commit fb2bdd32b2 upstream.
Compile-testing without CONFIG_OF shows that the of_match_ptr() macro
was used incorrectly here:
drivers/remoteproc/stm32_rproc.c:662:34: warning: unused variable 'stm32_rproc_match' [-Wunused-const-variable]
As in almost every driver, the solution is simply to remove the
use of this macro. The same thing happened with the deprecated
SIMPLE_DEV_PM_OPS(), but the corresponding warning was already shut
up with __maybe_unused annotations, so fix those as well by using the
correct DEFINE_SIMPLE_DEV_PM_OPS() macros and removing the extraneous
__maybe_unused modifiers. For completeness, also add a pm_ptr() to let
the PM ops be eliminated completely when CONFIG_PM is turned off.
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202307242300.ia82qBTp-lkp@intel.com
Fixes: 03bd158e15 ("remoteproc: stm32: use correct format strings on 64-bit")
Fixes: 410119ee29 ("remoteproc: stm32: wakeup the system by wdg irq")
Fixes: 13140de09c ("remoteproc: stm32: add an ST stm32_rproc driver")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
Link: https://lore.kernel.org/r/20230724195704.2432382-1-arnd@kernel.org
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit df7ecce842 ]
Clearing BSS should only be done once, at the very beginning.
efi_pe_entry() is the entrypoint from the firmware, which may not clear
BSS and so it is done explicitly. However, efi_pe_entry() is also used
as an entrypoint by the mixed mode startup code, in which case BSS will
already have been cleared, and doing it again at this point will corrupt
global variables holding the firmware's GDT/IDT and segment selectors.
So make the memset() conditional on whether the EFI stub is running in
native mode.
Fixes: b3810c5a2c ("x86/efistub: Clear decompressor BSS in native EFI entrypoint")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b3810c5a2c ]
The EFI stub on x86 no longer invokes the decompressor as a subsequent
boot stage, but calls into the decompression code directly while running
in the context of the EFI boot services.
This means that when using the native EFI entrypoint (as opposed to the
EFI handover protocol, which clears BSS explicitly), the firmware PE
image loader is being relied upon to ensure that BSS is zeroed before
the EFI stub is entered from the firmware.
As Radek's report proves, this is a bad idea. Not all loaders do this
correctly, which means some global variables that should be statically
initialized to 0x0 may have junk in them.
So clear BSS explicitly when entering via efi_pe_entry(). Note that
zeroing BSS from C code is not generally safe, but in this case, the
following assignment and dereference of a global pointer variable
ensures that the memset() cannot be deferred or reordered.
Cc: <stable@kernel.org> # v6.1+
Reported-by: Radek Podgorny <radek@podgorny.cz>
Closes: https://lore.kernel.org/all/a99a831a-8ad5-4cb0-bff9-be637311f771@podgorny.cz
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b4d78cfeb3 ]
It is possible to set up dm-integrity with smaller sector size than
the logical sector size of the underlying device. In this situation,
dm-integrity guarantees that the outgoing bios have the same alignment as
incoming bios (so, if you create a filesystem with 4k block size,
dm-integrity would send 4k-aligned bios to the underlying device).
This guarantee was broken when integrity_recheck was implemented.
integrity_recheck sends bio that is aligned to ic->sectors_per_block. So
if we set up integrity with 512-byte sector size on a device with logical
block size 4k, we would be sending unaligned bio. This triggered a bug in
one of our internal tests.
This commit fixes it by determining the actual alignment of the
incoming bio and then makes sure that the outgoing bio in
integrity_recheck has the same alignment.
Fixes: c88f5e553f ("dm-integrity: recheck the integrity tag after a failure")
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 6e5f0f6383 ]
Some IO will dispatch from kworker with different io_context settings
than the submitting task, we may need to specify a priority to avoid
losing priority.
Add IO priority parameter to dm_io() and update all callers.
Co-developed-by: Yibin Ding <yibin.ding@unisoc.com>
Signed-off-by: Yibin Ding <yibin.ding@unisoc.com>
Signed-off-by: Hongyu Jin <hongyu.jin@unisoc.com>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Stable-dep-of: b4d78cfeb3 ("dm-integrity: align the outgoing bio in integrity_recheck")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e4137851d4 ]
The tests send 100 pings in 0.1 second intervals and force a timeout of
11 seconds, which is borderline (especially on debug kernels), resulting
in random failures in netdev CI [1].
Fix by increasing the timeout to 20 seconds. It should not prolong the
test unless something is wrong, in which case the test will rightfully
fail.
[1]
# selftests: net/forwarding: vxlan_bridge_1d_port_8472_ipv6.sh
# INFO: Running tests with UDP port 8472
# TEST: ping: local->local [ OK ]
# TEST: ping: local->remote 1 [FAIL]
# Ping failed
[...]
Fixes: b07e9957f2 ("selftests: forwarding: Add VxLAN tests with a VLAN-unaware bridge for IPv6")
Fixes: 728b35259e ("selftests: forwarding: Add VxLAN tests with a VLAN-aware bridge for IPv6")
Reported-by: Paolo Abeni <pabeni@redhat.com>
Closes: https://lore.kernel.org/netdev/24a7051fdcd1f156c3704bca39e4b3c41dfc7c4b.camel@redhat.com/
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Hangbin Liu <liuhangbin@gmail.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://lore.kernel.org/r/20240320065717.4145325-1-idosch@nvidia.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit a20ad45008 ]
The TX buffer in spi_transfer can be a NULL pointer, so the interrupt
handler may end up writing to the invalid memory and cause crashes.
Add a check to trans->tx_buf before using it.
Fixes: 1ce24864bf ("spi: mediatek: Only do dma for 4-byte aligned buffers")
Signed-off-by: Fei Shao <fshao@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://msgid.link/r/20240321070942.1587146-2-fshao@chromium.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 69ddba9d17 ]
Currently, the MT753X switches treat frames with :01-0D and :0F MAC DAs as
regular multicast frames, therefore flooding them to user ports.
On page 205, section "8.6.3 Frame filtering" of the active standard, IEEE
Std 802.1Q™-2022, it is stated that frames with 01:80:C2:00:00:00-0F as MAC
DA must only be propagated to C-VLAN and MAC Bridge components. That means
VLAN-aware and VLAN-unaware bridges. On the switch designs with CPU ports,
these frames are supposed to be processed by the CPU (software). So we make
the switch only forward them to the CPU port. And if received from a CPU
port, forward to a single port. The software is responsible of making the
switch conform to the latter by setting a single port as destination port
on the special tag.
This switch intellectual property cannot conform to this part of the
standard fully. Whilst the REV_UN frame tag covers the remaining :04-0D and
:0F MAC DAs, it also includes :22-FF which the scope of propagation is not
supposed to be restricted for these MAC DAs.
Set frames with :01-03 MAC DAs to be trapped to the CPU port(s). Add a
comment for the remaining MAC DAs.
Note that the ingress port must have a PVID assigned to it for the switch
to forward untagged frames. A PVID is set by default on VLAN-aware and
VLAN-unaware ports. However, when the network interface that pertains to
the ingress port is attached to a vlan_filtering enabled bridge, the user
can remove the PVID assignment from it which would prevent the link-local
frames from being trapped to the CPU port. I am yet to see a way to forward
link-local frames while preventing other untagged frames from being
forwarded too.
Fixes: b8f126a8d5 ("net-next: dsa: add dsa support for Mediatek MT7530 switch")
Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e8bf353577 ]
Whether VLAN-aware or not, on every VID VLAN table entry that has the CPU
port as a member of it, frames are set to egress the CPU port with the VLAN
tag stacked. This is so that VLAN tags can be appended after hardware
special tag (called DSA tag in the context of Linux drivers).
For user ports on a VLAN-unaware bridge, frame ingressing the user port
egresses CPU port with only the special tag.
For user ports on a VLAN-aware bridge, frame ingressing the user port
egresses CPU port with the special tag and the VLAN tag.
This causes issues with link-local frames, specifically BPDUs, because the
software expects to receive them VLAN-untagged.
There are two options to make link-local frames egress untagged. Setting
CONSISTENT or UNTAGGED on the EG_TAG bits on the relevant register.
CONSISTENT means frames egress exactly as they ingress. That means
egressing with the VLAN tag they had at ingress or egressing untagged if
they ingressed untagged. Although link-local frames are not supposed to be
transmitted VLAN-tagged, if they are done so, when egressing through a CPU
port, the special tag field will be broken.
BPDU egresses CPU port with VLAN tag egressing stacked, received on
software:
00:01:25.104821 AF Unknown (382365846), length 106:
| STAG | | VLAN |
0x0000: 0000 6c27 614d 4143 0001 0000 8100 0001 ..l'aMAC........
0x0010: 0026 4242 0300 0000 0000 0000 6c27 614d .&BB........l'aM
0x0020: 4143 0000 0000 0000 6c27 614d 4143 0000 AC......l'aMAC..
0x0030: 0000 1400 0200 0f00 0000 0000 0000 0000 ................
BPDU egresses CPU port with VLAN tag egressing untagged, received on
software:
00:23:56.628708 AF Unknown (25215488), length 64:
| STAG |
0x0000: 0000 6c27 614d 4143 0001 0000 0026 4242 ..l'aMAC.....&BB
0x0010: 0300 0000 0000 0000 6c27 614d 4143 0000 ........l'aMAC..
0x0020: 0000 0000 6c27 614d 4143 0000 0000 1400 ....l'aMAC......
0x0030: 0200 0f00 0000 0000 0000 0000 ............
BPDU egresses CPU port with VLAN tag egressing tagged, received on
software:
00:01:34.311963 AF Unknown (25215488), length 64:
| Mess |
0x0000: 0000 6c27 614d 4143 0001 0001 0026 4242 ..l'aMAC.....&BB
0x0010: 0300 0000 0000 0000 6c27 614d 4143 0000 ........l'aMAC..
0x0020: 0000 0000 6c27 614d 4143 0000 0000 1400 ....l'aMAC......
0x0030: 0200 0f00 0000 0000 0000 0000 ............
To prevent confusing the software, force the frame to egress UNTAGGED
instead of CONSISTENT. This way, frames can't possibly be received TAGGED
by software which would have the special tag field broken.
VLAN Tag Egress Procedure
For all frames, one of these options set the earliest in this order will
apply to the frame:
- EG_TAG in certain registers for certain frames.
This will apply to frame with matching MAC DA or EtherType.
- EG_TAG in the address table.
This will apply to frame at its incoming port.
- EG_TAG in the PVC register.
This will apply to frame at its incoming port.
- EG_CON and [EG_TAG per port] in the VLAN table.
This will apply to frame at its outgoing port.
- EG_TAG in the PCR register.
This will apply to frame at its outgoing port.
EG_TAG in certain registers for certain frames:
PPPoE Discovery_ARP/RARP: PPP_EG_TAG and ARP_EG_TAG in the APC register.
IGMP_MLD: IGMP_EG_TAG and MLD_EG_TAG in the IMC register.
BPDU and PAE: BPDU_EG_TAG and PAE_EG_TAG in the BPC register.
REV_01 and REV_02: R01_EG_TAG and R02_EG_TAG in the RGAC1 register.
REV_03 and REV_0E: R03_EG_TAG and R0E_EG_TAG in the RGAC2 register.
REV_10 and REV_20: R10_EG_TAG and R20_EG_TAG in the RGAC3 register.
REV_21 and REV_UN: R21_EG_TAG and RUN_EG_TAG in the RGAC4 register.
With this change, it can be observed that a bridge interface with stp_state
and vlan_filtering enabled will properly block ports now.
Fixes: b8f126a8d5 ("net-next: dsa: add dsa support for Mediatek MT7530 switch")
Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 00bf631224 ]
When there are heavy load, cpumap kernel threads can be busy polling
packets from redirect queues and block out RCU tasks from reaching
quiescent states. It is insufficient to just call cond_resched() in such
context. Periodically raise a consolidated RCU QS before cond_resched
fixes the problem.
Fixes: 6710e11269 ("bpf: introduce new bpf cpu map type BPF_MAP_TYPE_CPUMAP")
Reviewed-by: Jesper Dangaard Brouer <hawk@kernel.org>
Signed-off-by: Yan Zhai <yan@cloudflare.com>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Acked-by: Jesper Dangaard Brouer <hawk@kernel.org>
Link: https://lore.kernel.org/r/c17b9f1517e19d813da3ede5ed33ee18496bb5d8.1710877680.git.yan@cloudflare.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d6dbbb1124 ]
NAPI threads can keep polling packets under load. Currently it is only
calling cond_resched() before repolling, but it is not sufficient to
clear out the holdout of RCU tasks, which prevent BPF tracing programs
from detaching for long period. This can be reproduced easily with
following set up:
ip netns add test1
ip netns add test2
ip -n test1 link add veth1 type veth peer name veth2 netns test2
ip -n test1 link set veth1 up
ip -n test1 link set lo up
ip -n test2 link set veth2 up
ip -n test2 link set lo up
ip -n test1 addr add 192.168.1.2/31 dev veth1
ip -n test1 addr add 1.1.1.1/32 dev lo
ip -n test2 addr add 192.168.1.3/31 dev veth2
ip -n test2 addr add 2.2.2.2/31 dev lo
ip -n test1 route add default via 192.168.1.3
ip -n test2 route add default via 192.168.1.2
for i in `seq 10 210`; do
for j in `seq 10 210`; do
ip netns exec test2 iptables -I INPUT -s 3.3.$i.$j -p udp --dport 5201
done
done
ip netns exec test2 ethtool -K veth2 gro on
ip netns exec test2 bash -c 'echo 1 > /sys/class/net/veth2/threaded'
ip netns exec test1 ethtool -K veth1 tso off
Then run an iperf3 client/server and a bpftrace script can trigger it:
ip netns exec test2 iperf3 -s -B 2.2.2.2 >/dev/null&
ip netns exec test1 iperf3 -c 2.2.2.2 -B 1.1.1.1 -u -l 1500 -b 3g -t 100 >/dev/null&
bpftrace -e 'kfunc:__napi_poll{@=count();} interval:s:1{exit();}'
Report RCU quiescent states periodically will resolve the issue.
Fixes: 29863d41bb ("net: implement threaded-able napi poll loop support")
Reviewed-by: Jesper Dangaard Brouer <hawk@kernel.org>
Signed-off-by: Yan Zhai <yan@cloudflare.com>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Acked-by: Jesper Dangaard Brouer <hawk@kernel.org>
Link: https://lore.kernel.org/r/4c3b0d3f32d3b18949d75b18e5e1d9f13a24f025.1710877680.git.yan@cloudflare.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 1a77557d48 ]
When under heavy load, network processing can run CPU-bound for many
tens of seconds. Even in preemptible kernels (non-RT kernel), this can
block RCU Tasks grace periods, which can cause trace-event removal to
take more than a minute, which is unacceptably long.
This commit therefore creates a new helper function that passes through
both RCU and RCU-Tasks quiescent states every 100 milliseconds. This
hard-coded value suffices for current workloads.
Suggested-by: Paul E. McKenney <paulmck@kernel.org>
Reviewed-by: Jesper Dangaard Brouer <hawk@kernel.org>
Signed-off-by: Yan Zhai <yan@cloudflare.com>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Acked-by: Jesper Dangaard Brouer <hawk@kernel.org>
Link: https://lore.kernel.org/r/90431d46ee112d2b0af04dbfe936faaca11810a5.1710877680.git.yan@cloudflare.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Stable-dep-of: d6dbbb1124 ("net: report RCU QS on threaded NAPI repolling")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b0e256f3dd ]
Clone already always provides a current view of the lookup table, use it
to destroy the set, otherwise it is possible to destroy elements twice.
This fix requires:
212ed75dc5 ("netfilter: nf_tables: integrate pipapo into commit protocol")
which came after:
9827a0e6e2 ("netfilter: nft_set_pipapo: release elements in clone from abort path").
Fixes: 9827a0e6e2 ("netfilter: nft_set_pipapo: release elements in clone from abort path")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 50e60de381 ]
For PF to AF interrupt vector and VF to AF vector same
interrupt handler is registered which is causing race condition.
When two interrupts are raised to two CPUs at same time
then two cores serve same event corrupting the data.
Fixes: 7304ac4567 ("octeontx2-af: Add mailbox IRQ and msg handlers")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit dfcf6355f5 ]
When PF sending link status messages to VF, it is possible
that by the time link_event_task work function is executed
VF might have brought down. Hence before sending VF link
status message check whether VF is up to receive it.
Fixes: ad513ed938 ("octeontx2-vf: Link event notification support")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 7558ce0d97 ]
Only one execution context for the workqueue used for PF and
VFs mailbox communication is incorrect since multiple works are
queued simultaneously by all the VFs and PF link UP messages.
Hence use default number of execution contexts by passing zero
as max_active to alloc_workqueue function. With this fix in place,
modify UP messages also to wait until completion.
Fixes: d424b6c024 ("octeontx2-pf: Enable SRIOV and added VF mbox handling")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 289f974674 ]
BACKGROUND
==========
When multiple work items are queued to a workqueue, their execution order
doesn't match the queueing order. They may get executed in any order and
simultaneously. When fully serialized execution - one by one in the queueing
order - is needed, an ordered workqueue should be used which can be created
with alloc_ordered_workqueue().
However, alloc_ordered_workqueue() was a later addition. Before it, an
ordered workqueue could be obtained by creating an UNBOUND workqueue with
@max_active==1. This originally was an implementation side-effect which was
broken by 4c16bd327c ("workqueue: restore WQ_UNBOUND/max_active==1 to be
ordered"). Because there were users that depended on the ordered execution,
5c0338c687 ("workqueue: restore WQ_UNBOUND/max_active==1 to be ordered")
made workqueue allocation path to implicitly promote UNBOUND workqueues w/
@max_active==1 to ordered workqueues.
While this has worked okay, overloading the UNBOUND allocation interface
this way creates other issues. It's difficult to tell whether a given
workqueue actually needs to be ordered and users that legitimately want a
min concurrency level wq unexpectedly gets an ordered one instead. With
planned UNBOUND workqueue updates to improve execution locality and more
prevalence of chiplet designs which can benefit from such improvements, this
isn't a state we wanna be in forever.
This patch series audits all callsites that create an UNBOUND workqueue w/
@max_active==1 and converts them to alloc_ordered_workqueue() as necessary.
WHAT TO LOOK FOR
================
The conversions are from
alloc_workqueue(WQ_UNBOUND | flags, 1, args..)
to
alloc_ordered_workqueue(flags, args...)
which don't cause any functional changes. If you know that fully ordered
execution is not ncessary, please let me know. I'll drop the conversion and
instead add a comment noting the fact to reduce confusion while conversion
is in progress.
If you aren't fully sure, it's completely fine to let the conversion
through. The behavior will stay exactly the same and we can always
reconsider later.
As there are follow-up workqueue core changes, I'd really appreciate if the
patch can be routed through the workqueue tree w/ your acks. Thanks.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Sunil Goutham <sgoutham@marvell.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Paolo Abeni <pabeni@redhat.com>
Cc: Ratheesh Kannoth <rkannoth@marvell.com>
Cc: Srujana Challa <schalla@marvell.com>
Cc: Geetha sowjanya <gakula@marvell.com>
Cc: netdev@vger.kernel.org
Stable-dep-of: 7558ce0d97 ("octeontx2-pf: Use default max_active works instead of one")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit a88e0f936b ]
A single line of interrupt is used to receive up notifications
and down reply messages from AF to PF (similarly from PF to its VF).
PF acts as bridge and forwards VF messages to AF and sends respsones
back from AF to VF. When an async event like link event is received
by up message when PF is in middle of forwarding VF message then
mailbox errors occur because PF state machine is corrupted.
Since VF is a separate driver or VF driver can be in a VM it is
not possible to serialize from the start of communication at VF.
Hence to differentiate between type of messages at PF this patch makes
sender to set mbox data register with distinct values for up and down
messages. Sender also checks whether previous interrupt is received
before triggering current interrupt by waiting for mailbox data register
to become zero.
Fixes: 5a6d7c9dae ("octeontx2-pf: Mailbox communication with AF")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 8299ffe3dc ]
CPT HW would trigger the CPT AF FLT interrupt when CPT engines
hits some uncorrectable errors and AF is the one which receives
the interrupt and recovers the engines.
This patch adds a mailbox for CPT VFs to request for CPT faulted
and recovered engines info.
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stable-dep-of: a88e0f936b ("octeontx2: Detect the mbox up or down message via register")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 9adb04ff62 ]
Optimize CPT PF identification in mbox handling for faster
mbox response by doing it at AF driver probe instead of
every mbox message.
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stable-dep-of: a88e0f936b ("octeontx2: Detect the mbox up or down message via register")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f58cf765e8 ]
On OcteonTX2 SoC, the admin function (AF) is the only one with all
priviliges to configure HW and alloc resources, PFs and it's VFs
have to request AF via mailbox for all their needs.
This patch adds a new mailbox for CPT VFs to request for CPT LF
reset.
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stable-dep-of: a88e0f936b ("octeontx2: Detect the mbox up or down message via register")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 07ea567d84 ]
When CPT engine has uncorrectable errors, it will get halted and
must be disabled and re-enabled. This patch adds code for the same.
Signed-off-by: Srujana Challa <schalla@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stable-dep-of: a88e0f936b ("octeontx2: Detect the mbox up or down message via register")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d27e2da94a ]
Fix race condition leading to system crash during EEH error handling
During EEH error recovery, the bnx2x driver's transmit timeout logic
could cause a race condition when handling reset tasks. The
bnx2x_tx_timeout() schedules reset tasks via bnx2x_sp_rtnl_task(),
which ultimately leads to bnx2x_nic_unload(). In bnx2x_nic_unload()
SGEs are freed using bnx2x_free_rx_sge_range(). However, this could
overlap with the EEH driver's attempt to reset the device using
bnx2x_io_slot_reset(), which also tries to free SGEs. This race
condition can result in system crashes due to accessing freed memory
locations in bnx2x_free_rx_sge()
799 static inline void bnx2x_free_rx_sge(struct bnx2x *bp,
800 struct bnx2x_fastpath *fp, u16 index)
801 {
802 struct sw_rx_page *sw_buf = &fp->rx_page_ring[index];
803 struct page *page = sw_buf->page;
....
where sw_buf was set to NULL after the call to dma_unmap_page()
by the preceding thread.
EEH: Beginning: 'slot_reset'
PCI 0011:01:00.0#10000: EEH: Invoking bnx2x->slot_reset()
bnx2x: [bnx2x_io_slot_reset:14228(eth1)]IO slot reset initializing...
bnx2x 0011:01:00.0: enabling device (0140 -> 0142)
bnx2x: [bnx2x_io_slot_reset:14244(eth1)]IO slot reset --> driver unload
Kernel attempted to read user page (0) - exploit attempt? (uid: 0)
BUG: Kernel NULL pointer dereference on read at 0x00000000
Faulting instruction address: 0xc0080000025065fc
Oops: Kernel access of bad area, sig: 11 [#1]
.....
Call Trace:
[c000000003c67a20] [c00800000250658c] bnx2x_io_slot_reset+0x204/0x610 [bnx2x] (unreliable)
[c000000003c67af0] [c0000000000518a8] eeh_report_reset+0xb8/0xf0
[c000000003c67b60] [c000000000052130] eeh_pe_report+0x180/0x550
[c000000003c67c70] [c00000000005318c] eeh_handle_normal_event+0x84c/0xa60
[c000000003c67d50] [c000000000053a84] eeh_event_handler+0xf4/0x170
[c000000003c67da0] [c000000000194c58] kthread+0x1c8/0x1d0
[c000000003c67e10] [c00000000000cf64] ret_from_kernel_thread+0x5c/0x64
To solve this issue, we need to verify page pool allocations before
freeing.
Fixes: 4cace675d6 ("bnx2x: Alloc 4k fragment for each rx ring buffer element")
Signed-off-by: Thinh Tran <thinhtr@linux.ibm.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://lore.kernel.org/r/20240315205535.1321-1-thinhtr@linux.ibm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 55e565c42d ]
Memory for the "checksums" pointer will leak if the data is rechecked
after checksum failure (because the associated kfree won't happen due
to 'goto skip_io').
Fix this by freeing the checksums memory before recheck, and just use
the "checksum_onstack" memory for storing checksum during recheck.
Fixes: c88f5e553f ("dm-integrity: recheck the integrity tag after a failure")
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 32fa4366cc ]
read_poll_timeout inside phy_read_poll_timeout can set val negative
in some cases (for example, __mdiobus_read inside phy_read can return
-EOPNOTSUPP).
Supposedly, commit 4ec7329517 ("net: phylib: fix phy_read*_poll_timeout()")
should fix problems with wrong-signed vals, but I do not see how
as val is sent to phy_read as is and __val = phy_read (not val)
is checked for sign.
Change val type for signed to allow better error handling as done in other
phy_read_poll_timeout callers. This will not fix any error handling
by itself, but allows, for example, to modify cond with appropriate
sign check or check resulting val separately.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Fixes: 014068dcb5 ("net: phy: genphy_loopback: add link speed configuration")
Signed-off-by: Nikita Kiryushin <kiryushin@ancud.ru>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://lore.kernel.org/r/20240315175052.8049-1-kiryushin@ancud.ru
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c9b3b81716 ]
Since the referenced commit, the xfrm_inner_extract_output() function
uses the protocol field to determine the address family. So not setting
it for IPv4 raw sockets meant that such packets couldn't be tunneled via
IPsec anymore.
IPv6 raw sockets are not affected as they already set the protocol since
9c9c9ad5fa ("ipv6: set skb->protocol on tcp, raw and ip6_append_data
genereated skbs").
Fixes: f4796398f2 ("xfrm: Remove inner/outer modes from output path")
Signed-off-by: Tobias Brunner <tobias@strongswan.org>
Reviewed-by: David Ahern <dsahern@kernel.org>
Reviewed-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/c5d9a947-eb19-4164-ac99-468ea814ce20@strongswan.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 3cf28cd492 ]
A failure during registration of the netdev notifier was not handled at
all. A failure during netlink initialization did not unregister the netdev
notifier.
Handle failures of netdev notifier registration and netlink initialization.
Both functions should only return negative values on failure and thereby
lead to the hsr module not being loaded.
Fixes: f421436a59 ("net/hsr: Add support for the High-availability Seamless Redundancy protocol (HSRv0)")
Signed-off-by: Felix Maurer <fmaurer@redhat.com>
Reviewed-by: Shigeru Yoshida <syoshida@redhat.com>
Reviewed-by: Breno Leitao <leitao@debian.org>
Link: https://lore.kernel.org/r/3ce097c15e3f7ace98fc7fd9bcbf299f092e63d1.1710504184.git.fmaurer@redhat.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 1422f28826 ]
acquire/release_in_xmit() work as bit lock in rds_send_xmit(), so they
are expected to ensure acquire/release memory ordering semantics.
However, test_and_set_bit/clear_bit() don't imply such semantics, on
top of this, following smp_mb__after_atomic() does not guarantee release
ordering (memory barrier actually should be placed before clear_bit()).
Instead, we use clear_bit_unlock/test_and_set_bit_lock() here.
Fixes: 0f4b1c7e89 ("rds: fix rds_send_xmit() serialization")
Fixes: 1f9ecd7eac ("RDS: Pass rds_conn_path to rds_send_xmit()")
Signed-off-by: Yewon Choi <woni9911@gmail.com>
Reviewed-by: Michal Kubiak <michal.kubiak@intel.com>
Link: https://lore.kernel.org/r/ZfQUxnNTO9AJmzwc@libra05
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 749a401683 ]
The MLX driver was not updating its control virtqueue size at set_vq_num
and instead always initialized to MLX5_CVQ_MAX_ENT (16) at
setup_cvq_vring.
Qemu would try to set the size to 64 by default, however, because the
CVQ size always was initialized to 16, an error would be thrown when
sending >16 control messages (as used-ring entry 17 is initialized to 0).
For example, starting a guest with x-svq=on and then executing the
following command would produce the error below:
# for i in {1..20}; do ifconfig eth0 hw ether XX:xx:XX:xx:XX:XX; done
qemu-system-x86_64: Insufficient written data (0)
[ 435.331223] virtio_net virtio0: Failed to set mac address by vq command.
SIOCSIFHWADDR: Invalid argument
Acked-by: Dragos Tatulea <dtatulea@nvidia.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Jonah Palmer <jonah.palmer@oracle.com>
Message-Id: <20240216142502.78095-1-jonah.palmer@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Fixes: 5262912ef3 ("vdpa/mlx5: Add support for control VQ and MAC setting")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 9588e7fc51 ]
vdpasim_do_reset sets running to true, which is wrong, as it allows
vdpasim_kick_vq to post work requests before the device has been
configured. To fix, do not set running until VIRTIO_CONFIG_S_DRIVER_OK
is set.
Fixes: 0c89e2a3a9 ("vdpa_sim: Implement suspend vdpa op")
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <1707517807-137331-1-git-send-email-steven.sistare@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 807f96abdf ]
As well noted by Pekka[1], the rounding of drm_fixp2int_round is wrong.
To round a number, you need to add 0.5 to the number and floor that,
drm_fixp2int_round() is adding 0.0000076. Make it add 0.5.
[1]: https://lore.kernel.org/all/20240301135327.22efe0dd.pekka.paalanen@collabora.com/
Fixes: 8b25320887 ("drm: Add fixed-point helper to get rounded integer values")
Suggested-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Melissa Wen <mwen@igalia.com>
Signed-off-by: Arthur Grillo <arthurgrillo@riseup.net>
Signed-off-by: Melissa Wen <melissa.srw@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240316-drm_fixed-v2-1-c1bc2665b5ed@riseup.net
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f490c492e9 ]
On MT7530, the HT_XTAL_FSEL field of the HWTRAP register stores a 2-bit
value that represents the frequency of the crystal oscillator connected to
the switch IC. The field is populated by the state of the ESW_P4_LED_0 and
ESW_P4_LED_0 pins, which is done right after reset is deasserted.
ESW_P4_LED_0 ESW_P3_LED_0 Frequency
-----------------------------------------
0 0 Reserved
0 1 20MHz
1 0 40MHz
1 1 25MHz
On MT7531, the XTAL25 bit of the STRAP register stores this. The LAN0LED0
pin is used to populate the bit. 25MHz when the pin is high, 40MHz when
it's low.
These pins are also used with LEDs, therefore, their state can be set to
something other than the bootstrapping configuration. For example, a link
may be established on port 3 before the DSA subdriver takes control of the
switch which would set ESW_P3_LED_0 to high.
Currently on mt7530_setup() and mt7531_setup(), 1000 - 1100 usec delay is
described between reset assertion and deassertion. Some switch ICs in real
life conditions cannot always have these pins set back to the bootstrapping
configuration before reset deassertion in this amount of delay. This causes
wrong crystal frequency to be selected which puts the switch in a
nonfunctional state after reset deassertion.
The tests below are conducted on an MT7530 with a 40MHz crystal oscillator
by Justin Swartz.
With a cable from an active peer connected to port 3 before reset, an
incorrect crystal frequency (0b11 = 25MHz) is selected:
[1] [3] [5]
: : :
_____________________________ __________________
ESW_P4_LED_0 |_______|
_____________________________
ESW_P3_LED_0 |__________________________
: : : :
: : [4]...:
: :
[2]................:
[1] Reset is asserted.
[2] Period of 1000 - 1100 usec.
[3] Reset is deasserted.
[4] Period of 315 usec. HWTRAP register is populated with incorrect
XTAL frequency.
[5] Signals reflect the bootstrapped configuration.
Increase the delay between reset_control_assert() and
reset_control_deassert(), and gpiod_set_value_cansleep(priv->reset, 0) and
gpiod_set_value_cansleep(priv->reset, 1) to 5000 - 5100 usec. This amount
ensures a higher possibility that the switch IC will have these pins back
to the bootstrapping configuration before reset deassertion.
With a cable from an active peer connected to port 3 before reset, the
correct crystal frequency (0b10 = 40MHz) is selected:
[1] [2-1] [3] [5]
: : : :
_____________________________ __________________
ESW_P4_LED_0 |_______|
___________________ _______
ESW_P3_LED_0 |_________| |__________________
: : : : :
: [2-2]...: [4]...:
[2]................:
[1] Reset is asserted.
[2] Period of 5000 - 5100 usec.
[2-1] ESW_P3_LED_0 goes low.
[2-2] Remaining period of 5000 - 5100 usec.
[3] Reset is deasserted.
[4] Period of 310 usec. HWTRAP register is populated with bootstrapped
XTAL frequency.
[5] Signals reflect the bootstrapped configuration.
ESW_P3_LED_0 low period before reset deassertion:
5000 usec
- 5100 usec
TEST RESET HOLD
# (usec)
---------------------
1 5410
2 5440
3 4375
4 5490
5 5475
6 4335
7 4370
8 5435
9 4205
10 4335
11 3750
12 3170
13 4395
14 4375
15 3515
16 4335
17 4220
18 4175
19 4175
20 4350
Min 3170
Max 5490
Median 4342.500
Avg 4466.500
Revert commit 2920dd92b9 ("net: dsa: mt7530: disable LEDs before reset").
Changing the state of pins via reset assertion is simpler and more
efficient than doing so by setting the LED controller off.
Fixes: b8f126a8d5 ("net-next: dsa: add dsa support for Mediatek MT7530 switch")
Fixes: c288575f78 ("net: dsa: mt7530: Add the support of MT7531 switch")
Co-developed-by: Justin Swartz <justin.swartz@risingedge.co.za>
Signed-off-by: Justin Swartz <justin.swartz@risingedge.co.za>
Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d7db7775ea ]
Commit d3256efd8e ("veth: allow enabling NAPI even without XDP") tried to fix
the fact that GRO was not possible without XDP, because veth did not use NAPI
without XDP. However, it also introduced the behaviour that GRO is always
enabled, when XDP is enabled.
While it might be desired for most cases, it is confusing for the user at best
as the GRO flag suddenly changes, when an XDP program is attached. It also
introduces some complexities in state management as was partially addressed in
commit fe9f801355 ("net: veth: clear GRO when clearing XDP even when down").
But the biggest problem is that it is not possible to disable GRO at all, when
an XDP program is attached, which might be needed for some use cases.
Fix this by not touching the GRO flag on XDP enable/disable as the code already
supports switching to NAPI if either GRO or XDP is requested.
Link: https://lore.kernel.org/lkml/20240311124015.38106-1-ignat@cloudflare.com/
Fixes: d3256efd8e ("veth: allow enabling NAPI even without XDP")
Fixes: fe9f801355 ("net: veth: clear GRO when clearing XDP even when down")
Signed-off-by: Ignat Korchagin <ignat@cloudflare.com>
Reviewed-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit ea80e3ed09 ]
A patch to resolve an issue was found in MediaTek's GPL-licensed SDK:
In the mtk_ppe_stop() function, the PPE scan mode is not disabled before
disabling the PPE. This can potentially lead to a hang during the process
of disabling the PPE.
Without this patch, the PPE may experience a hang during the reboot test.
Link: b40da332df
Fixes: ba37b7caf1 ("net: ethernet: mtk_eth_soc: add support for initializing the PPE")
Suggested-by: Bc-bocun Chen <bc-bocun.chen@mediatek.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f1b85ef15a ]
Clearing bit MAC_MCR_FORCE_LINK which forces the link down too early
can result in MAC ending up in a broken/blocked state.
Fix this by handling this bit in the .mac_link_up and .mac_link_down
calls instead of in .mac_finish.
Fixes: b8fc9f3082 ("net: ethernet: mediatek: Add basic PHYLINK support")
Suggested-by: Mason-cw Chang <Mason-cw.Chang@mediatek.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit de105068fe ]
We found a issue on production environment while using NVMe over RDMA,
admin_q reconnect failed forever while remote target and network is ok.
After dig into it, we found it may caused by a ABBA deadlock due to tag
allocation. In my case, the tag was hold by a keep alive request
waiting inside admin_q, as we quiesced admin_q while reset ctrl, so the
request maked as idle and will not process before reset success. As
fabric_q shares tagset with admin_q, while reconnect remote target, we
need a tag for connect command, but the only one reserved tag was held
by keep alive command which waiting inside admin_q. As a result, we
failed to reconnect admin_q forever. In order to fix this issue, I
think we should keep two reserved tags for admin queue.
Fixes: ed01fee283 ("nvme-fabrics: only reserve a single tag")
Signed-off-by: Chunguang Xu <chunguang.xu@shopee.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 93b24f579c ]
Add the apple shared tag workaround to nvme_alloc_io_tag_set to prepare
for using that helper in the PCIe driver.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Stable-dep-of: de105068fe ("nvme: fix reconnection fail due to reserved tag allocation")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b794d1c2ad ]
The reserved_tags are only needed for fabrics controllers. Right now only
fabrics drivers call this helper, so this is harmless, but we'll use it
in the PCIe driver soon.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Stable-dep-of: de105068fe ("nvme: fix reconnection fail due to reserved tag allocation")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 72ebb41b88 ]
A previous bugfix added a call to kcalloc(), which starting in gcc-14
causes a harmless warning about the argument order:
drivers/soc/fsl/dpio/dpio-service.c: In function 'dpaa2_io_service_enqueue_multiple_desc_fq':
drivers/soc/fsl/dpio/dpio-service.c:526:29: error: 'kcalloc' sizes specified with 'sizeof' in the earlier argument and not in the later argument [-Werror=calloc-transposed-args]
526 | ed = kcalloc(sizeof(struct qbman_eq_desc), 32, GFP_KERNEL);
| ^~~~~~
drivers/soc/fsl/dpio/dpio-service.c:526:29: note: earlier argument should specify number of elements, later size of each element
Since the two are only multiplied, the order does not change the
behavior, so just fix it now to shut up the compiler warning.
Dmity independently came up with the same fix.
Fixes: 5c4a5999b2 ("soc: fsl: dpio: avoid stack usage warning")
Reported-by: Dmitry Antipov <dmantipov@yandex.ru>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 367c50f784 ]
Current average steal timer calculation produces volatile and inflated
values. The only user of this value is KVM so far and it uses that to
decide whether or not to yield the vCPU which is seeing steal time.
KVM compares average steal timer to a threshold and if the threshold
is past then it does not allow CPU polling and yields it to host, else
it keeps the CPU by polling.
Since KVM's steal time threshold is very low by default (%10) it most
likely is not effected much by the bloated average steal timer values
because the operating region is pretty small. However there might be
new users in the future who might rely on this number. Fix average
steal timer calculation by changing the formula from:
avg_steal_timer = avg_steal_timer / 2 + steal_timer;
to the following:
avg_steal_timer = (avg_steal_timer + steal_timer) / 2;
This ensures that avg_steal_timer is actually a naive average of steal
timer values. It now closely follows steal timer values but of course
in a smoother manner.
Fixes: 152e9b8676 ("s390/vtime: steal time exponential moving average")
Signed-off-by: Mete Durlu <meted@linux.ibm.com>
Acked-by: Heiko Carstens <hca@linux.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e642921dfe ]
Use wake_up API instead of wake_up_interruptible, since
wait_event_timeout API is used for waiting on command completion.
Fixes: 1463f382f5 ("octeontx2-af: Add support for CGX link management")
Signed-off-by: Linu Cherian <lcherian@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 2a750d6a5b ]
syzkaller reported a warning of netns tracker [0] followed by KASAN
splat [1] and another ref tracker warning [1].
syzkaller could not find a repro, but in the log, the only suspicious
sequence was as follows:
18:26:22 executing program 1:
r0 = socket$inet6_mptcp(0xa, 0x1, 0x106)
...
connect$inet6(r0, &(0x7f0000000080)={0xa, 0x4001, 0x0, @loopback}, 0x1c) (async)
The notable thing here is 0x4001 in connect(), which is RDS_TCP_PORT.
So, the scenario would be:
1. unshare(CLONE_NEWNET) creates a per netns tcp listener in
rds_tcp_listen_init().
2. syz-executor connect()s to it and creates a reqsk.
3. syz-executor exit()s immediately.
4. netns is dismantled. [0]
5. reqsk timer is fired, and UAF happens while freeing reqsk. [1]
6. listener is freed after RCU grace period. [2]
Basically, reqsk assumes that the listener guarantees netns safety
until all reqsk timers are expired by holding the listener's refcount.
However, this was not the case for kernel sockets.
Commit 740ea3c4a0 ("tcp: Clean up kernel listener's reqsk in
inet_twsk_purge()") fixed this issue only for per-netns ehash.
Let's apply the same fix for the global ehash.
[0]:
ref_tracker: net notrefcnt@0000000065449cc3 has 1/1 users at
sk_alloc (./include/net/net_namespace.h:337 net/core/sock.c:2146)
inet6_create (net/ipv6/af_inet6.c:192 net/ipv6/af_inet6.c:119)
__sock_create (net/socket.c:1572)
rds_tcp_listen_init (net/rds/tcp_listen.c:279)
rds_tcp_init_net (net/rds/tcp.c:577)
ops_init (net/core/net_namespace.c:137)
setup_net (net/core/net_namespace.c:340)
copy_net_ns (net/core/net_namespace.c:497)
create_new_namespaces (kernel/nsproxy.c:110)
unshare_nsproxy_namespaces (kernel/nsproxy.c:228 (discriminator 4))
ksys_unshare (kernel/fork.c:3429)
__x64_sys_unshare (kernel/fork.c:3496)
do_syscall_64 (arch/x86/entry/common.c:52 arch/x86/entry/common.c:83)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:129)
...
WARNING: CPU: 0 PID: 27 at lib/ref_tracker.c:179 ref_tracker_dir_exit (lib/ref_tracker.c:179)
[1]:
BUG: KASAN: slab-use-after-free in inet_csk_reqsk_queue_drop (./include/net/inet_hashtables.h:180 net/ipv4/inet_connection_sock.c:952 net/ipv4/inet_connection_sock.c:966)
Read of size 8 at addr ffff88801b370400 by task swapper/0/0
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014
Call Trace:
<IRQ>
dump_stack_lvl (lib/dump_stack.c:107 (discriminator 1))
print_report (mm/kasan/report.c:378 mm/kasan/report.c:488)
kasan_report (mm/kasan/report.c:603)
inet_csk_reqsk_queue_drop (./include/net/inet_hashtables.h:180 net/ipv4/inet_connection_sock.c:952 net/ipv4/inet_connection_sock.c:966)
reqsk_timer_handler (net/ipv4/inet_connection_sock.c:979 net/ipv4/inet_connection_sock.c:1092)
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:2038)
run_timer_softirq (kernel/time/timer.c:2053)
__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:554)
irq_exit_rcu (kernel/softirq.c:427 kernel/softirq.c:632 kernel/softirq.c:644)
sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1076 (discriminator 14))
</IRQ>
Allocated by task 258 on cpu 0 at 83.612050s:
kasan_save_stack (mm/kasan/common.c:48)
kasan_save_track (mm/kasan/common.c:68)
__kasan_slab_alloc (mm/kasan/common.c:343)
kmem_cache_alloc (mm/slub.c:3813 mm/slub.c:3860 mm/slub.c:3867)
copy_net_ns (./include/linux/slab.h:701 net/core/net_namespace.c:421 net/core/net_namespace.c:480)
create_new_namespaces (kernel/nsproxy.c:110)
unshare_nsproxy_namespaces (kernel/nsproxy.c:228 (discriminator 4))
ksys_unshare (kernel/fork.c:3429)
__x64_sys_unshare (kernel/fork.c:3496)
do_syscall_64 (arch/x86/entry/common.c:52 arch/x86/entry/common.c:83)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:129)
Freed by task 27 on cpu 0 at 329.158864s:
kasan_save_stack (mm/kasan/common.c:48)
kasan_save_track (mm/kasan/common.c:68)
kasan_save_free_info (mm/kasan/generic.c:643)
__kasan_slab_free (mm/kasan/common.c:265)
kmem_cache_free (mm/slub.c:4299 mm/slub.c:4363)
cleanup_net (net/core/net_namespace.c:456 net/core/net_namespace.c:446 net/core/net_namespace.c:639)
process_one_work (kernel/workqueue.c:2638)
worker_thread (kernel/workqueue.c:2700 kernel/workqueue.c:2787)
kthread (kernel/kthread.c:388)
ret_from_fork (arch/x86/kernel/process.c:153)
ret_from_fork_asm (arch/x86/entry/entry_64.S:250)
The buggy address belongs to the object at ffff88801b370000
which belongs to the cache net_namespace of size 4352
The buggy address is located 1024 bytes inside of
freed 4352-byte region [ffff88801b370000, ffff88801b371100)
[2]:
WARNING: CPU: 0 PID: 95 at lib/ref_tracker.c:228 ref_tracker_free (lib/ref_tracker.c:228 (discriminator 1))
Modules linked in:
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014
RIP: 0010:ref_tracker_free (lib/ref_tracker.c:228 (discriminator 1))
...
Call Trace:
<IRQ>
__sk_destruct (./include/net/net_namespace.h:353 net/core/sock.c:2204)
rcu_core (./arch/x86/include/asm/preempt.h:26 kernel/rcu/tree.c:2165 kernel/rcu/tree.c:2433)
__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:554)
irq_exit_rcu (kernel/softirq.c:427 kernel/softirq.c:632 kernel/softirq.c:644)
sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1076 (discriminator 14))
</IRQ>
Reported-by: syzkaller <syzkaller@googlegroups.com>
Suggested-by: Eric Dumazet <edumazet@google.com>
Fixes: 467fa15356 ("RDS-TCP: Support multiple RDS-TCP listen endpoints, one per netns.")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20240308200122.64357-3-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Sasha Levin