Daniel Borkmann says:
====================
bpf 2021-12-08
We've added 12 non-merge commits during the last 22 day(s) which contain
a total of 29 files changed, 659 insertions(+), 80 deletions(-).
The main changes are:
1) Fix an off-by-two error in packet range markings and also add a batch of
new tests for coverage of these corner cases, from Maxim Mikityanskiy.
2) Fix a compilation issue on MIPS JIT for R10000 CPUs, from Johan Almbladh.
3) Fix two functional regressions and a build warning related to BTF kfunc
for modules, from Kumar Kartikeya Dwivedi.
4) Fix outdated code and docs regarding BPF's migrate_disable() use on non-
PREEMPT_RT kernels, from Sebastian Andrzej Siewior.
5) Add missing includes in order to be able to detangle cgroup vs bpf header
dependencies, from Jakub Kicinski.
6) Fix regression in BPF sockmap tests caused by missing detachment of progs
from sockets when they are removed from the map, from John Fastabend.
7) Fix a missing "no previous prototype" warning in x86 JIT caused by BPF
dispatcher, from Björn Töpel.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
bpf: Add selftests to cover packet access corner cases
bpf: Fix the off-by-two error in range markings
treewide: Add missing includes masked by cgroup -> bpf dependency
tools/resolve_btfids: Skip unresolved symbol warning for empty BTF sets
bpf: Fix bpf_check_mod_kfunc_call for built-in modules
bpf: Make CONFIG_DEBUG_INFO_BTF depend upon CONFIG_BPF_SYSCALL
mips, bpf: Fix reference to non-existing Kconfig symbol
bpf: Make sure bpf_disable_instrumentation() is safe vs preemption.
Documentation/locking/locktypes: Update migrate_disable() bits.
bpf, sockmap: Re-evaluate proto ops when psock is removed from sockmap
bpf, sockmap: Attach map progs to psock early for feature probes
bpf, x86: Fix "no previous prototype" warning
====================
Link: https://lore.kernel.org/r/20211208155125.11826-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
While preparing my patch series adding netns refcount tracking,
I spotted bugs in devlink_nl_cmd_reload()
Some error paths forgot to release a refcount on a netns.
To fix this, we can reduce the scope of get_net()/put_net()
section around the call to devlink_reload().
Fixes: ccdf07219d ("devlink: Add reload action option to devlink reload command")
Fixes: dc64cc7c63 ("devlink: Add devlink reload limit option")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Moshe Shemesh <moshe@mellanox.com>
Cc: Jacob Keller <jacob.e.keller@intel.com>
Cc: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Link: https://lore.kernel.org/r/20211205192822.1741045-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Each peer's endpoint contains a dst_cache entry that takes a reference
to another netdev. When the containing namespace exits, we take down the
socket and prevent future sockets from being created (by setting
creating_net to NULL), which removes that potential reference on the
netns. However, it doesn't release references to the netns that a netdev
cached in dst_cache might be taking, so the netns still might fail to
exit. Since the socket is gimped anyway, we can simply clear all the
dst_caches (by way of clearing the endpoint src), which will release all
references.
However, the current dst_cache_reset function only releases those
references lazily. But it turns out that all of our usages of
wg_socket_clear_peer_endpoint_src are called from contexts that are not
exactly high-speed or bottle-necked. For example, when there's
connection difficulty, or when userspace is reconfiguring the interface.
And in particular for this patch, when the netns is exiting. So for
those cases, it makes more sense to call dst_release immediately. For
that, we add a small helper function to dst_cache.
This patch also adds a test to netns.sh from Hangbin Liu to ensure this
doesn't regress.
Tested-by: Hangbin Liu <liuhangbin@gmail.com>
Reported-by: Xiumei Mu <xmu@redhat.com>
Cc: Toke Høiland-Jørgensen <toke@redhat.com>
Cc: Paolo Abeni <pabeni@redhat.com>
Fixes: 900575aa33 ("wireguard: device: avoid circular netns references")
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The kernel leaks memory when a `fib` rule is present in IPv6 nftables
firewall rules and a suppress_prefix rule is present in the IPv6 routing
rules (used by certain tools such as wg-quick). In such scenarios, every
incoming packet will leak an allocation in `ip6_dst_cache` slab cache.
After some hours of `bpftrace`-ing and source code reading, I tracked
down the issue to ca7a03c417 ("ipv6: do not free rt if
FIB_LOOKUP_NOREF is set on suppress rule").
The problem with that change is that the generic `args->flags` always have
`FIB_LOOKUP_NOREF` set[1][2] but the IPv6-specific flag
`RT6_LOOKUP_F_DST_NOREF` might not be, leading to `fib6_rule_suppress` not
decreasing the refcount when needed.
How to reproduce:
- Add the following nftables rule to a prerouting chain:
meta nfproto ipv6 fib saddr . mark . iif oif missing drop
This can be done with:
sudo nft create table inet test
sudo nft create chain inet test test_chain '{ type filter hook prerouting priority filter + 10; policy accept; }'
sudo nft add rule inet test test_chain meta nfproto ipv6 fib saddr . mark . iif oif missing drop
- Run:
sudo ip -6 rule add table main suppress_prefixlength 0
- Watch `sudo slabtop -o | grep ip6_dst_cache` to see memory usage increase
with every incoming ipv6 packet.
This patch exposes the protocol-specific flags to the protocol
specific `suppress` function, and check the protocol-specific `flags`
argument for RT6_LOOKUP_F_DST_NOREF instead of the generic
FIB_LOOKUP_NOREF when decreasing the refcount, like this.
[1]: ca7a03c417/net/ipv6/fib6_rules.c (L71)
[2]: ca7a03c417/net/ipv6/fib6_rules.c (L99)
Link: https://bugzilla.kernel.org/show_bug.cgi?id=215105
Fixes: ca7a03c417 ("ipv6: do not free rt if FIB_LOOKUP_NOREF is set on suppress rule")
Cc: stable@vger.kernel.org
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When IPv6 module gets initialized, but it's hitting an error in inet6_init()
where it then needs to undo all the prior initialization work, it also might
do a call to ndisc_cleanup() which then calls neigh_table_clear(). In there
is a missing timer cancellation of the table's managed_work item.
The kernel test robot explicitly triggered this error path and caused a UAF
crash similar to the below:
[...]
[ 28.833183][ C0] BUG: unable to handle page fault for address: f7a43288
[ 28.833973][ C0] #PF: supervisor write access in kernel mode
[ 28.834660][ C0] #PF: error_code(0x0002) - not-present page
[ 28.835319][ C0] *pde = 06b2c067 *pte = 00000000
[ 28.835853][ C0] Oops: 0002 [#1] PREEMPT
[ 28.836367][ C0] CPU: 0 PID: 303 Comm: sed Not tainted 5.16.0-rc1-00233-g83ff5faa0d3b #7
[ 28.837293][ C0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-1 04/01/2014
[ 28.838338][ C0] EIP: __run_timers.constprop.0+0x82/0x440
[...]
[ 28.845607][ C0] Call Trace:
[ 28.845942][ C0] <SOFTIRQ>
[ 28.846333][ C0] ? check_preemption_disabled.isra.0+0x2a/0x80
[ 28.846975][ C0] ? __this_cpu_preempt_check+0x8/0xa
[ 28.847570][ C0] run_timer_softirq+0xd/0x40
[ 28.848050][ C0] __do_softirq+0xf5/0x576
[ 28.848547][ C0] ? __softirqentry_text_start+0x10/0x10
[ 28.849127][ C0] do_softirq_own_stack+0x2b/0x40
[ 28.849749][ C0] </SOFTIRQ>
[ 28.850087][ C0] irq_exit_rcu+0x7d/0xc0
[ 28.850587][ C0] common_interrupt+0x2a/0x40
[ 28.851068][ C0] asm_common_interrupt+0x119/0x120
[...]
Note that IPv6 module cannot be unloaded as per 8ce4406103 ("ipv6: do not
allow ipv6 module to be removed") hence this can only be seen during module
initialization error. Tested with kernel test robot's reproducer.
Fixes: 7482e3841d ("net, neigh: Add NTF_MANAGED flag for managed neighbor entries")
Reported-by: kernel test robot <oliver.sang@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Li Zhijian <zhijianx.li@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a sock is added to a sock map we evaluate what proto op hooks need to
be used. However, when the program is removed from the sock map we have not
been evaluating if that changes the required program layout.
Before the patch listed in the 'fixes' tag this was not causing failures
because the base program set handles all cases. Specifically, the case with
a stream parser and the case with out a stream parser are both handled. With
the fix below we identified a race when running with a proto op that attempts
to read skbs off both the stream parser and the skb->receive_queue. Namely,
that a race existed where when the stream parser is empty checking the
skb->receive_queue from recvmsg at the precies moment when the parser is
paused and the receive_queue is not empty could result in skipping the stream
parser. This may break a RX policy depending on the parser to run.
The fix tag then loads a specific proto ops that resolved this race. But, we
missed removing that proto ops recv hook when the sock is removed from the
sockmap. The result is the stream parser is stopped so no more skbs will be
aggregated there, but the hook and BPF program continues to be attached on
the psock. User space will then get an EBUSY when trying to read the socket
because the recvmsg() handler is now waiting on a stopped stream parser.
To fix we rerun the proto ops init() function which will look at the new set
of progs attached to the psock and rest the proto ops hook to the correct
handlers. And in the above case where we remove the sock from the sock map
the RX prog will no longer be listed so the proto ops is removed.
Fixes: c5d2177a72 ("bpf, sockmap: Fix race in ingress receive verdict with redirect to self")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20211119181418.353932-3-john.fastabend@gmail.com
When a TCP socket is added to a sock map we look at the programs attached
to the map to determine what proto op hooks need to be changed. Before
the patch in the 'fixes' tag there were only two categories -- the empty
set of programs or a TX policy. In any case the base set handled the
receive case.
After the fix we have an optimized program for receive that closes a small,
but possible, race on receive. This program is loaded only when the map the
psock is being added to includes a RX policy. Otherwise, the race is not
possible so we don't need to handle the race condition.
In order for the call to sk_psock_init() to correctly evaluate the above
conditions all progs need to be set in the psock before the call. However,
in the current code this is not the case. We end up evaluating the
requirements on the old prog state. If your psock is attached to multiple
maps -- for example a tx map and rx map -- then the second update would pull
in the correct maps. But, the other pattern with a single rx enabled map
the correct receive hooks are not used. The result is the race fixed by the
patch in the fixes tag below may still be seen in this case.
To fix we simply set all psock->progs before doing the call into
sock_map_init(). With this the init() call gets the full list of programs
and chooses the correct proto ops on the first iteration instead of
requiring the second update to pull them in. This fixes the race case when
only a single map is used.
Fixes: c5d2177a72 ("bpf, sockmap: Fix race in ingress receive verdict with redirect to self")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20211119181418.353932-2-john.fastabend@gmail.com
The mlxsw driver calls to various devlink flash routines even before
users can get any access to the devlink instance itself. For example,
mlxsw_core_fw_rev_validate() one of such functions.
__mlxsw_core_bus_device_register
-> mlxsw_core_fw_rev_validate
-> mlxsw_core_fw_flash
-> mlxfw_firmware_flash
-> mlxfw_status_notify
-> devlink_flash_update_status_notify
-> __devlink_flash_update_notify
-> WARN_ON(...)
It causes to the WARN_ON to trigger warning about devlink not registered.
Fixes: cf53021740 ("devlink: Notify users when objects are accessible")
Reported-by: Danielle Ratson <danieller@nvidia.com>
Tested-by: Danielle Ratson <danieller@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit d00e60ee54.
As reported by Guillaume in [1]:
Enabling LPAE always enables CONFIG_ARCH_DMA_ADDR_T_64BIT
in 32-bit systems, which breaks the bootup proceess when a
ethernet driver is using page pool with PP_FLAG_DMA_MAP flag.
As we were hoping we had no active consumers for such system
when we removed the dma mapping support, and LPAE seems like
a common feature for 32 bits system, so revert it.
1. https://www.spinics.net/lists/netdev/msg779890.html
Fixes: d00e60ee54 ("page_pool: disable dma mapping support for 32-bit arch with 64-bit DMA")
Signed-off-by: Yunsheng Lin <linyunsheng@huawei.com>
Reported-by: "kernelci.org bot" <bot@kernelci.org>
Tested-by: "kernelci.org bot" <bot@kernelci.org>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann says:
====================
pull-request: bpf 2021-11-16
We've added 12 non-merge commits during the last 5 day(s) which contain
a total of 23 files changed, 573 insertions(+), 73 deletions(-).
The main changes are:
1) Fix pruning regression where verifier went overly conservative rejecting
previsouly accepted programs, from Alexei Starovoitov and Lorenz Bauer.
2) Fix verifier TOCTOU bug when using read-only map's values as constant
scalars during verification, from Daniel Borkmann.
3) Fix a crash due to a double free in XSK's buffer pool, from Magnus Karlsson.
4) Fix libbpf regression when cross-building runqslower, from Jean-Philippe Brucker.
5) Forbid use of bpf_ktime_get_coarse_ns() and bpf_timer_*() helpers in tracing
programs due to deadlock possibilities, from Dmitrii Banshchikov.
6) Fix checksum validation in sockmap's udp_read_sock() callback, from Cong Wang.
7) Various BPF sample fixes such as XDP stats in xdp_sample_user, from Alexander Lobakin.
8) Fix libbpf gen_loader error handling wrt fd cleanup, from Kumar Kartikeya Dwivedi.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
udp: Validate checksum in udp_read_sock()
bpf: Fix toctou on read-only map's constant scalar tracking
samples/bpf: Fix build error due to -isystem removal
selftests/bpf: Add tests for restricted helpers
bpf: Forbid bpf_ktime_get_coarse_ns and bpf_timer_* in tracing progs
libbpf: Perform map fd cleanup for gen_loader in case of error
samples/bpf: Fix incorrect use of strlen in xdp_redirect_cpu
tools/runqslower: Fix cross-build
samples/bpf: Fix summary per-sec stats in xdp_sample_user
selftests/bpf: Check map in map pruning
bpf: Fix inner map state pruning regression.
xsk: Fix crash on double free in buffer pool
====================
Link: https://lore.kernel.org/r/20211116141134.6490-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
sk_clone_lock() needs to call sock_inuse_add(1) before entering the
sk_free_unlock_clone() error path, for __sk_free() from sk_free() from
sk_free_unlock_clone() calls sock_inuse_add(-1).
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Fixes: 648845ab7e ("sock: Move the socket inuse to namespace.")
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 719c571970 ("net: make napi_disable() symmetric with
enable") accidentally introduced a bug sometimes leading to a kernel
BUG when bringing an iface up/down under heavy traffic load.
Prior to this commit, napi_disable() was polling n->state until
none of (NAPIF_STATE_SCHED | NAPIF_STATE_NPSVC) is set and then
always flip them. Now there's a possibility to get away with the
NAPIF_STATE_SCHE unset as 'continue' drops us to the cmpxchg()
call with an uninitialized variable, rather than straight to
another round of the state check.
Error path looks like:
napi_disable():
unsigned long val, new; /* new is uninitialized */
do {
val = READ_ONCE(n->state); /* NAPIF_STATE_NPSVC and/or
NAPIF_STATE_SCHED is set */
if (val & (NAPIF_STATE_SCHED | NAPIF_STATE_NPSVC)) { /* true */
usleep_range(20, 200);
continue; /* go straight to the condition check */
}
new = val | <...>
} while (cmpxchg(&n->state, val, new) != val); /* state == val, cmpxchg()
writes garbage */
napi_enable():
do {
val = READ_ONCE(n->state);
BUG_ON(!test_bit(NAPI_STATE_SCHED, &val)); /* 50/50 boom */
<...>
while the typical BUG splat is like:
[ 172.652461] ------------[ cut here ]------------
[ 172.652462] kernel BUG at net/core/dev.c:6937!
[ 172.656914] invalid opcode: 0000 [#1] PREEMPT SMP PTI
[ 172.661966] CPU: 36 PID: 2829 Comm: xdp_redirect_cp Tainted: G I 5.15.0 #42
[ 172.670222] Hardware name: Intel Corporation S2600WFT/S2600WFT, BIOS SE5C620.86B.02.01.0014.082620210524 08/26/2021
[ 172.680646] RIP: 0010:napi_enable+0x5a/0xd0
[ 172.684832] Code: 07 49 81 cc 00 01 00 00 4c 89 e2 48 89 d8 80 e6 fb f0 48 0f b1 55 10 48 39 c3 74 10 48 8b 5d 10 f6 c7 04 75 3d f6 c3 01 75 b4 <0f> 0b 5b 5d 41 5c c3 65 ff 05 b8 e5 61 53 48 c7 c6 c0 f3 34 ad 48
[ 172.703578] RSP: 0018:ffffa3c9497477a8 EFLAGS: 00010246
[ 172.708803] RAX: ffffa3c96615a014 RBX: 0000000000000000 RCX: ffff8a4b575301a0
< snip >
[ 172.782403] Call Trace:
[ 172.784857] <TASK>
[ 172.786963] ice_up_complete+0x6f/0x210 [ice]
[ 172.791349] ice_xdp+0x136/0x320 [ice]
[ 172.795108] ? ice_change_mtu+0x180/0x180 [ice]
[ 172.799648] dev_xdp_install+0x61/0xe0
[ 172.803401] dev_xdp_attach+0x1e0/0x550
[ 172.807240] dev_change_xdp_fd+0x1e6/0x220
[ 172.811338] do_setlink+0xee8/0x1010
[ 172.814917] rtnl_setlink+0xe5/0x170
[ 172.818499] ? bpf_lsm_binder_set_context_mgr+0x10/0x10
[ 172.823732] ? security_capable+0x36/0x50
< snip >
Fix this by replacing 'do { } while (cmpxchg())' with an "infinite"
for-loop with an explicit break.
From v1 [0]:
- just use a for-loop to simplify both the fix and the existing
code (Eric).
[0] https://lore.kernel.org/netdev/20211110191126.1214-1-alexandr.lobakin@intel.com
Fixes: 719c571970 ("net: make napi_disable() symmetric with enable")
Suggested-by: Eric Dumazet <edumazet@google.com> # for-loop
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20211110195605.1304-1-alexandr.lobakin@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The current conversion of skb->data_end reads like this:
; data_end = (void*)(long)skb->data_end;
559: (79) r1 = *(u64 *)(r2 +200) ; r1 = skb->data
560: (61) r11 = *(u32 *)(r2 +112) ; r11 = skb->len
561: (0f) r1 += r11
562: (61) r11 = *(u32 *)(r2 +116)
563: (1f) r1 -= r11
But similar to the case in 84f44df664 ("bpf: sock_ops sk access may stomp
registers when dst_reg = src_reg"), the code will read an incorrect skb->len
when src == dst. In this case we end up generating this xlated code:
; data_end = (void*)(long)skb->data_end;
559: (79) r1 = *(u64 *)(r1 +200) ; r1 = skb->data
560: (61) r11 = *(u32 *)(r1 +112) ; r11 = (skb->data)->len
561: (0f) r1 += r11
562: (61) r11 = *(u32 *)(r1 +116)
563: (1f) r1 -= r11
... where line 560 is the reading 4B of (skb->data + 112) instead of the
intended skb->len Here the skb pointer in r1 gets set to skb->data and the
later deref for skb->len ends up following skb->data instead of skb.
This fixes the issue similarly to the patch mentioned above by creating an
additional temporary variable and using to store the register when dst_reg =
src_reg. We name the variable bpf_temp_reg and place it in the cb context for
sk_skb. Then we restore from the temp to ensure nothing is lost.
Fixes: 16137b09a6 ("bpf: Compute data_end dynamically with JIT code")
Signed-off-by: Jussi Maki <joamaki@gmail.com>
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Link: https://lore.kernel.org/bpf/20211103204736.248403-6-john.fastabend@gmail.com
Strparser is reusing the qdisc_skb_cb struct to stash the skb message handling
progress, e.g. offset and length of the skb. First this is poorly named and
inherits a struct from qdisc that doesn't reflect the actual usage of cb[] at
this layer.
But, more importantly strparser is using the following to access its metadata.
(struct _strp_msg *)((void *)skb->cb + offsetof(struct qdisc_skb_cb, data))
Where _strp_msg is defined as:
struct _strp_msg {
struct strp_msg strp; /* 0 8 */
int accum_len; /* 8 4 */
/* size: 12, cachelines: 1, members: 2 */
/* last cacheline: 12 bytes */
};
So we use 12 bytes of ->data[] in struct. However in BPF code running parser
and verdict the user has read capabilities into the data[] array as well. Its
not too problematic, but we should not be exposing internal state to BPF
program. If its really needed then we can use the probe_read() APIs which allow
reading kernel memory. And I don't believe cb[] layer poses any API breakage by
moving this around because programs can't depend on cb[] across layers.
In order to fix another issue with a ctx rewrite we need to stash a temp
variable somewhere. To make this work cleanly this patch builds a cb struct
for sk_skb types called sk_skb_cb struct. Then we can use this consistently
in the strparser, sockmap space. Additionally we can start allowing ->cb[]
write access after this.
Fixes: 604326b41a ("bpf, sockmap: convert to generic sk_msg interface")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Jussi Maki <joamaki@gmail.com>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Link: https://lore.kernel.org/bpf/20211103204736.248403-5-john.fastabend@gmail.com
In order to fix an issue with sockets in TCP sockmap redirect cases we plan
to allow CLOSE state sockets to exist in the sockmap. However, the check in
bpf_sk_lookup_assign() currently only invalidates sockets in the
TCP_ESTABLISHED case relying on the checks on sockmap insert to ensure we
never SOCK_CLOSE state sockets in the map.
To prepare for this change we flip the logic in bpf_sk_lookup_assign() to
explicitly test for the accepted cases. Namely, a tcp socket in TCP_LISTEN
or a udp socket in TCP_CLOSE state. This also makes the code more resilent
to future changes.
Suggested-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Link: https://lore.kernel.org/bpf/20211103204736.248403-2-john.fastabend@gmail.com
Fix following coccicheck warning:
./net/core/devlink.c:69:6-10: WARNING use flexible-array member instead
Signed-off-by: Guo Zhengkui <guozhengkui@vivo.com>
Link: https://lore.kernel.org/r/20211103121607.27490-1-guozhengkui@vivo.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Track skbs containing only zerocopy data and avoid charging them to
kernel memory to correctly account the memory utilization for
msg_zerocopy. All of the data in such skbs is held in user pages which
are already accounted to user. Before this change, they are charged
again in kernel in __zerocopy_sg_from_iter. The charging in kernel is
excessive because data is not being copied into skb frags. This
excessive charging can lead to kernel going into memory pressure
state which impacts all sockets in the system adversely. Mark pure
zerocopy skbs with a SKBFL_PURE_ZEROCOPY flag and remove
charge/uncharge for data in such skbs.
Initially, an skb is marked pure zerocopy when it is empty and in
zerocopy path. skb can then change from a pure zerocopy skb to mixed
data skb (zerocopy and copy data) if it is at tail of write queue and
there is room available in it and non-zerocopy data is being sent in
the next sendmsg call. At this time sk_mem_charge is done for the pure
zerocopied data and the pure zerocopy flag is unmarked. We found that
this happens very rarely on workloads that pass MSG_ZEROCOPY.
A pure zerocopy skb can later be coalesced into normal skb if they are
next to each other in queue but this patch prevents coalescing from
happening. This avoids complexity of charging when skb downgrades from
pure zerocopy to mixed. This is also rare.
In sk_wmem_free_skb, if it is a pure zerocopy skb, an sk_mem_uncharge
for SKB_TRUESIZE(skb_end_offset(skb)) is done for sk_mem_charge in
tcp_skb_entail for an skb without data.
Testing with the msg_zerocopy.c benchmark between two hosts(100G nics)
with zerocopy showed that before this patch the 'sock' variable in
memory.stat for cgroup2 that tracks sum of sk_forward_alloc,
sk_rmem_alloc and sk_wmem_queued is around 1822720 and with this
change it is 0. This is due to no charge to sk_forward_alloc for
zerocopy data and shows memory utilization for kernel is lowered.
With this commit we don't see the warning we saw in previous commit
which resulted in commit 84882cf72c.
Signed-off-by: Talal Ahmad <talalahmad@google.com>
Acked-by: Arjun Roy <arjunroy@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While commit 097b9146c0 ("net: fix up truesize of cloned
skb in skb_prepare_for_shift()") fixed immediate issues found
when KFENCE was enabled/tested, there are still similar issues,
when tcp_trim_head() hits KFENCE while the master skb
is cloned.
This happens under heavy networking TX workloads,
when the TX completion might be delayed after incoming ACK.
This patch fixes the WARNING in sk_stream_kill_queues
when sk->sk_mem_queued/sk->sk_forward_alloc are not zero.
Fixes: d3fb45f370 ("mm, kfence: insert KFENCE hooks for SLAB")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Marco Elver <elver@google.com>
Link: https://lore.kernel.org/r/20211102004555.1359210-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Alexei Starovoitov says:
====================
pull-request: bpf-next 2021-11-01
We've added 181 non-merge commits during the last 28 day(s) which contain
a total of 280 files changed, 11791 insertions(+), 5879 deletions(-).
The main changes are:
1) Fix bpf verifier propagation of 64-bit bounds, from Alexei.
2) Parallelize bpf test_progs, from Yucong and Andrii.
3) Deprecate various libbpf apis including af_xdp, from Andrii, Hengqi, Magnus.
4) Improve bpf selftests on s390, from Ilya.
5) bloomfilter bpf map type, from Joanne.
6) Big improvements to JIT tests especially on Mips, from Johan.
7) Support kernel module function calls from bpf, from Kumar.
8) Support typeless and weak ksym in light skeleton, from Kumar.
9) Disallow unprivileged bpf by default, from Pawan.
10) BTF_KIND_DECL_TAG support, from Yonghong.
11) Various bpftool cleanups, from Quentin.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: (181 commits)
libbpf: Deprecate AF_XDP support
kbuild: Unify options for BTF generation for vmlinux and modules
selftests/bpf: Add a testcase for 64-bit bounds propagation issue.
bpf: Fix propagation of signed bounds from 64-bit min/max into 32-bit.
bpf: Fix propagation of bounds from 64-bit min/max into 32-bit and var_off.
selftests/bpf: Fix also no-alu32 strobemeta selftest
bpf: Add missing map_delete_elem method to bloom filter map
selftests/bpf: Add bloom map success test for userspace calls
bpf: Add alignment padding for "map_extra" + consolidate holes
bpf: Bloom filter map naming fixups
selftests/bpf: Add test cases for struct_ops prog
bpf: Add dummy BPF STRUCT_OPS for test purpose
bpf: Factor out helpers for ctx access checking
bpf: Factor out a helper to prepare trampoline for struct_ops prog
selftests, bpf: Fix broken riscv build
riscv, libbpf: Add RISC-V (RV64) support to bpf_tracing.h
tools, build: Add RISC-V to HOSTARCH parsing
riscv, bpf: Increase the maximum number of iterations
selftests, bpf: Add one test for sockmap with strparser
selftests, bpf: Fix test_txmsg_ingress_parser error
...
====================
Link: https://lore.kernel.org/r/20211102013123.9005-1-alexei.starovoitov@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Track skbs with only zerocopy data and avoid charging them to kernel
memory to correctly account the memory utilization for msg_zerocopy.
All of the data in such skbs is held in user pages which are already
accounted to user. Before this change, they are charged again in
kernel in __zerocopy_sg_from_iter. The charging in kernel is
excessive because data is not being copied into skb frags. This
excessive charging can lead to kernel going into memory pressure
state which impacts all sockets in the system adversely. Mark pure
zerocopy skbs with a SKBFL_PURE_ZEROCOPY flag and remove
charge/uncharge for data in such skbs.
Initially, an skb is marked pure zerocopy when it is empty and in
zerocopy path. skb can then change from a pure zerocopy skb to mixed
data skb (zerocopy and copy data) if it is at tail of write queue and
there is room available in it and non-zerocopy data is being sent in
the next sendmsg call. At this time sk_mem_charge is done for the pure
zerocopied data and the pure zerocopy flag is unmarked. We found that
this happens very rarely on workloads that pass MSG_ZEROCOPY.
A pure zerocopy skb can later be coalesced into normal skb if they are
next to each other in queue but this patch prevents coalescing from
happening. This avoids complexity of charging when skb downgrades from
pure zerocopy to mixed. This is also rare.
In sk_wmem_free_skb, if it is a pure zerocopy skb, an sk_mem_uncharge
for SKB_TRUESIZE(MAX_TCP_HEADER) is done for sk_mem_charge in
tcp_skb_entail for an skb without data.
Testing with the msg_zerocopy.c benchmark between two hosts(100G nics)
with zerocopy showed that before this patch the 'sock' variable in
memory.stat for cgroup2 that tracks sum of sk_forward_alloc,
sk_rmem_alloc and sk_wmem_queued is around 1822720 and with this
change it is 0. This is due to no charge to sk_forward_alloc for
zerocopy data and shows memory utilization for kernel is lowered.
Signed-off-by: Talal Ahmad <talalahmad@google.com>
Acked-by: Arjun Roy <arjunroy@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
If sockmap enable strparser, there are lose offset info in
sk_psock_skb_ingress(). If the length determined by parse_msg function is not
skb->len, the skb will be converted to sk_msg multiple times, and userspace
app will get the data multiple times.
Fix this by get the offset and length from strp_msg. And as Cong suggested,
add one bit in skb->_sk_redir to distinguish enable or disable strparser.
Fixes: 604326b41a ("bpf, sockmap: convert to generic sk_msg interface")
Signed-off-by: Liu Jian <liujian56@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Cong Wang <cong.wang@bytedance.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20211029141216.211899-1-liujian56@huawei.com
devlink compat code needs to drop rtnl_lock to take
devlink->lock to ensure correct lock ordering.
This is problematic because we're not strictly guaranteed
that the netdev will not disappear after we re-lock.
It may open a possibility of nested ->begin / ->complete
calls.
Instead of calling into devlink under rtnl_lock take
a ref on the devlink instance and make the call after
we've dropped rtnl_lock.
We (continue to) assume that netdevs have an implicit
reference on the devlink returned from ndo_get_devlink_port
Note that ndo_get_devlink_port will now get called
under rtnl_lock. That should be fine since none of
the drivers seem to be taking serious locks inside
ndo_get_devlink_port.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Allow those who hold implicit reference on a devlink instance
to try to take a full ref on it. This will be used from netdev
code which has an implicit ref because of driver call ordering.
Note that after recent changes devlink_unregister() may happen
before netdev unregister, but devlink_free() should still happen
after, so we are safe to try, but we can't just refcount_inc()
and assume it's not zero.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Don't take the lock in net/core/dev_ioctl.c,
we'll have things to do outside rtnl_lock soon.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a packet of a new flow arrives in openvswitch kernel module, it dissects
the packet and passes the extracted flow key to ovs-vswtichd daemon. If hw-
offload configuration is enabled, the daemon creates a new TC flower entry to
bypass openvswitch kernel module for the flow (TC flower can also offload flows
to NICs but this time that does not matter).
In this processing flow, I found the following issue in cases of GRE/IPIP
packets.
When ovs_flow_key_extract() in openvswitch module parses a packet of a new
GRE (or IPIP) flow received on non-tunneling vports, it extracts information
of the outer IP header for ip_proto/src_ip/dst_ip match keys.
This means ovs-vswitchd creates a TC flower entry with IP protocol/addresses
match keys whose values are those of the outer IP header. OTOH, TC flower,
which uses flow_dissector (different parser from openvswitch module), extracts
information of the inner IP header.
The following flow is an example to describe the issue in more detail.
<----------- Outer IP -----------------> <---------- Inner IP ---------->
+----------+--------------+--------------+----------+----------+----------+
| ip_proto | src_ip | dst_ip | ip_proto | src_ip | dst_ip |
| 47 (GRE) | 192.168.10.1 | 192.168.10.2 | 6 (TCP) | 10.0.0.1 | 10.0.0.2 |
+----------+--------------+--------------+----------+----------+----------+
In this case, TC flower entry and extracted information are shown as below:
- ovs-vswitchd creates TC flower entry with:
- ip_proto: 47
- src_ip: 192.168.10.1
- dst_ip: 192.168.10.2
- TC flower extracts below for IP header matches:
- ip_proto: 6
- src_ip: 10.0.0.1
- dst_ip: 10.0.0.2
Thus, GRE or IPIP packets never match the TC flower entry, as each
dissector behaves differently.
IMHO, the behavior of TC flower (flow dissector) does not look correct,
as ip_proto/src_ip/dst_ip in TC flower match means the outermost IP
header information except for GRE/IPIP cases. This patch adds a new
flow_dissector flag FLOW_DISSECTOR_F_STOP_BEFORE_ENCAP which skips
dissection of the encapsulated inner GRE/IPIP header in TC flower
classifier.
Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
devlink_alloc() and devlink_register() are both GPL.
A non-GPL module won't get far, so for consistency
we can make all symbols GPL without risking any real
life breakage.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change adds a new skb extension for MCTP, to represent a
request/response flow.
The intention is to use this in a later change to allow i2c controllers
to correctly configure a multiplexer over a flow.
Since we have a cleanup function in the core path (if an extension is
present), we'll need to make CONFIG_MCTP a bool, rather than a tristate.
Includes a fix for a build warning with clang:
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Reduce extra indirection from devlink_params_*() API. Such change
makes it clear that we can drop devlink->lock from these flows, because
everything is executed when the devlink is not registered yet.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There is a warning in xdp_rxq_info_unreg_mem_model() when reg_state isn't
equal to REG_STATE_REGISTERED, so the warning in xdp_rxq_info_unreg() is
redundant.
Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Link: https://lore.kernel.org/r/20211027013856.1866-1-yajun.deng@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This reverts commit 22849b5ea5 as it
revealed that mlxsw and netdevsim (copy/paste from mlxsw) reregisters
devlink objects during another devlink user triggered command.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This reverts commit 8bbeed4858 as it
revealed that mlxsw and netdevsim (copy/paste from mlxsw) reregisters
devlink objects during another devlink user triggered command.
Fixes: 22849b5ea5 ("devlink: Remove not-executed trap policer notifications")
Reported-by: syzbot+93d5accfaefceedf43c1@syzkaller.appspotmail.com
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
LRO and HW-GRO are mutually exclusive, this commit adds this restriction
in netdev_fix_feature. HW-GRO is preferred, that means in case both
HW-GRO and LRO features are requested, LRO is cleared.
Signed-off-by: Ben Ben-ishay <benishay@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Daniel Borkmann says:
====================
pull-request: bpf 2021-10-26
We've added 12 non-merge commits during the last 7 day(s) which contain
a total of 23 files changed, 118 insertions(+), 98 deletions(-).
The main changes are:
1) Fix potential race window in BPF tail call compatibility check, from Toke Høiland-Jørgensen.
2) Fix memory leak in cgroup fs due to missing cgroup_bpf_offline(), from Quanyang Wang.
3) Fix file descriptor reference counting in generic_map_update_batch(), from Xu Kuohai.
4) Fix bpf_jit_limit knob to the max supported limit by the arch's JIT, from Lorenz Bauer.
5) Fix BPF sockmap ->poll callbacks for UDP and AF_UNIX sockets, from Cong Wang and Yucong Sun.
6) Fix BPF sockmap concurrency issue in TCP on non-blocking sendmsg calls, from Liu Jian.
7) Fix build failure of INODE_STORAGE and TASK_STORAGE maps on !CONFIG_NET, from Tejun Heo.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
bpf: Fix potential race in tail call compatibility check
bpf: Move BPF_MAP_TYPE for INODE_STORAGE and TASK_STORAGE outside of CONFIG_NET
selftests/bpf: Use recv_timeout() instead of retries
net: Implement ->sock_is_readable() for UDP and AF_UNIX
skmsg: Extract and reuse sk_msg_is_readable()
net: Rename ->stream_memory_read to ->sock_is_readable
tcp_bpf: Fix one concurrency problem in the tcp_bpf_send_verdict function
cgroup: Fix memory leak caused by missing cgroup_bpf_offline
bpf: Fix error usage of map_fd and fdget() in generic_map_update_batch()
bpf: Prevent increasing bpf_jit_limit above max
bpf: Define bpf_jit_alloc_exec_limit for arm64 JIT
bpf: Define bpf_jit_alloc_exec_limit for riscv JIT
====================
Link: https://lore.kernel.org/r/20211026201920.11296-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
During a testing of an user-space application which transmits UDP
multicast datagrams and utilizes multicast routing to send the UDP
datagrams out of defined network interfaces, I've found a multicast
router does not fill-in UDP checksum into locally produced, looped-back
and forwarded UDP datagrams, if an original output NIC the datagrams
are sent to has UDP TX checksum offload enabled.
The datagrams are sent malformed out of the NIC the datagrams have been
forwarded to.
It is because:
1. If TX checksum offload is enabled on the output NIC, UDP checksum
is not calculated by kernel and is not filled into skb data.
2. dev_loopback_xmit(), which is called solely by
ip_mc_finish_output(), sets skb->ip_summed = CHECKSUM_UNNECESSARY
unconditionally.
3. Since 35fc92a9 ("[NET]: Allow forwarding of ip_summed except
CHECKSUM_COMPLETE"), the ip_summed value is preserved during
forwarding.
4. If ip_summed != CHECKSUM_PARTIAL, checksum is not calculated during
a packet egress.
The minimum fix in dev_loopback_xmit():
1. Preserves skb->ip_summed CHECKSUM_PARTIAL. This is the
case when the original output NIC has TX checksum offload enabled.
The effects are:
a) If the forwarding destination interface supports TX checksum
offloading, the NIC driver is responsible to fill-in the
checksum.
b) If the forwarding destination interface does NOT support TX
checksum offloading, checksums are filled-in by kernel before
skb is submitted to the NIC driver.
c) For local delivery, checksum validation is skipped as in the
case of CHECKSUM_UNNECESSARY, thanks to skb_csum_unnecessary().
2. Translates ip_summed CHECKSUM_NONE to CHECKSUM_UNNECESSARY. It
means, for CHECKSUM_NONE, the behavior is unmodified and is there
to skip a looped-back packet local delivery checksum validation.
Signed-off-by: Cyril Strejc <cyril.strejc@skoda.cz>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently in net_ns_get_ownership() it may not be able to set uid or gid
if make_kuid or make_kgid returns an invalid value, and an uninit-value
issue can be triggered by this.
This patch is to fix it by initializing the uid and gid before calling
net_ns_get_ownership(), as it does in kobject_get_ownership()
Fixes: e6dee9f389 ("net-sysfs: add netdev_change_owner()")
Reported-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Christian Brauner <christian.brauner@ubuntu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Drivers call netdev_set_num_tc() and then netdev_set_tc_queue()
to set the queue count and offset for each TC. So the queue count
and offset for the TCs may be zero for a short period after dev->num_tc
has been set. If a TX packet is being transmitted at this time in the
code path netdev_pick_tx() -> skb_tx_hash(), skb_tx_hash() may see
nonzero dev->num_tc but zero qcount for the TC. The while loop that
keeps looping while hash >= qcount will not end.
Fix it by checking the TC's qcount to be nonzero before using it.
Fixes: eadec877ce ("net: Add support for subordinate traffic classes to netdev_pick_tx")
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The parameters are registered before devlink_register() and all the
notifications are delayed. This patch removes not-possible parameters
notifications along with addition of code annotation logic.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The trap logic is registered before devlink_register() and all the
notifications are delayed. This patch removes not-possible trap group
notifications along with addition of code annotation logic.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The trap policer logic is registered before devlink_register() and all the
notifications are delayed. This patch removes not-possible notifications
along with addition of code annotation logic.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
