lkml.org 
[lkml]   [2018]   [Feb]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [BUG] lock_parent() breakage when used from shrink_dentry_list() (was Re: [PATCH v2 6/6] fs/dcache: Avoid remaining try_lock loop in shrink_dentry_list())
On Fri, Feb 23, 2018 at 01:35:52PM -0800, Linus Torvalds wrote:

> This is too subtle, and your fix to check d_lockref.count < 0 sounds
> wrong to me. If it's really gone, maybe it has been reused and the
> refcount is positive again, but it's something else than a dentry
> entirely?
>
> Hmm.
>
> No, you extended the rcu read section, so I guess your patch is fine.
> And lock_parent already has that pattern, soiit's not new.
>
> Ok, I agree, looks like lock_parent should just re-check that thing
> that it already checked earler, but that now might be true again
> because of we dropped d_lock.

IMO that's the right thing for backports; whether we keep it after
the getting rid of trylock loops is a different question. Note that
the only case where we do not have __dentry_kill() prevention
guaranteed by the caller (either by holding a reference, or by
holding onto ->i_lock all along) is in shrink_dentry_list().
And there we have more than enough of other subtle crap.

Moreover, there we have a good reason to treat "it had been moved"
as "kick it off the shrink list and free if it's already dead",
which might simplify the things. Below is a stab at that:

/*
* ONLY for shrink_dentry_list() - it returns false if it finds
* dentry grabbed, moved or killed, which is fine there but not
* anywhere else. OTOH, nobody else needs to deal with dentries
* getting killed under them.
*/
static bool shrink_lock_for_kill(struct dentry *dentry)
{
if (dentry->d_lockref.count)
return false;

inode = dentry->d_inode;
if (inode && unlikely(!spin_trylock(&inode->i_lock))) {
rcu_read_lock(); /* to protect inode */
spin_unlock(&dentry->d_lock);
spin_lock(&inode->i_lock);
spin_lock(&dentry->d_lock);
if (unlikely(dentry->d_lockref.count))
goto out;
/* changed inode means that somebody had grabbed it */
if (unlikely(inode != dentry->d_inode))
goto out;
rcu_read_unlock();
}

parent = dentry->d_parent;
if (IS_ROOT(dentry) || likely(spin_trylock(&parent->d_lock)))
return true;

rcu_read_lock(); /* to protect parent */
spin_unlock(&dentry->d_lock);
parent = READ_ONCE(dentry->d_parent);
spin_lock(&parent->d_lock);
if (unlikely(parent != dentry->d_parent)) {
spin_unlock(&parent->d_lock);
spin_lock(&dentry->d_lock);
goto out;
}
spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED);
if (likely(!dentry->d_lockref.count)) {
rcu_read_unlock();
return true;
}
spin_unlock(&parent->d_lock);
out:
spin_unlock(&inode->i_lock);
rcu_read_unlock();
return false;
}

static void shrink_dentry_list(struct list_head *list)
{
struct dentry *dentry, *parent;

while (!list_empty(list)) {
struct inode *inode;
dentry = list_entry(list->prev, struct dentry, d_lru);
spin_lock(&dentry->d_lock);
if (!shrink_lock_for_kill(dentry)) {
bool can_free = false;
d_shrink_del(dentry);
if (dentry->d_lockref.count < 0)
can_free = dentry->d_flags & DCACHE_MAY_FREE;
spin_unlock(&dentry->d_lock);
if (can_free)
dentry_free(dentry);
continue;
}
d_shrink_del(dentry);
parent = dentry->d_parent;
__dentry_kill(dentry);
if (dentry == parent)
continue;
dentry = parent;
....
same as now
....
}
}

\
 
 \ /
  Last update: 2018-02-24 01:25    [W:0.088 / U:0.588 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site