Skip to content

cloud-kernel
cloud-kernel

openanolis / cloud-kernel
1 年多 前同步成功

通知 160
Star 36
Fork 7
cloud-kernel
cloud-kernel
切换分支/标签
  1. 18 5月, 2017 7 次提交
  3. 22 4月, 2017 1 次提交
    • W
      net_sched: move the empty tp check from ->destroy() to ->delete() · 763dbf63
      由 WANG Cong 提交于 
      We could have a race condition where in ->classify() path we
dereference tp->root and meanwhile a parallel ->destroy() makes it
a NULL. Daniel cured this bug in commit d9363774
("net, sched: respect rcu grace period on cls destruction").

This happens when ->destroy() is called for deleting a filter to
check if we are the last one in tp, this tp is still linked and
visible at that time. The root cause of this problem is the semantic
of ->destroy(), it does two things (for non-force case):

1) check if tp is empty
2) if tp is empty we could really destroy it

and its caller, if cares, needs to check its return value to see if it
is really destroyed. Therefore we can't unlink tp unless we know it is
empty.

As suggested by Daniel, we could actually move the test logic to ->delete()
so that we can safely unlink tp after ->delete() tells us the last one is
just deleted and before ->destroy().

Fixes: 1e052be6 ("net_sched: destroy proto tp when all filters are gone")
Cc: Roi Dayan <roid@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: NCong Wang <xiyou.wangcong@gmail.com>
Acked-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      763dbf63
  5. 18 4月, 2017 1 次提交
  7. 14 4月, 2017 1 次提交
  9. 15 2月, 2017 1 次提交
  11. 11 2月, 2017 6 次提交
  13. 27 12月, 2016 1 次提交
    • D
      net, sched: fix soft lockup in tc_classify · 628185cf
      由 Daniel Borkmann 提交于 
      Shahar reported a soft lockup in tc_classify(), where we run into an
endless loop when walking the classifier chain due to tp->next == tp
which is a state we should never run into. The issue only seems to
trigger under load in the tc control path.

What happens is that in tc_ctl_tfilter(), thread A allocates a new
tp, initializes it, sets tp_created to 1, and calls into tp->ops->change()
with it. In that classifier callback we had to unlock/lock the rtnl
mutex and returned with -EAGAIN. One reason why we need to drop there
is, for example, that we need to request an action module to be loaded.

This happens via tcf_exts_validate() -> tcf_action_init/_1() meaning
after we loaded and found the requested action, we need to redo the
whole request so we don't race against others. While we had to unlock
rtnl in that time, thread B's request was processed next on that CPU.
Thread B added a new tp instance successfully to the classifier chain.
When thread A returned grabbing the rtnl mutex again, propagating -EAGAIN
and destroying its tp instance which never got linked, we goto replay
and redo A's request.

This time when walking the classifier chain in tc_ctl_tfilter() for
checking for existing tp instances we had a priority match and found
the tp instance that was created and linked by thread B. Now calling
again into tp->ops->change() with that tp was successful and returned
without error.

tp_created was never cleared in the second round, thus kernel thinks
that we need to link it into the classifier chain (once again). tp and
*back point to the same object due to the match we had earlier on. Thus
for thread B's already public tp, we reset tp->next to tp itself and
link it into the chain, which eventually causes the mentioned endless
loop in tc_classify() once a packet hits the data path.

Fix is to clear tp_created at the beginning of each request, also when
we replay it. On the paths that can cause -EAGAIN we already destroy
the original tp instance we had and on replay we really need to start
from scratch. It seems that this issue was first introduced in commit
12186be7 ("net_cls: fix unconfigured struct tcf_proto keeps chaining
and avoid kernel panic when we use cls_cgroup").

Fixes: 12186be7 ("net_cls: fix unconfigured struct tcf_proto keeps chaining and avoid kernel panic when we use cls_cgroup")
Reported-by: NShahar Klein <shahark@mellanox.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Tested-by: NShahar Klein <shahark@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      628185cf
  15. 03 12月, 2016 1 次提交
    • H
      net/sched: cls_flower: Add offload support using egress Hardware device · 7091d8c7
      由 Hadar Hen Zion 提交于 
      In order to support hardware offloading when the device given by the tc
rule is different from the Hardware underline device, extract the mirred
(egress) device from the tc action when a filter is added, using the new
tc_action_ops, get_dev().

Flower caches the information about the mirred device and use it for
calling ndo_setup_tc in filter change, update stats and delete.

Calling ndo_setup_tc of the mirred (egress) device instead of the
ingress device will allow a resolution between the software ingress
device and the underline hardware device.

The resolution will take place inside the offloading driver using
'egress_device' flag added to tc_to_netdev struct which is provided to
the offloading driver.
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Acked-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      7091d8c7
  17. 25 11月, 2016 1 次提交
  19. 18 11月, 2016 1 次提交
  21. 28 10月, 2016 1 次提交
    • J
      net sched filters: fix notification of filter delete with proper handle · 9ee78374
      由 Jamal Hadi Salim 提交于 
      Daniel says:

While trying out [1][2], I noticed that tc monitor doesn't show the
correct handle on delete:

$ tc monitor
qdisc clsact ffff: dev eno1 parent ffff:fff1
filter dev eno1 ingress protocol all pref 49152 bpf handle 0x2a [...]
deleted filter dev eno1 ingress protocol all pref 49152 bpf handle 0xf3be0c80

some context to explain the above:
The user identity of any tc filter is represented by a 32-bit
identifier encoded in tcm->tcm_handle. Example 0x2a in the bpf filter
above. A user wishing to delete, get or even modify a specific filter
uses this handle to reference it.
Every classifier is free to provide its own semantics for the 32 bit handle.
Example: classifiers like u32 use schemes like 800:1:801 to describe
the semantics of their filters represented as hash table, bucket and
node ids etc.
Classifiers also have internal per-filter representation which is different
from this externally visible identity. Most classifiers set this
internal representation to be a pointer address (which allows fast retrieval
of said filters in their implementations). This internal representation
is referenced with the "fh" variable in the kernel control code.

When a user successfuly deletes a specific filter, by specifying the correct
tcm->tcm_handle, an event is generated to user space which indicates
which specific filter was deleted.

Before this patch, the "fh" value was sent to user space as the identity.
As an example what is shown in the sample bpf filter delete event above
is 0xf3be0c80. This is infact a 32-bit truncation of 0xffff8807f3be0c80
which happens to be a 64-bit memory address of the internal filter
representation (address of the corresponding filter's struct cls_bpf_prog);

After this patch the appropriate user identifiable handle as encoded
in the originating request tcm->tcm_handle is generated in the event.
One of the cardinal rules of netlink rules is to be able to take an
event (such as a delete in this case) and reflect it back to the
kernel and successfully delete the filter. This patch achieves that.

Note, this issue has existed since the original TC action
infrastructure code patch back in 2004 as found in:
https://git.kernel.org/cgit/linux/kernel/git/history/history.git/commit/

[1] http://patchwork.ozlabs.org/patch/682828/
[2] http://patchwork.ozlabs.org/patch/682829/

Fixes: 4e54c4816bfe ("[NET]: Add tc extensions infrastructure.")
Reported-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NCong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: NJamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      9ee78374
  23. 13 10月, 2016 1 次提交
    • E
      net_sched: do not broadcast RTM_GETTFILTER result · fa59b27c
      由 Eric Dumazet 提交于 
      There are two ways to get tc filters from kernel to user space.

1) Full dump (tc_dump_tfilter())
2) RTM_GETTFILTER to get one precise filter, reducing overhead.

The second operation is unfortunately broadcasting its result,
polluting "tc monitor" users.

This patch makes sure only the requester gets the result, using
netlink_unicast() instead of rtnetlink_send()

Jamal cooked an iproute2 patch to implement "tc filter get" operation,
but other user space libraries already use RTM_GETTFILTER when a single
filter is queried, instead of dumping all filters.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Acked-by: NJamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      fa59b27c
  25. 20 9月, 2016 1 次提交
  27. 18 8月, 2016 1 次提交
  29. 17 6月, 2016 1 次提交
  31. 11 6月, 2016 1 次提交
    • D
      net, cls: allow for deleting all filters for given parent · ea7f8277
      由 Daniel Borkmann 提交于 
      Add a possibility where the user can just specify the parent and
all filters under that parent are then being purged. Currently,
for example for scripting, one needs to specify pref/prio to have
a well-defined number for 'tc filter del' command for addressing
the previously created instance or additionally filter handle in
case of priorities being the same. Improve usage by allowing the
option for tc to specify the parent and removing the whole chain
for that given parent.

Example usage after patch, no tc changes required:

  # tc qdisc replace dev foo clsact
  # tc filter add dev foo egress bpf da obj ./bpf.o
  # tc filter add dev foo egress bpf da obj ./bpf.o
  # tc filter show dev foo egress
  filter protocol all pref 49151 bpf
  filter protocol all pref 49151 bpf handle 0x1 bpf.o:[classifier] direct-action
  filter protocol all pref 49152 bpf
  filter protocol all pref 49152 bpf handle 0x1 bpf.o:[classifier] direct-action
  # tc filter del dev foo egress
  # tc filter show dev foo egress
  #

Previously, RTM_DELTFILTER requests with invalid prio of 0 were
rejected, so only netlink requests with RTM_NEWTFILTER and NLM_F_CREATE
flag were allowed where the kernel would auto-generate a pref/prio.
We can piggyback on that and use prio of 0 as a wildcard for
requests of RTM_DELTFILTER.

For notifying tc netlink monitoring users (e.g. libnl uses this
for caching), there are two options, that is, sending individual
tfilter_notify() notifications for each tcf_proto, or sending a
single one indicating wildcard removal. I tried both and there
are pros and cons for each, eventually I decided for sending
individual tfilter_notify(), so that user space can support this
seamlessly and there won't be a mess of changing each and every
application to make sure expectations from the kernel won't break
when they don't understand single notification. Since linear chains
don't really scale, I expect only a handful of classifiers to be
attached at max for a given parent anyway.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NJamal Hadi Salim <jhs@mojatatu.com>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      ea7f8277
  33. 08 6月, 2016 1 次提交
  35. 22 5月, 2015 1 次提交
    • D
      net: sched: fix call_rcu() race on classifier module unloads · c78e1746
      由 Daniel Borkmann 提交于 
      Vijay reported that a loop as simple as ...

  while true; do
    tc qdisc add dev foo root handle 1: prio
    tc filter add dev foo parent 1: u32 match u32 0 0  flowid 1
    tc qdisc del dev foo root
    rmmod cls_u32
  done

... will panic the kernel. Moreover, he bisected the change
apparently introducing it to 78fd1d0a ("netlink: Re-add
locking to netlink_lookup() and seq walker").

The removal of synchronize_net() from the netlink socket
triggering the qdisc to be removed, seems to have uncovered
an RCU resp. module reference count race from the tc API.
Given that RCU conversion was done after e341694e ("netlink:
Convert netlink_lookup() to use RCU protected hash table")
which added the synchronize_net() originally, occasion of
hitting the bug was less likely (not impossible though):

When qdiscs that i) support attaching classifiers and,
ii) have at least one of them attached, get deleted, they
invoke tcf_destroy_chain(), and thus call into ->destroy()
handler from a classifier module.

After RCU conversion, all classifier that have an internal
prio list, unlink them and initiate freeing via call_rcu()
deferral.

Meanhile, tcf_destroy() releases already reference to the
tp->ops->owner module before the queued RCU callback handler
has been invoked.

Subsequent rmmod on the classifier module is then not prevented
since all module references are already dropped.

By the time, the kernel invokes the RCU callback handler from
the module, that function address is then invalid.

One way to fix it would be to add an rcu_barrier() to
unregister_tcf_proto_ops() to wait for all pending call_rcu()s
to complete.

synchronize_rcu() is not appropriate as under heavy RCU
callback load, registered call_rcu()s could be deferred
longer than a grace period. In case we don't have any pending
call_rcu()s, the barrier is allowed to return immediately.

Since we came here via unregister_tcf_proto_ops(), there
are no users of a given classifier anymore. Further nested
call_rcu()s pointing into the module space are not being
done anywhere.

Only cls_bpf_delete_prog() may schedule a work item, to
unlock pages eventually, but that is not in the range/context
of cls_bpf anymore.

Fixes: 25d8c0d5 ("net: rcu-ify tcf_proto")
Fixes: 9888faef ("net: sched: cls_basic use RCU")
Reported-by: NVijay Subramanian <subramanian.vijay@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Cc: John Fastabend <john.r.fastabend@intel.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Thomas Graf <tgraf@suug.ch>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Alexei Starovoitov <ast@plumgrid.com>
Tested-by: NVijay Subramanian <subramanian.vijay@gmail.com>
Acked-by: NAlexei Starovoitov <ast@plumgrid.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      c78e1746
  37. 10 5月, 2015 1 次提交
  39. 10 3月, 2015 1 次提交
    • C
      net_sched: destroy proto tp when all filters are gone · 1e052be6
      由 Cong Wang 提交于 
      Kernel automatically creates a tp for each
(kind, protocol, priority) tuple, which has handle 0,
when we add a new filter, but it still is left there
after we remove our own, unless we don't specify the
handle (literally means all the filters under
the tuple). For example this one is left:

  # tc filter show dev eth0
  filter parent 8001: protocol arp pref 49152 basic

The user-space is hard to clean up these for kernel
because filters like u32 are organized in a complex way.
So kernel is responsible to remove it after all filters
are gone.  Each type of filter has its own way to
store the filters, so each type has to provide its
way to check if all filters are gone.

Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: NCong Wang <cwang@twopensource.com>
Signed-off-by: NCong Wang <xiyou.wangcong@gmail.com>
Acked-by: Jamal Hadi Salim<jhs@mojatatu.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      1e052be6
  41. 05 2月, 2015 1 次提交
  43. 09 10月, 2014 1 次提交
  45. 29 9月, 2014 1 次提交
  47. 14 9月, 2014 2 次提交
  49. 18 7月, 2014 1 次提交
  51. 03 5月, 2014 1 次提交
  53. 28 4月, 2014 2 次提交