platform/kernel/linux-rpi.git
20 months agomlx4: bpf_xdp_metadata_rx_hash add xdp rss hash type
Jesper Dangaard Brouer [Wed, 12 Apr 2023 19:48:55 +0000 (21:48 +0200)]
mlx4: bpf_xdp_metadata_rx_hash add xdp rss hash type

Update API for bpf_xdp_metadata_rx_hash() with arg for xdp rss hash type
via matching individual Completion Queue Entry (CQE) status bits.

Fixes: ab46182d0dcb ("net/mlx4_en: Support RX XDP metadata")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/r/168132893562.340624.12779118462402031248.stgit@firesoul
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
20 months agoveth: bpf_xdp_metadata_rx_hash add xdp rss hash type
Jesper Dangaard Brouer [Wed, 12 Apr 2023 19:48:50 +0000 (21:48 +0200)]
veth: bpf_xdp_metadata_rx_hash add xdp rss hash type

Update API for bpf_xdp_metadata_rx_hash() with arg for xdp rss hash type.

The veth driver currently only support XDP-hints based on SKB code path.
The SKB have lost information about the RSS hash type, by compressing
the information down to a single bitfield skb->l4_hash, that only knows
if this was a L4 hash value.

In preparation for veth, the xdp_rss_hash_type have an L4 indication
bit that allow us to return a meaningful L4 indication when working
with SKB based packets.

Fixes: 306531f0249f ("veth: Support RX XDP metadata")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/r/168132893055.340624.16209448340644513469.stgit@firesoul
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
20 months agomlx5: bpf_xdp_metadata_rx_hash add xdp rss hash type
Jesper Dangaard Brouer [Wed, 12 Apr 2023 19:48:45 +0000 (21:48 +0200)]
mlx5: bpf_xdp_metadata_rx_hash add xdp rss hash type

Update API for bpf_xdp_metadata_rx_hash() with arg for xdp rss hash type
via mapping table.

The mlx5 hardware can also identify and RSS hash IPSEC.  This indicate
hash includes SPI (Security Parameters Index) as part of IPSEC hash.

Extend xdp core enum xdp_rss_hash_type with IPSEC hash type.

Fixes: bc8d405b1ba9 ("net/mlx5e: Support RX XDP metadata")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/r/168132892548.340624.11185734579430124869.stgit@firesoul
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
20 months agoxdp: rss hash types representation
Jesper Dangaard Brouer [Wed, 12 Apr 2023 19:48:40 +0000 (21:48 +0200)]
xdp: rss hash types representation

The RSS hash type specifies what portion of packet data NIC hardware used
when calculating RSS hash value. The RSS types are focused on Internet
traffic protocols at OSI layers L3 and L4. L2 (e.g. ARP) often get hash
value zero and no RSS type. For L3 focused on IPv4 vs. IPv6, and L4
primarily TCP vs UDP, but some hardware supports SCTP.

Hardware RSS types are differently encoded for each hardware NIC. Most
hardware represent RSS hash type as a number. Determining L3 vs L4 often
requires a mapping table as there often isn't a pattern or sorting
according to ISO layer.

The patch introduce a XDP RSS hash type (enum xdp_rss_hash_type) that
contains both BITs for the L3/L4 types, and combinations to be used by
drivers for their mapping tables. The enum xdp_rss_type_bits get exposed
to BPF via BTF, and it is up to the BPF-programmer to match using these
defines.

This proposal change the kfunc API bpf_xdp_metadata_rx_hash() adding
a pointer value argument for provide the RSS hash type.
Change signature for all xmo_rx_hash calls in drivers to make it compile.

The RSS type implementations for each driver comes as separate patches.

Fixes: 3d76a4d3d4e5 ("bpf: XDP metadata RX kfuncs")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/r/168132892042.340624.582563003880565460.stgit@firesoul
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
20 months agoselftests/bpf: xdp_hw_metadata remove bpf_printk and add counters
Jesper Dangaard Brouer [Wed, 12 Apr 2023 19:48:35 +0000 (21:48 +0200)]
selftests/bpf: xdp_hw_metadata remove bpf_printk and add counters

The tool xdp_hw_metadata can be used by driver developers
implementing XDP-hints metadata kfuncs.

Remove all bpf_printk calls, as the tool already transfers all the
XDP-hints related information via metadata area to AF_XDP
userspace process.

Add counters for providing remaining information about failure and
skipped packet events.

Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/r/168132891533.340624.7313781245316405141.stgit@firesoul
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
20 months agoskbuff: Fix a race between coalescing and releasing SKBs
Liang Chen [Thu, 13 Apr 2023 09:03:53 +0000 (17:03 +0800)]
skbuff: Fix a race between coalescing and releasing SKBs

Commit 1effe8ca4e34 ("skbuff: fix coalescing for page_pool fragment
recycling") allowed coalescing to proceed with non page pool page and page
pool page when @from is cloned, i.e.

to->pp_recycle    --> false
from->pp_recycle  --> true
skb_cloned(from)  --> true

However, it actually requires skb_cloned(@from) to hold true until
coalescing finishes in this situation. If the other cloned SKB is
released while the merging is in process, from_shinfo->nr_frags will be
set to 0 toward the end of the function, causing the increment of frag
page _refcount to be unexpectedly skipped resulting in inconsistent
reference counts. Later when SKB(@to) is released, it frees the page
directly even though the page pool page is still in use, leading to
use-after-free or double-free errors. So it should be prohibited.

The double-free error message below prompted us to investigate:
BUG: Bad page state in process swapper/1  pfn:0e0d1
page:00000000c6548b28 refcount:-1 mapcount:0 mapping:0000000000000000
index:0x2 pfn:0xe0d1
flags: 0xfffffc0000000(node=0|zone=1|lastcpupid=0x1fffff)
raw: 000fffffc0000000 0000000000000000 ffffffff00000101 0000000000000000
raw: 0000000000000002 0000000000000000 ffffffffffffffff 0000000000000000
page dumped because: nonzero _refcount

CPU: 1 PID: 0 Comm: swapper/1 Tainted: G            E      6.2.0+
Call Trace:
 <IRQ>
dump_stack_lvl+0x32/0x50
bad_page+0x69/0xf0
free_pcp_prepare+0x260/0x2f0
free_unref_page+0x20/0x1c0
skb_release_data+0x10b/0x1a0
napi_consume_skb+0x56/0x150
net_rx_action+0xf0/0x350
? __napi_schedule+0x79/0x90
__do_softirq+0xc8/0x2b1
__irq_exit_rcu+0xb9/0xf0
common_interrupt+0x82/0xa0
</IRQ>
<TASK>
asm_common_interrupt+0x22/0x40
RIP: 0010:default_idle+0xb/0x20

Fixes: 53e0961da1c7 ("page_pool: add frag page recycling support in page pool")
Signed-off-by: Liang Chen <liangchen.linux@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20230413090353.14448-1-liangchen.linux@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonet: macb: fix a memory corruption in extended buffer descriptor mode
Roman Gushchin [Wed, 12 Apr 2023 23:21:44 +0000 (16:21 -0700)]
net: macb: fix a memory corruption in extended buffer descriptor mode

For quite some time we were chasing a bug which looked like a sudden
permanent failure of networking and mmc on some of our devices.
The bug was very sensitive to any software changes and even more to
any kernel debug options.

Finally we got a setup where the problem was reproducible with
CONFIG_DMA_API_DEBUG=y and it revealed the issue with the rx dma:

[   16.992082] ------------[ cut here ]------------
[   16.996779] DMA-API: macb ff0b0000.ethernet: device driver tries to free DMA memory it has not allocated [device address=0x0000000875e3e244] [size=1536 bytes]
[   17.011049] WARNING: CPU: 0 PID: 85 at kernel/dma/debug.c:1011 check_unmap+0x6a0/0x900
[   17.018977] Modules linked in: xxxxx
[   17.038823] CPU: 0 PID: 85 Comm: irq/55-8000f000 Not tainted 5.4.0 #28
[   17.045345] Hardware name: xxxxx
[   17.049528] pstate: 60000005 (nZCv daif -PAN -UAO)
[   17.054322] pc : check_unmap+0x6a0/0x900
[   17.058243] lr : check_unmap+0x6a0/0x900
[   17.062163] sp : ffffffc010003c40
[   17.065470] x29: ffffffc010003c40 x28: 000000004000c03c
[   17.070783] x27: ffffffc010da7048 x26: ffffff8878e38800
[   17.076095] x25: ffffff8879d22810 x24: ffffffc010003cc8
[   17.081407] x23: 0000000000000000 x22: ffffffc010a08750
[   17.086719] x21: ffffff8878e3c7c0 x20: ffffffc010acb000
[   17.092032] x19: 0000000875e3e244 x18: 0000000000000010
[   17.097343] x17: 0000000000000000 x16: 0000000000000000
[   17.102647] x15: ffffff8879e4a988 x14: 0720072007200720
[   17.107959] x13: 0720072007200720 x12: 0720072007200720
[   17.113261] x11: 0720072007200720 x10: 0720072007200720
[   17.118565] x9 : 0720072007200720 x8 : 000000000000022d
[   17.123869] x7 : 0000000000000015 x6 : 0000000000000098
[   17.129173] x5 : 0000000000000000 x4 : 0000000000000000
[   17.134475] x3 : 00000000ffffffff x2 : ffffffc010a1d370
[   17.139778] x1 : b420c9d75d27bb00 x0 : 0000000000000000
[   17.145082] Call trace:
[   17.147524]  check_unmap+0x6a0/0x900
[   17.151091]  debug_dma_unmap_page+0x88/0x90
[   17.155266]  gem_rx+0x114/0x2f0
[   17.158396]  macb_poll+0x58/0x100
[   17.161705]  net_rx_action+0x118/0x400
[   17.165445]  __do_softirq+0x138/0x36c
[   17.169100]  irq_exit+0x98/0xc0
[   17.172234]  __handle_domain_irq+0x64/0xc0
[   17.176320]  gic_handle_irq+0x5c/0xc0
[   17.179974]  el1_irq+0xb8/0x140
[   17.183109]  xiic_process+0x5c/0xe30
[   17.186677]  irq_thread_fn+0x28/0x90
[   17.190244]  irq_thread+0x208/0x2a0
[   17.193724]  kthread+0x130/0x140
[   17.196945]  ret_from_fork+0x10/0x20
[   17.200510] ---[ end trace 7240980785f81d6f ]---

[  237.021490] ------------[ cut here ]------------
[  237.026129] DMA-API: exceeded 7 overlapping mappings of cacheline 0x0000000021d79e7b
[  237.033886] WARNING: CPU: 0 PID: 0 at kernel/dma/debug.c:499 add_dma_entry+0x214/0x240
[  237.041802] Modules linked in: xxxxx
[  237.061637] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G        W         5.4.0 #28
[  237.068941] Hardware name: xxxxx
[  237.073116] pstate: 80000085 (Nzcv daIf -PAN -UAO)
[  237.077900] pc : add_dma_entry+0x214/0x240
[  237.081986] lr : add_dma_entry+0x214/0x240
[  237.086072] sp : ffffffc010003c30
[  237.089379] x29: ffffffc010003c30 x28: ffffff8878a0be00
[  237.094683] x27: 0000000000000180 x26: ffffff8878e387c0
[  237.099987] x25: 0000000000000002 x24: 0000000000000000
[  237.105290] x23: 000000000000003b x22: ffffffc010a0fa00
[  237.110594] x21: 0000000021d79e7b x20: ffffffc010abe600
[  237.115897] x19: 00000000ffffffef x18: 0000000000000010
[  237.121201] x17: 0000000000000000 x16: 0000000000000000
[  237.126504] x15: ffffffc010a0fdc8 x14: 0720072007200720
[  237.131807] x13: 0720072007200720 x12: 0720072007200720
[  237.137111] x11: 0720072007200720 x10: 0720072007200720
[  237.142415] x9 : 0720072007200720 x8 : 0000000000000259
[  237.147718] x7 : 0000000000000001 x6 : 0000000000000000
[  237.153022] x5 : ffffffc010003a20 x4 : 0000000000000001
[  237.158325] x3 : 0000000000000006 x2 : 0000000000000007
[  237.163628] x1 : 8ac721b3a7dc1c00 x0 : 0000000000000000
[  237.168932] Call trace:
[  237.171373]  add_dma_entry+0x214/0x240
[  237.175115]  debug_dma_map_page+0xf8/0x120
[  237.179203]  gem_rx_refill+0x190/0x280
[  237.182942]  gem_rx+0x224/0x2f0
[  237.186075]  macb_poll+0x58/0x100
[  237.189384]  net_rx_action+0x118/0x400
[  237.193125]  __do_softirq+0x138/0x36c
[  237.196780]  irq_exit+0x98/0xc0
[  237.199914]  __handle_domain_irq+0x64/0xc0
[  237.204000]  gic_handle_irq+0x5c/0xc0
[  237.207654]  el1_irq+0xb8/0x140
[  237.210789]  arch_cpu_idle+0x40/0x200
[  237.214444]  default_idle_call+0x18/0x30
[  237.218359]  do_idle+0x200/0x280
[  237.221578]  cpu_startup_entry+0x20/0x30
[  237.225493]  rest_init+0xe4/0xf0
[  237.228713]  arch_call_rest_init+0xc/0x14
[  237.232714]  start_kernel+0x47c/0x4a8
[  237.236367] ---[ end trace 7240980785f81d70 ]---

Lars was fast to find an explanation: according to the datasheet
bit 2 of the rx buffer descriptor entry has a different meaning in the
extended mode:
  Address [2] of beginning of buffer, or
  in extended buffer descriptor mode (DMA configuration register [28] = 1),
  indicates a valid timestamp in the buffer descriptor entry.

The macb driver didn't mask this bit while getting an address and it
eventually caused a memory corruption and a dma failure.

The problem is resolved by explicitly clearing the problematic bit
if hw timestamping is used.

Fixes: 7b4296148066 ("net: macb: Add support for PTP timestamps in DMA descriptors")
Signed-off-by: Roman Gushchin <roman.gushchin@linux.dev>
Co-developed-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20230412232144.770336-1-roman.gushchin@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoselftests: add the missing CONFIG_IP_SCTP in net config
Xin Long [Wed, 12 Apr 2023 15:13:06 +0000 (11:13 -0400)]
selftests: add the missing CONFIG_IP_SCTP in net config

The selftest sctp_vrf needs CONFIG_IP_SCTP set in config
when building the kernel, so add it.

Fixes: a61bd7b9fef3 ("selftests: add a selftest for sctp vrf")
Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Link: https://lore.kernel.org/r/61dddebc4d2dd98fe7fb145e24d4b2430e42b572.1681312386.git.lucien.xin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoudp6: fix potential access to stale information
Eric Dumazet [Wed, 12 Apr 2023 13:03:08 +0000 (13:03 +0000)]
udp6: fix potential access to stale information

lena wang reported an issue caused by udpv6_sendmsg()
mangling msg->msg_name and msg->msg_namelen, which
are later read from ____sys_sendmsg() :

/*
 * If this is sendmmsg() and sending to current destination address was
 * successful, remember it.
 */
if (used_address && err >= 0) {
used_address->name_len = msg_sys->msg_namelen;
if (msg_sys->msg_name)
memcpy(&used_address->name, msg_sys->msg_name,
       used_address->name_len);
}

udpv6_sendmsg() wants to pretend the remote address family
is AF_INET in order to call udp_sendmsg().

A fix would be to modify the address in-place, instead
of using a local variable, but this could have other side effects.

Instead, restore initial values before we return from udpv6_sendmsg().

Fixes: c71d8ebe7a44 ("net: Fix security_socket_sendmsg() bypass problem.")
Reported-by: lena wang <lena.wang@mediatek.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Maciej Żenczykowski <maze@google.com>
Link: https://lore.kernel.org/r/20230412130308.1202254-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoselftests: openvswitch: adjust datapath NL message declaration
Aaron Conole [Wed, 12 Apr 2023 11:58:28 +0000 (07:58 -0400)]
selftests: openvswitch: adjust datapath NL message declaration

The netlink message for creating a new datapath takes an array
of ports for the PID creation.  This shouldn't cause much issue
but correct it for future cases where we need to do decode of
datapath information that could include the per-cpu PID map.

Fixes: 25f16c873fb1 ("selftests: add openvswitch selftest suite")
Signed-off-by: Aaron Conole <aconole@redhat.com>
Link: https://lore.kernel.org/r/20230412115828.3991806-1-aconole@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge branch 'mptcp-more-fixes-for-6-3'
Jakub Kicinski [Thu, 13 Apr 2023 16:59:00 +0000 (09:59 -0700)]
Merge branch 'mptcp-more-fixes-for-6-3'

Matthieu Baerts says:

====================
mptcp: more fixes for 6.3

Patch 1 avoids scheduling the MPTCP worker on a closed socket on some
edge cases. It fixes issues that can be visible from v5.11.

Patch 2 makes sure the MPTCP worker doesn't try to manipulate
disconnected sockets. This is also a fix for an issue that can be
visible from v5.11.

Patch 3 fixes a NULL pointer dereference when MPTCP FastOpen is used
and an early fallback is done. A fix for v6.2.

Patch 4 improves the stability of the userspace PM selftest for a
subtest added in v6.2.
====================

Link: https://lore.kernel.org/r/20230411-upstream-net-20230411-mptcp-fixes-v1-0-ca540f3ef986@tessares.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoselftests: mptcp: userspace pm: uniform verify events
Matthieu Baerts [Tue, 11 Apr 2023 20:42:12 +0000 (22:42 +0200)]
selftests: mptcp: userspace pm: uniform verify events

Simply adding a "sleep" before checking something is usually not a good
idea because the time that has been picked can not be enough or too
much. The best is to wait for events with a timeout.

In this selftest, 'sleep 0.5' is used more than 40 times. It is always
used before calling a 'verify_*' function except for this
verify_listener_events which has been added later.

At the end, using all these 'sleep 0.5' seems to work: the slow CIs
don't complain so far. Also because it doesn't take too much time, we
can just add two more 'sleep 0.5' to uniform what is done before calling
a 'verify_*' function. For the same reasons, we can also delay a bigger
refactoring to replace all these 'sleep 0.5' by functions waiting for
events instead of waiting for a fix time and hope for the best.

Fixes: 6c73008aa301 ("selftests: mptcp: listener test for userspace PM")
Cc: stable@vger.kernel.org
Suggested-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agomptcp: fix NULL pointer dereference on fastopen early fallback
Paolo Abeni [Tue, 11 Apr 2023 20:42:11 +0000 (22:42 +0200)]
mptcp: fix NULL pointer dereference on fastopen early fallback

In case of early fallback to TCP, subflow_syn_recv_sock() deletes
the subflow context before returning the newly allocated sock to
the caller.

The fastopen path does not cope with the above unconditionally
dereferencing the subflow context.

Fixes: 36b122baf6a8 ("mptcp: add subflow_v(4,6)_send_synack()")
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agomptcp: stricter state check in mptcp_worker
Paolo Abeni [Tue, 11 Apr 2023 20:42:10 +0000 (22:42 +0200)]
mptcp: stricter state check in mptcp_worker

As reported by Christoph, the mptcp protocol can run the
worker when the relevant msk socket is in an unexpected state:

connect()
// incoming reset + fastclose
// the mptcp worker is scheduled
mptcp_disconnect()
// msk is now CLOSED
listen()
mptcp_worker()

Leading to the following splat:

divide error: 0000 [#1] PREEMPT SMP
CPU: 1 PID: 21 Comm: kworker/1:0 Not tainted 6.3.0-rc1-gde5e8fd0123c #11
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7 04/01/2014
Workqueue: events mptcp_worker
RIP: 0010:__tcp_select_window+0x22c/0x4b0 net/ipv4/tcp_output.c:3018
RSP: 0018:ffffc900000b3c98 EFLAGS: 00010293
RAX: 000000000000ffd7 RBX: 000000000000ffd7 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffffffff8214ce97 RDI: 0000000000000004
RBP: 000000000000ffd7 R08: 0000000000000004 R09: 0000000000010000
R10: 000000000000ffd7 R11: ffff888005afa148 R12: 000000000000ffd7
R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
FS:  0000000000000000(0000) GS:ffff88803ed00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000405270 CR3: 000000003011e006 CR4: 0000000000370ee0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <TASK>
 tcp_select_window net/ipv4/tcp_output.c:262 [inline]
 __tcp_transmit_skb+0x356/0x1280 net/ipv4/tcp_output.c:1345
 tcp_transmit_skb net/ipv4/tcp_output.c:1417 [inline]
 tcp_send_active_reset+0x13e/0x320 net/ipv4/tcp_output.c:3459
 mptcp_check_fastclose net/mptcp/protocol.c:2530 [inline]
 mptcp_worker+0x6c7/0x800 net/mptcp/protocol.c:2705
 process_one_work+0x3bd/0x950 kernel/workqueue.c:2390
 worker_thread+0x5b/0x610 kernel/workqueue.c:2537
 kthread+0x138/0x170 kernel/kthread.c:376
 ret_from_fork+0x2c/0x50 arch/x86/entry/entry_64.S:308
 </TASK>

This change addresses the issue explicitly checking for bad states
before running the mptcp worker.

Fixes: e16163b6e2b7 ("mptcp: refactor shutdown and close")
Cc: stable@vger.kernel.org
Reported-by: Christoph Paasch <cpaasch@apple.com>
Link: https://github.com/multipath-tcp/mptcp_net-next/issues/374
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Tested-by: Christoph Paasch <cpaasch@apple.com>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agomptcp: use mptcp_schedule_work instead of open-coding it
Paolo Abeni [Tue, 11 Apr 2023 20:42:09 +0000 (22:42 +0200)]
mptcp: use mptcp_schedule_work instead of open-coding it

Beyond reducing code duplication this also avoids scheduling
the mptcp_worker on a closed socket on some edge scenarios.

The addressed issue is actually older than the blamed commit
below, but this fix needs it as a pre-requisite.

Fixes: ba8f48f7a4d7 ("mptcp: introduce mptcp_schedule_work")
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonet: enetc: workaround for unresponsive pMAC after receiving express traffic
Vladimir Oltean [Tue, 11 Apr 2023 19:26:45 +0000 (22:26 +0300)]
net: enetc: workaround for unresponsive pMAC after receiving express traffic

I have observed an issue where the RX direction of the LS1028A ENETC pMAC
seems unresponsive. The minimal procedure to reproduce the issue is:

1. Connect ENETC port 0 with a loopback RJ45 cable to one of the Felix
   switch ports (0).

2. Bring the ports up (MAC Merge layer is not enabled on either end).

3. Send a large quantity of unidirectional (express) traffic from Felix
   to ENETC. I tried altering frame size and frame count, and it doesn't
   appear to be specific to either of them, but rather, to the quantity
   of octets received. Lowering the frame count, the minimum quantity of
   packets to reproduce relatively consistently seems to be around 37000
   frames at 1514 octets (w/o FCS) each.

4. Using ethtool --set-mm, enable the pMAC in the Felix and in the ENETC
   ports, in both RX and TX directions, and with verification on both
   ends.

5. Wait for verification to complete on both sides.

6. Configure a traffic class as preemptible on both ends.

7. Send some packets again.

The issue is at step 5, where the verification process of ENETC ends
(meaning that Felix responds with an SMD-R and ENETC sees the response),
but the verification process of Felix never ends (it remains VERIFYING).

If step 3 is skipped or if ENETC receives less traffic than
approximately that threshold, the test runs all the way through
(verification succeeds on both ends, preemptible traffic passes fine).

If, between step 4 and 5, the step below is also introduced:

4.1. Disable and re-enable PM0_COMMAND_CONFIG bit RX_EN

then again, the sequence of steps runs all the way through, and
verification succeeds, even if there was the previous RX traffic
injected into ENETC.

Traffic sent *by* the ENETC port prior to enabling the MAC Merge layer
does not seem to influence the verification result, only received
traffic does.

The LS1028A manual does not mention any relationship between
PM0_COMMAND_CONFIG and MMCSR, and the hardware people don't seem to
know for now either.

The bit that is toggled to work around the issue is also toggled
by enetc_mac_enable(), called from phylink's mac_link_down() and
mac_link_up() methods - which is how the workaround was found:
verification would work after a link down/up.

Fixes: c7b9e8086902 ("net: enetc: add support for MAC Merge layer")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20230411192645.1896048-1-vladimir.oltean@nxp.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
20 months agosctp: fix a potential overflow in sctp_ifwdtsn_skip
Xin Long [Mon, 10 Apr 2023 19:43:30 +0000 (15:43 -0400)]
sctp: fix a potential overflow in sctp_ifwdtsn_skip

Currently, when traversing ifwdtsn skips with _sctp_walk_ifwdtsn, it only
checks the pos against the end of the chunk. However, the data left for
the last pos may be < sizeof(struct sctp_ifwdtsn_skip), and dereference
it as struct sctp_ifwdtsn_skip may cause coverflow.

This patch fixes it by checking the pos against "the end of the chunk -
sizeof(struct sctp_ifwdtsn_skip)" in sctp_ifwdtsn_skip, similar to
sctp_fwdtsn_skip.

Fixes: 0fc2ea922c8a ("sctp: implement validate_ftsn for sctp_stream_interleave")
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Link: https://lore.kernel.org/r/2a71bffcd80b4f2c61fac6d344bb2f11c8fd74f7.1681155810.git.lucien.xin@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
20 months agonet: qrtr: Fix an uninit variable access bug in qrtr_tx_resume()
Ziyang Xuan [Mon, 10 Apr 2023 01:23:52 +0000 (09:23 +0800)]
net: qrtr: Fix an uninit variable access bug in qrtr_tx_resume()

Syzbot reported a bug as following:

=====================================================
BUG: KMSAN: uninit-value in qrtr_tx_resume+0x185/0x1f0 net/qrtr/af_qrtr.c:230
 qrtr_tx_resume+0x185/0x1f0 net/qrtr/af_qrtr.c:230
 qrtr_endpoint_post+0xf85/0x11b0 net/qrtr/af_qrtr.c:519
 qrtr_tun_write_iter+0x270/0x400 net/qrtr/tun.c:108
 call_write_iter include/linux/fs.h:2189 [inline]
 aio_write+0x63a/0x950 fs/aio.c:1600
 io_submit_one+0x1d1c/0x3bf0 fs/aio.c:2019
 __do_sys_io_submit fs/aio.c:2078 [inline]
 __se_sys_io_submit+0x293/0x770 fs/aio.c:2048
 __x64_sys_io_submit+0x92/0xd0 fs/aio.c:2048
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x63/0xcd

Uninit was created at:
 slab_post_alloc_hook mm/slab.h:766 [inline]
 slab_alloc_node mm/slub.c:3452 [inline]
 __kmem_cache_alloc_node+0x71f/0xce0 mm/slub.c:3491
 __do_kmalloc_node mm/slab_common.c:967 [inline]
 __kmalloc_node_track_caller+0x114/0x3b0 mm/slab_common.c:988
 kmalloc_reserve net/core/skbuff.c:492 [inline]
 __alloc_skb+0x3af/0x8f0 net/core/skbuff.c:565
 __netdev_alloc_skb+0x120/0x7d0 net/core/skbuff.c:630
 qrtr_endpoint_post+0xbd/0x11b0 net/qrtr/af_qrtr.c:446
 qrtr_tun_write_iter+0x270/0x400 net/qrtr/tun.c:108
 call_write_iter include/linux/fs.h:2189 [inline]
 aio_write+0x63a/0x950 fs/aio.c:1600
 io_submit_one+0x1d1c/0x3bf0 fs/aio.c:2019
 __do_sys_io_submit fs/aio.c:2078 [inline]
 __se_sys_io_submit+0x293/0x770 fs/aio.c:2048
 __x64_sys_io_submit+0x92/0xd0 fs/aio.c:2048
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x63/0xcd

It is because that skb->len requires at least sizeof(struct qrtr_ctrl_pkt)
in qrtr_tx_resume(). And skb->len equals to size in qrtr_endpoint_post().
But size is less than sizeof(struct qrtr_ctrl_pkt) when qrtr_cb->type
equals to QRTR_TYPE_RESUME_TX in qrtr_endpoint_post() under the syzbot
scenario. This triggers the uninit variable access bug.

Add size check when qrtr_cb->type equals to QRTR_TYPE_RESUME_TX in
qrtr_endpoint_post() to fix the bug.

Fixes: 5fdeb0d372ab ("net: qrtr: Implement outgoing flow control")
Reported-by: syzbot+4436c9630a45820fda76@syzkaller.appspotmail.com
Link: https://syzkaller.appspot.com/bug?id=c14607f0963d27d5a3d5f4c8639b500909e43540
Suggested-by: Manivannan Sadhasivam <mani@kernel.org>
Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Link: https://lore.kernel.org/r/20230410012352.3997823-1-william.xuanziyang@huawei.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
20 months agortnetlink: Restore RTM_NEW/DELLINK notification behavior
Martin Willi [Tue, 11 Apr 2023 07:43:19 +0000 (09:43 +0200)]
rtnetlink: Restore RTM_NEW/DELLINK notification behavior

The commits referenced below allows userspace to use the NLM_F_ECHO flag
for RTM_NEW/DELLINK operations to receive unicast notifications for the
affected link. Prior to these changes, applications may have relied on
multicast notifications to learn the same information without specifying
the NLM_F_ECHO flag.

For such applications, the mentioned commits changed the behavior for
requests not using NLM_F_ECHO. Multicast notifications are still received,
but now use the portid of the requester and the sequence number of the
request instead of zero values used previously. For the application, this
message may be unexpected and likely handled as a response to the
NLM_F_ACKed request, especially if it uses the same socket to handle
requests and notifications.

To fix existing applications relying on the old notification behavior,
set the portid and sequence number in the notification only if the
request included the NLM_F_ECHO flag. This restores the old behavior
for applications not using it, but allows unicasted notifications for
others.

Fixes: f3a63cce1b4f ("rtnetlink: Honour NLM_F_ECHO flag in rtnl_delete_link")
Fixes: d88e136cab37 ("rtnetlink: Honour NLM_F_ECHO flag in rtnl_newlink_create")
Signed-off-by: Martin Willi <martin@strongswan.org>
Acked-by: Guillaume Nault <gnault@redhat.com>
Acked-by: Hangbin Liu <liuhangbin@gmail.com>
Link: https://lore.kernel.org/r/20230411074319.24133-1-martin@strongswan.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonet: ti/cpsw: Add explicit platform_device.h and of_platform.h includes
Rob Herring [Mon, 10 Apr 2023 23:27:19 +0000 (18:27 -0500)]
net: ti/cpsw: Add explicit platform_device.h and of_platform.h includes

TI CPSW uses of_platform_* functions which are declared in of_platform.h.
of_platform.h gets implicitly included by of_device.h, but that is going
to be removed soon. Nothing else depends on of_device.h so it can be
dropped. of_platform.h also implicitly includes platform_device.h, so
add an explicit include for it, too.

Signed-off-by: Rob Herring <robh@kernel.org>
Reviewed-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agonet: wwan: iosm: Fix error handling path in ipc_pcie_probe()
Harshit Mogalapalli [Sat, 8 Apr 2023 19:43:21 +0000 (12:43 -0700)]
net: wwan: iosm: Fix error handling path in ipc_pcie_probe()

Smatch reports:
drivers/net/wwan/iosm/iosm_ipc_pcie.c:298 ipc_pcie_probe()
warn: missing unwind goto?

When dma_set_mask fails it directly returns without disabling pci
device and freeing ipc_pcie. Fix this my calling a correct goto label

As dma_set_mask returns either 0 or -EIO, we can use a goto label, as
it finally returns -EIO.

Add a set_mask_fail goto label which stands consistent with other goto
labels in this function..

Fixes: 035e3befc191 ("net: wwan: iosm: fix driver not working with INTEL_IOMMU disabled")
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agosmc: Fix use-after-free in tcp_write_timer_handler().
Kuniyuki Iwashima [Sat, 8 Apr 2023 18:49:43 +0000 (11:49 -0700)]
smc: Fix use-after-free in tcp_write_timer_handler().

With Eric's ref tracker, syzbot finally found a repro for
use-after-free in tcp_write_timer_handler() by kernel TCP
sockets. [0]

If SMC creates a kernel socket in __smc_create(), the kernel
socket is supposed to be freed in smc_clcsock_release() by
calling sock_release() when we close() the parent SMC socket.

However, at the end of smc_clcsock_release(), the kernel
socket's sk_state might not be TCP_CLOSE.  This means that
we have not called inet_csk_destroy_sock() in __tcp_close()
and have not stopped the TCP timers.

The kernel socket's TCP timers can be fired later, so we
need to hold a refcnt for net as we do for MPTCP subflows
in mptcp_subflow_create_socket().

[0]:
leaked reference.
 sk_alloc (./include/net/net_namespace.h:335 net/core/sock.c:2108)
 inet_create (net/ipv4/af_inet.c:319 net/ipv4/af_inet.c:244)
 __sock_create (net/socket.c:1546)
 smc_create (net/smc/af_smc.c:3269 net/smc/af_smc.c:3284)
 __sock_create (net/socket.c:1546)
 __sys_socket (net/socket.c:1634 net/socket.c:1618 net/socket.c:1661)
 __x64_sys_socket (net/socket.c:1672)
 do_syscall_64 (arch/x86/entry/common.c:50 arch/x86/entry/common.c:80)
 entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:120)
==================================================================
BUG: KASAN: slab-use-after-free in tcp_write_timer_handler (net/ipv4/tcp_timer.c:378 net/ipv4/tcp_timer.c:624 net/ipv4/tcp_timer.c:594)
Read of size 1 at addr ffff888052b65e0d by task syzrepro/18091

CPU: 0 PID: 18091 Comm: syzrepro Tainted: G        W          6.3.0-rc4-01174-gb5d54eb5899a #7
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-1.amzn2022.0.1 04/01/2014
Call Trace:
 <IRQ>
 dump_stack_lvl (lib/dump_stack.c:107)
 print_report (mm/kasan/report.c:320 mm/kasan/report.c:430)
 kasan_report (mm/kasan/report.c:538)
 tcp_write_timer_handler (net/ipv4/tcp_timer.c:378 net/ipv4/tcp_timer.c:624 net/ipv4/tcp_timer.c:594)
 tcp_write_timer (./include/linux/spinlock.h:390 net/ipv4/tcp_timer.c:643)
 call_timer_fn (./arch/x86/include/asm/jump_label.h:27 ./include/linux/jump_label.h:207 ./include/trace/events/timer.h:127 kernel/time/timer.c:1701)
 __run_timers.part.0 (kernel/time/timer.c:1752 kernel/time/timer.c:2022)
 run_timer_softirq (kernel/time/timer.c:2037)
 __do_softirq (./arch/x86/include/asm/jump_label.h:27 ./include/linux/jump_label.h:207 ./include/trace/events/irq.h:142 kernel/softirq.c:572)
 __irq_exit_rcu (kernel/softirq.c:445 kernel/softirq.c:650)
 irq_exit_rcu (kernel/softirq.c:664)
 sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1107 (discriminator 14))
 </IRQ>

Fixes: ac7138746e14 ("smc: establish new socket family")
Reported-by: syzbot+7e1e1bdb852961150198@syzkaller.appspotmail.com
Link: https://lore.kernel.org/netdev/000000000000a3f51805f8bcc43a@google.com/
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agoqlcnic: check pci_reset_function result
Denis Plotnikov [Fri, 7 Apr 2023 07:18:49 +0000 (10:18 +0300)]
qlcnic: check pci_reset_function result

Static code analyzer complains to unchecked return value.
The result of pci_reset_function() is unchecked.
Despite, the issue is on the FLR supported code path and in that
case reset can be done with pcie_flr(), the patch uses less invasive
approach by adding the result check of pci_reset_function().

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: 7e2cf4feba05 ("qlcnic: change driver hardware interface mechanism")
Signed-off-by: Denis Plotnikov <den-plotnikov@yandex-team.ru>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agoMerge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue
Jakub Kicinski [Wed, 12 Apr 2023 04:37:53 +0000 (21:37 -0700)]
Merge branch '40GbE' of git://git./linux/kernel/git/tnguy/net-queue

Tony Nguyen says:

====================
iavf: fix racing in VLANs

Ahmed Zaki says:

This patchset mainly fixes a racing issue in the iavf where the number of
VLANs in the vlan_filter_list might be more than the PF limit. To fix that,
we get rid of the cvlans and svlans bitmaps and keep all the required info
in the list.

The second patch adds two new states that are needed so that we keep the
VLAN info while the interface goes DOWN:
    -- DISABLE    (notify PF, but keep the filter in the list)
    -- INACTIVE   (dev is DOWN, filter is removed from PF)

Finally, the current code keeps each state in a separate bit field, which
is error prone. The first patch refactors that by replacing all bits with
a single enum. The changes are minimal where each bit change is replaced
with the new state value.

* '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
  iavf: remove active_cvlans and active_svlans bitmaps
  iavf: refactor VLAN filter states
====================

Link: https://lore.kernel.org/r/20230407210730.3046149-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge tag 'for-net-2023-04-10' of git://git.kernel.org/pub/scm/linux/kernel/git/bluet...
Jakub Kicinski [Wed, 12 Apr 2023 04:18:23 +0000 (21:18 -0700)]
Merge tag 'for-net-2023-04-10' of git://git./linux/kernel/git/bluetooth/bluetooth

Luiz Augusto von Dentz says:

====================
bluetooth pull request for net:

 - Fix not setting Dath Path for broadcast sink
 - Fix not cleaning up on LE Connection failure
 - SCO: Fix possible circular locking dependency
 - L2CAP: Fix use-after-free in l2cap_disconnect_{req,rsp}
 - Fix race condition in hidp_session_thread
 - btbcm: Fix logic error in forming the board name
 - btbcm: Fix use after free in btsdio_remove

* tag 'for-net-2023-04-10' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth:
  Bluetooth: L2CAP: Fix use-after-free in l2cap_disconnect_{req,rsp}
  Bluetooth: Set ISO Data Path on broadcast sink
  Bluetooth: hci_conn: Fix possible UAF
  Bluetooth: SCO: Fix possible circular locking dependency sco_sock_getsockopt
  Bluetooth: SCO: Fix possible circular locking dependency on sco_connect_cfm
  bluetooth: btbcm: Fix logic error in forming the board name.
  Bluetooth: btsdio: fix use after free bug in btsdio_remove due to race condition
  Bluetooth: Fix race condition in hidp_session_thread
  Bluetooth: Fix printing errors if LE Connection times out
  Bluetooth: hci_conn: Fix not cleaning up on LE Connection failure
====================

Link: https://lore.kernel.org/r/20230410172718.4067798-1-luiz.dentz@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonet: phy: nxp-c45-tja11xx: add remove callback
Radu Pirea (OSS) [Thu, 6 Apr 2023 09:59:04 +0000 (12:59 +0300)]
net: phy: nxp-c45-tja11xx: add remove callback

Unregister PTP clock when the driver is removed.
Purge the RX and TX skb queues.

Fixes: 514def5dd339 ("phy: nxp-c45-tja11xx: add timestamping support")
CC: stable@vger.kernel.org # 5.15+
Signed-off-by: Radu Pirea (OSS) <radu-nicolae.pirea@oss.nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20230406095904.75456-1-radu-nicolae.pirea@oss.nxp.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
20 months agoBluetooth: L2CAP: Fix use-after-free in l2cap_disconnect_{req,rsp}
Luiz Augusto von Dentz [Thu, 6 Apr 2023 16:33:09 +0000 (09:33 -0700)]
Bluetooth: L2CAP: Fix use-after-free in l2cap_disconnect_{req,rsp}

Similar to commit d0be8347c623 ("Bluetooth: L2CAP: Fix use-after-free
caused by l2cap_chan_put"), just use l2cap_chan_hold_unless_zero to
prevent referencing a channel that is about to be destroyed.

Cc: stable@kernel.org
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Min Li <lm0963hack@gmail.com>
20 months agoBluetooth: Set ISO Data Path on broadcast sink
Claudia Draghicescu [Wed, 5 Apr 2023 11:19:18 +0000 (14:19 +0300)]
Bluetooth: Set ISO Data Path on broadcast sink

This patch enables ISO data rx on broadcast sink.

Fixes: eca0ae4aea66 ("Bluetooth: Add initial implementation of BIS connections")
Signed-off-by: Claudia Draghicescu <claudia.rosu@nxp.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agoBluetooth: hci_conn: Fix possible UAF
Luiz Augusto von Dentz [Mon, 3 Apr 2023 21:19:14 +0000 (14:19 -0700)]
Bluetooth: hci_conn: Fix possible UAF

This fixes the following trace:

==================================================================
BUG: KASAN: slab-use-after-free in hci_conn_del+0xba/0x3a0
Write of size 8 at addr ffff88800208e9c8 by task iso-tester/31

CPU: 0 PID: 31 Comm: iso-tester Not tainted 6.3.0-rc2-g991aa4a69a47
 #4716
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.1-2.fc36
04/01/2014
Call Trace:
 <TASK>
 dump_stack_lvl+0x1d/0x70
 print_report+0xce/0x610
 ? __virt_addr_valid+0xd4/0x150
 ? hci_conn_del+0xba/0x3a0
 kasan_report+0xdd/0x110
 ? hci_conn_del+0xba/0x3a0
 hci_conn_del+0xba/0x3a0
 hci_conn_hash_flush+0xf2/0x120
 hci_dev_close_sync+0x388/0x920
 hci_unregister_dev+0x122/0x260
 vhci_release+0x4f/0x90
 __fput+0x102/0x430
 task_work_run+0xf1/0x160
 ? __pfx_task_work_run+0x10/0x10
 ? mark_held_locks+0x24/0x90
 exit_to_user_mode_prepare+0x170/0x180
 syscall_exit_to_user_mode+0x19/0x50
 do_syscall_64+0x4e/0x90
 entry_SYSCALL_64_after_hwframe+0x70/0xda

Fixes: 0f00cd322d22 ("Bluetooth: Free potentially unfreed SCO connection")
Link: https://syzkaller.appspot.com/bug?extid=8bb72f86fc823817bc5d
Cc: <stable@vger.kernel.org>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agoBluetooth: SCO: Fix possible circular locking dependency sco_sock_getsockopt
Luiz Augusto von Dentz [Thu, 30 Mar 2023 21:45:03 +0000 (14:45 -0700)]
Bluetooth: SCO: Fix possible circular locking dependency sco_sock_getsockopt

This attempts to fix the following trace:

======================================================
WARNING: possible circular locking dependency detected
6.3.0-rc2-g68fcb3a7bf97 #4706 Not tainted
------------------------------------------------------
sco-tester/31 is trying to acquire lock:
ffff8880025b8070 (&hdev->lock){+.+.}-{3:3}, at:
sco_sock_getsockopt+0x1fc/0xa90

but task is already holding lock:
ffff888001eeb130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at:
sco_sock_getsockopt+0x104/0xa90

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #2 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}:
       lock_sock_nested+0x32/0x80
       sco_connect_cfm+0x118/0x4a0
       hci_sync_conn_complete_evt+0x1e6/0x3d0
       hci_event_packet+0x55c/0x7c0
       hci_rx_work+0x34c/0xa00
       process_one_work+0x575/0x910
       worker_thread+0x89/0x6f0
       kthread+0x14e/0x180
       ret_from_fork+0x2b/0x50

-> #1 (hci_cb_list_lock){+.+.}-{3:3}:
       __mutex_lock+0x13b/0xcc0
       hci_sync_conn_complete_evt+0x1ad/0x3d0
       hci_event_packet+0x55c/0x7c0
       hci_rx_work+0x34c/0xa00
       process_one_work+0x575/0x910
       worker_thread+0x89/0x6f0
       kthread+0x14e/0x180
       ret_from_fork+0x2b/0x50

-> #0 (&hdev->lock){+.+.}-{3:3}:
       __lock_acquire+0x18cc/0x3740
       lock_acquire+0x151/0x3a0
       __mutex_lock+0x13b/0xcc0
       sco_sock_getsockopt+0x1fc/0xa90
       __sys_getsockopt+0xe9/0x190
       __x64_sys_getsockopt+0x5b/0x70
       do_syscall_64+0x42/0x90
       entry_SYSCALL_64_after_hwframe+0x70/0xda

other info that might help us debug this:

Chain exists of:
  &hdev->lock --> hci_cb_list_lock --> sk_lock-AF_BLUETOOTH-BTPROTO_SCO

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(sk_lock-AF_BLUETOOTH-BTPROTO_SCO);
                               lock(hci_cb_list_lock);
                               lock(sk_lock-AF_BLUETOOTH-BTPROTO_SCO);
  lock(&hdev->lock);

 *** DEADLOCK ***

1 lock held by sco-tester/31:
 #0: ffff888001eeb130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0},
 at: sco_sock_getsockopt+0x104/0xa90

Fixes: 248733e87d50 ("Bluetooth: Allow querying of supported offload codecs over SCO socket")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agoBluetooth: SCO: Fix possible circular locking dependency on sco_connect_cfm
Luiz Augusto von Dentz [Thu, 30 Mar 2023 21:15:50 +0000 (14:15 -0700)]
Bluetooth: SCO: Fix possible circular locking dependency on sco_connect_cfm

This attempts to fix the following trace:

======================================================
WARNING: possible circular locking dependency detected
6.3.0-rc2-g0b93eeba4454 #4703 Not tainted
------------------------------------------------------
kworker/u3:0/46 is trying to acquire lock:
ffff888001fd9130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at:
sco_connect_cfm+0x118/0x4a0

but task is already holding lock:
ffffffff831e3340 (hci_cb_list_lock){+.+.}-{3:3}, at:
hci_sync_conn_complete_evt+0x1ad/0x3d0

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #2 (hci_cb_list_lock){+.+.}-{3:3}:
       __mutex_lock+0x13b/0xcc0
       hci_sync_conn_complete_evt+0x1ad/0x3d0
       hci_event_packet+0x55c/0x7c0
       hci_rx_work+0x34c/0xa00
       process_one_work+0x575/0x910
       worker_thread+0x89/0x6f0
       kthread+0x14e/0x180
       ret_from_fork+0x2b/0x50

-> #1 (&hdev->lock){+.+.}-{3:3}:
       __mutex_lock+0x13b/0xcc0
       sco_sock_connect+0xfc/0x630
       __sys_connect+0x197/0x1b0
       __x64_sys_connect+0x37/0x50
       do_syscall_64+0x42/0x90
       entry_SYSCALL_64_after_hwframe+0x70/0xda

-> #0 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}:
       __lock_acquire+0x18cc/0x3740
       lock_acquire+0x151/0x3a0
       lock_sock_nested+0x32/0x80
       sco_connect_cfm+0x118/0x4a0
       hci_sync_conn_complete_evt+0x1e6/0x3d0
       hci_event_packet+0x55c/0x7c0
       hci_rx_work+0x34c/0xa00
       process_one_work+0x575/0x910
       worker_thread+0x89/0x6f0
       kthread+0x14e/0x180
       ret_from_fork+0x2b/0x50

other info that might help us debug this:

Chain exists of:
  sk_lock-AF_BLUETOOTH-BTPROTO_SCO --> &hdev->lock --> hci_cb_list_lock

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(hci_cb_list_lock);
                               lock(&hdev->lock);
                               lock(hci_cb_list_lock);
  lock(sk_lock-AF_BLUETOOTH-BTPROTO_SCO);

 *** DEADLOCK ***

4 locks held by kworker/u3:0/46:
 #0: ffff8880028d1130 ((wq_completion)hci0#2){+.+.}-{0:0}, at:
 process_one_work+0x4c0/0x910
 #1: ffff8880013dfde0 ((work_completion)(&hdev->rx_work)){+.+.}-{0:0},
 at: process_one_work+0x4c0/0x910
 #2: ffff8880025d8070 (&hdev->lock){+.+.}-{3:3}, at:
 hci_sync_conn_complete_evt+0xa6/0x3d0
 #3: ffffffffb79e3340 (hci_cb_list_lock){+.+.}-{3:3}, at:
 hci_sync_conn_complete_evt+0x1ad/0x3d0

Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agobluetooth: btbcm: Fix logic error in forming the board name.
Sasha Finkelstein [Fri, 10 Mar 2023 10:28:42 +0000 (11:28 +0100)]
bluetooth: btbcm: Fix logic error in forming the board name.

This patch fixes an incorrect loop exit condition in code that replaces
'/' symbols in the board name. There might also be a memory corruption
issue here, but it is unlikely to be a real problem.

Cc: <stable@vger.kernel.org>
Signed-off-by: Sasha Finkelstein <fnkl.kernel@gmail.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agoBluetooth: btsdio: fix use after free bug in btsdio_remove due to race condition
Zheng Wang [Wed, 8 Mar 2023 16:45:01 +0000 (00:45 +0800)]
Bluetooth: btsdio: fix use after free bug in btsdio_remove due to race condition

In btsdio_probe, the data->work is bound with btsdio_work. It will be
started in btsdio_send_frame.

If the btsdio_remove runs with a unfinished work, there may be a race
condition that hdev is freed but used in btsdio_work. Fix it by
canceling the work before do cleanup in btsdio_remove.

Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agoBluetooth: Fix race condition in hidp_session_thread
Min Li [Sat, 4 Mar 2023 14:23:30 +0000 (22:23 +0800)]
Bluetooth: Fix race condition in hidp_session_thread

There is a potential race condition in hidp_session_thread that may
lead to use-after-free. For instance, the timer is active while
hidp_del_timer is called in hidp_session_thread(). After hidp_session_put,
then 'session' will be freed, causing kernel panic when hidp_idle_timeout
is running.

The solution is to use del_timer_sync instead of del_timer.

Here is the call trace:

? hidp_session_probe+0x780/0x780
call_timer_fn+0x2d/0x1e0
__run_timers.part.0+0x569/0x940
hidp_session_probe+0x780/0x780
call_timer_fn+0x1e0/0x1e0
ktime_get+0x5c/0xf0
lapic_next_deadline+0x2c/0x40
clockevents_program_event+0x205/0x320
run_timer_softirq+0xa9/0x1b0
__do_softirq+0x1b9/0x641
__irq_exit_rcu+0xdc/0x190
irq_exit_rcu+0xe/0x20
sysvec_apic_timer_interrupt+0xa1/0xc0

Cc: stable@vger.kernel.org
Signed-off-by: Min Li <lm0963hack@gmail.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agoBluetooth: Fix printing errors if LE Connection times out
Luiz Augusto von Dentz [Fri, 24 Mar 2023 20:18:20 +0000 (13:18 -0700)]
Bluetooth: Fix printing errors if LE Connection times out

This fixes errors like bellow when LE Connection times out since that
is actually not a controller error:

 Bluetooth: hci0: Opcode 0x200d failed: -110
 Bluetooth: hci0: request failed to create LE connection: err -110

Instead the code shall properly detect if -ETIMEDOUT is returned and
send HCI_OP_LE_CREATE_CONN_CANCEL to give up on the connection.

Link: https://github.com/bluez/bluez/issues/340
Fixes: 8e8b92ee60de ("Bluetooth: hci_sync: Add hci_le_create_conn_sync")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agoBluetooth: hci_conn: Fix not cleaning up on LE Connection failure
Luiz Augusto von Dentz [Fri, 24 Mar 2023 17:57:55 +0000 (10:57 -0700)]
Bluetooth: hci_conn: Fix not cleaning up on LE Connection failure

hci_connect_le_scan_cleanup shall always be invoked to cleanup the
states and re-enable passive scanning if necessary, otherwise it may
cause the pending action to stay active causing multiple attempts to
connect.

Fixes: 9b3628d79b46 ("Bluetooth: hci_sync: Cleanup hci_conn if it cannot be aborted")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
20 months agoMerge branch 'sfp-eeprom'
David S. Miller [Sun, 9 Apr 2023 15:01:28 +0000 (16:01 +0100)]
Merge branch 'sfp-eeprom'

Ivan Bornyakov says:

====================
net: fix EEPROM read of absent SFP module

The patchset is to improve EEPROM read requests when SFP module is
absent.

ChangeLog:
v1:
https://lore.kernel.org/netdev/20230405153900.747-1-i.bornyakov@metrotek.ru/
v2:
  * reword commit message of "net: sfp: initialize sfp->i2c_block_size
    at sfp allocation"
  * add second patch to eliminate excessive I2C transfers in
    sfp_module_eeprom() and sfp_module_eeprom_by_page()
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agonet: sfp: avoid EEPROM read of absent SFP module
Ivan Bornyakov [Thu, 6 Apr 2023 13:08:33 +0000 (16:08 +0300)]
net: sfp: avoid EEPROM read of absent SFP module

If SFP module is not present, it is sensible to fail sfp_module_eeprom()
and sfp_module_eeprom_by_page() early to avoid excessive I2C transfers
which are garanteed to fail.

Suggested-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Ivan Bornyakov <i.bornyakov@metrotek.ru>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agonet: sfp: initialize sfp->i2c_block_size at sfp allocation
Ivan Bornyakov [Thu, 6 Apr 2023 13:08:32 +0000 (16:08 +0300)]
net: sfp: initialize sfp->i2c_block_size at sfp allocation

sfp->i2c_block_size is initialized at SFP module insertion in
sfp_sm_mod_probe(). Because of that, if SFP module was never inserted
since boot, sfp_read() call will lead to zero-length I2C read attempt,
and not all I2C controllers are happy with zero-length reads.

One way to issue sfp_read() on empty SFP cage is to execute ethtool -m.
If SFP module was never plugged since boot, there will be a zero-length
I2C read attempt.

  # ethtool -m xge0
  i2c i2c-3: adapter quirk: no zero length (addr 0x0050, size 0, read)
  Cannot get Module EEPROM data: Operation not supported

If SFP module was plugged then removed at least once,
sfp->i2c_block_size will be initialized and ethtool -m will fail with
different exit code and without I2C error

  # ethtool -m xge0
  Cannot get Module EEPROM data: Remote I/O error

Fix this by initializing sfp->i2_block_size at struct sfp allocation
stage so no wild sfp_read() could issue zero-length I2C read.

Signed-off-by: Ivan Bornyakov <i.bornyakov@metrotek.ru>
Fixes: 0d035bed2a4a ("net: sfp: VSOL V2801F / CarlitoxxPro CPGOS03-0490 v2.0 workaround")
Cc: stable@vger.kernel.org
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agor8152: Add __GFP_NOWARN to big allocations
Douglas Anderson [Fri, 7 Apr 2023 00:14:26 +0000 (17:14 -0700)]
r8152: Add __GFP_NOWARN to big allocations

When memory is a little tight on my system, it's pretty easy to see
warnings that look like this.

  ksoftirqd/0: page allocation failure: order:3, mode:0x40a20(GFP_ATOMIC|__GFP_COMP), nodemask=(null),cpuset=/,mems_allowed=0
  ...
  Call trace:
   dump_backtrace+0x0/0x1e8
   show_stack+0x20/0x2c
   dump_stack_lvl+0x60/0x78
   dump_stack+0x18/0x38
   warn_alloc+0x104/0x174
   __alloc_pages+0x588/0x67c
   alloc_rx_agg+0xa0/0x190 [r8152 ...]
   r8152_poll+0x270/0x760 [r8152 ...]
   __napi_poll+0x44/0x1ec
   net_rx_action+0x100/0x300
   __do_softirq+0xec/0x38c
   run_ksoftirqd+0x38/0xec
   smpboot_thread_fn+0xb8/0x248
   kthread+0x134/0x154
   ret_from_fork+0x10/0x20

On a fragmented system it's normal that order 3 allocations will
sometimes fail, especially atomic ones. The driver handles these
failures fine and the WARN just creates spam in the logs for this
case. The __GFP_NOWARN flag is exactly for this situation, so add it
to the allocation.

NOTE: my testing is on a 5.15 system, but there should be no reason
that this would be fundamentally different on a mainline kernel.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Hayes Wang <hayeswang@realtek.com>
Link: https://lore.kernel.org/r/20230406171411.1.I84dbef45786af440fd269b71e9436a96a8e7a152@changeid
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonet: phy: nxp-c45-tja11xx: fix unsigned long multiplication overflow
Radu Pirea (OSS) [Thu, 6 Apr 2023 09:59:53 +0000 (12:59 +0300)]
net: phy: nxp-c45-tja11xx: fix unsigned long multiplication overflow

Any multiplication between GENMASK(31, 0) and a number bigger than 1
will be truncated because of the overflow, if the size of unsigned long
is 32 bits.

Replaced GENMASK with GENMASK_ULL to make sure that multiplication will
be between 64 bits values.

Cc: <stable@vger.kernel.org> # 5.15+
Fixes: 514def5dd339 ("phy: nxp-c45-tja11xx: add timestamping support")
Signed-off-by: Radu Pirea (OSS) <radu-nicolae.pirea@oss.nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20230406095953.75622-1-radu-nicolae.pirea@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonet: openvswitch: fix race on port output
Felix Huettner [Wed, 5 Apr 2023 07:53:41 +0000 (07:53 +0000)]
net: openvswitch: fix race on port output

assume the following setup on a single machine:
1. An openvswitch instance with one bridge and default flows
2. two network namespaces "server" and "client"
3. two ovs interfaces "server" and "client" on the bridge
4. for each ovs interface a veth pair with a matching name and 32 rx and
   tx queues
5. move the ends of the veth pairs to the respective network namespaces
6. assign ip addresses to each of the veth ends in the namespaces (needs
   to be the same subnet)
7. start some http server on the server network namespace
8. test if a client in the client namespace can reach the http server

when following the actions below the host has a chance of getting a cpu
stuck in a infinite loop:
1. send a large amount of parallel requests to the http server (around
   3000 curls should work)
2. in parallel delete the network namespace (do not delete interfaces or
   stop the server, just kill the namespace)

there is a low chance that this will cause the below kernel cpu stuck
message. If this does not happen just retry.
Below there is also the output of bpftrace for the functions mentioned
in the output.

The series of events happening here is:
1. the network namespace is deleted calling
   `unregister_netdevice_many_notify` somewhere in the process
2. this sets first `NETREG_UNREGISTERING` on both ends of the veth and
   then runs `synchronize_net`
3. it then calls `call_netdevice_notifiers` with `NETDEV_UNREGISTER`
4. this is then handled by `dp_device_event` which calls
   `ovs_netdev_detach_dev` (if a vport is found, which is the case for
   the veth interface attached to ovs)
5. this removes the rx_handlers of the device but does not prevent
   packages to be sent to the device
6. `dp_device_event` then queues the vport deletion to work in
   background as a ovs_lock is needed that we do not hold in the
   unregistration path
7. `unregister_netdevice_many_notify` continues to call
   `netdev_unregister_kobject` which sets `real_num_tx_queues` to 0
8. port deletion continues (but details are not relevant for this issue)
9. at some future point the background task deletes the vport

If after 7. but before 9. a packet is send to the ovs vport (which is
not deleted at this point in time) which forwards it to the
`dev_queue_xmit` flow even though the device is unregistering.
In `skb_tx_hash` (which is called in the `dev_queue_xmit`) path there is
a while loop (if the packet has a rx_queue recorded) that is infinite if
`dev->real_num_tx_queues` is zero.

To prevent this from happening we update `do_output` to handle devices
without carrier the same as if the device is not found (which would
be the code path after 9. is done).

Additionally we now produce a warning in `skb_tx_hash` if we will hit
the infinite loop.

bpftrace (first word is function name):

__dev_queue_xmit server: real_num_tx_queues: 1, cpu: 2, pid: 28024, tid: 28024, skb_addr: 0xffff9edb6f207000, reg_state: 1
netdev_core_pick_tx server: addr: 0xffff9f0a46d4a000 real_num_tx_queues: 1, cpu: 2, pid: 28024, tid: 28024, skb_addr: 0xffff9edb6f207000, reg_state: 1
dp_device_event server: real_num_tx_queues: 1 cpu 9, pid: 21024, tid: 21024, event 2, reg_state: 1
synchronize_rcu_expedited: cpu 9, pid: 21024, tid: 21024
synchronize_rcu_expedited: cpu 9, pid: 21024, tid: 21024
synchronize_rcu_expedited: cpu 9, pid: 21024, tid: 21024
synchronize_rcu_expedited: cpu 9, pid: 21024, tid: 21024
dp_device_event server: real_num_tx_queues: 1 cpu 9, pid: 21024, tid: 21024, event 6, reg_state: 2
ovs_netdev_detach_dev server: real_num_tx_queues: 1 cpu 9, pid: 21024, tid: 21024, reg_state: 2
netdev_rx_handler_unregister server: real_num_tx_queues: 1, cpu: 9, pid: 21024, tid: 21024, reg_state: 2
synchronize_rcu_expedited: cpu 9, pid: 21024, tid: 21024
netdev_rx_handler_unregister ret server: real_num_tx_queues: 1, cpu: 9, pid: 21024, tid: 21024, reg_state: 2
dp_device_event server: real_num_tx_queues: 1 cpu 9, pid: 21024, tid: 21024, event 27, reg_state: 2
dp_device_event server: real_num_tx_queues: 1 cpu 9, pid: 21024, tid: 21024, event 22, reg_state: 2
dp_device_event server: real_num_tx_queues: 1 cpu 9, pid: 21024, tid: 21024, event 18, reg_state: 2
netdev_unregister_kobject: real_num_tx_queues: 1, cpu: 9, pid: 21024, tid: 21024
synchronize_rcu_expedited: cpu 9, pid: 21024, tid: 21024
ovs_vport_send server: real_num_tx_queues: 0, cpu: 2, pid: 28024, tid: 28024, skb_addr: 0xffff9edb6f207000, reg_state: 2
__dev_queue_xmit server: real_num_tx_queues: 0, cpu: 2, pid: 28024, tid: 28024, skb_addr: 0xffff9edb6f207000, reg_state: 2
netdev_core_pick_tx server: addr: 0xffff9f0a46d4a000 real_num_tx_queues: 0, cpu: 2, pid: 28024, tid: 28024, skb_addr: 0xffff9edb6f207000, reg_state: 2
broken device server: real_num_tx_queues: 0, cpu: 2, pid: 28024, tid: 28024
ovs_dp_detach_port server: real_num_tx_queues: 0 cpu 9, pid: 9124, tid: 9124, reg_state: 2
synchronize_rcu_expedited: cpu 9, pid: 33604, tid: 33604

stuck message:

watchdog: BUG: soft lockup - CPU#5 stuck for 26s! [curl:1929279]
Modules linked in: veth pktgen bridge stp llc ip_set_hash_net nft_counter xt_set nft_compat nf_tables ip_set_hash_ip ip_set nfnetlink_cttimeout nfnetlink openvswitch nsh nf_conncount nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 tls binfmt_misc nls_iso8859_1 input_leds joydev serio_raw dm_multipath scsi_dh_rdac scsi_dh_emc scsi_dh_alua sch_fq_codel drm efi_pstore virtio_rng ip_tables x_tables autofs4 btrfs blake2b_generic zstd_compress raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid1 raid0 multipath linear hid_generic usbhid hid crct10dif_pclmul crc32_pclmul ghash_clmulni_intel aesni_intel virtio_net ahci net_failover crypto_simd cryptd psmouse libahci virtio_blk failover
CPU: 5 PID: 1929279 Comm: curl Not tainted 5.15.0-67-generic #74-Ubuntu
Hardware name: OpenStack Foundation OpenStack Nova, BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014
RIP: 0010:netdev_pick_tx+0xf1/0x320
Code: 00 00 8d 48 ff 0f b7 c1 66 39 ca 0f 86 e9 01 00 00 45 0f b7 ff 41 39 c7 0f 87 5b 01 00 00 44 29 f8 41 39 c7 0f 87 4f 01 00 00 <eb> f2 0f 1f 44 00 00 49 8b 94 24 28 04 00 00 48 85 d2 0f 84 53 01
RSP: 0018:ffffb78b40298820 EFLAGS: 00000246
RAX: 0000000000000000 RBX: ffff9c8773adc2e0 RCX: 000000000000083f
RDX: 0000000000000000 RSI: ffff9c8773adc2e0 RDI: ffff9c870a25e000
RBP: ffffb78b40298858 R08: 0000000000000001 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: ffff9c870a25e000
R13: ffff9c870a25e000 R14: ffff9c87fe043480 R15: 0000000000000000
FS:  00007f7b80008f00(0000) GS:ffff9c8e5f740000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f7b80f6a0b0 CR3: 0000000329d66000 CR4: 0000000000350ee0
Call Trace:
 <IRQ>
 netdev_core_pick_tx+0xa4/0xb0
 __dev_queue_xmit+0xf8/0x510
 ? __bpf_prog_exit+0x1e/0x30
 dev_queue_xmit+0x10/0x20
 ovs_vport_send+0xad/0x170 [openvswitch]
 do_output+0x59/0x180 [openvswitch]
 do_execute_actions+0xa80/0xaa0 [openvswitch]
 ? kfree+0x1/0x250
 ? kfree+0x1/0x250
 ? kprobe_perf_func+0x4f/0x2b0
 ? flow_lookup.constprop.0+0x5c/0x110 [openvswitch]
 ovs_execute_actions+0x4c/0x120 [openvswitch]
 ovs_dp_process_packet+0xa1/0x200 [openvswitch]
 ? ovs_ct_update_key.isra.0+0xa8/0x120 [openvswitch]
 ? ovs_ct_fill_key+0x1d/0x30 [openvswitch]
 ? ovs_flow_key_extract+0x2db/0x350 [openvswitch]
 ovs_vport_receive+0x77/0xd0 [openvswitch]
 ? __htab_map_lookup_elem+0x4e/0x60
 ? bpf_prog_680e8aff8547aec1_kfree+0x3b/0x714
 ? trace_call_bpf+0xc8/0x150
 ? kfree+0x1/0x250
 ? kfree+0x1/0x250
 ? kprobe_perf_func+0x4f/0x2b0
 ? kprobe_perf_func+0x4f/0x2b0
 ? __mod_memcg_lruvec_state+0x63/0xe0
 netdev_port_receive+0xc4/0x180 [openvswitch]
 ? netdev_port_receive+0x180/0x180 [openvswitch]
 netdev_frame_hook+0x1f/0x40 [openvswitch]
 __netif_receive_skb_core.constprop.0+0x23d/0xf00
 __netif_receive_skb_one_core+0x3f/0xa0
 __netif_receive_skb+0x15/0x60
 process_backlog+0x9e/0x170
 __napi_poll+0x33/0x180
 net_rx_action+0x126/0x280
 ? ttwu_do_activate+0x72/0xf0
 __do_softirq+0xd9/0x2e7
 ? rcu_report_exp_cpu_mult+0x1b0/0x1b0
 do_softirq+0x7d/0xb0
 </IRQ>
 <TASK>
 __local_bh_enable_ip+0x54/0x60
 ip_finish_output2+0x191/0x460
 __ip_finish_output+0xb7/0x180
 ip_finish_output+0x2e/0xc0
 ip_output+0x78/0x100
 ? __ip_finish_output+0x180/0x180
 ip_local_out+0x5e/0x70
 __ip_queue_xmit+0x184/0x440
 ? tcp_syn_options+0x1f9/0x300
 ip_queue_xmit+0x15/0x20
 __tcp_transmit_skb+0x910/0x9c0
 ? __mod_memcg_state+0x44/0xa0
 tcp_connect+0x437/0x4e0
 ? ktime_get_with_offset+0x60/0xf0
 tcp_v4_connect+0x436/0x530
 __inet_stream_connect+0xd4/0x3a0
 ? kprobe_perf_func+0x4f/0x2b0
 ? aa_sk_perm+0x43/0x1c0
 inet_stream_connect+0x3b/0x60
 __sys_connect_file+0x63/0x70
 __sys_connect+0xa6/0xd0
 ? setfl+0x108/0x170
 ? do_fcntl+0xe8/0x5a0
 __x64_sys_connect+0x18/0x20
 do_syscall_64+0x5c/0xc0
 ? __x64_sys_fcntl+0xa9/0xd0
 ? exit_to_user_mode_prepare+0x37/0xb0
 ? syscall_exit_to_user_mode+0x27/0x50
 ? do_syscall_64+0x69/0xc0
 ? __sys_setsockopt+0xea/0x1e0
 ? exit_to_user_mode_prepare+0x37/0xb0
 ? syscall_exit_to_user_mode+0x27/0x50
 ? __x64_sys_setsockopt+0x1f/0x30
 ? do_syscall_64+0x69/0xc0
 ? irqentry_exit+0x1d/0x30
 ? exc_page_fault+0x89/0x170
 entry_SYSCALL_64_after_hwframe+0x61/0xcb
RIP: 0033:0x7f7b8101c6a7
Code: 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2a 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 18 89 54 24 0c 48 89 34 24 89
RSP: 002b:00007ffffd6b2198 EFLAGS: 00000246 ORIG_RAX: 000000000000002a
RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f7b8101c6a7
RDX: 0000000000000010 RSI: 00007ffffd6b2360 RDI: 0000000000000005
RBP: 0000561f1370d560 R08: 00002795ad21d1ac R09: 0030312e302e302e
R10: 00007ffffd73f080 R11: 0000000000000246 R12: 0000561f1370c410
R13: 0000000000000000 R14: 0000000000000005 R15: 0000000000000000
 </TASK>

Fixes: 7f8a436eaa2c ("openvswitch: Add conntrack action")
Co-developed-by: Luca Czesla <luca.czesla@mail.schwarz>
Signed-off-by: Luca Czesla <luca.czesla@mail.schwarz>
Signed-off-by: Felix Huettner <felix.huettner@mail.schwarz>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Link: https://lore.kernel.org/r/ZC0pBXBAgh7c76CA@kernel-bug-kernel-bug
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf
Jakub Kicinski [Sat, 8 Apr 2023 01:23:36 +0000 (18:23 -0700)]
Merge tag 'for-netdev' of https://git./linux/kernel/git/bpf/bpf

Daniel Borkmann says:

====================
pull-request: bpf 2023-04-08

We've added 4 non-merge commits during the last 11 day(s) which contain
a total of 5 files changed, 39 insertions(+), 6 deletions(-).

The main changes are:

1) Fix BPF TCP socket iterator to use correct helper for dropping
   socket's refcount, that is, sock_gen_put instead of sock_put,
   from Martin KaFai Lau.

2) Fix a BTI exception splat in BPF trampoline-generated code on arm64,
   from Xu Kuohai.

3) Fix a LongArch JIT error from missing BPF_NOSPEC no-op, from George Guo.

4) Fix dynamic XDP feature detection of veth in xdp_redirect selftest,
   from Lorenzo Bianconi.

* tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
  selftests/bpf: fix xdp_redirect xdp-features selftest for veth driver
  bpf, arm64: Fixed a BTI error on returning to patched function
  LoongArch, bpf: Fix jit to skip speculation barrier opcode
  bpf: tcp: Use sock_gen_put instead of sock_put in bpf_iter_tcp
====================

Link: https://lore.kernel.org/r/20230407224642.30906-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoiavf: remove active_cvlans and active_svlans bitmaps
Ahmed Zaki [Thu, 6 Apr 2023 21:35:28 +0000 (15:35 -0600)]
iavf: remove active_cvlans and active_svlans bitmaps

The VLAN filters info is currently being held in a list and 2 bitmaps
(active_cvlans and active_svlans). We are experiencing some racing where
data is not in sync in the list and bitmaps. For example, the VLAN is
initially added to the list but only when the PF replies, it is added to
the bitmap. If a user adds many V2 VLANS before the PF responds:

    while [ $((i++)) ]
        ip l add l eth0 name eth0.$i type vlan id $i

we might end up with more VLAN list entries than the designated limit.
Also, The "ip link show" will show more links added than the PF limit.

On the other and, the bitmaps are only used to check the number of VLAN
filters and to re-enable the filters when the interface goes from DOWN to
UP.

This patch gets rid of the bitmaps and uses the list only. To do that,
the states of the VLAN filter are modified:
1 - IAVF_VLAN_REMOVE: the entry needs to be totally removed after informing
  the PF. This is the "ip link del eth0.$i" path.
2 - IAVF_VLAN_DISABLE: (new) the netdev went down. The filter needs to be
  removed from the PF and then marked INACTIVE.
3 - IAVF_VLAN_INACTIVE: (new) no PF filter exists, but the user did not
  delete the VLAN.

Fixes: 48ccc43ecf10 ("iavf: Add support VIRTCHNL_VF_OFFLOAD_VLAN_V2 during netdev config")
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
20 months agoiavf: refactor VLAN filter states
Ahmed Zaki [Thu, 6 Apr 2023 21:35:27 +0000 (15:35 -0600)]
iavf: refactor VLAN filter states

The VLAN filter states are currently being saved as individual bits.
This is error prone as multiple bits might be mistakenly set.

Fix by replacing the bits with a single state enum. Also, add an
"ACTIVE" state for filters that are accepted by the PF.

Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
20 months agoMerge branch 'bonding-ns-validation-fixes'
David S. Miller [Fri, 7 Apr 2023 07:47:20 +0000 (08:47 +0100)]
Merge branch 'bonding-ns-validation-fixes'

Hangbin Liu says:

====================
bonding: fix ns validation on backup slaves

The first patch fixed a ns validation issue on backup slaves. The second
patch re-format the bond option test and add a test lib file. The third
patch add the arp validate regression test for the kernel patch.

Here is the new bonding option test without the kernel fix:

]# ./bond_options.sh
TEST: prio (active-backup miimon primary_reselect 0)           [ OK ]
TEST: prio (active-backup miimon primary_reselect 1)           [ OK ]
TEST: prio (active-backup miimon primary_reselect 2)           [ OK ]
TEST: prio (active-backup arp_ip_target primary_reselect 0)    [ OK ]
TEST: prio (active-backup arp_ip_target primary_reselect 1)    [ OK ]
TEST: prio (active-backup arp_ip_target primary_reselect 2)    [ OK ]
TEST: prio (active-backup ns_ip6_target primary_reselect 0)    [ OK ]
TEST: prio (active-backup ns_ip6_target primary_reselect 1)    [ OK ]
TEST: prio (active-backup ns_ip6_target primary_reselect 2)    [ OK ]
TEST: prio (balance-tlb miimon primary_reselect 0)             [ OK ]
TEST: prio (balance-tlb miimon primary_reselect 1)             [ OK ]
TEST: prio (balance-tlb miimon primary_reselect 2)             [ OK ]
TEST: prio (balance-tlb arp_ip_target primary_reselect 0)      [ OK ]
TEST: prio (balance-tlb arp_ip_target primary_reselect 1)      [ OK ]
TEST: prio (balance-tlb arp_ip_target primary_reselect 2)      [ OK ]
TEST: prio (balance-tlb ns_ip6_target primary_reselect 0)      [ OK ]
TEST: prio (balance-tlb ns_ip6_target primary_reselect 1)      [ OK ]
TEST: prio (balance-tlb ns_ip6_target primary_reselect 2)      [ OK ]
TEST: prio (balance-alb miimon primary_reselect 0)             [ OK ]
TEST: prio (balance-alb miimon primary_reselect 1)             [ OK ]
TEST: prio (balance-alb miimon primary_reselect 2)             [ OK ]
TEST: prio (balance-alb arp_ip_target primary_reselect 0)      [ OK ]
TEST: prio (balance-alb arp_ip_target primary_reselect 1)      [ OK ]
TEST: prio (balance-alb arp_ip_target primary_reselect 2)      [ OK ]
TEST: prio (balance-alb ns_ip6_target primary_reselect 0)      [ OK ]
TEST: prio (balance-alb ns_ip6_target primary_reselect 1)      [ OK ]
TEST: prio (balance-alb ns_ip6_target primary_reselect 2)      [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 0)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 1)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 2)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 3)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 4)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 5)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 6)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 0)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 1)  [ OK ]
TEST: arp_validate (interface eth1 mii_status DOWN)                 [FAIL]
TEST: arp_validate (interface eth2 mii_status DOWN)                 [FAIL]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 2)  [FAIL]
TEST: arp_validate (interface eth1 mii_status DOWN)                 [FAIL]
TEST: arp_validate (interface eth2 mii_status DOWN)                 [FAIL]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 3)  [FAIL]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 4)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 5)  [ OK ]
TEST: arp_validate (interface eth1 mii_status DOWN)                 [FAIL]
TEST: arp_validate (interface eth2 mii_status DOWN)                 [FAIL]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 6)  [FAIL]

Here is the test result after the kernel fix:
TEST: arp_validate (active-backup arp_ip_target arp_validate 0)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 1)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 2)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 3)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 4)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 5)  [ OK ]
TEST: arp_validate (active-backup arp_ip_target arp_validate 6)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 0)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 1)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 2)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 3)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 4)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 5)  [ OK ]
TEST: arp_validate (active-backup ns_ip6_target arp_validate 6)  [ OK ]
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agoselftests: bonding: add arp validate test
Hangbin Liu [Thu, 6 Apr 2023 08:23:52 +0000 (16:23 +0800)]
selftests: bonding: add arp validate test

This patch add bonding arp validate tests with mode active backup,
monitor arp_ip_target and ns_ip6_target. It also checks mii_status
to make sure all slaves are UP.

Acked-by: Jonathan Toppins <jtoppins@redhat.com>
Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agoselftests: bonding: re-format bond option tests
Hangbin Liu [Thu, 6 Apr 2023 08:23:51 +0000 (16:23 +0800)]
selftests: bonding: re-format bond option tests

To improve the testing process for bond options, A new bond topology lib
is added to our testing setup. The current option_prio.sh file will be
renamed to bond_options.sh so that all bonding options can be tested here.
Specifically, for priority testing, we will run all tests using modes
1, 5, and 6. These changes will help us streamline the testing process
and ensure that our bond options are rigorously evaluated.

Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Acked-by: Jonathan Toppins <jtoppins@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agobonding: fix ns validation on backup slaves
Hangbin Liu [Thu, 6 Apr 2023 08:23:50 +0000 (16:23 +0800)]
bonding: fix ns validation on backup slaves

When arp_validate is set to 2, 3, or 6, validation is performed for
backup slaves as well. As stated in the bond documentation, validation
involves checking the broadcast ARP request sent out via the active
slave. This helps determine which slaves are more likely to function in
the event of an active slave failure.

However, when the target is an IPv6 address, the NS message sent from
the active interface is not checked on backup slaves. Additionally,
based on the bond_arp_rcv() rule b, we must reverse the saddr and daddr
when checking the NS message.

Note that when checking the NS message, the destination address is a
multicast address. Therefore, we must convert the target address to
solicited multicast in the bond_get_targets_ip6() function.

Prior to the fix, the backup slaves had a mii status of "down", but
after the fix, all of the slaves' mii status was updated to "UP".

Fixes: 4e24be018eb9 ("bonding: add new parameter ns_targets")
Reviewed-by: Jonathan Toppins <jtoppins@redhat.com>
Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agotcp: restrict net.ipv4.tcp_app_win
YueHaibing [Thu, 6 Apr 2023 06:34:50 +0000 (14:34 +0800)]
tcp: restrict net.ipv4.tcp_app_win

UBSAN: shift-out-of-bounds in net/ipv4/tcp_input.c:555:23
shift exponent 255 is too large for 32-bit type 'int'
CPU: 1 PID: 7907 Comm: ssh Not tainted 6.3.0-rc4-00161-g62bad54b26db-dirty #206
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014
Call Trace:
 <TASK>
 dump_stack_lvl+0x136/0x150
 __ubsan_handle_shift_out_of_bounds+0x21f/0x5a0
 tcp_init_transfer.cold+0x3a/0xb9
 tcp_finish_connect+0x1d0/0x620
 tcp_rcv_state_process+0xd78/0x4d60
 tcp_v4_do_rcv+0x33d/0x9d0
 __release_sock+0x133/0x3b0
 release_sock+0x58/0x1b0

'maxwin' is int, shifting int for 32 or more bits is undefined behaviour.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agoniu: Fix missing unwind goto in niu_alloc_channels()
Harshit Mogalapalli [Thu, 6 Apr 2023 06:31:18 +0000 (23:31 -0700)]
niu: Fix missing unwind goto in niu_alloc_channels()

Smatch reports: drivers/net/ethernet/sun/niu.c:4525
niu_alloc_channels() warn: missing unwind goto?

If niu_rbr_fill() fails, then we are directly returning 'err' without
freeing the channels.

Fix this by changing direct return to a goto 'out_err'.

Fixes: a3138df9f20e ("[NIU]: Add Sun Neptune ethernet driver.")
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20 months agoMerge tag 'net-6.3-rc6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 6 Apr 2023 18:39:07 +0000 (11:39 -0700)]
Merge tag 'net-6.3-rc6-2' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Including fixes from wireless and can.

  Current release - regressions:

   - wifi: mac80211:
      - fix potential null pointer dereference
      - fix receiving mesh packets in forwarding=0 networks
      - fix mesh forwarding

  Current release - new code bugs:

   - virtio/vsock: fix leaks due to missing skb owner

  Previous releases - regressions:

   - raw: fix NULL deref in raw_get_next().

   - sctp: check send stream number after wait_for_sndbuf

   - qrtr:
      - fix a refcount bug in qrtr_recvmsg()
      - do not do DEL_SERVER broadcast after DEL_CLIENT

   - wifi: brcmfmac: fix SDIO suspend/resume regression

   - wifi: mt76: fix use-after-free in fw features query.

   - can: fix race between isotp_sendsmg() and isotp_release()

   - eth: mtk_eth_soc: fix remaining throughput regression

   - eth: ice: reset FDIR counter in FDIR init stage

  Previous releases - always broken:

   - core: don't let netpoll invoke NAPI if in xmit context

   - icmp: guard against too small mtu

   - ipv6: fix an uninit variable access bug in __ip6_make_skb()

   - wifi: mac80211: fix the size calculation of
     ieee80211_ie_len_eht_cap()

   - can: fix poll() to not report false EPOLLOUT events

   - eth: gve: secure enough bytes in the first TX desc for all TCP
     pkts"

* tag 'net-6.3-rc6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (47 commits)
  net: stmmac: check fwnode for phy device before scanning for phy
  net: stmmac: Add queue reset into stmmac_xdp_open() function
  selftests: net: rps_default_mask.sh: delete veth link specifically
  net: fec: make use of MDIO C45 quirk
  can: isotp: fix race between isotp_sendsmg() and isotp_release()
  can: isotp: isotp_ops: fix poll() to not report false EPOLLOUT events
  can: isotp: isotp_recvmsg(): use sock_recv_cmsgs() to get SOCK_RXQ_OVFL infos
  can: j1939: j1939_tp_tx_dat_new(): fix out-of-bounds memory access
  gve: Secure enough bytes in the first TX desc for all TCP pkts
  netlink: annotate lockless accesses to nlk->max_recvmsg_len
  ethtool: reset #lanes when lanes is omitted
  ping: Fix potentail NULL deref for /proc/net/icmp.
  raw: Fix NULL deref in raw_get_next().
  ice: Reset FDIR counter in FDIR init stage
  ice: fix wrong fallback logic for FDIR
  net: stmmac: fix up RX flow hash indirection table when setting channels
  net: ethernet: ti: am65-cpsw: Fix mdio cleanup in probe
  wifi: mt76: ignore key disable commands
  wifi: ath11k: reduce the MHI timeout to 20s
  ipv6: Fix an uninit variable access bug in __ip6_make_skb()
  ...

20 months agoMerge tag 'linux-kselftest-fixes-6.3-rc6' of git://git.kernel.org/pub/scm/linux/kerne...
Linus Torvalds [Thu, 6 Apr 2023 18:34:18 +0000 (11:34 -0700)]
Merge tag 'linux-kselftest-fixes-6.3-rc6' of git://git./linux/kernel/git/shuah/linux-kselftest

Pull Kselftest fixes from Shuah Khan:
 "One single fix to mount_setattr_test build failure"

* tag 'linux-kselftest-fixes-6.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  selftests mount: Fix mount_setattr_test builds failed

20 months agoMerge tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg...
Linus Torvalds [Thu, 6 Apr 2023 18:27:21 +0000 (11:27 -0700)]
Merge tag 'for-linus-iommufd' of git://git./linux/kernel/git/jgg/iommufd

Pull iommufd fixes from Jason Gunthorpe:

 - An invalid VA range can be be put in a pages and eventually trigger
   WARN_ON, reject it early

 - Use of the wrong start index value when doing the complex batch carry
   scheme

 - Wrong store ordering resulting in corrupting data used in a later
   calculation that corrupted the batch structure during carry

* tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg/iommufd:
  iommufd: Do not corrupt the pfn list when doing batch carry
  iommufd: Fix unpinning of pages when an access is present
  iommufd: Check for uptr overflow

20 months agoMerge tag 'pwm/for-6.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry...
Linus Torvalds [Thu, 6 Apr 2023 18:08:03 +0000 (11:08 -0700)]
Merge tag 'pwm/for-6.3-rc6' of git://git./linux/kernel/git/thierry.reding/linux-pwm

Pull pwm fixes from Thierry Reding:
 "These are some fixes to make sure the PWM state structure is always
  initialized to a known state.

  Prior to this it could happen in some situations that random data from
  the stack would leak into the data structure and cause subtle bugs"

* tag 'pwm/for-6.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm:
  pwm: Zero-initialize the pwm_state passed to driver's .get_state()
  pwm: meson: Explicitly set .polarity in .get_state()
  pwm: sprd: Explicitly set .polarity in .get_state()
  pwm: iqs620a: Explicitly set .polarity in .get_state()
  pwm: cros-ec: Explicitly set .polarity in .get_state()
  pwm: hibvt: Explicitly set .polarity in .get_state()

20 months agoMerge tag 'drm-fixes-2023-04-06' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Thu, 6 Apr 2023 17:25:27 +0000 (10:25 -0700)]
Merge tag 'drm-fixes-2023-04-06' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Daniel Vetter:
 "Mostly i915 fixes: dp mst for compression/dsc, perf ioctl uaf, ctx rpm
  accounting, gt reset vs huc loading.

  And a few individual driver fixes: ivpu dma fence&suspend, panfrost
  mmap, nouveau color depth"

* tag 'drm-fixes-2023-04-06' of git://anongit.freedesktop.org/drm/drm:
  accel/ivpu: Fix S3 system suspend when not idle
  accel/ivpu: Add dma fence to command buffers only
  drm/i915: Fix context runtime accounting
  drm/i915: fix race condition UAF in i915_perf_add_config_ioctl
  drm/i915: Use compressed bpp when calculating m/n value for DP MST DSC
  drm/i915/huc: Cancel HuC delayed load timer on reset.
  drm/i915/ttm: fix sparse warning
  drm/panfrost: Fix the panfrost_mmu_map_fault_addr() error path
  drm/nouveau/disp: Support more modes by checking with lower bpc

20 months agoMerge tag 'sound-6.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Thu, 6 Apr 2023 17:19:30 +0000 (10:19 -0700)]
Merge tag 'sound-6.3-rc6' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "The majority of changes here are various fixes for Intel drivers,
  and there is a change in ASoC PCM core for the format constraints.

  In addition, a workaround for HD-audio HDMI regressions and usual
  HD-audio quirks are found"

* tag 'sound-6.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda/hdmi: Preserve the previous PCM device upon re-enablement
  ALSA: hda/realtek: Add quirk for Clevo X370SNW
  ALSA: hda/realtek: fix mute/micmute LEDs for a HP ProBook
  ASoC: SOF: avoid a NULL dereference with unsupported widgets
  ASoC: da7213.c: add missing pm_runtime_disable()
  ASoC: hdac_hdmi: use set_stream() instead of set_tdm_slots()
  ASoC: codecs: lpass: fix the order or clks turn off during suspend
  ASoC: Intel: bytcr_rt5640: Add quirk for the Acer Iconia One 7 B1-750
  ASoC: SOF: ipc4: Ensure DSP is in D0I0 during sof_ipc4_set_get_data()
  ASoC: amd: yc: Add DMI entries to support Victus by HP Laptop 16-e1xxx (8A22)
  ASoC: soc-pcm: fix hw->formats cleared by soc_pcm_hw_init() for dpcm
  ASoC: Intel: soc-acpi: add table for Intel 'Rooks County' NUC M15
  ASOC: Intel: sof_sdw: add quirk for Intel 'Rooks County' NUC M15

20 months agoMerge tag 'platform-drivers-x86-v6.3-5' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 6 Apr 2023 17:13:23 +0000 (10:13 -0700)]
Merge tag 'platform-drivers-x86-v6.3-5' of git://git./linux/kernel/git/pdx86/platform-drivers-x86

Pull x86 platform driver fixes from Hans de Goede:

 -  more think-lmi fixes

 -  one DMI quirk addition

* tag 'platform-drivers-x86-v6.3-5' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  platform/x86: thinkpad_acpi: Add missing T14s Gen1 type to s2idle quirk list
  platform/x86: think-lmi: Clean up display of current_value on Thinkstation
  platform/x86: think-lmi: Fix memory leaks when parsing ThinkStation WMI strings
  platform/x86: think-lmi: Fix memory leak when showing current settings

20 months agoMerge tag 'asm-generic-fixes-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 6 Apr 2023 16:51:04 +0000 (09:51 -0700)]
Merge tag 'asm-generic-fixes-6.3' of git://git./linux/kernel/git/arnd/asm-generic

Pull asm-generic fixes from Arnd Bergmann:
 "These are minor fixes to address false-positive build warnings:

  Some of the less common I/O accessors are missing __force casts and
  cause sparse warnings for their implied byteswap, and a recent change
  to __generic_cmpxchg_local() causes a warning about constant integer
  truncation"

* tag 'asm-generic-fixes-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic:
  asm-generic: avoid __generic_cmpxchg_local warnings
  asm-generic/io.h: suppress endianness warnings for relaxed accessors
  asm-generic/io.h: suppress endianness warnings for readq() and writeq()

20 months agoselftests/bpf: fix xdp_redirect xdp-features selftest for veth driver
Lorenzo Bianconi [Thu, 6 Apr 2023 10:40:19 +0000 (12:40 +0200)]
selftests/bpf: fix xdp_redirect xdp-features selftest for veth driver

xdp-features supported by veth driver are no more static, but they
depends on veth configuration (e.g. if GRO is enabled/disabled or
TX/RX queue configuration). Take it into account in xdp_redirect
xdp-features selftest for veth driver.

Fixes: fccca038f300 ("veth: take into account device reconfiguration for xdp_features flag")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://lore.kernel.org/r/bc35455cfbb1d4f7f52536955ded81ad47d8dc54.1680777371.git.lorenzo@kernel.org
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
20 months agonet: stmmac: check fwnode for phy device before scanning for phy
Michael Sit Wei Hong [Thu, 6 Apr 2023 02:45:41 +0000 (10:45 +0800)]
net: stmmac: check fwnode for phy device before scanning for phy

Some DT devices already have phy device configured in the DT/ACPI.
Current implementation scans for a phy unconditionally even though
there is a phy listed in the DT/ACPI and already attached.

We should check the fwnode if there is any phy device listed in
fwnode and decide whether to scan for a phy to attach to.

Fixes: fe2cfbc96803 ("net: stmmac: check if MAC needs to attach to a PHY")
Reported-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Link: https://lore.kernel.org/lkml/20230403212434.296975-1-martin.blumenstingl@googlemail.com/
Tested-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Shahab Vahedi <shahab@synopsys.com>
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com>
Tested-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Suggested-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Michael Sit Wei Hong <michael.wei.hong.sit@intel.com>
Link: https://lore.kernel.org/r/20230406024541.3556305-1-michael.wei.hong.sit@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonet: stmmac: Add queue reset into stmmac_xdp_open() function
Song Yoong Siang [Tue, 4 Apr 2023 04:48:23 +0000 (12:48 +0800)]
net: stmmac: Add queue reset into stmmac_xdp_open() function

Queue reset was moved out from __init_dma_rx_desc_rings() and
__init_dma_tx_desc_rings() functions. Thus, the driver fails to transmit
and receive packet after XDP prog setup.

This commit adds the missing queue reset into stmmac_xdp_open() function.

Fixes: f9ec5723c3db ("net: ethernet: stmicro: stmmac: move queue reset to dedicated functions")
Cc: <stable@vger.kernel.org> # 6.0+
Signed-off-by: Song Yoong Siang <yoong.siang.song@intel.com>
Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>
Link: https://lore.kernel.org/r/20230404044823.3226144-1-yoong.siang.song@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoselftests: net: rps_default_mask.sh: delete veth link specifically
Hangbin Liu [Tue, 4 Apr 2023 07:24:11 +0000 (15:24 +0800)]
selftests: net: rps_default_mask.sh: delete veth link specifically

When deleting the netns and recreating a new one while re-adding the
veth interface, there is a small window of time during which the old
veth interface has not yet been removed. This can cause the new addition
to fail. To resolve this issue, we can either wait for a short while to
ensure that the old veth interface is deleted, or we can specifically
remove the veth interface.

Before this patch:
  # ./rps_default_mask.sh
  empty rps_default_mask                                      [ ok ]
  changing rps_default_mask dont affect existing devices      [ ok ]
  changing rps_default_mask dont affect existing netns        [ ok ]
  changing rps_default_mask affect newly created devices      [ ok ]
  changing rps_default_mask don't affect newly child netns[II][ ok ]
  rps_default_mask is 0 by default in child netns             [ ok ]
  RTNETLINK answers: File exists
  changing rps_default_mask in child ns don't affect the main one[ ok ]
  cat: /sys/class/net/vethC11an1/queues/rx-0/rps_cpus: No such file or directory
  changing rps_default_mask in child ns affects new childns devices./rps_default_mask.sh: line 36: [: -eq: unary operator expected
  [fail] expected 1 found
  changing rps_default_mask in child ns don't affect existing devices[ ok ]

After this patch:
  # ./rps_default_mask.sh
  empty rps_default_mask                                      [ ok ]
  changing rps_default_mask dont affect existing devices      [ ok ]
  changing rps_default_mask dont affect existing netns        [ ok ]
  changing rps_default_mask affect newly created devices      [ ok ]
  changing rps_default_mask don't affect newly child netns[II][ ok ]
  rps_default_mask is 0 by default in child netns             [ ok ]
  changing rps_default_mask in child ns don't affect the main one[ ok ]
  changing rps_default_mask in child ns affects new childns devices[ ok ]
  changing rps_default_mask in child ns don't affect existing devices[ ok ]

Fixes: 3a7d84eae03b ("self-tests: more rps self tests")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
Link: https://lore.kernel.org/r/20230404072411.879476-1-liuhangbin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonet: fec: make use of MDIO C45 quirk
Greg Ungerer [Tue, 4 Apr 2023 05:22:07 +0000 (15:22 +1000)]
net: fec: make use of MDIO C45 quirk

Not all fec MDIO bus drivers support C45 mode transactions. The older fec
hardware block in many ColdFire SoCs does not appear to support them, at
least according to most of the different ColdFire SoC reference manuals.
The bits used to generate C45 access on the iMX parts, in the OP field
of the MMFR register, are documented as generating non-compliant MII
frames (it is not documented as to exactly how they are non-compliant).

Commit 8d03ad1ab0b0 ("net: fec: Separate C22 and C45 transactions")
means the fec driver will always register c45 MDIO read and write
methods. During probe these will always be accessed now generating
non-compliant MII accesses on ColdFire based devices.

Add a quirk define, FEC_QUIRK_HAS_MDIO_C45, that can be used to
distinguish silicon that supports MDIO C45 framing or not. Add this to
all the existing iMX quirks, so they will be behave as they do now (*).

(*) it seems that some iMX parts may not support C45 transactions either.
    The iMX25 and iMX50 Reference Manuals contain similar wording to
    the ColdFire Reference Manuals on this.

Fixes: 8d03ad1ab0b0 ("net: fec: Separate C22 and C45 transactions")
Signed-off-by: Greg Ungerer <gerg@linux-m68k.org>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20230404052207.3064861-1-gerg@linux-m68k.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge tag 'wireless-2023-04-05' of git://git.kernel.org/pub/scm/linux/kernel/git...
Jakub Kicinski [Thu, 6 Apr 2023 00:24:26 +0000 (17:24 -0700)]
Merge tag 'wireless-2023-04-05' of git://git./linux/kernel/git/wireless/wireless

Kalle Valo says:

====================
wireless fixes for v6.3

mt76 has a fix for leaking cleartext frames on a certain scenario and
two firmware file handling related fixes. For brcmfmac we have a fix
for an older SDIO suspend regression and for ath11k avoiding a kernel
crash during hibernation with SUSE kernels.

* tag 'wireless-2023-04-05' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless:
  wifi: mt76: ignore key disable commands
  wifi: ath11k: reduce the MHI timeout to 20s
  wifi: mt76: mt7921: fix fw used for offload check for mt7922
  wifi: mt76: mt7921: Fix use-after-free in fw features query.
  wifi: brcmfmac: Fix SDIO suspend/resume regression
====================

Link: https://lore.kernel.org/r/20230405105536.4E946C433D2@smtp.kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge tag 'linux-can-fixes-for-6.3-20230405' of git://git.kernel.org/pub/scm/linux...
Jakub Kicinski [Thu, 6 Apr 2023 00:22:06 +0000 (17:22 -0700)]
Merge tag 'linux-can-fixes-for-6.3-20230405' of git://git./linux/kernel/git/mkl/linux-can

Marc Kleine-Budde says:

====================
pull-request: can 2023-04-05

The first patch is by Oleksij Rempel and fixes a out-of-bounds memory
access in the j1939 protocol.

The remaining 3 patches target the ISOTP protocol. Oliver Hartkopp
fixes the ISOTP protocol to pass information about dropped PDUs to the
user space via control messages. Michal Sojka's patch fixes poll() to
not forward false EPOLLOUT events. And Oliver Hartkopp fixes a race
condition between isotp_sendsmg() and isotp_release().

* tag 'linux-can-fixes-for-6.3-20230405' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can:
  can: isotp: fix race between isotp_sendsmg() and isotp_release()
  can: isotp: isotp_ops: fix poll() to not report false EPOLLOUT events
  can: isotp: isotp_recvmsg(): use sock_recv_cmsgs() to get SOCK_RXQ_OVFL infos
  can: j1939: j1939_tp_tx_dat_new(): fix out-of-bounds memory access
====================

Link: https://lore.kernel.org/r/20230405092444.1802340-1-mkl@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net...
Jakub Kicinski [Thu, 6 Apr 2023 00:10:32 +0000 (17:10 -0700)]
Merge branch '100GbE' of git://git./linux/kernel/git/tnguy/net-queue

Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2023-04-04 (ice)

This series contains updates to ice driver only.

Simei adjusts error path on adding VF Flow Director filters that were
not releasing all resources.

Lingyu adds setting/resetting of VF Flow Director filters counters
during initialization.

* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
  ice: Reset FDIR counter in FDIR init stage
  ice: fix wrong fallback logic for FDIR
====================

Link: https://lore.kernel.org/r/20230404172306.450880-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge tag 'drm-misc-fixes-2023-04-05' of git://anongit.freedesktop.org/drm/drm-misc...
Daniel Vetter [Wed, 5 Apr 2023 19:06:27 +0000 (21:06 +0200)]
Merge tag 'drm-misc-fixes-2023-04-05' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

Short summary of fixes pull:

 * ivpu: DMA fence and suspend fixes
 * nouveau: Color-depth fixes
 * panfrost: Fix mmap error handling

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230405182855.GA1551@linux-uq9g
20 months agoMerge tag 'trace-v6.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/trace...
Linus Torvalds [Wed, 5 Apr 2023 16:11:08 +0000 (09:11 -0700)]
Merge tag 'trace-v6.3-rc5' of git://git./linux/kernel/git/trace/linux-trace

Pull tracing fixes from Steven Rostedt:

 - Fix timerlat notification, as it was not triggering the notify to
   users when a new max latency was hit.

 - Do not trigger max latency if the tracing is off.

   When tracing is off, the ring buffer is not updated, it does not make
   sense to notify when there's a new max latency detected by the
   tracer, as why that latency happened is not available. The tracing
   logic still runs when the ring buffer is disabled, but it should not
   be triggering notifications.

 - Fix race on freeing the synthetic event "last_cmd" variable by adding
   a mutex around it.

 - Fix race between reader and writer of the ring buffer by adding
   memory barriers. When the writer is still on the reader page it must
   have its content visible on the buffer before it moves the commit
   index that the reader uses to know how much content is on the page.

 - Make get_lock_parent_ip() always inlined, as it uses _THIS_IP_ and
   _RET_IP_, which gets broken if it is not inlined.

 - Make __field(int, arr[5]) in a TRACE_EVENT() macro fail to build.

   The field formats of trace events are calculated by using
   sizeof(type) and other means by what is passed into the structure
   macros like __field(). The __field() macro is only meant for atom
   types like int, long, short, pointer, etc. It is not meant for
   arrays.

   The code will currently compile with arrays, but then the format
   produced will be inaccurate, and user space parsing tools will break.

   Two bugs have already been fixed, now add code that will make the
   kernel fail to build if another trace event includes this buggy field
   format.

 - Fix boot up snapshot code:

   Boot snapshots were triggering when not even asked for on the kernel
   command line. This was caused by two bugs:

    1) It would trigger a snapshot on any instance if one was created
       from the kernel command line.

    2) The error handling would only affect the top level instance.
       So the fact that a snapshot was done on a instance that didn't
       allocate a buffer triggered a warning written into the top level
       buffer, and worse yet, disabled the top level buffer.

 - Fix memory leak that was caused when an error was logged in a trace
   buffer instance, and then the buffer instance was removed.

   The allocated error log messages still needed to be freed.

* tag 'trace-v6.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing: Free error logs of tracing instances
  tracing: Fix ftrace_boot_snapshot command line logic
  tracing: Have tracing_snapshot_instance_cond() write errors to the appropriate instance
  tracing: Error if a trace event has an array for a __field()
  tracing/osnoise: Fix notify new tracing_max_latency
  tracing/timerlat: Notify new max thread latency
  ftrace: Mark get_lock_parent_ip() __always_inline
  ring-buffer: Fix race while reader and writer are on the same page
  tracing/synthetic: Fix races on freeing last_cmd

20 months agotracing: Free error logs of tracing instances
Steven Rostedt (Google) [Tue, 4 Apr 2023 23:45:04 +0000 (19:45 -0400)]
tracing: Free error logs of tracing instances

When a tracing instance is removed, the error messages that hold errors
that occurred in the instance needs to be freed. The following reports a
memory leak:

 # cd /sys/kernel/tracing
 # mkdir instances/foo
 # echo 'hist:keys=x' > instances/foo/events/sched/sched_switch/trigger
 # cat instances/foo/error_log
 [  117.404795] hist:sched:sched_switch: error: Couldn't find field
   Command: hist:keys=x
                      ^
 # rmdir instances/foo

Then check for memory leaks:

 # echo scan > /sys/kernel/debug/kmemleak
 # cat /sys/kernel/debug/kmemleak
unreferenced object 0xffff88810d8ec700 (size 192):
  comm "bash", pid 869, jiffies 4294950577 (age 215.752s)
  hex dump (first 32 bytes):
    60 dd 68 61 81 88 ff ff 60 dd 68 61 81 88 ff ff  `.ha....`.ha....
    a0 30 8c 83 ff ff ff ff 26 00 0a 00 00 00 00 00  .0......&.......
  backtrace:
    [<00000000dae26536>] kmalloc_trace+0x2a/0xa0
    [<00000000b2938940>] tracing_log_err+0x277/0x2e0
    [<000000004a0e1b07>] parse_atom+0x966/0xb40
    [<0000000023b24337>] parse_expr+0x5f3/0xdb0
    [<00000000594ad074>] event_hist_trigger_parse+0x27f8/0x3560
    [<00000000293a9645>] trigger_process_regex+0x135/0x1a0
    [<000000005c22b4f2>] event_trigger_write+0x87/0xf0
    [<000000002cadc509>] vfs_write+0x162/0x670
    [<0000000059c3b9be>] ksys_write+0xca/0x170
    [<00000000f1cddc00>] do_syscall_64+0x3e/0xc0
    [<00000000868ac68c>] entry_SYSCALL_64_after_hwframe+0x72/0xdc
unreferenced object 0xffff888170c35a00 (size 32):
  comm "bash", pid 869, jiffies 4294950577 (age 215.752s)
  hex dump (first 32 bytes):
    0a 20 20 43 6f 6d 6d 61 6e 64 3a 20 68 69 73 74  .  Command: hist
    3a 6b 65 79 73 3d 78 0a 00 00 00 00 00 00 00 00  :keys=x.........
  backtrace:
    [<000000006a747de5>] __kmalloc+0x4d/0x160
    [<000000000039df5f>] tracing_log_err+0x29b/0x2e0
    [<000000004a0e1b07>] parse_atom+0x966/0xb40
    [<0000000023b24337>] parse_expr+0x5f3/0xdb0
    [<00000000594ad074>] event_hist_trigger_parse+0x27f8/0x3560
    [<00000000293a9645>] trigger_process_regex+0x135/0x1a0
    [<000000005c22b4f2>] event_trigger_write+0x87/0xf0
    [<000000002cadc509>] vfs_write+0x162/0x670
    [<0000000059c3b9be>] ksys_write+0xca/0x170
    [<00000000f1cddc00>] do_syscall_64+0x3e/0xc0
    [<00000000868ac68c>] entry_SYSCALL_64_after_hwframe+0x72/0xdc

The problem is that the error log needs to be freed when the instance is
removed.

Link: https://lore.kernel.org/lkml/76134d9f-a5ba-6a0d-37b3-28310b4a1e91@alu.unizg.hr/
Link: https://lore.kernel.org/linux-trace-kernel/20230404194504.5790b95f@gandalf.local.home
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Thorsten Leemhuis <regressions@leemhuis.info>
Cc: Ulf Hansson <ulf.hansson@linaro.org>
Cc: Eric Biggers <ebiggers@kernel.org>
Fixes: 2f754e771b1a6 ("tracing: Have the error logs show up in the proper instances")
Reported-by: Mirsad Goran Todorovac <mirsad.todorovac@alu.unizg.hr>
Tested-by: Mirsad Todorovac <mirsad.todorovac@alu.unizg.hr>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
20 months agocan: isotp: fix race between isotp_sendsmg() and isotp_release()
Oliver Hartkopp [Fri, 31 Mar 2023 13:19:35 +0000 (15:19 +0200)]
can: isotp: fix race between isotp_sendsmg() and isotp_release()

As discussed with Dae R. Jeong and Hillf Danton here [1] the sendmsg()
function in isotp.c might get into a race condition when restoring the
former tx.state from the old_state.

Remove the old_state concept and implement proper locking for the
ISOTP_IDLE transitions in isotp_sendmsg(), inspired by a
simplification idea from Hillf Danton.

Introduce a new tx.state ISOTP_SHUTDOWN and use the same locking
mechanism from isotp_release() which resolves a potential race between
isotp_sendsmg() and isotp_release().

[1] https://lore.kernel.org/linux-can/ZB%2F93xJxq%2FBUqAgG@dragonet

v1: https://lore.kernel.org/all/20230331102114.15164-1-socketcan@hartkopp.net
v2: https://lore.kernel.org/all/20230331123600.3550-1-socketcan@hartkopp.net
    take care of signal interrupts for wait_event_interruptible() in
    isotp_release()
v3: https://lore.kernel.org/all/20230331130654.9886-1-socketcan@hartkopp.net
    take care of signal interrupts for wait_event_interruptible() in
    isotp_sendmsg() in the wait_tx_done case
v4: https://lore.kernel.org/all/20230331131935.21465-1-socketcan@hartkopp.net
    take care of signal interrupts for wait_event_interruptible() in
    isotp_sendmsg() in ALL cases

Cc: Dae R. Jeong <threeearcat@gmail.com>
Cc: Hillf Danton <hdanton@sina.com>
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Fixes: 4f027cba8216 ("can: isotp: split tx timer into transmission and timeout")
Link: https://lore.kernel.org/all/20230331131935.21465-1-socketcan@hartkopp.net
Cc: stable@vger.kernel.org
[mkl: rephrase commit message]
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
20 months agoMerge tag 'drm-intel-fixes-2023-04-05' of git://anongit.freedesktop.org/drm/drm-intel...
Daniel Vetter [Wed, 5 Apr 2023 09:14:18 +0000 (11:14 +0200)]
Merge tag 'drm-intel-fixes-2023-04-05' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

drm/i915 fixes for v6.3-rc6:
- Fix DP MST DSC M/N calculation to use compressed bpp
- Fix racy use-after-free in perf ioctl
- Fix context runtime accounting
- Fix handling of GT reset during HuC loading
- Fix use of unsigned vm_fault_t for error values

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87zg7mzomz.fsf@intel.com
20 months agocan: isotp: isotp_ops: fix poll() to not report false EPOLLOUT events
Michal Sojka [Fri, 31 Mar 2023 12:55:11 +0000 (14:55 +0200)]
can: isotp: isotp_ops: fix poll() to not report false EPOLLOUT events

When using select()/poll()/epoll() with a non-blocking ISOTP socket to
wait for when non-blocking write is possible, a false EPOLLOUT event
is sometimes returned. This can happen at least after sending a
message which must be split to multiple CAN frames.

The reason is that isotp_sendmsg() returns -EAGAIN when tx.state is
not equal to ISOTP_IDLE and this behavior is not reflected in
datagram_poll(), which is used in isotp_ops.

This is fixed by introducing ISOTP-specific poll function, which
suppresses the EPOLLOUT events in that case.

v2: https://lore.kernel.org/all/20230302092812.320643-1-michal.sojka@cvut.cz
v1: https://lore.kernel.org/all/20230224010659.48420-1-michal.sojka@cvut.cz
    https://lore.kernel.org/all/b53a04a2-ba1f-3858-84c1-d3eb3301ae15@hartkopp.net

Signed-off-by: Michal Sojka <michal.sojka@cvut.cz>
Reported-by: Jakub Jira <jirajak2@fel.cvut.cz>
Tested-by: Oliver Hartkopp <socketcan@hartkopp.net>
Acked-by: Oliver Hartkopp <socketcan@hartkopp.net>
Fixes: e057dd3fc20f ("can: add ISO 15765-2:2016 transport protocol")
Link: https://lore.kernel.org/all/20230331125511.372783-1-michal.sojka@cvut.cz
Cc: stable@vger.kernel.org
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
20 months agocan: isotp: isotp_recvmsg(): use sock_recv_cmsgs() to get SOCK_RXQ_OVFL infos
Oliver Hartkopp [Thu, 30 Mar 2023 17:02:48 +0000 (19:02 +0200)]
can: isotp: isotp_recvmsg(): use sock_recv_cmsgs() to get SOCK_RXQ_OVFL infos

isotp.c was still using sock_recv_timestamp() which does not provide
control messages to detect dropped PDUs in the receive path.

Fixes: e057dd3fc20f ("can: add ISO 15765-2:2016 transport protocol")
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Link: https://lore.kernel.org/all/20230330170248.62342-1-socketcan@hartkopp.net
Cc: stable@vger.kernel.org
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
20 months agocan: j1939: j1939_tp_tx_dat_new(): fix out-of-bounds memory access
Oleksij Rempel [Tue, 4 Apr 2023 07:31:28 +0000 (09:31 +0200)]
can: j1939: j1939_tp_tx_dat_new(): fix out-of-bounds memory access

In the j1939_tp_tx_dat_new() function, an out-of-bounds memory access
could occur during the memcpy() operation if the size of skb->cb is
larger than the size of struct j1939_sk_buff_cb. This is because the
memcpy() operation uses the size of skb->cb, leading to a read beyond
the struct j1939_sk_buff_cb.

Updated the memcpy() operation to use the size of struct
j1939_sk_buff_cb instead of the size of skb->cb. This ensures that the
memcpy() operation only reads the memory within the bounds of struct
j1939_sk_buff_cb, preventing out-of-bounds memory access.

Additionally, add a BUILD_BUG_ON() to check that the size of skb->cb
is greater than or equal to the size of struct j1939_sk_buff_cb. This
ensures that the skb->cb buffer is large enough to hold the
j1939_sk_buff_cb structure.

Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol")
Reported-by: Shuangpeng Bai <sjb7183@psu.edu>
Tested-by: Shuangpeng Bai <sjb7183@psu.edu>
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://groups.google.com/g/syzkaller/c/G_LL-C3plRs/m/-8xCi6dCAgAJ
Link: https://lore.kernel.org/all/20230404073128.3173900-1-o.rempel@pengutronix.de
Cc: stable@vger.kernel.org
[mkl: rephrase commit message]
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
20 months agoaccel/ivpu: Fix S3 system suspend when not idle
Jacek Lawrynowicz [Fri, 31 Mar 2023 11:36:03 +0000 (13:36 +0200)]
accel/ivpu: Fix S3 system suspend when not idle

Wait for VPU to be idle in ivpu_pm_suspend_cb() before powering off
the device, so jobs are not lost and TDRs are not triggered after
resume.

Fixes: 852be13f3bd3 ("accel/ivpu: Add PM support")
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230331113603.2802515-3-stanislaw.gruszka@linux.intel.com
20 months agoaccel/ivpu: Add dma fence to command buffers only
Karol Wachowski [Fri, 31 Mar 2023 11:36:02 +0000 (13:36 +0200)]
accel/ivpu: Add dma fence to command buffers only

Currently job->done_fence is added to every BO handle within a job. If job
handle (command buffer) is shared between multiple submits, KMD will add
the fence in each of them. Then bo_wait_ioctl() executed on command buffer
will exit only when all jobs containing that handle are done.

This creates deadlock scenario for user mode driver in case when job handle
is added as dependency of another job, because bo_wait_ioctl() of first job
will wait until second job finishes, and second job can not finish before
first one.

Having fences added only to job buffer handle allows user space to execute
bo_wait_ioctl() on the job even if it's handle is submitted with other job.

Fixes: cd7272215c44 ("accel/ivpu: Add command buffer submission logic")
Signed-off-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230331113603.2802515-2-stanislaw.gruszka@linux.intel.com
20 months agotracing: Fix ftrace_boot_snapshot command line logic
Steven Rostedt (Google) [Wed, 5 Apr 2023 02:21:15 +0000 (22:21 -0400)]
tracing: Fix ftrace_boot_snapshot command line logic

The kernel command line ftrace_boot_snapshot by itself is supposed to
trigger a snapshot at the end of boot up of the main top level trace
buffer. A ftrace_boot_snapshot=foo will do the same for an instance called
foo that was created by trace_instance=foo,...

The logic was broken where if ftrace_boot_snapshot was by itself, it would
trigger a snapshot for all instances that had tracing enabled, regardless
if it asked for a snapshot or not.

When a snapshot is requested for a buffer, the buffer's
tr->allocated_snapshot is set to true. Use that to know if a trace buffer
wants a snapshot at boot up or not.

Since the top level buffer is part of the ftrace_trace_arrays list,
there's no reason to treat it differently than the other buffers. Just
iterate the list if ftrace_boot_snapshot was specified.

Link: https://lkml.kernel.org/r/20230405022341.895334039@goodmis.org
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ross Zwisler <zwisler@google.com>
Fixes: 9c1c251d670bc ("tracing: Allow boot instances to have snapshot buffers")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
20 months agotracing: Have tracing_snapshot_instance_cond() write errors to the appropriate instance
Steven Rostedt (Google) [Wed, 5 Apr 2023 02:21:14 +0000 (22:21 -0400)]
tracing: Have tracing_snapshot_instance_cond() write errors to the appropriate instance

If a trace instance has a failure with its snapshot code, the error
message is to be written to that instance's buffer. But currently, the
message is written to the top level buffer. Worse yet, it may also disable
the top level buffer and not the instance that had the issue.

Link: https://lkml.kernel.org/r/20230405022341.688730321@goodmis.org
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ross Zwisler <zwisler@google.com>
Fixes: 2824f50332486 ("tracing: Make the snapshot trigger work with instances")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
20 months agogve: Secure enough bytes in the first TX desc for all TCP pkts
Shailend Chand [Mon, 3 Apr 2023 17:28:09 +0000 (10:28 -0700)]
gve: Secure enough bytes in the first TX desc for all TCP pkts

Non-GSO TCP packets whose SKBs' linear portion did not include the
entire TCP header were not populating the first Tx descriptor with
as many bytes as the vNIC expected. This change ensures that all
TCP packets populate the first descriptor with the correct number of
bytes.

Fixes: 893ce44df565 ("gve: Add basic driver framework for Compute Engine Virtual NIC")
Signed-off-by: Shailend Chand <shailend@google.com>
Link: https://lore.kernel.org/r/20230403172809.2939306-1-shailend@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agonetlink: annotate lockless accesses to nlk->max_recvmsg_len
Eric Dumazet [Mon, 3 Apr 2023 21:46:43 +0000 (21:46 +0000)]
netlink: annotate lockless accesses to nlk->max_recvmsg_len

syzbot reported a data-race in data-race in netlink_recvmsg() [1]

Indeed, netlink_recvmsg() can be run concurrently,
and netlink_dump() also needs protection.

[1]
BUG: KCSAN: data-race in netlink_recvmsg / netlink_recvmsg

read to 0xffff888141840b38 of 8 bytes by task 23057 on cpu 0:
netlink_recvmsg+0xea/0x730 net/netlink/af_netlink.c:1988
sock_recvmsg_nosec net/socket.c:1017 [inline]
sock_recvmsg net/socket.c:1038 [inline]
__sys_recvfrom+0x1ee/0x2e0 net/socket.c:2194
__do_sys_recvfrom net/socket.c:2212 [inline]
__se_sys_recvfrom net/socket.c:2208 [inline]
__x64_sys_recvfrom+0x78/0x90 net/socket.c:2208
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x63/0xcd

write to 0xffff888141840b38 of 8 bytes by task 23037 on cpu 1:
netlink_recvmsg+0x114/0x730 net/netlink/af_netlink.c:1989
sock_recvmsg_nosec net/socket.c:1017 [inline]
sock_recvmsg net/socket.c:1038 [inline]
____sys_recvmsg+0x156/0x310 net/socket.c:2720
___sys_recvmsg net/socket.c:2762 [inline]
do_recvmmsg+0x2e5/0x710 net/socket.c:2856
__sys_recvmmsg net/socket.c:2935 [inline]
__do_sys_recvmmsg net/socket.c:2958 [inline]
__se_sys_recvmmsg net/socket.c:2951 [inline]
__x64_sys_recvmmsg+0xe2/0x160 net/socket.c:2951
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x63/0xcd

value changed: 0x0000000000000000 -> 0x0000000000001000

Reported by Kernel Concurrency Sanitizer on:
CPU: 1 PID: 23037 Comm: syz-executor.2 Not tainted 6.3.0-rc4-syzkaller-00195-g5a57b48fdfcb #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/02/2023

Fixes: 9063e21fb026 ("netlink: autosize skb lengthes")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Link: https://lore.kernel.org/r/20230403214643.768555-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoethtool: reset #lanes when lanes is omitted
Andy Roulin [Mon, 3 Apr 2023 21:20:53 +0000 (14:20 -0700)]
ethtool: reset #lanes when lanes is omitted

If the number of lanes was forced and then subsequently the user
omits this parameter, the ksettings->lanes is reset. The driver
should then reset the number of lanes to the device's default
for the specified speed.

However, although the ksettings->lanes is set to 0, the mod variable
is not set to true to indicate the driver and userspace should be
notified of the changes.

The consequence is that the same ethtool operation will produce
different results based on the initial state.

If the initial state is:
$ ethtool swp1 | grep -A 3 'Speed: '
        Speed: 500000Mb/s
        Lanes: 2
        Duplex: Full
        Auto-negotiation: on

then executing 'ethtool -s swp1 speed 50000 autoneg off' will yield:
$ ethtool swp1 | grep -A 3 'Speed: '
        Speed: 500000Mb/s
        Lanes: 2
        Duplex: Full
        Auto-negotiation: off

While if the initial state is:
$ ethtool swp1 | grep -A 3 'Speed: '
        Speed: 500000Mb/s
        Lanes: 1
        Duplex: Full
        Auto-negotiation: off

executing the same 'ethtool -s swp1 speed 50000 autoneg off' results in:
$ ethtool swp1 | grep -A 3 'Speed: '
        Speed: 500000Mb/s
        Lanes: 1
        Duplex: Full
        Auto-negotiation: off

This patch fixes this behavior. Omitting lanes will always results in
the driver choosing the default lane width for the chosen speed. In this
scenario, regardless of the initial state, the end state will be, e.g.,

$ ethtool swp1 | grep -A 3 'Speed: '
        Speed: 500000Mb/s
        Lanes: 2
        Duplex: Full
        Auto-negotiation: off

Fixes: 012ce4dd3102 ("ethtool: Extend link modes settings uAPI with lanes")
Signed-off-by: Andy Roulin <aroulin@nvidia.com>
Reviewed-by: Danielle Ratson <danieller@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://lore.kernel.org/r/ac238d6b-8726-8156-3810-6471291dbc7f@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge branch 'raw-ping-fix-locking-in-proc-net-raw-icmp'
Jakub Kicinski [Wed, 5 Apr 2023 01:56:59 +0000 (18:56 -0700)]
Merge branch 'raw-ping-fix-locking-in-proc-net-raw-icmp'

Kuniyuki Iwashima says:

====================
raw/ping: Fix locking in /proc/net/{raw,icmp}.

The first patch fixes a NULL deref for /proc/net/raw and second one fixes
the same issue for ping sockets.

The first patch also converts hlist_nulls to hlist, but this is because
the current code uses sk_nulls_for_each() for lockless readers, instead
of sk_nulls_for_each_rcu() which adds memory barrier, but raw sockets
does not use the nulls marker nor SLAB_TYPESAFE_BY_RCU in the first place.

OTOH, the ping sockets already uses sk_nulls_for_each_rcu(), and such
conversion can be posted later for net-next.
====================

Link: https://lore.kernel.org/r/20230403194959.48928-1-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoping: Fix potentail NULL deref for /proc/net/icmp.
Kuniyuki Iwashima [Mon, 3 Apr 2023 19:49:59 +0000 (12:49 -0700)]
ping: Fix potentail NULL deref for /proc/net/icmp.

After commit dbca1596bbb0 ("ping: convert to RCU lookups, get rid
of rwlock"), we use RCU for ping sockets, but we should use spinlock
for /proc/net/icmp to avoid a potential NULL deref mentioned in
the previous patch.

Let's go back to using spinlock there.

Note we can convert ping sockets to use hlist instead of hlist_nulls
because we do not use SLAB_TYPESAFE_BY_RCU for ping sockets.

Fixes: dbca1596bbb0 ("ping: convert to RCU lookups, get rid of rwlock")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoraw: Fix NULL deref in raw_get_next().
Kuniyuki Iwashima [Mon, 3 Apr 2023 19:49:58 +0000 (12:49 -0700)]
raw: Fix NULL deref in raw_get_next().

Dae R. Jeong reported a NULL deref in raw_get_next() [0].

It seems that the repro was running these sequences in parallel so
that one thread was iterating on a socket that was being freed in
another netns.

  unshare(0x40060200)
  r0 = syz_open_procfs(0x0, &(0x7f0000002080)='net/raw\x00')
  socket$inet_icmp_raw(0x2, 0x3, 0x1)
  pread64(r0, &(0x7f0000000000)=""/10, 0xa, 0x10000000007f)

After commit 0daf07e52709 ("raw: convert raw sockets to RCU"), we
use RCU and hlist_nulls_for_each_entry() to iterate over SOCK_RAW
sockets.  However, we should use spinlock for slow paths to avoid
the NULL deref.

Also, SOCK_RAW does not use SLAB_TYPESAFE_BY_RCU, and the slab object
is not reused during iteration in the grace period.  In fact, the
lockless readers do not check the nulls marker with get_nulls_value().
So, SOCK_RAW should use hlist instead of hlist_nulls.

Instead of adding an unnecessary barrier by sk_nulls_for_each_rcu(),
let's convert hlist_nulls to hlist and use sk_for_each_rcu() for
fast paths and sk_for_each() and spinlock for /proc/net/raw.

[0]:
general protection fault, probably for non-canonical address 0xdffffc0000000005: 0000 [#1] PREEMPT SMP KASAN
KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
CPU: 2 PID: 20952 Comm: syz-executor.0 Not tainted 6.2.0-g048ec869bafd-dirty #7
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS:  00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055bb9614b35f CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <TASK>
 seq_read_iter+0x4c6/0x10f0 fs/seq_file.c:225
 seq_read+0x224/0x320 fs/seq_file.c:162
 pde_read fs/proc/inode.c:316 [inline]
 proc_reg_read+0x23f/0x330 fs/proc/inode.c:328
 vfs_read+0x31e/0xd30 fs/read_write.c:468
 ksys_pread64 fs/read_write.c:665 [inline]
 __do_sys_pread64 fs/read_write.c:675 [inline]
 __se_sys_pread64 fs/read_write.c:672 [inline]
 __x64_sys_pread64+0x1e9/0x280 fs/read_write.c:672
 do_syscall_x64 arch/x86/entry/common.c:51 [inline]
 do_syscall_64+0x4e/0xa0 arch/x86/entry/common.c:82
 entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x478d29
Code: f7 d8 64 89 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 bc ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f843ae8dbe8 EFLAGS: 00000246 ORIG_RAX: 0000000000000011
RAX: ffffffffffffffda RBX: 0000000000791408 RCX: 0000000000478d29
RDX: 000000000000000a RSI: 0000000020000000 RDI: 0000000000000003
RBP: 00000000f477909a R08: 0000000000000000 R09: 0000000000000000
R10: 000010000000007f R11: 0000000000000246 R12: 0000000000791740
R13: 0000000000791414 R14: 0000000000791408 R15: 00007ffc2eb48a50
 </TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---
RIP: 0010:read_pnet include/net/net_namespace.h:383 [inline]
RIP: 0010:sock_net include/net/sock.h:649 [inline]
RIP: 0010:raw_get_next net/ipv4/raw.c:974 [inline]
RIP: 0010:raw_get_idx net/ipv4/raw.c:986 [inline]
RIP: 0010:raw_seq_start+0x431/0x800 net/ipv4/raw.c:995
Code: ef e8 33 3d 94 f7 49 8b 6d 00 4c 89 ef e8 b7 65 5f f7 49 89 ed 49 83 c5 98 0f 84 9a 00 00 00 48 83 c5 c8 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 00 3d 94 f7 4c 8b 7d 00 48 89 ef
RSP: 0018:ffffc9001154f9b0 EFLAGS: 00010206
RAX: 0000000000000005 RBX: 1ffff1100302c8fd RCX: 0000000000000000
RDX: 0000000000000028 RSI: ffffc9001154f988 RDI: ffffc9000f77a338
RBP: 0000000000000029 R08: ffffffff8a50ffb4 R09: fffffbfff24b6bd9
R10: fffffbfff24b6bd9 R11: 0000000000000000 R12: ffff88801db73b78
R13: fffffffffffffff9 R14: dffffc0000000000 R15: 0000000000000030
FS:  00007f843ae8e700(0000) GS:ffff888063700000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f92ff166000 CR3: 000000003c672000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400

Fixes: 0daf07e52709 ("raw: convert raw sockets to RCU")
Reported-by: syzbot <syzkaller@googlegroups.com>
Reported-by: Dae R. Jeong <threeearcat@gmail.com>
Link: https://lore.kernel.org/netdev/ZCA2mGV_cmq7lIfV@dragonet/
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
20 months agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Tue, 4 Apr 2023 18:29:37 +0000 (11:29 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull kvm fixes from Paolo Bonzini:
 "PPC:
   - Hide KVM_CAP_IRQFD_RESAMPLE if XIVE is enabled

  s390:
   - Fix handling of external interrupts in protected guests

  x86:
   - Resample the pending state of IOAPIC interrupts when unmasking them

   - Fix usage of Hyper-V "enlightened TLB" on AMD

   - Small fixes to real mode exceptions

   - Suppress pending MMIO write exits if emulator detects exception

  Documentation:
   - Fix rST syntax"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  docs: kvm: x86: Fix broken field list
  KVM: PPC: Make KVM_CAP_IRQFD_RESAMPLE platform dependent
  KVM: s390: pv: fix external interruption loop not always detected
  KVM: nVMX: Do not report error code when synthesizing VM-Exit from Real Mode
  KVM: x86: Clear "has_error_code", not "error_code", for RM exception injection
  KVM: x86: Suppress pending MMIO write exits if emulator detects exception
  KVM: x86/ioapic: Resample the pending state of an IRQ when unmasking
  KVM: irqfd: Make resampler_list an RCU list
  KVM: SVM: Flush Hyper-V TLB when required

20 months agoMerge tag 'nfsd-6.3-5' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux
Linus Torvalds [Tue, 4 Apr 2023 18:20:55 +0000 (11:20 -0700)]
Merge tag 'nfsd-6.3-5' of git://git./linux/kernel/git/cel/linux

Pull nfsd fixes from Chuck Lever:

 - Fix a crash and a resource leak in NFSv4 COMPOUND processing

 - Fix issues with AUTH_SYS credential handling

 - Try again to address an NFS/NFSD/SUNRPC build dependency regression

* tag 'nfsd-6.3-5' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
  NFSD: callback request does not use correct credential for AUTH_SYS
  NFS: Remove "select RPCSEC_GSS_KRB5
  sunrpc: only free unix grouplist after RCU settles
  nfsd: call op_release, even when op_func returns an error
  NFSD: Avoid calling OPDESC() with ops->opnum == OP_ILLEGAL

20 months agodocs: kvm: x86: Fix broken field list
Takahiro Itazuri [Fri, 31 Mar 2023 09:31:16 +0000 (10:31 +0100)]
docs: kvm: x86: Fix broken field list

Add a missing ":" to fix a broken field list.

Signed-off-by: Takahiro Itazuri <itazur@amazon.com>
Fixes: ba7bb663f554 ("KVM: x86: Provide per VM capability for disabling PMU virtualization")
Message-Id: <20230331093116.99820-1-itazur@amazon.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
20 months agoasm-generic: avoid __generic_cmpxchg_local warnings
Arnd Bergmann [Thu, 2 Mar 2023 08:53:31 +0000 (09:53 +0100)]
asm-generic: avoid __generic_cmpxchg_local warnings

Code that passes a 32-bit constant into cmpxchg() produces a harmless
sparse warning because of the truncation in the branch that is not taken:

fs/erofs/zdata.c: note: in included file (through /home/arnd/arm-soc/arch/arm/include/asm/cmpxchg.h, /home/arnd/arm-soc/arch/arm/include/asm/atomic.h, /home/arnd/arm-soc/include/linux/atomic.h, ...):
include/asm-generic/cmpxchg-local.h:29:33: warning: cast truncates bits from constant value (5f0ecafe becomes fe)
include/asm-generic/cmpxchg-local.h:33:34: warning: cast truncates bits from constant value (5f0ecafe becomes cafe)
include/asm-generic/cmpxchg-local.h:29:33: warning: cast truncates bits from constant value (5f0ecafe becomes fe)
include/asm-generic/cmpxchg-local.h:30:42: warning: cast truncates bits from constant value (5f0edead becomes ad)
include/asm-generic/cmpxchg-local.h:33:34: warning: cast truncates bits from constant value (5f0ecafe becomes cafe)
include/asm-generic/cmpxchg-local.h:34:44: warning: cast truncates bits from constant value (5f0edead becomes dead)

This was reported as a regression to Matt's recent __generic_cmpxchg_local
patch, though this patch only added more warnings on top of the ones
that were already there.

Rewording the truncation to use an explicit bitmask instead of a cast
to a smaller type avoids the warning but otherwise leaves the code
unchanged.

I had another look at why the cast is even needed for atomic_cmpxchg(),
and as Matt describes the problem here is that atomic_t contains a
signed 'int', but cmpxchg() takes an 'unsigned long' argument, and
converting between the two leads to a 64-bit sign-extension of
negative 32-bit atomics.

I checked the other implementations of arch_cmpxchg() and did not find
any others that run into the same problem as __generic_cmpxchg_local(),
but it's easy to be on the safe side here and always convert the
signed int into an unsigned int when calling arch_cmpxchg(), as this
will work even when any of the arch_cmpxchg() implementations run
into the same problem.

Fixes: 624654152284 ("locking/atomic: cmpxchg: Make __generic_cmpxchg_local compare against zero-extended 'old' value")
Reviewed-by: Matt Evans <mev@rivosinc.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
20 months agoasm-generic/io.h: suppress endianness warnings for relaxed accessors
Vladimir Oltean [Mon, 9 Jan 2023 13:11:53 +0000 (15:11 +0200)]
asm-generic/io.h: suppress endianness warnings for relaxed accessors

Copy the forced type casts from the normal MMIO accessors to suppress
the sparse warnings that point out __raw_readl() returns a native endian
word (just like readl()).

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
20 months agoasm-generic/io.h: suppress endianness warnings for readq() and writeq()
Vladimir Oltean [Mon, 9 Jan 2023 13:11:52 +0000 (15:11 +0200)]
asm-generic/io.h: suppress endianness warnings for readq() and writeq()

Commit c1d55d50139b ("asm-generic/io.h: Fix sparse warnings on
big-endian architectures") missed fixing the 64-bit accessors.

Arnd explains in the attached link why the casts are necessary, even if
__raw_readq() and __raw_writeq() do not take endian-specific types.

Link: https://lore.kernel.org/lkml/9105d6fc-880b-4734-857d-e3d30b87ccf6@app.fastmail.com/
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
20 months agoice: Reset FDIR counter in FDIR init stage
Lingyu Liu [Tue, 28 Mar 2023 10:49:11 +0000 (10:49 +0000)]
ice: Reset FDIR counter in FDIR init stage

Reset the FDIR counters when FDIR inits. Without this patch,
when VF initializes or resets, all the FDIR counters are not
cleaned, which may cause unexpected behaviors for future FDIR
rule create (e.g., rule conflict).

Fixes: 1f7ea1cd6a37 ("ice: Enable FDIR Configure for AVF")
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
Signed-off-by: Lingyu Liu <lingyu.liu@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
20 months agoice: fix wrong fallback logic for FDIR
Simei Su [Wed, 22 Mar 2023 02:24:15 +0000 (10:24 +0800)]
ice: fix wrong fallback logic for FDIR

When adding a FDIR filter, if ice_vc_fdir_set_irq_ctx returns failure,
the inserted fdir entry will not be removed and if ice_vc_fdir_write_fltr
returns failure, the fdir context info for irq handler will not be cleared
which may lead to inconsistent or memory leak issue. This patch refines
failure cases to resolve this issue.

Fixes: 1f7ea1cd6a37 ("ice: Enable FDIR Configure for AVF")
Signed-off-by: Simei Su <simei.su@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
20 months agoNFSD: callback request does not use correct credential for AUTH_SYS
Dai Ngo [Sat, 1 Apr 2023 20:22:08 +0000 (13:22 -0700)]
NFSD: callback request does not use correct credential for AUTH_SYS

Currently callback request does not use the credential specified in
CREATE_SESSION if the security flavor for the back channel is AUTH_SYS.

Problem was discovered by pynfs 4.1 DELEG5 and DELEG7 test with error:
DELEG5   st_delegation.testCBSecParms     : FAILURE
           expected callback with uid, gid == 17, 19, got 0, 0

Signed-off-by: Dai Ngo <dai.ngo@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Fixes: 8276c902bbe9 ("SUNRPC: remove uid and gid from struct auth_cred")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
20 months agoNFS: Remove "select RPCSEC_GSS_KRB5
Chuck Lever [Tue, 28 Mar 2023 17:47:58 +0000 (13:47 -0400)]
NFS: Remove "select RPCSEC_GSS_KRB5

If CONFIG_CRYPTO=n (e.g. arm/shmobile_defconfig):

   WARNING: unmet direct dependencies detected for RPCSEC_GSS_KRB5
     Depends on [n]: NETWORK_FILESYSTEMS [=y] && SUNRPC [=y] && CRYPTO [=n]
     Selected by [y]:
     - NFS_V4 [=y] && NETWORK_FILESYSTEMS [=y] && NFS_FS [=y]

As NFSv4 can work without crypto enabled, remove the RPCSEC_GSS_KRB5
dependency altogether.

Trond says:
> It is possible to use the NFSv4.1 client with just AUTH_SYS, and
> in fact there are plenty of people out there using only that. The
> fact that RFC5661 gets its knickers in a twist about RPCSEC_GSS
> support is largely irrelevant to those people.
>
> The other issue is that ’select’ enforces the strict dependency
> that if the NFS client is compiled into the kernel, then the
> RPCSEC_GSS and kerberos code needs to be compiled in as well: they
> cannot exist as modules.

Fixes: e57d06527738 ("NFS & NFSD: Update GSS dependencies")
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Suggested-by: Trond Myklebust <trondmy@hammerspace.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
20 months agosunrpc: only free unix grouplist after RCU settles
Jeff Layton [Thu, 30 Mar 2023 18:24:27 +0000 (14:24 -0400)]
sunrpc: only free unix grouplist after RCU settles

While the unix_gid object is rcu-freed, the group_info list that it
contains is not. Ensure that we only put the group list reference once
we are really freeing the unix_gid object.

Reported-by: Zhi Li <yieli@redhat.com>
Link: https://bugzilla.redhat.com/show_bug.cgi?id=2183056
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Fixes: fd5d2f78261b ("SUNRPC: Make server side AUTH_UNIX use lockless lookups")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
20 months agonet: stmmac: fix up RX flow hash indirection table when setting channels
Corinna Vinschen [Mon, 3 Apr 2023 12:11:20 +0000 (14:11 +0200)]
net: stmmac: fix up RX flow hash indirection table when setting channels

stmmac_reinit_queues() fails to fix up the RX hash.  Even if the number
of channels gets restricted, the output of `ethtool -x' indicates that
all RX queues are used:

  $ ethtool -l enp0s29f2
  Channel parameters for enp0s29f2:
  Pre-set maximums:
  RX: 8
  TX: 8
  Other: n/a
  Combined: n/a
  Current hardware settings:
  RX: 8
  TX: 8
  Other: n/a
  Combined: n/a
  $ ethtool -x enp0s29f2
  RX flow hash indirection table for enp0s29f2 with 8 RX ring(s):
      0:      0     1     2     3     4     5     6     7
      8:      0     1     2     3     4     5     6     7
  [...]
  $ ethtool -L enp0s29f2 rx 3
  $ ethtool -x enp0s29f2
  RX flow hash indirection table for enp0s29f2 with 3 RX ring(s):
      0:      0     1     2     3     4     5     6     7
      8:      0     1     2     3     4     5     6     7
  [...]

Fix this by setting the indirection table according to the number
of specified queues.  The result is now as expected:

  $ ethtool -L enp0s29f2 rx 3
  $ ethtool -x enp0s29f2
  RX flow hash indirection table for enp0s29f2 with 3 RX ring(s):
      0:      0     1     2     0     1     2     0     1
      8:      2     0     1     2     0     1     2     0
  [...]

Tested on Intel Elkhart Lake.

Fixes: 0366f7e06a6b ("net: stmmac: add ethtool support for get/set channels")
Signed-off-by: Corinna Vinschen <vinschen@redhat.com>
Link: https://lore.kernel.org/r/20230403121120.489138-1-vinschen@redhat.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
20 months agoiommufd: Do not corrupt the pfn list when doing batch carry
Jason Gunthorpe [Fri, 31 Mar 2023 15:32:26 +0000 (12:32 -0300)]
iommufd: Do not corrupt the pfn list when doing batch carry

If batch->end is 0 then setting npfns[0] before computing the new value of
pfns will fail to adjust the pfn and result in various page accounting
corruptions. It should be ordered after.

This seems to result in various kinds of page meta-data corruption related
failures:

  WARNING: CPU: 1 PID: 527 at mm/gup.c:75 try_grab_folio+0x503/0x740
  Modules linked in:
  CPU: 1 PID: 527 Comm: repro Not tainted 6.3.0-rc2-eeac8ede1755+ #1
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014
  RIP: 0010:try_grab_folio+0x503/0x740
  Code: e3 01 48 89 de e8 6d c1 dd ff 48 85 db 0f 84 7c fe ff ff e8 4f bf dd ff 49 8d 47 ff 48 89 45 d0 e9 73 fe ff ff e8 3d bf dd ff <0f> 0b 31 db e9 d0 fc ff ff e8 2f bf dd ff 48 8b 5d c8 31 ff 48 89
  RSP: 0018:ffffc90000f37908 EFLAGS: 00010046
  RAX: 0000000000000000 RBX: 00000000fffffc02 RCX: ffffffff81504c26
  RDX: 0000000000000000 RSI: ffff88800d030000 RDI: 0000000000000002
  RBP: ffffc90000f37948 R08: 000000000003ca24 R09: 0000000000000008
  R10: 000000000003ca00 R11: 0000000000000023 R12: ffffea000035d540
  R13: 0000000000000001 R14: 0000000000000000 R15: ffffea000035d540
  FS:  00007fecbf659740(0000) GS:ffff88807dd00000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 00000000200011c3 CR3: 000000000ef66006 CR4: 0000000000770ee0
  PKRU: 55555554
  Call Trace:
   <TASK>
   internal_get_user_pages_fast+0xd32/0x2200
   pin_user_pages_fast+0x65/0x90
   pfn_reader_user_pin+0x376/0x390
   pfn_reader_next+0x14a/0x7b0
   pfn_reader_first+0x140/0x1b0
   iopt_area_fill_domain+0x74/0x210
   iopt_table_add_domain+0x30e/0x6e0
   iommufd_device_selftest_attach+0x7f/0x140
   iommufd_test+0x10ff/0x16f0
   iommufd_fops_ioctl+0x206/0x330
   __x64_sys_ioctl+0x10e/0x160
   do_syscall_64+0x3b/0x90
   entry_SYSCALL_64_after_hwframe+0x72/0xdc

Cc: <stable@vger.kernel.org>
Fixes: f394576eb11d ("iommufd: PFN handling for iopt_pages")
Link: https://lore.kernel.org/r/3-v1-ceab6a4d7d7a+94-iommufd_syz_jgg@nvidia.com
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reported-by: Pengfei Xu <pengfei.xu@intel.com>
Tested-by: Pengfei Xu <pengfei.xu@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
20 months agoiommufd: Fix unpinning of pages when an access is present
Jason Gunthorpe [Fri, 31 Mar 2023 15:32:25 +0000 (12:32 -0300)]
iommufd: Fix unpinning of pages when an access is present

syzkaller found that the calculation of batch_last_index should use
'start_index' since at input to this function the batch is either empty or
it has already been adjusted to cross any accesses so it will start at the
point we are unmapping from.

Getting this wrong causes the unmap to run over the end of the pages
which corrupts pages that were never mapped. In most cases this triggers
the num pinned debugging:

  WARNING: CPU: 0 PID: 557 at drivers/iommu/iommufd/pages.c:294 __iopt_area_unfill_domain+0x152/0x560
  Modules linked in:
  CPU: 0 PID: 557 Comm: repro Not tainted 6.3.0-rc2-eeac8ede1755 #1
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014
  RIP: 0010:__iopt_area_unfill_domain+0x152/0x560
  Code: d2 0f ff 44 8b 64 24 54 48 8b 44 24 48 31 ff 44 89 e6 48 89 44 24 38 e8 fc d3 0f ff 45 85 e4 0f 85 eb 01 00 00 e8 0e d2 0f ff <0f> 0b e8 07 d2 0f ff 48 8b 44 24 38 89 5c 24 58 89 18 8b 44 24 54
  RSP: 0018:ffffc9000108baf0 EFLAGS: 00010246
  RAX: 0000000000000000 RBX: 00000000ffffffff RCX: ffffffff821e3f85
  RDX: 0000000000000000 RSI: ffff88800faf0000 RDI: 0000000000000002
  RBP: ffffc9000108bd18 R08: 000000000003ca25 R09: 0000000000000014
  R10: 000000000003ca00 R11: 0000000000000024 R12: 0000000000000004
  R13: 0000000000000801 R14: 00000000000007ff R15: 0000000000000800
  FS:  00007f3499ce1740(0000) GS:ffff88807dc00000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 0000000020000243 CR3: 00000000179c2001 CR4: 0000000000770ef0
  PKRU: 55555554
  Call Trace:
   <TASK>
   iopt_area_unfill_domain+0x32/0x40
   iopt_table_remove_domain+0x23f/0x4c0
   iommufd_device_selftest_detach+0x3a/0x90
   iommufd_selftest_destroy+0x55/0x70
   iommufd_object_destroy_user+0xce/0x130
   iommufd_destroy+0xa2/0xc0
   iommufd_fops_ioctl+0x206/0x330
   __x64_sys_ioctl+0x10e/0x160
   do_syscall_64+0x3b/0x90
   entry_SYSCALL_64_after_hwframe+0x72/0xdc

Also add some useful WARN_ON sanity checks.

Cc: <stable@vger.kernel.org>
Fixes: 8d160cd4d506 ("iommufd: Algorithms for PFN storage")
Link: https://lore.kernel.org/r/2-v1-ceab6a4d7d7a+94-iommufd_syz_jgg@nvidia.com
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reported-by: Pengfei Xu <pengfei.xu@intel.com>
Tested-by: Pengfei Xu <pengfei.xu@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
20 months agoiommufd: Check for uptr overflow
Jason Gunthorpe [Fri, 31 Mar 2023 15:32:24 +0000 (12:32 -0300)]
iommufd: Check for uptr overflow

syzkaller found that setting up a map with a user VA that wraps past zero
can trigger WARN_ONs, particularly from pin_user_pages weirdly returning 0
due to invalid arguments.

Prevent creating a pages with a uptr and size that would math overflow.

  WARNING: CPU: 0 PID: 518 at drivers/iommu/iommufd/pages.c:793 pfn_reader_user_pin+0x2e6/0x390
  Modules linked in:
  CPU: 0 PID: 518 Comm: repro Not tainted 6.3.0-rc2-eeac8ede1755+ #1
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014
  RIP: 0010:pfn_reader_user_pin+0x2e6/0x390
  Code: b1 11 e9 25 fe ff ff e8 28 e4 0f ff 31 ff 48 89 de e8 2e e6 0f ff 48 85 db 74 0a e8 14 e4 0f ff e9 4d ff ff ff e8 0a e4 0f ff <0f> 0b bb f2 ff ff ff e9 3c ff ff ff e8 f9 e3 0f ff ba 01 00 00 00
  RSP: 0018:ffffc90000f9fa30 EFLAGS: 00010246
  RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffffffff821e2b72
  RDX: 0000000000000000 RSI: ffff888014184680 RDI: 0000000000000002
  RBP: ffffc90000f9fa78 R08: 00000000000000ff R09: 0000000079de6f4e
  R10: ffffc90000f9f790 R11: ffff888014185418 R12: ffffc90000f9fc60
  R13: 0000000000000002 R14: ffff888007879800 R15: 0000000000000000
  FS:  00007f4227555740(0000) GS:ffff88807dc00000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 0000000020000043 CR3: 000000000e748005 CR4: 0000000000770ef0
  PKRU: 55555554
  Call Trace:
   <TASK>
   pfn_reader_next+0x14a/0x7b0
   ? interval_tree_double_span_iter_update+0x11a/0x140
   pfn_reader_first+0x140/0x1b0
   iopt_pages_rw_slow+0x71/0x280
   ? __this_cpu_preempt_check+0x20/0x30
   iopt_pages_rw_access+0x2b2/0x5b0
   iommufd_access_rw+0x19f/0x2f0
   iommufd_test+0xd11/0x16f0
   ? write_comp_data+0x2f/0x90
   iommufd_fops_ioctl+0x206/0x330
   __x64_sys_ioctl+0x10e/0x160
   ? __pfx_iommufd_fops_ioctl+0x10/0x10
   do_syscall_64+0x3b/0x90
   entry_SYSCALL_64_after_hwframe+0x72/0xdc

Cc: <stable@vger.kernel.org>
Fixes: 8d160cd4d506 ("iommufd: Algorithms for PFN storage")
Link: https://lore.kernel.org/r/1-v1-ceab6a4d7d7a+94-iommufd_syz_jgg@nvidia.com
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reported-by: Pengfei Xu <pengfei.xu@intel.com>
Tested-by: Pengfei Xu <pengfei.xu@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>