We've added 7 non-merge commits during the last 3 day(s) which contain
a total of 10 files changed, 174 insertions(+), 48 deletions(-).
The main changes are:
1) Various sockmap fixes, from John and Jussi.
2) Fix out-of-bound issue with bpf_pseudo_func, from Martin.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
bpf, sockmap: sk_skb data_end access incorrect when src_reg = dst_reg
bpf: sockmap, strparser, and tls are reusing qdisc_skb_cb and colliding
bpf, sockmap: Fix race in ingress receive verdict with redirect to self
bpf, sockmap: Remove unhash handler for BPF sockmap usage
bpf, sockmap: Use stricter sk state checks in sk_lookup_assign
bpf: selftest: Trigger a DCE on the whole subprog
bpf: Stop caching subprog index in the bpf_pseudo_func insn
====================
Arnd Bergmann [Mon, 8 Nov 2021 11:12:24 +0000 (12:12 +0100)]
amt: add IPV6 Kconfig dependency
This driver cannot be built-in if IPV6 is a loadable module:
x86_64-linux-ld: drivers/net/amt.o: in function `amt_build_mld_gq':
amt.c:(.text+0x2e7d): undefined reference to `ipv6_dev_get_saddr'
Add the idiomatic Kconfig dependency that all such modules
have.
Fixes: c33242ef4b12 ("amt: add control plane of amt interface") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Taehee Yoo <ap420073@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Dan Carpenter [Tue, 9 Nov 2021 11:47:36 +0000 (14:47 +0300)]
gve: Fix off by one in gve_tx_timeout()
The priv->ntfy_blocks[] has "priv->num_ntfy_blks" elements so this >
needs to be >= to prevent an off by one bug. The priv->ntfy_blocks[]
array is allocated in gve_alloc_notify_blocks().
Fixes: bf9217f78b7a ("gve: Recover from queue stall due to missed IRQ") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: David S. Miller <davem@davemloft.net>
According to the patch "defer ax25 kfree after unregister_netdev", this
patch reorder the kfree after the unregister_netdev to avoid the possible
UAF as the unregister_netdev() is well synchronized and won't return if
there is a running routine.
Signed-off-by: Lin Ma <linma@zju.edu.cn> Signed-off-by: David S. Miller <davem@davemloft.net>
Even though there are two synchronization primitives before the kfree:
1. wait_for_completion(&ax->dead). This can prevent the race with
routines from mkiss_ioctl. However, it cannot stop the routine coming
from upper layer, i.e., the ax25_sendmsg.
2. netif_stop_queue(ax->dev). It seems that this line of code aims to
halt the transmit queue but it fails to stop the routine that already
being xmit.
This patch reorder the kfree after the unregister_netdev to avoid the
possible UAF as the unregister_netdev() is well synchronized and won't
return if there is a running routine.
Signed-off-by: Lin Ma <linma@zju.edu.cn> Signed-off-by: David S. Miller <davem@davemloft.net>
Jean Sacren [Mon, 8 Nov 2021 06:59:41 +0000 (23:59 -0700)]
net: sungem_phy: fix code indentation
Remove extra space in front of the return statement.
Fixes: cd1fce78e046 ("sungem_phy: support bcm5461 phy, autoneg.") Signed-off-by: Jean Sacren <sakiwit@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
But similar to the case in da7d69a49af1 ("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:
... 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: efac30de1b894 ("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
John Fastabend [Wed, 3 Nov 2021 20:47:35 +0000 (13:47 -0700)]
bpf: sockmap, strparser, and tls are reusing qdisc_skb_cb and colliding
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.
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: 38506f4bbc9de ("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
John Fastabend [Wed, 3 Nov 2021 20:47:34 +0000 (13:47 -0700)]
bpf, sockmap: Fix race in ingress receive verdict with redirect to self
A socket in a sockmap may have different combinations of programs attached
depending on configuration. There can be no programs in which case the socket
acts as a sink only. There can be a TX program in this case a BPF program is
attached to sending side, but no RX program is attached. There can be an RX
program only where sends have no BPF program attached, but receives are hooked
with BPF. And finally, both TX and RX programs may be attached. Giving us the
permutations:
None, Tx, Rx, and TxRx
To date most of our use cases have been TX case being used as a fast datapath
to directly copy between local application and a userspace proxy. Or Rx cases
and TxRX applications that are operating an in kernel based proxy. The traffic
in the first case where we hook applications into a userspace application looks
like this:
In this case all traffic from AppA (after 3whs) is copied into the AppB
ingress queue and no traffic is ever on the TCP recieive_queue.
In the second case the application never receives, except in some rare error
cases, traffic on the actual user space socket. Instead the send happens in
the kernel.
AppProxy socket pool
sk0 ------------->{sk1,sk2, skn}
^ |
| |
| v
ingress lb egress
TCP TCP
Here because traffic is never read off the socket with userspace recv() APIs
there is only ever one reader on the sk receive_queue. Namely the BPF programs.
However, we've started to introduce a third configuration where the BPF program
on receive should process the data, but then the normal case is to push the
data into the receive queue of AppB.
This is different from the App{A,B} redirect because traffic is first received
on the sk->receive_queue.
Now for the issue. The tcp_bpf_recvmsg() handler first checks the ingress_msg
queue for any data handled by the BPF rx program and returned with PASS code
so that it was enqueued on the ingress msg queue. Then if no data exists on
that queue it checks the socket receive queue. Unfortunately, this is the same
receive_queue the BPF program is reading data off of. So we get a race. Its
possible for the recvmsg() hook to pull data off the receive_queue before the
BPF hook has a chance to read it. It typically happens when an application is
banging on recv() and getting EAGAINs. Until they manage to race with the RX
BPF program.
To fix this we note that before this patch at attach time when the socket is
loaded into the map we check if it needs a TX program or just the base set of
proto bpf hooks. Then it uses the above general RX hook regardless of if we
have a BPF program attached at rx or not. This patch now extends this check to
handle all cases enumerated above, TX, RX, TXRX, and none. And to fix above
race when an RX program is attached we use a new hook that is nearly identical
to the old one except now we do not let the recv() call skip the RX BPF program.
Now only the BPF program pulls data from sk->receive_queue and recv() only
pulls data from the ingress msgQ post BPF program handling.
With this resolved our AppB from above has been up and running for many hours
without detecting any errors. We do this by correlating counters in RX BPF
events and the AppB to ensure data is never skipping the BPF program. Selftests,
was not able to detect this because we only run them for a short period of time
on well ordered send/recvs so we don't get any of the noise we see in real
application environments.
Fixes: 040dda931bce0 ("bpf: skb_verdict, support SK_PASS on RX BPF path") 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/20211103204736.248403-4-john.fastabend@gmail.com
John Fastabend [Wed, 3 Nov 2021 20:47:33 +0000 (13:47 -0700)]
bpf, sockmap: Remove unhash handler for BPF sockmap usage
We do not need to handle unhash from BPF side we can simply wait for the
close to happen. The original concern was a socket could transition from
ESTABLISHED state to a new state while the BPF hook was still attached.
But, we convinced ourself this is no longer possible and we also improved
BPF sockmap to handle listen sockets so this is no longer a problem.
More importantly though there are cases where unhash is called when data is
in the receive queue. The BPF unhash logic will flush this data which is
wrong. To be correct it should keep the data in the receive queue and allow
a receiving application to continue reading the data. This may happen when
tcp_abort() is received for example. Instead of complicating the logic in
unhash simply moving all this to tcp_close() hook solves this.
Fixes: 040dda931bce0 ("bpf: skb_verdict, support SK_PASS on RX BPF path") 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/20211103204736.248403-3-john.fastabend@gmail.com
John Fastabend [Wed, 3 Nov 2021 20:47:32 +0000 (13:47 -0700)]
bpf, sockmap: Use stricter sk state checks in sk_lookup_assign
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.
litex_liteeth: Fix a double free in the remove function
'netdev' is a managed resource allocated in the probe using
'devm_alloc_etherdev()'.
It must not be freed explicitly in the remove function.
Fixes: 1896d58e447d ("net: Add driver for LiteX's LiteETH network interface") Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
Chengfeng Ye [Fri, 5 Nov 2021 13:36:36 +0000 (06:36 -0700)]
nfc: pn533: Fix double free when pn533_fill_fragment_skbs() fails
skb is already freed by dev_kfree_skb in pn533_fill_fragment_skbs,
but follow error handler branch when pn533_fill_fragment_skbs()
fails, skb is freed again, results in double free issue. Fix this
by not free skb in error path of pn533_fill_fragment_skbs.
Fixes: 61ffe638b35c ("NFC: pn533: Split large Tx frames in chunks") Fixes: fb68b74b833c ("NFC: pn533: Target mode Tx fragmentation support") Signed-off-by: Chengfeng Ye <cyeaa@connect.ust.hk> Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Anders Roxell [Fri, 5 Nov 2021 16:45:11 +0000 (17:45 +0100)]
selftests: net: tls: remove unused variable and code
When building selftests/net with clang, the compiler warn about the
function abs() see below:
tls.c:657:15: warning: variable 'len_compared' set but not used [-Wunused-but-set-variable]
unsigned int len_compared = 0;
^
Rework to remove the unused variable and the for-loop where the variable
'len_compared' was assinged.
Fixes: fe64170d1ffc ("selftests: tls: add selftests for TLS sockets") Signed-off-by: Anders Roxell <anders.roxell@linaro.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
The prestera FW v4.0 support commit has been merged
accidentally w/o review comments addressed and waiting
for the final patch set to be uploaded. So, fix the remaining
comments related to structure laid out and build issues.
Reported-by: kernel test robot <lkp@intel.com> Fixes: 96bd7bd661a4 ("net: marvell: prestera: add firmware v4.0 support") Signed-off-by: Volodymyr Mytnyk <vmytnyk@marvell.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sctp: remove unreachable code from sctp_sf_violation_chunk()
sctp_sf_violation_chunk() is not called with asoc argument equal to NULL,
but if that happens it would lead to NULL pointer dereference
in sctp_vtag_verify().
The patch removes code that handles NULL asoc in sctp_sf_violation_chunk().
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru> Proposed-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Fri, 5 Nov 2021 21:42:14 +0000 (14:42 -0700)]
llc: fix out-of-bound array index in llc_sk_dev_hash()
Both ifindex and LLC_SK_DEV_HASH_ENTRIES are signed.
This means that (ifindex % LLC_SK_DEV_HASH_ENTRIES) is negative
if @ifindex is negative.
We could simply make LLC_SK_DEV_HASH_ENTRIES unsigned.
In this patch I chose to use hash_32() to get more entropy
from @ifindex, like llc_sk_laddr_hashfn().
UBSAN: array-index-out-of-bounds in ./include/net/llc.h:75:26
index -43 is out of range for type 'hlist_head [64]'
CPU: 1 PID: 20999 Comm: syz-executor.3 Not tainted 5.15.0-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
ubsan_epilogue+0xb/0x5a lib/ubsan.c:151
__ubsan_handle_out_of_bounds.cold+0x62/0x6c lib/ubsan.c:291
llc_sk_dev_hash include/net/llc.h:75 [inline]
llc_sap_add_socket+0x49c/0x520 net/llc/llc_conn.c:697
llc_ui_bind+0x680/0xd70 net/llc/af_llc.c:404
__sys_bind+0x1e9/0x250 net/socket.c:1693
__do_sys_bind net/socket.c:1704 [inline]
__se_sys_bind net/socket.c:1702 [inline]
__x64_sys_bind+0x6f/0xb0 net/socket.c:1702
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7fa503407ae9
Fixes: 92ac30a1af46 ("llc: use a device based hash table to speed up multicast delivery") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Fix kernel-doc warnings and spacing in hns3_ethtool.c:
hns3_ethtool.c:246: warning: No description found for return value of 'hns3_lp_run_test'
hns3_ethtool.c:408: warning: expecting prototype for hns3_nic_self_test(). Prototype was for hns3_self_test() instead
Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Reported-by: kernel test robot <lkp@intel.com> Cc: Peng Li <lipeng321@huawei.com> Cc: Guangbin Huang <huangguangbin2@huawei.com> Cc: Yisen Zhuang <yisen.zhuang@huawei.com> Cc: Salil Mehta <salil.mehta@huawei.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: Jakub Kicinski <kuba@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
nfc: port100: lower verbosity of cancelled URB messages
It is not an error to receive an URB with -ENOENT because it can come
from regular user operations, e.g. pressing CTRL+C when running nfctool
from neard. Make it a debugging message, not an error.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> Signed-off-by: David S. Miller <davem@davemloft.net>
this is a pull request of 8 patches for net/master.
The first 3 patches are by Zhang Changzhong and fix 3 standard
conformance problems in the j1939 CAN stack.
The next patch is by Vincent Mailhol and fixes a memory leak in the
leak error path of the etas_es58x CAN driver.
Stephane Grosjean contributes 2 patches for the peak_usb driver to fix
the bus error handling and update the order of printed information
regarding firmware version and available updates.
The last 2 patches are by me and fixes a packet starvation problem in
the bus off case and the error handling in the mcp251xfd_chip_start()
function.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
can: mcp251xfd: mcp251xfd_irq(): add missing can_rx_offload_threaded_irq_finish() in case of bus off
The function can_rx_offload_threaded_irq_finish() is needed to trigger
the NAPI thread to deliver read CAN frames to the networking stack.
This patch adds the missing call to can_rx_offload_threaded_irq_finish()
in case of a bus off, before leaving the interrupt handler to avoid
packet starvation.
Link: https://lore.kernel.org/all/20211106201526.44292-1-mkl@pengutronix.de Fixes: bd74060fe56b ("can: rx-offload: can_rx_offload_threaded_irq_finish(): add new function to be called from threaded interrupt") Cc: stable@vger.kernel.org Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
This patch adds a test to trigger the DCE to remove
the whole subprog to ensure the verifier does not
depend on a stable subprog index. The DCE is done
by testing a global const.
bpf: Stop caching subprog index in the bpf_pseudo_func insn
This patch is to fix an out-of-bound access issue when jit-ing the
bpf_pseudo_func insn (i.e. ld_imm64 with src_reg == BPF_PSEUDO_FUNC)
In jit_subprog(), it currently reuses the subprog index cached in
insn[1].imm. This subprog index is an index into a few array related
to subprogs. For example, in jit_subprog(), it is an index to the newly
allocated 'struct bpf_prog **func' array.
The subprog index was cached in insn[1].imm after add_subprog(). However,
this could become outdated (and too big in this case) if some subprogs
are completely removed during dead code elimination (in
adjust_subprog_starts_after_remove). The cached index in insn[1].imm
is not updated accordingly and causing out-of-bound issue in the later
jit_subprog().
Unlike bpf_pseudo_'func' insn, the current bpf_pseudo_'call' insn
is handling the DCE properly by calling find_subprog(insn->imm) to
figure out the index instead of caching the subprog index.
The existing bpf_adj_branches() will adjust the insn->imm
whenever insn is added or removed.
Instead of having two ways handling subprog index,
this patch is to make bpf_pseudo_func works more like
bpf_pseudo_call.
First change is to stop caching the subprog index result
in insn[1].imm after add_subprog(). The verification
process will use find_subprog(insn->imm) to figure
out the subprog index.
Second change is in bpf_adj_branches() and have it to
adjust the insn->imm for the bpf_pseudo_func insn also
whenever insn is added or removed.
Third change is in jit_subprog(). Like the bpf_pseudo_call handling,
bpf_pseudo_func temporarily stores the find_subprog() result
in insn->off. It is fine because the prog's insn has been finalized
at this point. insn->off will be reset back to 0 later to avoid
confusing the userspace prog dump tool.
can: peak_usb: always ask for BERR reporting for PCAN-USB devices
Since for the PCAN-USB, the management of the transition to the
ERROR_WARNING or ERROR_PASSIVE state is done according to the error
counters, these must be requested unconditionally.
Link: https://lore.kernel.org/all/20211021081505.18223-2-s.grosjean@peak-system.com Fixes: 10a74d0a38b3 ("can: peak_usb: pcan_usb_decode_error(): upgrade handling of bus state changes") Cc: stable@vger.kernel.org Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Vincent Mailhol [Tue, 26 Oct 2021 18:07:40 +0000 (03:07 +0900)]
can: etas_es58x: es58x_rx_err_msg(): fix memory leak in error path
In es58x_rx_err_msg(), if can->do_set_mode() fails, the function
directly returns without calling netif_rx(skb). This means that the
skb previously allocated by alloc_can_err_skb() is not freed. In other
terms, this is a memory leak.
This patch simply removes the return statement in the error branch and
let the function continue.
Issue was found with GCC -fanalyzer, please follow the link below for
details.
Zhang Changzhong [Thu, 28 Oct 2021 14:38:26 +0000 (22:38 +0800)]
can: j1939: j1939_can_recv(): ignore messages with invalid source address
According to SAE-J1939-82 2015 (A.3.6 Row 2), a receiver should never
send TP.CM_CTS to the global address, so we can add a check in
j1939_can_recv() to drop messages with invalid source address.
Nghia Le [Thu, 4 Nov 2021 14:37:40 +0000 (21:37 +0700)]
ipv6: remove useless assignment to newinet in tcp_v6_syn_recv_sock()
The newinet value is initialized with inet_sk() in a block code to
handle sockets for the ETH_P_IP protocol. Along this code path,
newinet is never read. Thus, assignment to newinet is needless and
can be removed.
We've added 15 non-merge commits during the last 3 day(s) which contain
a total of 14 files changed, 199 insertions(+), 90 deletions(-).
The main changes are:
1) Fix regression from stack spill/fill of <8 byte scalars, from Martin KaFai Lau.
2) Fix perf's build of bpftool's bootstrap version due to missing libbpf
headers, from Quentin Monnet.
3) Fix riscv{32,64} BPF exception tables build errors and warnings, from Björn Töpel.
4) Fix bpf fs to allow RENAME_EXCHANGE support for atomic upgrades on sk_lookup
control planes, from Lorenz Bauer.
5) Fix libbpf's error reporting in bpf_map_lookup_and_delete_elem_flags() due to
missing libbpf_err_errno(), from Mehrdad Arshad Rad.
6) Various fixes to make xdp_redirect_multi selftest more reliable, from Hangbin Liu.
7) Fix netcnt selftest to make it run serial and thus avoid conflicts with other
cgroup/skb selftests run in parallel that could cause flakes, from Andrii Nakryiko.
8) Fix reuseport_bpf_numa networking selftest to skip unavailable NUMA nodes,
from Kleber Sacilotto de Souza.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
riscv, bpf: Fix RV32 broken build, and silence RV64 warning
selftests/bpf/xdp_redirect_multi: Limit the tests in netns
selftests/bpf/xdp_redirect_multi: Give tcpdump a chance to terminate cleanly
selftests/bpf/xdp_redirect_multi: Use arping to accurate the arp number
selftests/bpf/xdp_redirect_multi: Put the logs to tmp folder
libbpf: Fix lookup_and_delete_elem_flags error reporting
bpftool: Install libbpf headers for the bootstrap version, too
selftests/net: Fix reuseport_bpf_numa by skipping unavailable nodes
selftests/bpf: Verifier test on refill from a smaller spill
bpf: Do not reject when the stack read size is different from the tracked scalar size
selftests/bpf: Make netcnt selftests serial to avoid spurious failures
selftests/bpf: Test RENAME_EXCHANGE and RENAME_NOREPLACE on bpffs
selftests/bpf: Convert test_bpffs to ASSERT macros
libfs: Support RENAME_EXCHANGE in simple_rename()
libfs: Move shmem_exchange to simple_rename_exchange
====================
Björn Töpel [Wed, 3 Nov 2021 11:54:53 +0000 (12:54 +0100)]
riscv, bpf: Fix RV32 broken build, and silence RV64 warning
Commit 082319ac4849 ("riscv, bpf: Add BPF exception tables") only addressed
RV64, and broke the RV32 build [1]. Fix by gating the exception tables code
with CONFIG_ARCH_RV64I.
Further, silence a "-Wmissing-prototypes" warning [2] in the RV64 BPF JIT.
Hangbin Liu [Wed, 27 Oct 2021 03:35:53 +0000 (11:35 +0800)]
selftests/bpf/xdp_redirect_multi: Limit the tests in netns
As I want to test both DEVMAP and DEVMAP_HASH in XDP multicast redirect, I
limited DEVMAP max entries to a small value for performace. When the test
runs after amount of interface creating/deleting tests. The interface index
will exceed the map max entries and xdp_redirect_multi will error out with
"Get interfacesInterface index to large".
Fix this issue by limit the tests in netns and specify the ifindex when
creating interfaces.
Fixes: a6bc42d13dd6 ("selftests/bpf: Add xdp_redirect_multi test") Reported-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Link: https://lore.kernel.org/bpf/20211027033553.962413-5-liuhangbin@gmail.com
Hangbin Liu [Wed, 27 Oct 2021 03:35:51 +0000 (11:35 +0800)]
selftests/bpf/xdp_redirect_multi: Use arping to accurate the arp number
The arp request number triggered by ping none exist address is not accurate,
which may lead the test false negative/positive. Change to use arping to
accurate the arp number. Also do not use grep pattern match for dot.
Fixes: a6bc42d13dd6 ("selftests/bpf: Add xdp_redirect_multi test") Suggested-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Link: https://lore.kernel.org/bpf/20211027033553.962413-3-liuhangbin@gmail.com
Quentin Monnet [Fri, 5 Nov 2021 01:58:13 +0000 (01:58 +0000)]
bpftool: Install libbpf headers for the bootstrap version, too
We recently changed bpftool's Makefile to make it install libbpf's
headers locally instead of pulling them from the source directory of the
library. Although bpftool needs two versions of libbpf, a "regular" one
and a "bootstrap" version, we would only install headers for the regular
libbpf build. Given that this build always occurs before the bootstrap
build when building bpftool, this is enough to ensure that the bootstrap
bpftool will have access to the headers exported through the regular
libbpf build.
However, this did not account for the case when we only want the
bootstrap version of bpftool, through the "bootstrap" target. For
example, perf needs the bootstrap version only, to generate BPF
skeletons. In that case, when are the headers installed? For some time,
the issue has been masked, because we had a step (the installation of
headers internal to libbpf) which would depend on the regular build of
libbpf and hence trigger the export of the headers, just for the sake of
creating a directory. But this changed with commit efe283687b59
("bpftool: Remove Makefile dep. on $(LIBBPF) for
$(LIBBPF_INTERNAL_HDRS)"), where we cleaned up that stage and removed
the dependency on the regular libbpf build. As a result, when we only
want the bootstrap bpftool version, the regular libbpf is no longer
built. The bootstrap libbpf version is built, but headers are not
exported, and the bootstrap bpftool build fails because of the missing
headers.
To fix this, we also install the library headers for the bootstrap
version of libbpf, to use them for the bootstrap bpftool and for
generating the skeletons.
Fixes: 22a91e67dc29 ("bpftool: Install libbpf headers instead of including the dir") Reported-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Quentin Monnet <quentin@isovalent.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Link: https://lore.kernel.org/bpf/20211105015813.6171-1-quentin@isovalent.com
Arnd Bergmann [Fri, 5 Nov 2021 09:29:40 +0000 (10:29 +0100)]
octeontx2-nicvf: fix ioctl callback
The mii ioctls are now handled by the ndo_eth_ioctl() callback,
not the old ndo_do_ioctl(), but octeontx2-nicvf introduced the
function for the old way.
Move it over to ndo_eth_ioctl() to actually allow calling it from
user space.
Fixes: ad8d98ccc7c5 ("octeontx2-nicvf: Add PTP hardware clock support to NIX VF") Fixes: 8d944ab8cc20 ("dev_ioctl: split out ndo_eth_ioctl") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Arnd Bergmann [Fri, 5 Nov 2021 09:29:39 +0000 (10:29 +0100)]
ax88796c: fix ioctl callback
The timestamp ioctls are now handled by the ndo_eth_ioctl() callback,
not the old ndo_do_ioctl(), but oax88796 introduced the
function for the old way.
Move it over to ndo_eth_ioctl() to actually allow calling it from
user space.
Fixes: 41969ffea93b ("net: ax88796c: ASIX AX88796C SPI Ethernet Adapter Driver") Fixes: 8d944ab8cc20 ("dev_ioctl: split out ndo_eth_ioctl") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Lukasz Stelmach <l.stelmach@samsung.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Yang Li [Fri, 5 Nov 2021 07:12:50 +0000 (15:12 +0800)]
amt: Fix NULL but dereferenced coccicheck error
Eliminate the following coccicheck warning:
./drivers/net/amt.c:2795:6-9: ERROR: amt is NULL but dereferenced.
Reported-by: Abaci Robot <abaci@linux.alibaba.com> Signed-off-by: Yang Li <yang.lee@linux.alibaba.com> Signed-off-by: David S. Miller <davem@davemloft.net>
The only reference to this array is wrapped in of_match_ptr().
Reported-by: kernel test robot <lkp@intel.com> Fixes: 41969ffea93b ("net: ax88796c: ASIX AX88796C SPI Ethernet Adapter Driver") Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Menglong Dong [Fri, 5 Nov 2021 07:35:41 +0000 (15:35 +0800)]
net: udp: correct the document for udp_mem
udp_mem is a vector of 3 INTEGERs, which is used to limit the number of
pages allowed for queueing by all UDP sockets.
However, sk_has_memory_pressure() in __sk_mem_raise_allocated() always
return false for udp, as memory pressure is not supported by udp, which
means that __sk_mem_raise_allocated() will fail once pages allocated
for udp socket exceeds udp_mem[0].
Therefor, udp_mem[0] is the only one that limit the number of pages.
However, the document of udp_mem just express that udp_mem[2] is the
limitation. So, just fix it.
Signed-off-by: Menglong Dong <imagedong@tencent.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Arnd Bergmann [Thu, 4 Nov 2021 13:34:42 +0000 (14:34 +0100)]
octeontx2-pf: select CONFIG_NET_DEVLINK
The octeontx2 pf nic driver failsz to link when the devlink support
is not reachable:
aarch64-linux-ld: drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.o: in function `otx2_dl_mcam_count_get':
otx2_devlink.c:(.text+0x10): undefined reference to `devlink_priv'
aarch64-linux-ld: drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.o: in function `otx2_dl_mcam_count_validate':
otx2_devlink.c:(.text+0x50): undefined reference to `devlink_priv'
aarch64-linux-ld: drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.o: in function `otx2_dl_mcam_count_set':
otx2_devlink.c:(.text+0xd0): undefined reference to `devlink_priv'
aarch64-linux-ld: drivers/net/ethernet/marvell/octeontx2/nic/otx2_devlink.o: in function `otx2_devlink_info_get':
otx2_devlink.c:(.text+0x150): undefined reference to `devlink_priv'
This is already selected by the admin function driver, but not the
actual nic, which might be built-in when the af driver is not.
Fixes: c51a515f129f ("octeontx2-pf: devlink params support to set mcam entry count") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Huang Guobin [Tue, 2 Nov 2021 09:37:33 +0000 (17:37 +0800)]
bonding: Fix a use-after-free problem when bond_sysfs_slave_add() failed
When I do fuzz test for bonding device interface, I got the following
use-after-free Calltrace:
==================================================================
BUG: KASAN: use-after-free in bond_enslave+0x1521/0x24f0
Read of size 8 at addr ffff88825bc11c00 by task ifenslave/7365
Last potentially related work creation:
kasan_save_stack+0x23/0x50
kasan_record_aux_stack+0xb7/0xd0
insert_work+0x43/0x190
__queue_work+0x2e3/0x970
delayed_work_timer_fn+0x3e/0x50
call_timer_fn+0x148/0x470
run_timer_softirq+0x8a8/0xc50
__do_softirq+0x107/0x55f
Second to last potentially related work creation:
kasan_save_stack+0x23/0x50
kasan_record_aux_stack+0xb7/0xd0
insert_work+0x43/0x190
__queue_work+0x2e3/0x970
__queue_delayed_work+0x130/0x180
queue_delayed_work_on+0xa7/0xb0
bond_enslave+0xe25/0x24f0
bond_do_ioctl+0x3e0/0x450
dev_ifsioc+0x2ba/0x970
dev_ioctl+0x112/0x710
sock_do_ioctl+0x118/0x1b0
sock_ioctl+0x2e0/0x490
__x64_sys_ioctl+0x118/0x150
do_syscall_64+0x35/0xb0
entry_SYSCALL_64_after_hwframe+0x44/0xae
Memory state around the buggy address: ffff88825bc11b00: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc ffff88825bc11b80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>ffff88825bc11c00: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
^ ffff88825bc11c80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ffff88825bc11d00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================
Put new_slave in bond_sysfs_slave_add() will cause use-after-free problems
when new_slave is accessed in the subsequent error handling process. Since
new_slave will be put in the subsequent error handling process, remove the
unnecessary put to fix it.
In addition, when sysfs_create_file() fails, if some files have been crea-
ted successfully, we need to call sysfs_remove_file() to remove them.
Since there are sysfs_create_files() & sysfs_remove_files() can be used,
use these two functions instead.
Fixes: 72b0b06f3f49 (bonding: use kobject_put instead of _del after kobject_add) Signed-off-by: Huang Guobin <huangguobin4@huawei.com> Reviewed-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
This pair of patches introduces checks for padding fields of struct
sockaddr_mctp/sockaddr_mctp_ext to ease their re-use for possible
extensions in the future; as well as zeroing of these fields
in the respective sockaddr filling routines. While the first commit
is definitely an ABI breakage, it is proposed in hopes that the change
is made soon enough (the interface appeared only in Linux 5.15)
to avoid affecting any existing user space.
====================
mctp: handle the struct sockaddr_mctp padding fields
In order to have the padding fields actually usable in the future,
there have to be checks that user space doesn't supply non-zero garbage
there. It is also worth setting these padding fields to zero, unless
it is known that they have been already zeroed.
Cc: stable@vger.kernel.org # v5.15 Fixes: e8f4bcac129916b1 ("mctp: Be explicit about struct sockaddr_mctp padding") Signed-off-by: Eugene Syromiatnikov <esyr@redhat.com> Acked-by: Jeremy Kerr <jk@codeconstruct.com.au> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jakub Kicinski [Thu, 4 Nov 2021 23:51:49 +0000 (16:51 -0700)]
Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2021-11-03
Brett fixes issues with promiscuous mode settings not being properly
enabled and removes setting of VF antispoof along with promiscuous
mode. He also ensures that VF Tx queues are always disabled and resolves
a race between virtchnl handling and VF related ndo ops.
Sylwester fixes an issue where a VF MAC could not be set to its primary
MAC if the address is already present.
* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
ice: Fix race conditions between virtchnl handling and VF ndo ops
ice: Fix not stopping Tx queues for VFs
ice: Fix replacing VF hardware MAC to existing MAC filter
ice: Remove toggling of antispoof for VF trusted promiscuous mode
ice: Fix VF true promiscuous mode
====================
Heiner Kallweit [Wed, 3 Nov 2021 21:08:28 +0000 (22:08 +0100)]
net: phy: fix duplex out of sync problem while changing settings
As reported by Zhang there's a small issue if in forced mode the duplex
mode changes with the link staying up [0]. In this case the MAC isn't
notified about the change.
The proposed patch relies on the phylib state machine and ignores the
fact that there are drivers that uses phylib but not the phylib state
machine. So let's don't change the behavior for such drivers and fix
it w/o re-adding state PHY_FORCING for the case that phylib state
machine is used.
selftests/net: Fix reuseport_bpf_numa by skipping unavailable nodes
In some platforms the numa node numbers are not necessarily consecutive,
meaning that not all nodes from 0 to the value returned by numa_max_node()
are available on the system. Using node numbers which are not available
results on errors from libnuma such as:
---- IPv4 UDP ----
send node 0, receive socket 0
libnuma: Warning: Cannot read node cpumask from sysfs
./reuseport_bpf_numa: failed to pin to node: No such file or directory
Fix it by checking if the node number bit is set on numa_nodes_ptr, which
is defined on libnuma as "Set with all nodes the kernel has exposed to
userspace".
Fixes: 196619f605c5 ("net: add new socket option SO_RESERVE_MEM") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Acked-by: Wei Wang <weiwan@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Leonard Crestez [Wed, 3 Nov 2021 22:17:51 +0000 (00:17 +0200)]
tcp: Use BIT() for OPTION_* constants
Extending these flags using the existing (1 << x) pattern triggers
complaints from checkpatch. Instead of ignoring checkpatch modify the
existing values to use BIT(x) style in a separate commit.
Signed-off-by: Leonard Crestez <cdleonard@gmail.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Andrea Righi [Thu, 4 Nov 2021 10:46:13 +0000 (11:46 +0100)]
selftests: net: properly support IPv6 in GSO GRE test
Explicitly pass -6 to netcat when the test is using IPv6 to prevent
failures.
Also make sure to pass "-N" to netcat to close the socket after EOF on
the client side, otherwise we would always hit the timeout and the test
would fail.
TEST: GREv6/v4 - copy file w/ TSO [ OK ]
TEST: GREv6/v4 - copy file w/ GSO [ OK ]
TEST: GREv6/v6 - copy file w/ TSO [ OK ]
TEST: GREv6/v6 - copy file w/ GSO [ OK ]
Fixes: 628f434d2129 ("selftests: add simple GSO GRE test") Signed-off-by: Andrea Righi <andrea.righi@canonical.com> Signed-off-by: David S. Miller <davem@davemloft.net>
ice: Fix race conditions between virtchnl handling and VF ndo ops
The VF can be configured via the PF's ndo ops at the same time the PF is
receiving/handling virtchnl messages. This has many issues, with
one of them being the ndo op could be actively resetting a VF (i.e.
resetting it to the default state and deleting/re-adding the VF's VSI)
while a virtchnl message is being handled. The following error was seen
because a VF ndo op was used to change a VF's trust setting while the
VIRTCHNL_OP_CONFIG_VSI_QUEUES was ongoing:
[35274.192484] ice 0000:88:00.0: Failed to set LAN Tx queue context, error: ICE_ERR_PARAM
[35274.193074] ice 0000:88:00.0: VF 0 failed opcode 6, retval: -5
[35274.193640] iavf 0000:88:01.0: PF returned error -5 (IAVF_ERR_PARAM) to our request 6
Fix this by making sure the virtchnl handling and VF ndo ops that
trigger VF resets cannot run concurrently. This is done by adding a
struct mutex cfg_lock to each VF structure. For VF ndo ops, the mutex
will be locked around the critical operations and VFR. Since the ndo ops
will trigger a VFR, the virtchnl thread will use mutex_trylock(). This
is done because if any other thread (i.e. VF ndo op) has the mutex, then
that means the current VF message being handled is no longer valid, so
just ignore it.
This issue can be seen using the following commands:
When a VF is removed and/or reset its Tx queues need to be
stopped from the PF. This is done by calling the ice_dis_vf_qs()
function, which calls ice_vsi_stop_lan_tx_rings(). Currently
ice_dis_vf_qs() is protected by the VF state bit ICE_VF_STATE_QS_ENA.
Unfortunately, this is causing the Tx queues to not be disabled in some
cases and when the VF tries to re-enable/reconfigure its Tx queues over
virtchnl the op is failing. This is because a VF can be reset and/or
removed before the ICE_VF_STATE_QS_ENA bit is set, but the Tx queues
were already configured via ice_vsi_cfg_single_txq() in the
VIRTCHNL_OP_CONFIG_VSI_QUEUES op. However, the ICE_VF_STATE_QS_ENA bit
is set on a successful VIRTCHNL_OP_ENABLE_QUEUES, which will always
happen after the VIRTCHNL_OP_CONFIG_VSI_QUEUES op.
This was causing the following error message when loading the ice
driver, creating VFs, and modifying VF trust in an endless loop:
[35274.192484] ice 0000:88:00.0: Failed to set LAN Tx queue context, error: ICE_ERR_PARAM
[35274.193074] ice 0000:88:00.0: VF 0 failed opcode 6, retval: -5
[35274.193640] iavf 0000:88:01.0: PF returned error -5 (IAVF_ERR_PARAM) to our request 6
Fix this by always calling ice_dis_vf_qs() and silencing the error
message in ice_vsi_stop_tx_ring() since the calling code ignores the
return anyway. Also, all other places that call ice_vsi_stop_tx_ring()
catch the error, so this doesn't affect those flows since there was no
change to the values the function returns.
Other solutions were considered (i.e. tracking which VF queues had been
"started/configured" in VIRTCHNL_OP_CONFIG_VSI_QUEUES, but it seemed
more complicated than it was worth. This solution also brings in the
chance for other unexpected conditions due to invalid state bit checks.
So, the proposed solution seemed like the best option since there is no
harm in failing to stop Tx queues that were never started.
This issue can be seen using the following commands:
ip link set ens785f1 vf 0 trust on
ip link set ens785f0 vf 0 trust on
done
Fixes: 519d5f19ed82 ("ice: add support for virtchnl_queue_select.[tx|rx]_queues bitmap") Signed-off-by: Brett Creeley <brett.creeley@intel.com> Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
ice: Fix replacing VF hardware MAC to existing MAC filter
VF was not able to change its hardware MAC address in case
the new address was already present in the MAC filter list.
Change the handling of VF add mac request to not return
if requested MAC address is already present on the list
and check if its hardware MAC needs to be updated in this case.
Fixes: a419d1c30fd4 ("ice: Enable ip link show on the PF to display VF unicast MAC(s)") Signed-off-by: Sylwester Dziedziuch <sylwesterx.dziedziuch@intel.com> Tested-by: Tony Brelinski <tony.brelinski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Brett Creeley [Wed, 5 May 2021 21:17:57 +0000 (14:17 -0700)]
ice: Remove toggling of antispoof for VF trusted promiscuous mode
Currently when a trusted VF enables promiscuous mode spoofchk will be
disabled. This is wrong and should only be modified from the
ndo_set_vf_spoofchk callback. Fix this by removing the call to toggle
spoofchk for trusted VFs.
Fixes: 3580aa26254f ("ice: Add VF promiscuous support") Signed-off-by: Brett Creeley <brett.creeley@intel.com> Tested-by: Tony Brelinski <tony.brelinski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Brett Creeley [Fri, 26 Feb 2021 21:19:23 +0000 (13:19 -0800)]
ice: Fix VF true promiscuous mode
When a VF requests promiscuous mode and it's trusted and true promiscuous
mode is enabled the PF driver attempts to enable unicast and/or
multicast promiscuous mode filters based on the request. This is fine,
but there are a couple issues with the current code.
[1] The define to configure the unicast promiscuous mode mask also
includes bits to configure the multicast promiscuous mode mask, which
causes multicast to be set/cleared unintentionally.
[2] All 4 cases for enable/disable unicast/multicast mode are not
handled in the promiscuous mode message handler, which causes
unexpected results regarding the current promiscuous mode settings.
To fix [1] make sure any promiscuous mask defines include the correct
bits for each of the promiscuous modes.
To fix [2] make sure that all 4 cases are handled since there are 2 bits
(FLAG_VF_UNICAST_PROMISC and FLAG_VF_MULTICAST_PROMISC) that can be
either set or cleared. Also, since either unicast and/or multicast
promiscuous configuration can fail, introduce two separate error values
to handle each of these cases.
Fixes: 3580aa26254f ("ice: Add VF promiscuous support") Signed-off-by: Brett Creeley <brett.creeley@intel.com> Tested-by: Tony Brelinski <tony.brelinski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
selftests/bpf: Verifier test on refill from a smaller spill
This patch adds a verifier test to ensure the verifier can read 8 bytes
from the stack after two 32bit write at fp-4 and fp-8. The test is similar
to the reported case from bcc [0].
bpf: Do not reject when the stack read size is different from the tracked scalar size
Below is a simplified case from a report in bcc [0]:
r4 = 20
*(u32 *)(r10 -4) = r4
*(u32 *)(r10 -8) = r4 /* r4 state is tracked */
r4 = *(u64 *)(r10 -8) /* Read more than the tracked 32bit scalar.
* verifier rejects as 'corrupted spill memory'.
*/
After commit 645b4075d71b ("bpf: Support <8-byte scalar spill and refill"),
the 8-byte aligned 32bit spill is also tracked by the verifier and the
register state is stored.
However, if 8 bytes are read from the stack instead of the tracked 4 byte
scalar, then verifier currently rejects the program as "corrupted spill
memory". This patch fixes this case by allowing it to read but marks the
register as unknown.
Also note that, if the prog is trying to corrupt/leak an earlier spilled
pointer by spilling another <8 bytes register on top, this has already
been rejected in the check_stack_write_fixed_off().
[0] https://github.com/iovisor/bcc/pull/3683
Fixes: 645b4075d71b ("bpf: Support <8-byte scalar spill and refill") Reported-by: Hengqi Chen <hengqi.chen@gmail.com> Reported-by: Yonghong Song <yhs@gmail.com> Signed-off-by: Martin KaFai Lau <kafai@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Hengqi Chen <hengqi.chen@gmail.com> Acked-by: Yonghong Song <yhs@fb.com> Link: https://lore.kernel.org/bpf/20211102064535.316018-1-kafai@fb.com
Andrii Nakryiko [Wed, 3 Nov 2021 05:41:13 +0000 (22:41 -0700)]
selftests/bpf: Make netcnt selftests serial to avoid spurious failures
When running `./test_progs -j` test_netcnt fails with a very high
probability, undercounting number of packets received (9999 vs expected
10000). It seems to be conflicting with other cgroup/skb selftests. So
make it serial for now to make parallel mode more robust.
Lorenz Bauer [Thu, 28 Oct 2021 09:47:24 +0000 (10:47 +0100)]
selftests/bpf: Test RENAME_EXCHANGE and RENAME_NOREPLACE on bpffs
Add tests to exercise the behaviour of RENAME_EXCHANGE and RENAME_NOREPLACE
on bpffs. The former checks that after an exchange the inode of two
directories has changed. The latter checks that the source still exists
after a failed rename. Generally, having support for renameat2(RENAME_EXCHANGE)
in bpffs fixes atomic upgrades of our sk_lookup control plane.
Vladimir Oltean [Tue, 2 Nov 2021 19:31:22 +0000 (21:31 +0200)]
net: dsa: felix: fix broken VLAN-tagged PTP under VLAN-aware bridge
Normally it is expected that the dsa_device_ops :: rcv() method finishes
parsing the DSA tag and consumes it, then never looks at it again.
But commit 30b955a3e0f9 ("net: dsa: ocelot: add hardware timestamping
support for Felix") added support for RX timestamping in a very
unconventional way. On this switch, a partial timestamp is available in
the DSA header, but the driver got away with not parsing that timestamp
right away, but instead delayed that parsing for a little longer:
dsa_switch_rcv():
nskb = cpu_dp->rcv(skb, dev); <------------- not here
-> ocelot_rcv()
...
if (dsa_skb_defer_rx_timestamp(p, skb)) <--- but here
-> felix_rxtstamp()
return 0;
When in felix_rxtstamp(), this driver accounted for the fact that
eth_type_trans() happened in the meanwhile, so it got a hold of the
extraction header again by subtracting (ETH_HLEN + OCELOT_TAG_LEN) bytes
from the current skb->data.
This worked for quite some time but was quite fragile from the very
beginning. Not to mention that having DSA tag parsing split in two
different files, under different folders (net/dsa/tag_ocelot.c vs
drivers/net/dsa/ocelot/felix.c) made it quite non-obvious for patches to
come that they might break this.
Finally, the blamed commit does the following: at the end of
ocelot_rcv(), it checks whether the skb payload contains a VLAN header.
If it does, and this port is under a VLAN-aware bridge, that VLAN ID
might not be correct in the sense that the packet might have suffered
VLAN rewriting due to TCAM rules (VCAP IS1). So we consume the VLAN ID
from the skb payload using __skb_vlan_pop(), and take the classified
VLAN ID from the DSA tag, and construct a hwaccel VLAN tag with the
classified VLAN, and the skb payload is VLAN-untagged.
aka it moves the Ethernet header 4 bytes to the right, and pulls 4 bytes
from the skb headroom (effectively also moving skb->data, by definition).
So for felix_rxtstamp()'s fragile logic, all bets are off now.
Instead of having the "extraction" pointer point to the DSA header,
it actually points to 4 bytes _inside_ the extraction header.
Corollary, the last 4 bytes of the "extraction" header are in fact 4
stale bytes of the destination MAC address from the Ethernet header,
from prior to the __skb_vlan_pop() movement.
So of course, RX timestamps are completely bogus when the system is
configured in this way.
The fix is actually very simple: just don't structure the code like that.
For better or worse, the DSA PTP timestamping API does not offer a
straightforward way for drivers to present their RX timestamps, but
other drivers (sja1105) have established a simple mechanism to carry
their RX timestamp from dsa_device_ops :: rcv() all the way to
dsa_switch_ops :: port_rxtstamp() and even later. That mechanism is to
simply save the partial timestamp to the skb->cb, and complete it later.
Question: why don't we simply populate the skb's struct
skb_shared_hwtstamps from ocelot_rcv(), and bother with this
complication of propagating the timestamp to felix_rxtstamp()?
Answer: dsa_switch_ops :: port_rxtstamp() answers the question whether
PTP packets need sleepable context to retrieve the full RX timestamp.
Currently felix_rxtstamp() answers "no, thanks" to that question, and
calls ocelot_ptp_gettime64() from softirq atomic context. This is
understandable, since Felix VSC9959 is a PCIe memory-mapped switch, so
hardware access does not require sleeping. But the felix driver is
preparing for the introduction of other switches where hardware access
is over a slow bus like SPI or MDIO:
https://lore.kernel.org/lkml/20210814025003.2449143-1-colin.foster@in-advantage.com/
So I would like to keep this code structure, so the rework needed when
that driver will need PTP support will be minimal (answer "yes, I need
deferred context for this skb's RX timestamp", then the partial
timestamp will still be found in the skb->cb.
Fixes: 8c873a201195 ("net: dsa: tag_ocelot: use VLAN information from tagging header when available") Reported-by: Po Liu <po.liu@nxp.com> Cc: Yangbo Lu <yangbo.lu@nxp.com> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Ansuel Smith [Tue, 2 Nov 2021 18:30:41 +0000 (19:30 +0100)]
net: dsa: qca8k: make sure PAD0 MAC06 exchange is disabled
Some device set MAC06 exchange in the bootloader. This cause some
problem as we don't support this strange mode and we just set the port6
as the primary CPU port. With MAC06 exchange, PAD0 reg configure port6
instead of port0. Add an extra check and explicitly disable MAC06 exchange
to correctly configure the port PAD config.
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com> Fixes: 52ec49c332ad ("net: dsa: qca8k: add support for cpu port 6") Reviewed-by: Vladimir Oltean <olteanv@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Ziyang Xuan [Tue, 2 Nov 2021 02:12:18 +0000 (10:12 +0800)]
net: vlan: fix a UAF in vlan_dev_real_dev()
The real_dev of a vlan net_device may be freed after
unregister_vlan_dev(). Access the real_dev continually by
vlan_dev_real_dev() will trigger the UAF problem for the
real_dev like following:
Move the put_device(real_dev) to vlan_dev_free(). Ensure
real_dev not be freed before vlan_dev unregistered.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: syzbot+e4df4e1389e28972e955@syzkaller.appspotmail.com Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Menglong Dong [Wed, 3 Nov 2021 08:28:43 +0000 (16:28 +0800)]
net: udp6: replace __UDP_INC_STATS() with __UDP6_INC_STATS()
__UDP_INC_STATS() is used in udpv6_queue_rcv_one_skb() when encap_rcv()
fails. __UDP6_INC_STATS() should be used here, so replace it with
__UDP6_INC_STATS().
Signed-off-by: Menglong Dong <imagedong@tencent.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jakub Kicinski [Tue, 2 Nov 2021 22:02:36 +0000 (15:02 -0700)]
ethtool: fix ethtool msg len calculation for pause stats
ETHTOOL_A_PAUSE_STAT_MAX is the MAX attribute id,
so we need to subtract non-stats and add one to
get a count (IOW -2+1 == -1).
Otherwise we'll see:
ethnl cmd 21: calculated reply length 40, but consumed 52
Fixes: 72da277b46bd ("ethtool: add standard pause stats") Signed-off-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Saeed Mahameed <saeedm@nvidia.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Talal Ahmad [Wed, 3 Nov 2021 02:58:44 +0000 (22:58 -0400)]
net: avoid double accounting for pure zerocopy skbs
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.
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>
Lin Ma [Tue, 2 Nov 2021 08:10:21 +0000 (16:10 +0800)]
NFC: add necessary privilege flags in netlink layer
The CAP_NET_ADMIN checks are needed to prevent attackers faking a
device under NCIUARTSETDRIVER and exploit privileged commands.
This patch add GENL_ADMIN_PERM flags in genl_ops to fulfill the check.
Except for commands like NFC_CMD_GET_DEVICE, NFC_CMD_GET_TARGET,
NFC_CMD_LLC_GET_PARAMS, and NFC_CMD_GET_SE, which are mainly information-
read operations.
Signed-off-by: Lin Ma <linma@zju.edu.cn> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Wed, 3 Nov 2021 11:09:21 +0000 (11:09 +0000)]
Merge branch 'sctp-=security-hook-fixes'
Xin Long says:
====================
security: fixups for the security hooks in sctp
There are a couple of problems in the currect security hooks in sctp:
1. The hooks incorrectly treat sctp_endpoint in SCTP as request_sock in
TCP, while it's in fact no more than an extension of the sock, and
represents the local host. It is created when sock is created, not
when a conn request comes. sctp_association is actually the correct
one to represent the connection, and created when a conn request
arrives.
2. security_sctp_assoc_request() hook should also be called in processing
COOKIE ECHO, as that's the place where the real assoc is created and
used in the future.
The problems above may cause accept sk, peeloff sk or client sk having
the incorrect security labels.
So this patchset is to change some hooks and pass asoc into them and save
these secids into asoc, as well as add the missing sctp_assoc_request
hook into the COOKIE ECHO processing.
v1->v2:
- See each patch, and thanks the help from Ondrej, Paul and Richard.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Xin Long [Tue, 2 Nov 2021 12:02:50 +0000 (08:02 -0400)]
security: implement sctp_assoc_established hook in selinux
Different from selinux_inet_conn_established(), it also gives the
secid to asoc->peer_secid in selinux_sctp_assoc_established(),
as one UDP-type socket may have more than one asocs.
Note that peer_secid in asoc will save the peer secid for this
asoc connection, and peer_sid in sksec will just keep the peer
secid for the latest connection. So the right use should be do
peeloff for UDP-type socket if there will be multiple asocs in
one socket, so that the peeloff socket has the right label for
its asoc.
v1->v2:
- call selinux_inet_conn_established() to reduce some code
duplication in selinux_sctp_assoc_established(), as Ondrej
suggested.
- when doing peeloff, it calls sock_create() where it actually
gets secid for socket from socket_sockcreate_sid(). So reuse
SECSID_WILD to ensure the peeloff socket keeps using that
secid after calling selinux_sctp_sk_clone() for client side.
Fixes: b75380427693 ("security: Add support for SCTP security hooks") Reported-by: Prashanth Prahlad <pprahlad@redhat.com> Reviewed-by: Richard Haines <richard_c_haines@btinternet.com> Tested-by: Richard Haines <richard_c_haines@btinternet.com> Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Xin Long [Tue, 2 Nov 2021 12:02:49 +0000 (08:02 -0400)]
security: add sctp_assoc_established hook
security_sctp_assoc_established() is added to replace
security_inet_conn_established() called in
sctp_sf_do_5_1E_ca(), so that asoc can be accessed in security
subsystem and save the peer secid to asoc->peer_secid.
v1->v2:
- fix the return value of security_sctp_assoc_established() in
security.h, found by kernel test robot and Ondrej.
Fixes: b75380427693 ("security: Add support for SCTP security hooks") Reported-by: Prashanth Prahlad <pprahlad@redhat.com> Reviewed-by: Richard Haines <richard_c_haines@btinternet.com> Tested-by: Richard Haines <richard_c_haines@btinternet.com> Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Xin Long [Tue, 2 Nov 2021 12:02:48 +0000 (08:02 -0400)]
security: call security_sctp_assoc_request in sctp_sf_do_5_1D_ce
The asoc created when receives the INIT chunk is a temporary one, it
will be deleted after INIT_ACK chunk is replied. So for the real asoc
created in sctp_sf_do_5_1D_ce() when the COOKIE_ECHO chunk is received,
security_sctp_assoc_request() should also be called.
v1->v2:
- fix some typo and grammar errors, noticed by Ondrej.
Fixes: b75380427693 ("security: Add support for SCTP security hooks") Reported-by: Prashanth Prahlad <pprahlad@redhat.com> Reviewed-by: Richard Haines <richard_c_haines@btinternet.com> Tested-by: Richard Haines <richard_c_haines@btinternet.com> Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Xin Long [Tue, 2 Nov 2021 12:02:47 +0000 (08:02 -0400)]
security: pass asoc to sctp_assoc_request and sctp_sk_clone
This patch is to move secid and peer_secid from endpoint to association,
and pass asoc to sctp_assoc_request and sctp_sk_clone instead of ep. As
ep is the local endpoint and asoc represents a connection, and in SCTP
one sk/ep could have multiple asoc/connection, saving secid/peer_secid
for new asoc will overwrite the old asoc's.
Note that since asoc can be passed as NULL, security_sctp_assoc_request()
is moved to the place right after the new_asoc is created in
sctp_sf_do_5_1B_init() and sctp_sf_do_unexpected_init().
v1->v2:
- fix the description of selinux_netlbl_skbuff_setsid(), as Jakub noticed.
- fix the annotation in selinux_sctp_assoc_request(), as Richard Noticed.
Fixes: b75380427693 ("security: Add support for SCTP security hooks") Reported-by: Prashanth Prahlad <pprahlad@redhat.com> Reviewed-by: Richard Haines <richard_c_haines@btinternet.com> Tested-by: Richard Haines <richard_c_haines@btinternet.com> Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Hangbin Liu [Wed, 3 Nov 2021 02:44:59 +0000 (10:44 +0800)]
kselftests/net: add missed toeplitz.sh/toeplitz_client.sh to Makefile
When generating the selftests to another folder, the toeplitz.sh
and toeplitz_client.sh are missing as they are not in Makefile, e.g.
make -C tools/testing/selftests/ install \
TARGETS="net" INSTALL_PATH=/tmp/kselftests
Making them under TEST_PROGS_EXTENDED as they test NIC hardware features
and are not intended to be run from kselftests.
Fixes: b3a4036cc56c ("selftests/net: toeplitz test") Reviewed-by: Willem de Bruijn <willemb@google.com> Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Hangbin Liu [Wed, 3 Nov 2021 02:44:58 +0000 (10:44 +0800)]
kselftests/net: add missed vrf_strict_mode_test.sh test to Makefile
When generating the selftests to another folder, the
vrf_strict_mode_test.sh test will miss as it is not in Makefile, e.g.
make -C tools/testing/selftests/ install \
TARGETS="net" INSTALL_PATH=/tmp/kselftests
Fixes: d2562496feaa ("selftests: add selftest for the VRF strict mode") Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Hangbin Liu [Wed, 3 Nov 2021 02:44:57 +0000 (10:44 +0800)]
kselftests/net: add missed SRv6 tests
When generating the selftests to another folder, the SRv6 tests are
missing as they are not in Makefile, e.g.
make -C tools/testing/selftests/ install \
TARGETS="net" INSTALL_PATH=/tmp/kselftests
Fixes: 7cde78511c9d ("selftests: seg6: add selftest for SRv6 End.DT46 Behavior") Fixes: b3b6ecbdb063 ("selftests: add selftest for the SRv6 End.DT4 behavior") Fixes: cbe74f5c9653 ("selftests: add selftest for the SRv6 End.DT6 (VRF) behavior") Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Hangbin Liu [Wed, 3 Nov 2021 02:44:56 +0000 (10:44 +0800)]
kselftests/net: add missed setup_loopback.sh/setup_veth.sh to Makefile
When generating the selftests to another folder, the include file
setup_loopback.sh/setup_veth.sh for gro.sh/gre_gro.sh are missing as
they are not in Makefile, e.g.
make -C tools/testing/selftests/ install \
TARGETS="net" INSTALL_PATH=/tmp/kselftests
Fixes: 3f52169c8f70 ("selftests/net: GRO coalesce test") Fixes: 43de96217ee8 ("selftests/net: allow GRO coalesce test on veth") Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Hangbin Liu [Wed, 3 Nov 2021 02:44:55 +0000 (10:44 +0800)]
kselftests/net: add missed icmp.sh test to Makefile
When generating the selftests to another folder, the icmp.sh test will
miss as it is not in Makefile, e.g.
make -C tools/testing/selftests/ install \
TARGETS="net" INSTALL_PATH=/tmp/kselftests
Fixes: 9df041c4a1e0 ("selftests/net: Add icmp.sh for testing ICMP dummy address responses") Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Tue, 2 Nov 2021 00:45:55 +0000 (17:45 -0700)]
net: add and use skb_unclone_keeptruesize() helper
While commit 6d0be49b56d7 ("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.
In function ‘prestera_hw_build_tests’,
inlined from ‘prestera_hw_switch_init’ at drivers/net/ethernet/marvell/prestera/prestera_hw.c:788:2:
././include/linux/compiler_types.h:335:38: error: call to ‘__compiletime_assert_345’ declared with attribute error: BUILD_BUG_ON failed: sizeof(struct prestera_msg_switch_attr_req) != 16
...
The driver assumes structure members are naturally aligned, but does not
add explicit padding, thus breaking architectures where integral values
are not always naturally aligned (e.g. on m68k, __alignof(int) is 2, not
4).
projects / kernel.git / log