_sb, head, p); if (!inode) goto out; @@ -243,7 +241,6 @@ static int proc_sys_readdir(struct file struct inode *inode = dentry->d_inode; struct ctl_table_header *head = grab_header(inode); struct ctl_table *table = PROC_I(inode)->sysctl_entry; - struct ctl_table_header *h = NULL; unsigned long pos; int ret = -EINVAL; @@ -277,14 +274,13 @@ static int proc_sys_readdir(struct file if (ret) goto out; - for (h = sysctl_head_next(NULL); h; h = sysctl_head_next(h)) { - if (h->attached_to != table) + sysctl_head_finish(head); + for (head = sysctl_head_next(NULL); head; head = sysctl_head_next(head)) { + if (head->attached_to != table) continue; - ret = scan(h, h->attached_by, &pos, filp, dirent, filldir); - if (ret) { - sysctl_head_finish(h); + ret = scan(head, head->attached_by, &pos, filp, dirent, filldir); + if (ret) break; - } } ret = 1; out: _ Patches currently in -mm which might be from ebiederm@xxxxxxxxxxxxxxxxxx are sysctl-dont-take-the-use-count-of-multiple-heads-at-a-time.patch sysctl-lockdep-support-for-sysctl-reference-counting.patch sysctl-lockdep-support-for-sysctl-reference-counting-fix.patch -- To unsubscribe from this list: send the line "unsubscribe mm-commits" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html