linux-stable.git/net/core/netpoll.c, branch linux-3.6.y

netpoll: revert 6bdb7fe3104 and fix be_poll() instead

2012-08-29T19:03:23+00:00

Against -net.

In the patch "netpoll: re-enable irq in poll_napi()", I tried to
fix the following warning:

[100718.051041] ------------[ cut here ]------------
[100718.051048] WARNING: at kernel/softirq.c:159 local_bh_enable_ip+0x7d/0xb0()
(Not tainted)
[100718.051049] Hardware name: ProLiant BL460c G7
...
[100718.051068] Call Trace:
[100718.051073]  [] ? warn_slowpath_common+0x87/0xc0
[100718.051075]  [] ? warn_slowpath_null+0x1a/0x20
[100718.051077]  [] ? local_bh_enable_ip+0x7d/0xb0
[100718.051080]  [] ? _spin_unlock_bh+0x1b/0x20
[100718.051085]  [] ? be_process_mcc+0x74/0x230 [be2net]
[100718.051088]  [] ? be_poll_tx_mcc+0x16c/0x290 [be2net]
[100718.051090]  [] ? netpoll_poll_dev+0xd6/0x490
[100718.051095]  [] ? bond_poll_controller+0x75/0x80 [bonding]
[100718.051097]  [] ? netpoll_poll_dev+0x45/0x490
[100718.051100]  [] ? ksize+0x19/0x80
[100718.051102]  [] ? netpoll_send_skb_on_dev+0x157/0x240

by reenabling IRQ before calling ->poll, but it seems more
problems are introduced after that patch:

http://ozlabs.org/~akpm/stuff/IMG_20120824_122054.jpg
http://marc.info/?l=linux-netdev&m=134563282530588&w=2

So it is safe to fix be2net driver code directly.

This patch reverts the offending commit and fixes be_poll() by
avoid disabling BH there, this is okay because be_poll()
can be called either by poll_napi() which already disables
IRQ, or by net_rx_action() which already disables BH.

Reported-by: Andrew Morton 
Reported-by: Sylvain Munaut 
Cc: Sylvain Munaut 
Cc: Andrew Morton 
Cc: David Miller 
Cc: Sathya Perla 
Cc: Subbu Seetharaman 
Cc: Ajit Khaparde 
Signed-off-by: Cong Wang 
Tested-by: Sylvain Munaut 
Signed-off-by: David S. Miller

netpoll: re-enable irq in poll_napi()

2012-08-14T21:33:33+00:00

napi->poll() needs IRQ enabled, so we have to re-enable IRQ before
calling it.

Cc: David Miller 
Signed-off-by: Cong Wang 
Signed-off-by: David S. Miller

netpoll: handle vlan tags in netpoll tx and rx path

2012-08-14T21:33:32+00:00

Without this patch, I can't get netconsole logs remotely over
vlan. The reason is probably we don't handle vlan tags in either
netpoll tx or rx path.

I am not sure if I use these vlan functions correctly, at
least this patch works.

Cc: Benjamin LaHaise 
Cc: Patrick McHardy 
Cc: David Miller 
Signed-off-by: Cong Wang 
Signed-off-by: David S. Miller

netpoll: take rcu_read_lock_bh() in netpoll_send_skb_on_dev()

2012-08-14T21:33:31+00:00

This patch fixes several problems in the call path of
netpoll_send_skb_on_dev():

1. Disable IRQ's before calling netpoll_send_skb_on_dev().

2. All the callees of netpoll_send_skb_on_dev() should use
   rcu_dereference_bh() to dereference ->npinfo.

3. Rename arp_reply() to netpoll_arp_reply(), the former is too generic.

Cc: "David S. Miller" 
Signed-off-by: Cong Wang 
Signed-off-by: David S. Miller

netpoll: take rcu_read_lock_bh() in netpoll_rx()

2012-08-14T21:33:31+00:00

In __netpoll_rx(), it dereferences ->npinfo without rcu_dereference_bh(),
this patch fixes it by using the 'npinfo' passed from netpoll_rx()
where it is already dereferenced with rcu_dereference_bh().

Cc: "David S. Miller" 
Signed-off-by: Cong Wang 
Signed-off-by: David S. Miller

netpoll: make __netpoll_cleanup non-block

2012-08-14T21:33:30+00:00

Like the previous patch, slave_disable_netpoll() and __netpoll_cleanup()
may be called with read_lock() held too, so we should make them
non-block, by moving the cleanup and kfree() to call_rcu_bh() callbacks.

Cc: "David S. Miller" 
Signed-off-by: Cong Wang 
Signed-off-by: David S. Miller

netpoll: use GFP_ATOMIC in slave_enable_netpoll() and __netpoll_setup()

2012-08-14T21:33:30+00:00

slave_enable_netpoll() and __netpoll_setup() may be called
with read_lock() held, so should use GFP_ATOMIC to allocate
memory. Eric suggested to pass gfp flags to __netpoll_setup().

Cc: Eric Dumazet 
Cc: "David S. Miller" 
Reported-by: Dan Carpenter 
Signed-off-by: Eric Dumazet 
Signed-off-by: Cong Wang 
Signed-off-by: David S. Miller

netpoll: move np->dev and np->dev_name init into __netpoll_setup()

2012-07-17T16:02:36+00:00

Signed-off-by: Jiri Pirko 
Signed-off-by: David S. Miller

netpoll: fix netpoll_send_udp() bugs

2012-06-13T22:57:31+00:00

Bogdan Hamciuc diagnosed and fixed following bug in netpoll_send_udp() :

"skb->len += len;" instead of "skb_put(skb, len);"

Meaning that _if_ a network driver needs to call skb_realloc_headroom(),
only packet headers would be copied, leaving garbage in the payload.

However the skb_realloc_headroom() must be avoided as much as possible
since it requires memory and netpoll tries hard to work even if memory
is exhausted (using a pool of preallocated skbs)

It appears netpoll_send_udp() reserved 16 bytes for the ethernet header,
which happens to work for typicall drivers but not all.

Right thing is to use LL_RESERVED_SPACE(dev)
(And also add dev->needed_tailroom of tailroom)

This patch combines both fixes.

Many thanks to Bogdan for raising this issue.

Reported-by: Bogdan Hamciuc 
Signed-off-by: Eric Dumazet 
Tested-by: Bogdan Hamciuc 
Cc: Herbert Xu 
Cc: Neil Horman 
Reviewed-by: Neil Horman 
Reviewed-by: Cong Wang 
Signed-off-by: David S. Miller

Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

2012-02-19T21:03:15+00:00

Conflicts:
	drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c

Small minor conflict in bnx2x, wherein one commit changed how
statistics were stored in software, and another commit
fixed endianness bugs wrt. reading the values provided by
the chip in memory.

Signed-off-by: David S. Miller