Since UDP no more uses sk->destructor, we can clear completely
the skb head state before enqueuing. Amend and use
skb_release_head_state() for that.
All head states share a single cacheline, which is not
normally used/accesses on dequeue. We can avoid entirely accessing
such cacheline implementing and using in the UDP code a specialized
skb free helper which ignores the skb head state.
This saves a cacheline miss at skb deallocation time.
v1 -> v2:
replaced secpath_reset() with skb_release_head_state()
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The same code is replicated in 3 different places; move it to a
common helper.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is a defense-in-depth measure in response to bugs like
4d6fa57b4d ("macsec: avoid heap overflow in skb_to_sgvec"). There's
not only a potential overflow of sglist items, but also a stack overflow
potential, so we fix this by limiting the amount of recursion this function
is allowed to do. Not actually providing a bounded base case is a future
disaster that we can easily avoid here.
As a small matter of house keeping, we take this opportunity to move the
documentation comment over the actual function the documentation is for.
While this could be implemented by using an explicit stack of skbuffs,
when implementing this, the function complexity increased considerably,
and I don't think such complexity and bloat is actually worth it. So,
instead I built this and tested it on x86, x86_64, ARM, ARM64, and MIPS,
and measured the stack usage there. I also reverted the recent MIPS
changes that give it a separate IRQ stack, so that I could experience
some worst-case situations. I found that limiting it to 24 layers deep
yielded a good stack usage with room for safety, as well as being much
deeper than any driver actually ever creates.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: David Howells <dhowells@redhat.com>
Cc: Sabrina Dubroca <sd@queasysnail.net>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add SOF_TIMESTAMPING_OPT_TX_SWHW option to allow an outgoing packet to
be looped to the socket's error queue with a software timestamp even
when a hardware transmit timestamp is expected to be provided by the
driver.
Applications using this option will receive two separate messages from
the error queue, one with a software timestamp and the other with a
hardware timestamp. As the hardware timestamp is saved to the shared skb
info, which may happen before the first message with software timestamp
is received by the application, the hardware timestamp is copied to the
SCM_TIMESTAMPING control message only when the skb has no software
timestamp or it is an incoming packet.
While changing sw_tx_timestamp(), inline it in skb_tx_timestamp() as
there are no other users.
CC: Richard Cochran <richardcochran@gmail.com>
CC: Willem de Bruijn <willemb@google.com>
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since commit b68581778c ("net: Make skb->skb_iif always track
skb->dev") skbs don't have the original index of the interface which
received the packet. This information is now needed for a new control
message related to hardware timestamping.
Instead of adding a new field to skb, we can find the device by the NAPI
ID if it is available, i.e. CONFIG_NET_RX_BUSY_POLL is enabled and the
driver is using NAPI. Add dev_get_by_napi_id() and also skb_napi_id() to
hide the CONFIG_NET_RX_BUSY_POLL ifdef.
CC: Richard Cochran <richardcochran@gmail.com>
Suggested-by: Willem de Bruijn <willemb@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add FCoE to the list of protocols that can set CHECKSUM_UNNECESSARY; add a
note to CHECKSUM_COMPLETE section to specify that it does not apply to SCTP
and FCoE protocols.
Suggested-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Acked-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
skb_csum_hwoffload_help() uses netdev features and skb->csum_not_inet to
determine if skb needs software computation of Internet Checksum or crc32c
(or nothing, if this computation can be done by the hardware). Use it in
place of skb_checksum_help() in validate_xmit_skb() to avoid corruption
of non-GSO SCTP packets having skb->ip_summed equal to CHECKSUM_PARTIAL.
While at it, remove references to skb_csum_off_chk* functions, since they
are not present anymore in Linux _ see commit cf53b1da73 ("Revert
"net: Add driver helper functions to determine checksum offloadability"").
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
skb->csum_not_inet carries the indication on which algorithm is needed to
compute checksum on skb in the transmit path, when skb->ip_summed is equal
to CHECKSUM_PARTIAL. If skb carries a SCTP packet and crc32c hasn't been
yet written in L4 header, skb->csum_not_inet is assigned to 1; otherwise,
assume Internet Checksum is needed and thus set skb->csum_not_inet to 0.
Suggested-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Acked-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This bit was introduced with commit 5a21232983 ("net: Support for
csum_bad in skbuff") to reduce the stack workload when processing RX
packets carrying a wrong Internet Checksum. Up to now, only one driver and
GRO core are setting it.
Suggested-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
skb_crc32c_csum_help is like skb_checksum_help, but it is designed for
checksumming SCTP packets using crc32c (see RFC3309), provided that
libcrc32c.ko has been loaded before. In case libcrc32c is not loaded,
invoking skb_crc32c_csum_help on a skb results in one the following
printouts:
warn_crc32c_csum_update: attempt to compute crc32c without libcrc32c.ko
warn_crc32c_csum_combine: attempt to compute crc32c without libcrc32c.ko
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
sctp_compute_checksum requires crc32c symbol (provided by libcrc32c), so
it can't be used in net core. Like it has been done previously with other
symbols (e.g. ipv6_dst_lookup), introduce a stub struct skb_checksum_ops
to allow computation of crc32c checksum in net core after sctp.ko (and thus
libcrc32c) has been loaded.
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TCP Timestamps option is defined in RFC 7323
Traditionally on linux, it has been tied to the internal
'jiffies' variable, because it had been a cheap and good enough
generator.
For TCP flows on the Internet, 1 ms resolution would be much better
than 4ms or 10ms (HZ=250 or HZ=100 respectively)
For TCP flows in the DC, Google has used usec resolution for more
than two years with great success [1]
Receive size autotuning (DRS) is indeed more precise and converges
faster to optimal window size.
This patch converts tp->tcp_mstamp to a plain u64 value storing
a 1 usec TCP clock.
This choice will allow us to upstream the 1 usec TS option as
discussed in IETF 97.
[1] https://www.ietf.org/proceedings/97/slides/slides-97-tcpm-tcp-options-for-low-latency-00.pdf
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
And update __sk_queue_drop_skb() to work on the specified queue.
This will help the udp protocol to use an additional private
rx queue in a later patch.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking updates from David Millar:
"Here are some highlights from the 2065 networking commits that
happened this development cycle:
1) XDP support for IXGBE (John Fastabend) and thunderx (Sunil Kowuri)
2) Add a generic XDP driver, so that anyone can test XDP even if they
lack a networking device whose driver has explicit XDP support
(me).
3) Sparc64 now has an eBPF JIT too (me)
4) Add a BPF program testing framework via BPF_PROG_TEST_RUN (Alexei
Starovoitov)
5) Make netfitler network namespace teardown less expensive (Florian
Westphal)
6) Add symmetric hashing support to nft_hash (Laura Garcia Liebana)
7) Implement NAPI and GRO in netvsc driver (Stephen Hemminger)
8) Support TC flower offload statistics in mlxsw (Arkadi Sharshevsky)
9) Multiqueue support in stmmac driver (Joao Pinto)
10) Remove TCP timewait recycling, it never really could possibly work
well in the real world and timestamp randomization really zaps any
hint of usability this feature had (Soheil Hassas Yeganeh)
11) Support level3 vs level4 ECMP route hashing in ipv4 (Nikolay
Aleksandrov)
12) Add socket busy poll support to epoll (Sridhar Samudrala)
13) Netlink extended ACK support (Johannes Berg, Pablo Neira Ayuso,
and several others)
14) IPSEC hw offload infrastructure (Steffen Klassert)"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (2065 commits)
tipc: refactor function tipc_sk_recv_stream()
tipc: refactor function tipc_sk_recvmsg()
net: thunderx: Optimize page recycling for XDP
net: thunderx: Support for XDP header adjustment
net: thunderx: Add support for XDP_TX
net: thunderx: Add support for XDP_DROP
net: thunderx: Add basic XDP support
net: thunderx: Cleanup receive buffer allocation
net: thunderx: Optimize CQE_TX handling
net: thunderx: Optimize RBDR descriptor handling
net: thunderx: Support for page recycling
ipx: call ipxitf_put() in ioctl error path
net: sched: add helpers to handle extended actions
qed*: Fix issues in the ptp filter config implementation.
qede: Fix concurrency issue in PTP Tx path processing.
stmmac: Add support for SIMATIC IOT2000 platform
net: hns: fix ethtool_get_strings overflow in hns driver
tcp: fix wraparound issue in tcp_lp
bpf, arm64: fix jit branch offset related to ldimm64
bpf, arm64: implement jiting of BPF_XADD
...
All available gso_type flags are currently in use, so
extend gso_type from 'unsigned short' to 'unsigned int'
to be able to add further flags.
We reorder the struct skb_shared_info to use
two bytes of the four byte hole before dataref.
All fields before dataref are cleared, i.e.
four bytes more than before the change.
The remaining two byte hole is moved to the
beginning of the structure, this protects us
from immediate overwites on out of bound writes
to the sk_buff head.
Structure layout on x86-64 before the change:
struct skb_shared_info {
unsigned char nr_frags; /* 0 1 */
__u8 tx_flags; /* 1 1 */
short unsigned int gso_size; /* 2 2 */
short unsigned int gso_segs; /* 4 2 */
short unsigned int gso_type; /* 6 2 */
struct sk_buff * frag_list; /* 8 8 */
struct skb_shared_hwtstamps hwtstamps; /* 16 8 */
u32 tskey; /* 24 4 */
__be32 ip6_frag_id; /* 28 4 */
atomic_t dataref; /* 32 4 */
/* XXX 4 bytes hole, try to pack */
void * destructor_arg; /* 40 8 */
skb_frag_t frags[17]; /* 48 272 */
/* --- cacheline 5 boundary (320 bytes) --- */
/* size: 320, cachelines: 5, members: 12 */
/* sum members: 316, holes: 1, sum holes: 4 */
};
Structure layout on x86-64 after the change:
struct skb_shared_info {
short unsigned int _unused; /* 0 2 */
unsigned char nr_frags; /* 2 1 */
__u8 tx_flags; /* 3 1 */
short unsigned int gso_size; /* 4 2 */
short unsigned int gso_segs; /* 6 2 */
struct sk_buff * frag_list; /* 8 8 */
struct skb_shared_hwtstamps hwtstamps; /* 16 8 */
unsigned int gso_type; /* 24 4 */
u32 tskey; /* 28 4 */
__be32 ip6_frag_id; /* 32 4 */
atomic_t dataref; /* 36 4 */
void * destructor_arg; /* 40 8 */
skb_frag_t frags[17]; /* 48 272 */
/* --- cacheline 5 boundary (320 bytes) --- */
/* size: 320, cachelines: 5, members: 13 */
};
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We are going to split <linux/sched/clock.h> out of <linux/sched.h>, which
will have to be picked up from other headers and .c files.
Create a trivial placeholder <linux/sched/clock.h> file that just
maps to <linux/sched.h> to make this patch obviously correct and
bisectable.
Include the new header in the files that are going to need it.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Introduce skb_mac_offset() that could be used to get mac header offset.
Signed-off-by: Amir Vadai <amir@vadai.me>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add new skbuff flag to allow protocols to confirm neighbour.
When same struct dst_entry can be used for many different
neighbours we can not use it for pending confirmations.
Add sock_confirm_neigh() helper to confirm the neighbour and
use it for IPv4, IPv6 and VRF before dst_neigh_output.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pablo Neira Ayuso says:
====================
Netfilter updates for net-next
The following patchset contains Netfilter updates for your net-next
tree, they are:
1) Stash ctinfo 3-bit field into pointer to nf_conntrack object from
sk_buff so we only access one single cacheline in the conntrack
hotpath. Patchset from Florian Westphal.
2) Don't leak pointer to internal structures when exporting x_tables
ruleset back to userspace, from Willem DeBruijn. This includes new
helper functions to copy data to userspace such as xt_data_to_user()
as well as conversions of our ip_tables, ip6_tables and arp_tables
clients to use it. Not surprinsingly, ebtables requires an ad-hoc
update. There is also a new field in x_tables extensions to indicate
the amount of bytes that we copy to userspace.
3) Add nf_log_all_netns sysctl: This new knob allows you to enable
logging via nf_log infrastructure for all existing netnamespaces.
Given the effort to provide pernet syslog has been discontinued,
let's provide a way to restore logging using netfilter kernel logging
facilities in trusted environments. Patch from Michal Kubecek.
4) Validate SCTP checksum from conntrack helper, from Davide Caratti.
5) Merge UDPlite conntrack and NAT helpers into UDP, this was mostly
a copy&paste from the original helper, from Florian Westphal.
6) Reset netfilter state when duplicating packets, also from Florian.
7) Remove unnecessary check for broadcast in IPv6 in pkttype match and
nft_meta, from Liping Zhang.
8) Add missing code to deal with loopback packets from nft_meta when
used by the netdev family, also from Liping.
9) Several cleanups on nf_tables, one to remove unnecessary check from
the netlink control plane path to add table, set and stateful objects
and code consolidation when unregister chain hooks, from Gao Feng.
10) Fix harmless reference counter underflow in IPVS that, however,
results in problems with the introduction of the new refcount_t
type, from David Windsor.
11) Enable LIBCRC32C from nf_ct_sctp instead of nf_nat_sctp,
from Davide Caratti.
12) Missing documentation on nf_tables uapi header, from Liping Zhang.
13) Use rb_entry() helper in xt_connlimit, from Geliang Tang.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
After this change conntrack operations (lookup, creation, matching from
ruleset) only access one instead of two sk_buff cache lines.
This works for normal conntracks because those are allocated from a slab
that guarantees hw cacheline or 8byte alignment (whatever is larger)
so the 3 bits needed for ctinfo won't overlap with nf_conn addresses.
Template allocation now does manual address alignment (see previous change)
on arches that don't have sufficent kmalloc min alignment.
Some spots intentionally use skb->_nfct instead of skb_nfct() helpers,
this is to avoid undoing the skb_nfct() use when we remove untracked
conntrack object in the future.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Followup patch renames skb->nfct and changes its type so add a helper to
avoid intrusive rename change later.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Patch series "Page fragment updates", v4.
This patch series takes care of a few cleanups for the page fragments
API.
First we do some renames so that things are much more consistent. First
we move the page_frag_ portion of the name to the front of the functions
names. Secondly we split out the cache specific functions from the
other page fragment functions by adding the word "cache" to the name.
Finally I added a bit of documentation that will hopefully help to
explain some of this. I plan to revisit this later as we get things
more ironed out in the near future with the changes planned for the DMA
setup to support eXpress Data Path.
This patch (of 3):
This patch renames the page frag functions to be more consistent with
other APIs. Specifically we place the name page_frag first in the name
and then have either an alloc or free call name that we append as the
suffix. This makes it a bit clearer in terms of naming.
In addition we drop the leading double underscores since we are
technically no longer a backing interface and instead the front end that
is called from the networking APIs.
Link: http://lkml.kernel.org/r/20170104023854.13451.67390.stgit@localhost.localdomain
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The tc_from field fulfills two roles. It encodes whether a packet was
redirected by an act_mirred device and, if so, whether act_mirred was
called on ingress or egress. Split it into separate fields.
The information is needed by the special IFB loop, where packets are
taken out of the normal path by act_mirred, forwarded to IFB, then
reinjected at their original location (ingress or egress) by IFB.
The IFB device cannot use skb->tc_at_ingress, because that may have
been overwritten as the packet travels from act_mirred to ifb_xmit,
when it passes through tc_classify on the IFB egress path. Cache this
value in skb->tc_from_ingress.
That field is valid only if a packet arriving at ifb_xmit came from
act_mirred. Other packets can be crafted to reach ifb_xmit. These
must be dropped. Set tc_redirected on redirection and drop all packets
that do not have this bit set.
Both fields are set only on cloned skbs in tc actions, so original
packet sources do not have to clear the bit when reusing packets
(notably, pktgen and octeon).
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Field tc_at is used only within tc actions to distinguish ingress from
egress processing. A single bit is sufficient for this purpose.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Extract the remaining two fields from tc_verd and remove the __u16
completely. TC_AT and TC_FROM are converted to equivalent two-bit
integer fields tc_at and tc_from. Where possible, use existing
helper skb_at_tc_ingress when reading tc_at. Introduce helper
skb_reset_tc to clear fields.
Not documenting tc_from and tc_at, because they will be replaced
with single bit fields in follow-on patches.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Packets sent by the IFB device skip subsequent tc classification.
A single bit governs this state. Move it out of tc_verd in
anticipation of removing that __u16 completely.
The new bitfield tc_skip_classify temporarily uses one bit of a
hole, until tc_verd is removed completely in a follow-up patch.
Remove the bit hole comment. It could be 2, 3, 4 or 5 bits long.
With that many options, little value in documenting it.
Introduce a helper function to deduplicate the logic in the two
sites that check this bit.
The field tc_skip_classify is set only in IFB on skbs cloned in
act_mirred, so original packet sources do not have to clear the
bit when reusing packets (notably, pktgen and octeon).
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ktime_set(S,N) was required for the timespec storage type and is still
useful for situations where a Seconds and Nanoseconds part of a time value
needs to be converted. For anything where the Seconds argument is 0, this
is pointless and can be replaced with a simple assignment.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Pull vfs updates from Al Viro:
- more ->d_init() stuff (work.dcache)
- pathname resolution cleanups (work.namei)
- a few missing iov_iter primitives - copy_from_iter_full() and
friends. Either copy the full requested amount, advance the iterator
and return true, or fail, return false and do _not_ advance the
iterator. Quite a few open-coded callers converted (and became more
readable and harder to fuck up that way) (work.iov_iter)
- several assorted patches, the big one being logfs removal
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
logfs: remove from tree
vfs: fix put_compat_statfs64() does not handle errors
namei: fold should_follow_link() with the step into not-followed link
namei: pass both WALK_GET and WALK_MORE to should_follow_link()
namei: invert WALK_PUT logics
namei: shift interpretation of LOOKUP_FOLLOW inside should_follow_link()
namei: saner calling conventions for mountpoint_last()
namei.c: get rid of user_path_parent()
switch getfrag callbacks to ..._full() primitives
make skb_add_data,{_nocache}() and skb_copy_to_page_nocache() advance only on success
[iov_iter] new primitives - copy_from_iter_full() and friends
don't open-code file_inode()
ceph: switch to use of ->d_init()
ceph: unify dentry_operations instances
lustre: switch to use of ->d_init()
In UDP RX handler, we currently clear skb->dev before skb
is added to receive queue, because device pointer is no longer
available once we exit from RCU section.
Since this first cache line is always hot, lets reuse this space
to store skb->truesize and thus avoid a cache line miss at
udp_recvmsg()/udp_skb_destructor time while receive queue
spinlock is held.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Under UDP flood, many softirq producers try to add packets to
UDP receive queue, and one user thread is burning one cpu trying
to dequeue packets as fast as possible.
Two parts of the per packet cost are :
- copying payload from kernel space to user space,
- freeing memory pieces associated with skb.
If socket is under pressure, softirq handler(s) can try to pull in
skb->head the payload of the packet if it fits.
Meaning the softirq handler(s) can free/reuse the page fragment
immediately, instead of letting udp_recvmsg() do this hundreds of usec
later, possibly from another node.
Additional gains :
- We reduce skb->truesize and thus can store more packets per SO_RCVBUF
- We avoid cache line misses at copyout() time and consume_skb() time,
and avoid one put_page() with potential alien freeing on NUMA hosts.
This comes at the cost of a copy, bounded to available tail room, which
is usually small. (We might have to fix GRO_MAX_HEAD which looks bigger
than necessary)
This patch gave me about 5 % increase in throughput in my tests.
skb_condense() helper could probably used in other contexts.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
1) cast to "int" is unnecessary:
u8 will be promoted to int before decrementing,
small positive numbers fit into "int", so their values won't be changed
during promotion.
Once everything is int including loop counters, signedness doesn't
matter: 32-bit operations will stay 32-bit operations.
But! Someone tried to make this loop smart by making everything of
the same type apparently in an attempt to optimise it.
Do the optimization, just differently.
Do the cast where it matters. :^)
2) frag size is unsigned entity and sum of fragments sizes is also
unsigned.
Make everything unsigned, leave no MOVSX instruction behind.
add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-4 (-4)
function old new delta
skb_cow_data 835 834 -1
ip_do_fragment 2549 2548 -1
ip6_fragment 3130 3128 -2
Total: Before=154865032, After=154865028, chg -0.00%
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A new argument is added to __skb_recv_datagram to provide
an explicit skb destructor, invoked under the receive queue
lock.
The UDP protocol uses such argument to perform memory
reclaiming on dequeue, so that the UDP protocol does not
set anymore skb->desctructor.
Instead explicit memory reclaiming is performed at close() time and
when skbs are removed from the receive queue.
The in kernel UDP protocol users now need to call a
skb_recv_udp() variant instead of skb_recv_datagram() to
properly perform memory accounting on dequeue.
Overall, this allows acquiring only once the receive queue
lock on dequeue.
Tested using pktgen with random src port, 64 bytes packet,
wire-speed on a 10G link as sender and udp_sink as the receiver,
using an l4 tuple rxhash to stress the contention, and one or more
udp_sink instances with reuseport.
nr sinks vanilla patched
1 440 560
3 2150 2300
6 3650 3800
9 4450 4600
12 6250 6450
v1 -> v2:
- do rmem and allocated memory scheduling under the receive lock
- do bulk scheduling in first_packet_length() and in udp_destruct_sock()
- avoid the typdef for the dequeue callback
Suggested-by: Eric Dumazet <edumazet@google.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Mostly simple overlapping changes.
For example, David Ahern's adjacency list revamp in 'net-next'
conflicted with an adjacency list traversal bug fix in 'net'.
Signed-off-by: David S. Miller <davem@davemloft.net>
The skbuff and sock structure both had missing parameter annotation
values.
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull VFS splice updates from Al Viro:
"There's a bunch of branches this cycle, both mine and from other folks
and I'd rather send pull requests separately.
This one is the conversion of ->splice_read() to ITER_PIPE iov_iter
(and introduction of such). Gets rid of a lot of code in fs/splice.c
and elsewhere; there will be followups, but these are for the next
cycle... Some pipe/splice-related cleanups from Miklos in the same
branch as well"
* 'work.splice_read' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
pipe: fix comment in pipe_buf_operations
pipe: add pipe_buf_steal() helper
pipe: add pipe_buf_confirm() helper
pipe: add pipe_buf_release() helper
pipe: add pipe_buf_get() helper
relay: simplify relay_file_read()
switch default_file_splice_read() to use of pipe-backed iov_iter
switch generic_file_splice_read() to use of ->read_iter()
new iov_iter flavour: pipe-backed
fuse_dev_splice_read(): switch to add_to_pipe()
skb_splice_bits(): get rid of callback
new helper: add_to_pipe()
splice: lift pipe_lock out of splice_to_pipe()
splice: switch get_iovec_page_array() to iov_iter
splice_to_pipe(): don't open-code wakeup_pipe_readers()
consistent treatment of EFAULT on O_DIRECT read/write
since pipe_lock is the outermost now, we don't need to drop/regain
socket locks around the call of splice_to_pipe() from skb_splice_bits(),
which kills the need to have a socket-specific callback; we can just
call splice_to_pipe() and be done with that.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
This exports the functionality of extracting the tag from the payload,
without moving next vlan tag into hw accel tag.
Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This work implements direct packet access for helpers and direct packet
write in a similar fashion as already available for XDP types via commits
4acf6c0b84 ("bpf: enable direct packet data write for xdp progs") and
6841de8b0d ("bpf: allow helpers access the packet directly"), and as a
complementary feature to the already available direct packet read for tc
(cls/act) programs.
For enabling this, we need to introduce two helpers, bpf_skb_pull_data()
and bpf_csum_update(). The first is generally needed for both, read and
write, because they would otherwise only be limited to the current linear
skb head. Usually, when the data_end test fails, programs just bail out,
or, in the direct read case, use bpf_skb_load_bytes() as an alternative
to overcome this limitation. If such data sits in non-linear parts, we
can just pull them in once with the new helper, retest and eventually
access them.
At the same time, this also makes sure the skb is uncloned, which is, of
course, a necessary condition for direct write. As this needs to be an
invariant for the write part only, the verifier detects writes and adds
a prologue that is calling bpf_skb_pull_data() to effectively unclone the
skb from the very beginning in case it is indeed cloned. The heuristic
makes use of a similar trick that was done in 233577a220 ("net: filter:
constify detection of pkt_type_offset"). This comes at zero cost for other
programs that do not use the direct write feature. Should a program use
this feature only sparsely and has read access for the most parts with,
for example, drop return codes, then such write action can be delegated
to a tail called program for mitigating this cost of potential uncloning
to a late point in time where it would have been paid similarly with the
bpf_skb_store_bytes() as well. Advantage of direct write is that the
writes are inlined whereas the helper cannot make any length assumptions
and thus needs to generate a call to memcpy() also for small sizes, as well
as cost of helper call itself with sanity checks are avoided. Plus, when
direct read is already used, we don't need to cache or perform rechecks
on the data boundaries (due to verifier invalidating previous checks for
helpers that change skb->data), so more complex programs using rewrites
can benefit from switching to direct read plus write.
For direct packet access to helpers, we save the otherwise needed copy into
a temp struct sitting on stack memory when use-case allows. Both facilities
are enabled via may_access_direct_pkt_data() in verifier. For now, we limit
this to map helpers and csum_diff, and can successively enable other helpers
where we find it makes sense. Helpers that definitely cannot be allowed for
this are those part of bpf_helper_changes_skb_data() since they can change
underlying data, and those that write into memory as this could happen for
packet typed args when still cloned. bpf_csum_update() helper accommodates
for the fact that we need to fixup checksum_complete when using direct write
instead of bpf_skb_store_bytes(), meaning the programs can use available
helpers like bpf_csum_diff(), and implement csum_add(), csum_sub(),
csum_block_add(), csum_block_sub() equivalents in eBPF together with the
new helper. A usage example will be provided for iproute2's examples/bpf/
directory.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Over the years, TCP BDP has increased by several orders of magnitude,
and some people are considering to reach the 2 Gbytes limit.
Even with current window scale limit of 14, ~1 Gbytes maps to ~740,000
MSS.
In presence of packet losses (or reorders), TCP stores incoming packets
into an out of order queue, and number of skbs sitting there waiting for
the missing packets to be received can be in the 10^5 range.
Most packets are appended to the tail of this queue, and when
packets can finally be transferred to receive queue, we scan the queue
from its head.
However, in presence of heavy losses, we might have to find an arbitrary
point in this queue, involving a linear scan for every incoming packet,
throwing away cpu caches.
This patch converts it to a RB tree, to get bounded latencies.
Yaogong wrote a preliminary patch about 2 years ago.
Eric did the rebase, added ofo_last_skb cache, polishing and tests.
Tested with network dropping between 1 and 10 % packets, with good
success (about 30 % increase of throughput in stress tests)
Next step would be to also use an RB tree for the write queue at sender
side ;)
Signed-off-by: Yaogong Wang <wygivan@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Acked-By: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
switchdev_port_fwd_mark_set() is used to set the 'offload_fwd_mark' of
port netdevs so that packets being flooded by the device won't be
flooded twice.
It works by assigning a unique identifier (the ifindex of the first
bridge port) to bridge ports sharing the same parent ID. This prevents
packets from being flooded twice by the same switch, but will flood
packets through bridge ports belonging to a different switch.
This method is problematic when stacked devices are taken into account,
such as VLANs. In such cases, a physical port netdev can have upper
devices being members in two different bridges, thus requiring two
different 'offload_fwd_mark's to be configured on the port netdev, which
is impossible.
The main problem is that packet and netdev marking is performed at the
physical netdev level, whereas flooding occurs between bridge ports,
which are not necessarily port netdevs.
Instead, packet and netdev marking should really be done in the bridge
driver with the switch driver only telling it which packets it already
forwarded. The bridge driver will mark such packets using the mark
assigned to the ingress bridge port and will prevent the packet from
being forwarded through any bridge port sharing the same mark (i.e.
having the same parent ID).
Remove the current switchdev 'offload_fwd_mark' implementation and
instead implement the proposed method. In addition, make rocker - the
sole user of the mark - use the proposed method.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This work adds a bpf_skb_change_tail() helper for tc BPF programs. The
basic idea is to expand or shrink the skb in a controlled manner. The
eBPF program can then rewrite the rest via helpers like bpf_skb_store_bytes(),
bpf_lX_csum_replace() and others rather than passing a raw buffer for
writing here.
bpf_skb_change_tail() is really a slow path helper and intended for
replies with f.e. ICMP control messages. Concept is similar to other
helpers like bpf_skb_change_proto() helper to keep the helper without
protocol specifics and let the BPF program mangle the remaining parts.
A flags field has been added and is reserved for now should we extend
the helper in future.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
bpf_skb_store_bytes() invocations above L2 header need BPF_F_RECOMPUTE_CSUM
flag for updates, so that CHECKSUM_COMPLETE will be fixed up along the way.
Where we ran into an issue with bpf_skb_store_bytes() is when we did a
single-byte update on the IPv6 hoplimit despite using BPF_F_RECOMPUTE_CSUM
flag; simple ping via ICMPv6 triggered a hw csum failure as a result. The
underlying issue has been tracked down to a buffer alignment issue.
Meaning, that csum_partial() computations via skb_postpull_rcsum() and
skb_postpush_rcsum() pair invoked had a wrong result since they operated on
an odd address for the hoplimit, while other computations were done on an
even address. This mix doesn't work as-is with skb_postpull_rcsum(),
skb_postpush_rcsum() pair as it always expects at least half-word alignment
of input buffers, which is normally the case. Thus, instead of these helpers
using csum_sub() and (implicitly) csum_add(), we need to use csum_block_sub(),
csum_block_add(), respectively. For unaligned offsets, they rotate the sum
to align it to a half-word boundary again, otherwise they work the same as
csum_sub() and csum_add().
Adding __skb_postpull_rcsum(), __skb_postpush_rcsum() variants that take the
offset as an input and adapting bpf_skb_store_bytes() to them fixes the hw
csum failures again. The skb_postpull_rcsum(), skb_postpush_rcsum() helpers
use a 0 constant for offset so that the compiler optimizes the offset & 1
test away and generates the same code as with csum_sub()/_add().
Fixes: 608cd71a9c ("tc: bpf: generalize pedit action")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/mellanox/mlx5/core/en.h
drivers/net/ethernet/mellanox/mlx5/core/en_main.c
drivers/net/usb/r8152.c
All three conflicts were overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
Suggested-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Similar to commit 9b368814b3 ("net: fix bridge multicast packet checksum validation")
we need to fixup the checksum for CHECKSUM_COMPLETE when
pushing skb on RX path. Otherwise we get similar splats.
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Tom Herbert <tom@herbertland.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
