On Sun, Jan 20, 2008 at 02:30:27AM -0800, David Miller ([EMAIL PROTECTED])
wrote:
> From: Andrew Morton <[EMAIL PROTECTED]>
> Date: Sat, 19 Jan 2008 16:58:02 -0800
>
> > ouch.
>
> Yep, several people are hitting this it seems.
>
> If Pavel doesn't provide a fix or direction soon I'll just revert.
It looks like patch is still valid.
Here is a problem description as I undestood.
When new device (let's talk about ethernet, since that is what I tested)
is being turned on, it gets neigh_parms entry allocated for it via
inetdev_init(), which is called for NETDEV_REGISTER inetdev event.
This entry is stored in arp_tbl table and is in_dev->arp_parms.
When later new arp entry is created, device is provided into
arp_constructor(), which clones (increase reference counter) device's
in_dev->arp_parms and puts it into provided neighbour entry.
When later we remove device, its in_dev->arp_parms's reference counter
is high enough (it is equal to number of arp entries found on given
device plu one), so neigh_parms_destroy() is not called. Later all
neighbour entries are flushed by garbage collector and reference counter
for that parm hits zero and device can be removed.
I will think about how to fix the problem nicely or if this patch still
can be simplified/dropped, but so far it looks valid. Maybe this
analysis will help someone to fix problem first.
Here is debug dmesg:
[ 21.835595] inetdev_init: allocating parms.
[ 21.839829] neigh_parms_alloc: parms: ffff81003d8e8df0, dev: eth0, refcnt:
1, dev_refcnt: 2.
...
[ 30.251576] r8169: eth0: link up
[ 31.067079] NET: Registered protocol family 10
[ 31.072055] neigh_parms_alloc: parms: ffff81003efc72a8, dev: lo, refcnt: 1,
dev_refcnt: 9.
[ 31.080891] neigh_alloc: parms: ffffffff8812afe8, dev: <NULL>, refcnt: 2.
[ 31.087816] neigh_parms_alloc: parms: ffff81003efc7210, dev: eth0, refcnt:
1, dev_refcnt: 9.
[ 31.097335] neigh_alloc: parms: ffffffff804deb88, dev: <NULL>, refcnt: 2.
[ 31.104172] arp_constructor: parms: ffff81003f8c3be8, dev: lo, refcnt: 2.
[ 31.500348] neigh_alloc: parms: ffffffff8812afe8, dev: <NULL>, refcnt: 2.
[ 32.499628] neigh_alloc: parms: ffffffff8812afe8, dev: <NULL>, refcnt: 2.
[ 102.827796] neigh_destroy: parms: ffff81003efc7210, dev: eth0, refcnt: 3,
dev_refcnt: 13.
[ 106.828843] neigh_destroy: parms: ffff81003f8c3be8, dev: lo, refcnt: 2,
dev_refcnt: 78.
[ 109.810987] neigh_alloc: parms: ffffffff804deb88, dev: <NULL>, refcnt: 2.
First arp entry for eth0 device, bump the counter:
[ 109.817827] arp_constructor: parms: ffff81003d8e8df0, dev: eth0, refcnt: 2.
[ 109.831811] neigh_alloc: parms: ffffffff804deb88, dev: <NULL>, refcnt: 2.
[ 109.838661] arp_constructor: parms: ffff81003f8c3be8, dev: lo, refcnt: 2.
[ 110.837894] neigh_destroy: parms: ffff81003efc7210, dev: eth0, refcnt: 2,
dev_refcnt: 15.
Can not release that neigh parm:
[ 113.638228] neigh_parms_release: parms: ffff81003d8e8df0, dev: eth0, refcnt:
2, dev_refcnt: 5.
Can release some other (for ipv6):
[ 113.649380] neigh_parms_release: parms: ffff81003efc7210, dev: eth0, refcnt:
1, dev_refcnt: 5.
[ 113.671806] neigh_parms_destroy: parms: ffff81003efc7210, dev: eth0,
dev_refcnt: 3.
[ 123.916250] unregister_netdevice: waiting for eth0 to become free. Usage
count = 1
GC hits us:
[ 124.839572] neigh_destroy: parms: ffff81003d8e8df0, dev: eth0, refcnt: 1,
dev_refcnt: 11.
[ 124.847813] neigh_parms_destroy: parms: ffff81003d8e8df0, dev: eth0,
dev_refcnt: 1.
[ 124.952026] ACPI: PCI interrupt for device 0000:02:0d.0 disabled
--
Evgeniy Polyakov
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
