linux-toradex.git/net/sched, branch v3.15-rc8

net_sched: fix an oops in tcindex filter

2014-05-21T20:47:13+00:00

Kelly reported the following crash:

        IP: [] tcf_action_exec+0x46/0x90
        PGD 3009067 PUD 300c067 PMD 11ff30067 PTE 800000011634b060
        Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
        CPU: 1 PID: 639 Comm: dhclient Not tainted 3.15.0-rc4+ #342
        Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
        task: ffff8801169ecd00 ti: ffff8800d21b8000 task.ti: ffff8800d21b8000
        RIP: 0010:[]  [] tcf_action_exec+0x46/0x90
        RSP: 0018:ffff8800d21b9b90  EFLAGS: 00010283
        RAX: 00000000ffffffff RBX: ffff88011634b8e8 RCX: ffff8800cf7133d8
        RDX: ffff88011634b900 RSI: ffff8800cf7133e0 RDI: ffff8800d210f840
        RBP: ffff8800d21b9bb0 R08: ffffffff8287bf60 R09: 0000000000000001
        R10: ffff8800d2b22b24 R11: 0000000000000001 R12: ffff8800d210f840
        R13: ffff8800d21b9c50 R14: ffff8800cf7133e0 R15: ffff8800cad433d8
        FS:  00007f49723e1840(0000) GS:ffff88011a800000(0000) knlGS:0000000000000000
        CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
        CR2: ffff88011634b8f0 CR3: 00000000ce469000 CR4: 00000000000006e0
        Stack:
         ffff8800d2170188 ffff8800d210f840 ffff8800d2171b90 0000000000000000
         ffff8800d21b9be8 ffffffff817c55bb ffff8800d21b9c50 ffff8800d2171b90
         ffff8800d210f840 ffff8800d21b0300 ffff8800d21b9c50 ffff8800d21b9c18
        Call Trace:
         [] tcindex_classify+0x88/0x9b
         [] tc_classify_compat+0x3e/0x7b
         [] tc_classify+0x25/0x9f
         [] htb_enqueue+0x55/0x27a
         [] dsmark_enqueue+0x165/0x1a4
         [] __dev_queue_xmit+0x35e/0x536
         [] dev_queue_xmit+0x10/0x12
         [] packet_sendmsg+0xb26/0xb9a
         [] ? __lock_acquire+0x3ae/0xdf3
         [] __sock_sendmsg_nosec+0x25/0x27
         [] sock_aio_write+0xd0/0xe7
         [] do_sync_write+0x59/0x78
         [] vfs_write+0xb5/0x10a
         [] SyS_write+0x49/0x7f
         [] system_call_fastpath+0x16/0x1b

This is because we memcpy struct tcindex_filter_result which contains
struct tcf_exts, obviously struct list_head can not be simply copied.
This is a regression introduced by commit 33be627159913b094bb578
(net_sched: act: use standard struct list_head).

It's not very easy to fix it as the code is a mess:

       if (old_r)
               memcpy(&cr, r, sizeof(cr));
       else {
               memset(&cr, 0, sizeof(cr));
               tcf_exts_init(&cr.exts, TCA_TCINDEX_ACT, TCA_TCINDEX_POLICE);
       }
       ...
       tcf_exts_change(tp, &cr.exts, &e);
       ...
       memcpy(r, &cr, sizeof(cr));

the above code should equal to:

        tcindex_filter_result_init(&cr);
        if (old_r)
               cr.res = r->res;
        ...
        if (old_r)
               tcf_exts_change(tp, &r->exts, &e);
        else
               tcf_exts_change(tp, &cr.exts, &e);
        ...
        r->res = cr.res;

after this change, since there is no need to copy struct tcf_exts.

And it also fixes other places zero'ing struct's contains struct tcf_exts.

Fixes: commit 33be627159913b0 (net_sched: act: use standard struct list_head)
Reported-by: Kelly Anderson 
Tested-by: Kelly Anderson 
Cc: David S. Miller 
Signed-off-by: Cong Wang 
Signed-off-by: David S. Miller

net: sched: lock imbalance in hhf qdisc

2014-05-04T23:41:45+00:00

hhf_change() takes the sch_tree_lock and releases it but misses the
error cases. Fix the missed case here.

To reproduce try a command like this,

# tc qdisc change dev p3p2 root hhf quantum 40960 non_hh_weight 300000

Signed-off-by: John Fastabend 
Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller

net: Use netlink_ns_capable to verify the permisions of netlink messages

2014-04-24T17:44:54+00:00

It is possible by passing a netlink socket to a more privileged
executable and then to fool that executable into writing to the socket
data that happens to be valid netlink message to do something that
privileged executable did not intend to do.

To keep this from happening replace bare capable and ns_capable calls
with netlink_capable, netlink_net_calls and netlink_ns_capable calls.
Which act the same as the previous calls except they verify that the
opener of the socket had the desired permissions as well.

Reported-by: Andy Lutomirski 
Signed-off-by: "Eric W. Biederman" 
Signed-off-by: David S. Miller

net-sysfs: expose number of carrier on/off changes

2014-03-31T20:24:52+00:00

This allows to monitor carrier on/off transitions and detect link
flapping issues:
 - new /sys/class/net/X/carrier_changes
 - new rtnetlink IFLA_CARRIER_CHANGES (getlink)

Tested:
  - grep . /sys/class/net/*/carrier_changes
    + ip link set dev X down/up
    + plug/unplug cable
  - updated iproute2: prints IFLA_CARRIER_CHANGES
  - iproute2 20121211-2 (debian): unchanged behavior

Signed-off-by: David Decotigny 
Signed-off-by: David S. Miller

net: sched: use no more than one page in struct fw_head

2014-03-18T18:17:55+00:00

In commit b4e9b520ca5d ("[NET_SCHED]: Add mask support to fwmark
classifier") Patrick added an u32 field in fw_head, making it slightly
bigger than one page.

Lets use 256 slots to make fw_hash() more straight forward, and move
@mask to the beginning of the structure as we often use a small number
of skb->mark. @mask and first hash buckets share the same cache line.

This brings back the memory usage to less than 4000 bytes, and permits
John to add a rcu_head at the end of the structure later without any
worry.

Signed-off-by: Eric Dumazet 
Cc: Thomas Graf 
Cc: John Fastabend 
Acked-by: Thomas Graf 
Signed-off-by: David S. Miller

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

2014-03-15T02:31:55+00:00

Conflicts:
	drivers/net/usb/r8152.c
	drivers/net/xen-netback/netback.c

Both the r8152 and netback conflicts were simple overlapping
changes.

Signed-off-by: David S. Miller

net_sched: return nla_nest_end() instead of skb->len

2014-03-13T19:39:20+00:00

nla_nest_end() already has return skb->len, so replace
return skb->len with return nla_nest_end instead().

Signed-off-by: Yang Yingliang 
Signed-off-by: David S. Miller

pkt_sched: add cond_resched() to class and qdisc dump

2014-03-12T03:54:23+00:00

We have seen delays of more than 50ms in class or qdisc dumps, in case
device is under high TX stress, even with the prior 4KB per skb limit.

Add cond_resched() to give a chance to higher prio tasks to get cpu.

Signed-off-by; Eric Dumazet 
Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller

pkt_sched: do not use rcu in tc_dump_qdisc()

2014-03-12T03:54:23+00:00

Like all rtnetlink dump operations, we hold RTNL in tc_dump_qdisc(),
so we do not need to use rcu protection to protect list of netdevices.

This will allow preemption to occur, thus reducing latencies.
Following patch adds explicit cond_resched() calls.

Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller

pkt_sched: fq: do not hold qdisc lock while allocating memory

2014-03-10T20:17:52+00:00

Resizing fq hash table allocates memory while holding qdisc spinlock,
with BH disabled.

This is definitely not good, as allocation might sleep.

We can drop the lock and get it when needed, we hold RTNL so no other
changes can happen at the same time.

Signed-off-by: Eric Dumazet 
Fixes: afe4fd062416 ("pkt_sched: fq: Fair Queue packet scheduler")
Signed-off-by: David S. Miller