[PATCH] fix waitqueue leak in devfs_d_revalidate_wait

[PATCH] fix waitqueue leak in devfs_d_revalidate_wait

[Christoph Hellwig]

devfs_d_revalidate_wait adds to a waitqueue but never removes from it
again so we there's one entry full of reused stack space added on
each call (I wonder how this ever worked).

The function has a few more bugs (it effectivly does a sleep_on instead
of checking for the actual even and can't deal with negative dentries
at all), but I just had breakfast and don't want to poke into devfs
internals deeper - I still hope Adam's smalldevfs will get merged
anyway..


--- 1.74/fs/devfs/base.c	Mon Mar 17 01:33:08 2003
+++ edited/fs/devfs/base.c	Tue Mar 18 09:37:29 2003
@@ -2336,6 +2232,8 @@
     wait_queue_head_t wait_queue;
 };
 
+/* XXX: this doesn't handle the case where we got a negative dentry
+        but a devfs entry has been registered in the meanwhile */
 static int devfs_d_revalidate_wait (struct dentry *dentry, int flags)
 {
     struct inode *dir = dentry->d_parent->d_inode;
@@ -2380,6 +2278,7 @@
 	add_wait_queue (&lookup_info->wait_queue, &wait);
 	read_unlock (&parent->u.dir.lock);
 	schedule ();
+	remove_wait_queue (&lookup_info->wait_queue, &wait);
     }
     else read_unlock (&parent->u.dir.lock);
     return 1;

Re: [PATCH] fix waitqueue leak in devfs_d_revalidate_wait

[Andrew Morton]

Christoph Hellwig <hch@lst.de> wrote:
>
> I still hope Adam's smalldevfs will get merged anyway..

Well me too, but smalldevfs is a bit stalled at present.

It seems to work OK now, but there are a few small incompatibilities with the
current devfs.  People need to make adjustments to their initscripts, or to
untar a tarball-of-device-nodes into smalldevfs at boot time.

Like all other kernel developers I do not use devfs and am not really able to
evaluate how serious these problems will be for people.

Adam is disinclined to address these administrative incompatibilites in his
user-space tools.  smalldevfs is presently at risk of getting dropped out.

Re: [PATCH] fix waitqueue leak in devfs_d_revalidate_wait

[Helge Hafting]

Andrew Morton wrote:
[...]
> It seems to work OK now, but there are a few small incompatibilities with the
> current devfs.  People need to make adjustments to their initscripts, or to
> untar a tarball-of-device-nodes into smalldevfs at boot time.

I assume smalldevfs is what you get when you configure devfs in 2.5.64-mm8?
It works fine for me, on smp with debian testing. I changed nothing
at all, and noticed that devfsd didn't run.  But it don't seem necessary,
it seems to work even without those compatibility symlinks.

> Adam is disinclined to address these administrative incompatibilites in his
> user-space tools.  smalldevfs is presently at risk of getting dropped out.

Please continue, it works and I understand the code is an improvement.
I'll begin testing on my work pc too.

Helge Hafting