==> Regarding Re: [PATCH 1/3] autofs4 - expiring filesystem from under process; [EMAIL PROTECTED] adds:
raven> On Mon, 11 Apr 2005, Jeff Moyer wrote: >> ==> Regarding [PATCH 1/3] autofs4 - expiring filesystem from under >> process; [EMAIL PROTECTED] adds: >> >> Could also please explain how the following is handled: >> >> expire process runs and issues AUTOFS_EXPIRE_MULTI, which sets >> AUTOFS_INF_EXPIRING in flags. While the expire is in progress, another >> process access the directory in question, causing a call to >> try_to_fill_dentry. try_to_fill_dentry sees the AUTOFS_INF_EXPIRING >> flag is set, and so calls autofs4_wait with notify set to NFY_NONE. >> However, when we take the wq sem, we find that the expire has finished, >> and thus create a new wq entry. Because NFY_NONE is set, we don't tell >> the daemon. >> >> So how will this process ever get woken up? >> raven> I've thought about this for quite a while and I think all that's raven> needed is to recognise that we're about to expire a dentry that's raven> not mounted anymore. raven> Can you think of a case for which this patch fails? I don't see how the patch you provided addresses the race between an expire event and a lookup. The real issue here is that the checking of AUTOFS_INF_EXPIRING and the list operations associated therewith need to happen atomically. Until this happens, we will have races. The attached patch is more in line with how I think the problem should be fixed. I have not yet tested or even compiled this. Could you please look this over and comment? Do you have a reproducer test case for this, by the way? That would be vastly useful. Thanks, Jeff --- linux-2.6.9/fs/autofs4/waitq.c.orig 2005-04-20 14:34:40.746320400 -0400 +++ linux-2.6.9/fs/autofs4/waitq.c 2005-04-20 16:39:34.543090128 -0400 @@ -161,6 +161,7 @@ int autofs4_wait(struct autofs_sb_info * enum autofs_notify notify) { struct autofs_wait_queue *wq; + struct autofs_info *de_info = autofs4_dentry_ino(dentry); char *name; int len, status; @@ -183,6 +184,22 @@ int autofs4_wait(struct autofs_sb_info * return -EINTR; } + /* + * The following two notify values are special: + * + * The only case in which we don:t want notification is when we + * are trying to fill a dentry and there is an expiry in process. + * So, we check if we are expiring inside the wq_sem to avoid + * races. Notice that we moved the setting of AUTOFS_INF_EXPIRING + * to inside the lock, as well. + */ + if (notify & NFY_EXPIRE) + de_info->d_flags |= AUTOFS_INF_EXPIRING; + + if (notify & NFY_NONE) + if (!de_info->d_flags & AUTOFS_INF_EXPIRING) + return NFY_NONE; + for (wq = sbi->queues ; wq ; wq = wq->next) { if (wq->hash == dentry->d_name.hash && wq->len == len && @@ -208,6 +225,7 @@ int autofs4_wait(struct autofs_sb_info * wq->hash = dentry->d_name.hash; wq->name = name; wq->len = len; + wq->info = de_info; wq->status = -EINTR; /* Status return if interrupted */ atomic_set(&wq->wait_ctr, 2); up(&sbi->wq_sem); @@ -287,6 +305,8 @@ int autofs4_wait_release(struct autofs_s } *wql = wq->next; /* Unlink from chain */ + /* at this point, expiries have finished */ + wq->info->flags &= ~AUTOFS_INF_EXPIRING; up(&sbi->wq_sem); kfree(wq->name); wq->name = NULL; /* Do not wait on this queue */ --- linux-2.6.9/fs/autofs4/expire.c.orig 2005-04-20 15:11:41.940647536 -0400 +++ linux-2.6.9/fs/autofs4/expire.c 2005-04-20 16:00:13.627003928 -0400 @@ -343,13 +343,9 @@ int autofs4_expire_multi(struct super_bl return -EFAULT; if ((dentry = autofs4_expire(sb, mnt, sbi, do_now)) != NULL) { - struct autofs_info *de_info = autofs4_dentry_ino(dentry); - /* This is synchronous because it makes the daemon a little easier */ - de_info->flags |= AUTOFS_INF_EXPIRING; ret = autofs4_wait(sbi, dentry, NFY_EXPIRE); - de_info->flags &= ~AUTOFS_INF_EXPIRING; dput(dentry); } --- linux-2.6.9/fs/autofs4/autofs_i.h.orig 2005-04-20 15:36:49.956394296 -0400 +++ linux-2.6.9/fs/autofs4/autofs_i.h 2005-04-20 16:41:50.617403688 -0400 @@ -82,6 +82,7 @@ struct autofs_wait_queue { int hash; int len; char *name; + struct autofs_info *info; /* This is for status reporting upon return */ int status; atomic_t wait_ctr; - To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html