IEP driver supports both perout and pps signal generation
but perout feature is faulty with half-cooked support
due to some missing configuration. Remove perout
support from the driver and reject perout requests with
"not supported" error code.
Fixes: c1e0230eea ("net: ti: icss-iep: Add IEP driver")
Signed-off-by: Meghana Malladi <m-malladi@ti.com>
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Link: https://patch.msgid.link/20250227092441.1848419-1-m-malladi@ti.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
There is an issue with one-step timestamp based on UDP/IP. The peer will
discard the sync packet because of the wrong UDP checksum. For ENETC v1,
the software needs to update the UDP checksum when updating the
originTimestamp field, so that the hardware can correctly update the UDP
checksum when updating the correction field. Otherwise, the UDP checksum
in the sync packet will be wrong.
Fixes: 7294380c52 ("enetc: support PTP Sync packet one-step timestamping")
Cc: stable@vger.kernel.org
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://patch.msgid.link/20250224111251.1061098-6-wei.fang@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Actually ENETC VFs do not support HWTSTAMP_TX_ONESTEP_SYNC because only
ENETC PF can access PMa_SINGLE_STEP registers. And there will be a crash
if VFs are used to test one-step timestamp, the crash log as follows.
[ 129.110909] Unable to handle kernel paging request at virtual address 00000000000080c0
[ 129.287769] Call trace:
[ 129.290219] enetc_port_mac_wr+0x30/0xec (P)
[ 129.294504] enetc_start_xmit+0xda4/0xe74
[ 129.298525] enetc_xmit+0x70/0xec
[ 129.301848] dev_hard_start_xmit+0x98/0x118
Fixes: 41514737ec ("enetc: add get_ts_info interface for ethtool")
Cc: stable@vger.kernel.org
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://patch.msgid.link/20250224111251.1061098-5-wei.fang@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We have recently seen reports of lockdep circular lock dependency warnings
when loading the iAVF driver:
[ 1504.790308] ======================================================
[ 1504.790309] WARNING: possible circular locking dependency detected
[ 1504.790310] 6.13.0 #net_next_rt.c2933b2befe2.el9 Not tainted
[ 1504.790311] ------------------------------------------------------
[ 1504.790312] kworker/u128:0/13566 is trying to acquire lock:
[ 1504.790313] ffff97d0e4738f18 (&dev->lock){+.+.}-{4:4}, at: register_netdevice+0x52c/0x710
[ 1504.790320]
[ 1504.790320] but task is already holding lock:
[ 1504.790321] ffff97d0e47392e8 (&adapter->crit_lock){+.+.}-{4:4}, at: iavf_finish_config+0x37/0x240 [iavf]
[ 1504.790330]
[ 1504.790330] which lock already depends on the new lock.
[ 1504.790330]
[ 1504.790330]
[ 1504.790330] the existing dependency chain (in reverse order) is:
[ 1504.790331]
[ 1504.790331] -> #1 (&adapter->crit_lock){+.+.}-{4:4}:
[ 1504.790333] __lock_acquire+0x52d/0xbb0
[ 1504.790337] lock_acquire+0xd9/0x330
[ 1504.790338] mutex_lock_nested+0x4b/0xb0
[ 1504.790341] iavf_finish_config+0x37/0x240 [iavf]
[ 1504.790347] process_one_work+0x248/0x6d0
[ 1504.790350] worker_thread+0x18d/0x330
[ 1504.790352] kthread+0x10e/0x250
[ 1504.790354] ret_from_fork+0x30/0x50
[ 1504.790357] ret_from_fork_asm+0x1a/0x30
[ 1504.790361]
[ 1504.790361] -> #0 (&dev->lock){+.+.}-{4:4}:
[ 1504.790364] check_prev_add+0xf1/0xce0
[ 1504.790366] validate_chain+0x46a/0x570
[ 1504.790368] __lock_acquire+0x52d/0xbb0
[ 1504.790370] lock_acquire+0xd9/0x330
[ 1504.790371] mutex_lock_nested+0x4b/0xb0
[ 1504.790372] register_netdevice+0x52c/0x710
[ 1504.790374] iavf_finish_config+0xfa/0x240 [iavf]
[ 1504.790379] process_one_work+0x248/0x6d0
[ 1504.790381] worker_thread+0x18d/0x330
[ 1504.790383] kthread+0x10e/0x250
[ 1504.790385] ret_from_fork+0x30/0x50
[ 1504.790387] ret_from_fork_asm+0x1a/0x30
[ 1504.790389]
[ 1504.790389] other info that might help us debug this:
[ 1504.790389]
[ 1504.790389] Possible unsafe locking scenario:
[ 1504.790389]
[ 1504.790390] CPU0 CPU1
[ 1504.790391] ---- ----
[ 1504.790391] lock(&adapter->crit_lock);
[ 1504.790393] lock(&dev->lock);
[ 1504.790394] lock(&adapter->crit_lock);
[ 1504.790395] lock(&dev->lock);
[ 1504.790397]
[ 1504.790397] *** DEADLOCK ***
This appears to be caused by the change in commit 5fda3f3534 ("net: make
netdev_lock() protect netdev->reg_state"), which added a netdev_lock() in
register_netdevice.
The iAVF driver calls register_netdevice() from iavf_finish_config(), as a
final stage of its state machine post-probe. It currently takes the RTNL
lock, then the netdev lock, and then the device critical lock. This pattern
is used throughout the driver. Thus there is a strong dependency that the
crit_lock should not be acquired before the net device lock. The change to
register_netdevice creates an ABBA lock order violation because the iAVF
driver is holding the crit_lock while calling register_netdevice, which
then takes the netdev_lock.
It seems likely that future refactors could result in netdev APIs which
hold the netdev_lock while calling into the driver. This means that we
should not re-order the locks so that netdev_lock is acquired after the
device private crit_lock.
Instead, notice that we already release the netdev_lock prior to calling
the register_netdevice. This flow only happens during the early driver
initialization as we transition through the __IAVF_STARTUP,
__IAVF_INIT_VERSION_CHECK, __IAVF_INIT_GET_RESOURCES, etc.
Analyzing the places where we take crit_lock in the driver there are two
sources:
a) several of the work queue tasks including adminq_task, watchdog_task,
reset_task, and the finish_config task.
b) various callbacks which ultimately stem back to .ndo operations or
ethtool operations.
The latter cannot be triggered until after the netdevice registration is
completed successfully.
The iAVF driver uses alloc_ordered_workqueue, which is an unbound workqueue
that has a max limit of 1, and thus guarantees that only a single work item
on the queue is executing at any given time, so none of the other work
threads could be executing due to the ordered workqueue guarantees.
The iavf_finish_config() function also does not do anything else after
register_netdevice, unless it fails. It seems unlikely that the driver
private crit_lock is protecting anything that register_netdevice() itself
touches.
Thus, to fix this ABBA lock violation, lets simply release the
adapter->crit_lock as well as netdev_lock prior to calling
register_netdevice(). We do still keep holding the RTNL lock as required by
the function. If we do fail to register the netdevice, then we re-acquire
the adapter critical lock to finish the transition back to
__IAVF_INIT_CONFIG_ADAPTER.
This ensures every call where both netdev_lock and the adapter->crit_lock
are acquired under the same ordering.
Fixes: afc664987a ("eth: iavf: extend the netdev_lock usage")
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://patch.msgid.link/20250224190647.3601930-5-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
If ice_ena_vfs() fails after calling ice_create_vf_entries(), it frees
all VFs without removing them from snapshot PF-VF mailbox list, leading
to list corruption.
Reproducer:
devlink dev eswitch set $PF1_PCI mode switchdev
ip l s $PF1 up
ip l s $PF1 promisc on
sleep 1
echo 1 > /sys/class/net/$PF1/device/sriov_numvfs
sleep 1
echo 1 > /sys/class/net/$PF1/device/sriov_numvfs
Trace (minimized):
list_add corruption. next->prev should be prev (ffff8882e241c6f0), but was 0000000000000000. (next=ffff888455da1330).
kernel BUG at lib/list_debug.c:29!
RIP: 0010:__list_add_valid_or_report+0xa6/0x100
ice_mbx_init_vf_info+0xa7/0x180 [ice]
ice_initialize_vf_entry+0x1fa/0x250 [ice]
ice_sriov_configure+0x8d7/0x1520 [ice]
? __percpu_ref_switch_mode+0x1b1/0x5d0
? __pfx_ice_sriov_configure+0x10/0x10 [ice]
Sometimes a KASAN report can be seen instead with a similar stack trace:
BUG: KASAN: use-after-free in __list_add_valid_or_report+0xf1/0x100
VFs are added to this list in ice_mbx_init_vf_info(), but only removed
in ice_free_vfs(). Move the removing to ice_free_vf_entries(), which is
also being called in other places where VFs are being removed (including
ice_free_vfs() itself).
Fixes: 8cd8a6b17d ("ice: move VF overflow message count into struct ice_mbx_vf_info")
Reported-by: Sujai Buvaneswaran <sujai.buvaneswaran@intel.com>
Closes: https://lore.kernel.org/intel-wired-lan/PH0PR11MB50138B635F2E5CEB7075325D961F2@PH0PR11MB5013.namprd11.prod.outlook.com
Reviewed-by: Martyna Szapar-Mudlaw <martyna.szapar-mudlaw@linux.intel.com>
Signed-off-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Sujai Buvaneswaran <sujai.buvaneswaran@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://patch.msgid.link/20250224190647.3601930-2-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Stats calculations involve a RMW to add the stat update to the existing
value. This is currently not protected by any synchronization mechanism,
so data races are possible. Add a spinlock to protect the update. The
reader side could be protected using u64_stats, but we would still need
a spinlock for the update side anyway. And we always do an update
immediately before reading the stats anyway.
Fixes: 89e5785fc8 ("[PATCH] Atmel MACB ethernet driver")
Signed-off-by: Sean Anderson <sean.anderson@linux.dev>
Link: https://patch.msgid.link/20250220162950.95941-1-sean.anderson@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The external PHY will undergo a soft reset twice during the resume process
when it wake up from suspend. The first reset occurs when the axienet
driver calls phylink_of_phy_connect(), and the second occurs when
mdio_bus_phy_resume() invokes phy_init_hw(). The second soft reset of the
external PHY does not reinitialize the internal PHY, which causes issues
with the internal PHY, resulting in the PHY link being down. To prevent
this, setting the mac_managed_pm flag skips the mdio_bus_phy_resume()
function.
Fixes: a129b41fe0 ("Revert "net: phy: dp83867: perform soft reset and retain established link"")
Signed-off-by: Nick Hu <nick.hu@sifive.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://patch.msgid.link/20250217055843.19799-1-nick.hu@sifive.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Before this patch the NETDEV_XDP_ACT_NDO_XMIT XDP feature flag is set by
default as part of driver initialization, and is never cleared. However,
this flag differs from others in that it is used as an indicator for
whether the driver is ready to perform the ndo_xdp_xmit operation as
part of an XDP_REDIRECT. Kernel helpers
xdp_features_(set|clear)_redirect_target exist to convey this meaning.
This patch ensures that the netdev is only reported as a redirect target
when XDP queues exist to forward traffic.
Fixes: 39a7f4aa3e ("gve: Add XDP REDIRECT support for GQI-QPL format")
Cc: stable@vger.kernel.org
Reviewed-by: Praveen Kaligineedi <pkaligineedi@google.com>
Reviewed-by: Jeroen de Borst <jeroendb@google.com>
Signed-off-by: Joshua Washington <joshwash@google.com>
Link: https://patch.msgid.link/20250214224417.1237818-1-joshwash@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Previously, after successfully flushing the xmit buffer to VIOS,
the tx_bytes stat was incremented by the length of the skb.
It is invalid to access the skb memory after sending the buffer to
the VIOS because, at any point after sending, the VIOS can trigger
an interrupt to free this memory. A race between reading skb->len
and freeing the skb is possible (especially during LPM) and will
result in use-after-free:
==================================================================
BUG: KASAN: slab-use-after-free in ibmvnic_xmit+0x75c/0x1808 [ibmvnic]
Read of size 4 at addr c00000024eb48a70 by task hxecom/14495
<...>
Call Trace:
[c000000118f66cf0] [c0000000018cba6c] dump_stack_lvl+0x84/0xe8 (unreliable)
[c000000118f66d20] [c0000000006f0080] print_report+0x1a8/0x7f0
[c000000118f66df0] [c0000000006f08f0] kasan_report+0x128/0x1f8
[c000000118f66f00] [c0000000006f2868] __asan_load4+0xac/0xe0
[c000000118f66f20] [c0080000046eac84] ibmvnic_xmit+0x75c/0x1808 [ibmvnic]
[c000000118f67340] [c0000000014be168] dev_hard_start_xmit+0x150/0x358
<...>
Freed by task 0:
kasan_save_stack+0x34/0x68
kasan_save_track+0x2c/0x50
kasan_save_free_info+0x64/0x108
__kasan_mempool_poison_object+0x148/0x2d4
napi_skb_cache_put+0x5c/0x194
net_tx_action+0x154/0x5b8
handle_softirqs+0x20c/0x60c
do_softirq_own_stack+0x6c/0x88
<...>
The buggy address belongs to the object at c00000024eb48a00 which
belongs to the cache skbuff_head_cache of size 224
==================================================================
Fixes: 032c5e8284 ("Driver for IBM System i/p VNIC protocol")
Signed-off-by: Nick Child <nnac123@linux.ibm.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250214155233.235559-1-nnac123@linux.ibm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
For successful XDP_TX and XDP_REDIRECT cases, the packet was received
successfully so update RX statistics. Use original received
packet length for that.
TX packets statistics are incremented on TX completion so don't
update it while TX queueing.
If xdp_convert_buff_to_frame() fails, increment tx_dropped.
Signed-off-by: Roger Quadros <rogerq@kernel.org>
Fixes: 8acacc40f7 ("net: ethernet: ti: am65-cpsw: Add minimal XDP support")
Link: https://patch.msgid.link/20250210-am65-cpsw-xdp-fixes-v1-2-ec6b1f7f1aca@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
If the XDP program doesn't result in XDP_PASS then we leak the
memory allocated by am65_cpsw_build_skb().
It is pointless to allocate SKB memory before running the XDP
program as we would be wasting CPU cycles for cases other than XDP_PASS.
Move the SKB allocation after evaluating the XDP program result.
This fixes the memleak. A performance boost is seen for XDP_DROP test.
XDP_DROP test:
Before: 460256 rx/s 0 err/s
After: 784130 rx/s 0 err/s
Fixes: 8acacc40f7 ("net: ethernet: ti: am65-cpsw: Add minimal XDP support")
Signed-off-by: Roger Quadros <rogerq@kernel.org>
Link: https://patch.msgid.link/20250210-am65-cpsw-xdp-fixes-v1-1-ec6b1f7f1aca@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2025-02-11 (idpf, ixgbe, igc)
For idpf:
Sridhar fixes a couple issues in handling of RSC packets.
Josh adds a call to set_real_num_queues() to keep queue count in sync.
For ixgbe:
Piotr removes missed IS_ERR() removal when ERR_PTR usage was removed.
For igc:
Zdenek Bouska fixes reporting of Rx timestamp with AF_XDP.
Siang sets buffer type on empty frame to ensure proper handling.
* '200GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
igc: Set buffer type for empty frames in igc_init_empty_frame
igc: Fix HW RX timestamp when passed by ZC XDP
ixgbe: Fix possible skb NULL pointer dereference
idpf: call set_real_num_queues in idpf_open
idpf: record rx queue in skb for RSC packets
idpf: fix handling rsc packet with a single segment
====================
Link: https://patch.msgid.link/20250211214343.4092496-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Fixes HW RX timestamp in the following scenario:
- AF_PACKET socket with enabled HW RX timestamps is created
- AF_XDP socket with enabled zero copy is created
- frame is forwarded to the BPF program, where the timestamp should
still be readable (extracted by igc_xdp_rx_timestamp(), kfunc
behind bpf_xdp_metadata_rx_timestamp())
- the frame got XDP_PASS from BPF program, redirecting to the stack
- AF_PACKET socket receives the frame with HW RX timestamp
Moves the skb timestamp setting from igc_dispatch_skb_zc() to
igc_construct_skb_zc() so that igc_construct_skb_zc() is similar to
igc_construct_skb().
This issue can also be reproduced by running:
# tools/testing/selftests/bpf/xdp_hw_metadata enp1s0
When a frame with the wrong port 9092 (instead of 9091) is used:
# echo -n xdp | nc -u -q1 192.168.10.9 9092
then the RX timestamp is missing and xdp_hw_metadata prints:
skb hwtstamp is not found!
With this fix or when copy mode is used:
# tools/testing/selftests/bpf/xdp_hw_metadata -c enp1s0
then RX timestamp is found and xdp_hw_metadata prints:
found skb hwtstamp = 1736509937.852786132
Fixes: 069b142f58 ("igc: Add support for PTP .getcyclesx64()")
Signed-off-by: Zdenek Bouska <zdenek.bouska@siemens.com>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Florian Bezdeka <florian.bezdeka@siemens.com>
Reviewed-by: Song Yoong Siang <yoong.siang.song@intel.com>
Tested-by: Mor Bar-Gabay <morx.bar.gabay@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
On initial driver load, alloc_etherdev_mqs is called with whatever max
queue values are provided by the control plane. However, if the driver
is loaded on a system where num_online_cpus() returns less than the max
queues, the netdev will think there are more queues than are actually
available. Only num_online_cpus() will be allocated, but
skb_get_queue_mapping(skb) could possibly return an index beyond the
range of allocated queues. Consequently, the packet is silently dropped
and it appears as if TX is broken.
Set the real number of queues during open so the netdev knows how many
queues will be allocated.
Fixes: 1c325aac10 ("idpf: configure resources for TX queues")
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Handle rsc packet with a single segment same as a multi
segment rsc packet so that CHECKSUM_PARTIAL is set in the
skb->ip_summed field. The current code is passing CHECKSUM_NONE
resulting in TCP GRO layer doing checksum in SW and hiding the
issue. This will fail when using dmabufs as payload buffers as
skb frag would be unreadable.
Fixes: 3a8845af66 ("idpf: add RX splitq napi poll support")
Signed-off-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Commit df542f6693 ("net: stmmac: Switch to zero-copy in
non-XDP RX path") makes DMA write received frame into buffer at offset
of NET_SKB_PAD and sets page pool parameters to sync from offset of
NET_SKB_PAD. But when Header Payload Split is enabled, the header is
written at offset of NET_SKB_PAD, while the payload is written at
offset of zero. Uncorrect offset parameter for the payload breaks dma
coherence [1] since both CPU and DMA touch the page buffer from offset
of zero which is not handled by the page pool sync parameter.
And in case the DMA cannot split the received frame, for example,
a large L2 frame, pp_params.max_len should grow to match the tail
of entire frame.
[1] https://lore.kernel.org/netdev/d465f277-bac7-439f-be1d-9a47dfe2d951@nvidia.com/
Reported-by: Jon Hunter <jonathanh@nvidia.com>
Reported-by: Brad Griffis <bgriffis@nvidia.com>
Suggested-by: Ido Schimmel <idosch@idosch.org>
Fixes: df542f6693 ("net: stmmac: Switch to zero-copy in non-XDP RX path")
Signed-off-by: Furong Xu <0x1207@gmail.com>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Tested-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250207085639.13580-1-0x1207@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This reverts commit 8865d22656, which caused breakage for platforms
which are not using xgmac2 or gmac4. Only these two cores have the
capability of providing the FIFO sizes from hardware capability fields
(which are provided in priv->dma_cap.[tr]x_fifo_size.)
All other cores can not, which results in these two fields containing
zero. We also have platforms that do not provide a value in
priv->plat->[tr]x_fifo_size, resulting in these also being zero.
This causes the new tests introduced by the reverted commit to fail,
and produce e.g.:
stmmaceth f0804000.eth: Can't specify Rx FIFO size
An example of such a platform which fails is QEMU's npcm750-evb.
This uses dwmac1000 which, as noted above, does not have the capability
to provide the FIFO sizes from hardware.
Therefore, revert the commit to maintain compatibility with the way
the driver used to work.
Reported-by: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/4e98f967-f636-46fb-9eca-d383b9495b86@roeck-us.net
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Steven Price <steven.price@arm.com>
Fixes: 8865d22656 ("net: stmmac: Specify hardware capability value when FIFO size isn't specified")
Link: https://patch.msgid.link/E1tfeyR-003YGJ-Gb@rmk-PC.armlinux.org.uk
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Tony Nguyen says:
====================
ice: fix Rx data path for heavy 9k MTU traffic
Maciej Fijalkowski says:
This patchset fixes a pretty nasty issue that was reported by RedHat
folks which occurred after ~30 minutes (this value varied, just trying
here to state that it was not observed immediately but rather after a
considerable longer amount of time) when ice driver was tortured with
jumbo frames via mix of iperf traffic executed simultaneously with
wrk/nginx on client/server sides (HTTP and TCP workloads basically).
The reported splats were spanning across all the bad things that can
happen to the state of page - refcount underflow, use-after-free, etc.
One of these looked as follows:
[ 2084.019891] BUG: Bad page state in process swapper/34 pfn:97fcd0
[ 2084.025990] page:00000000a60ee772 refcount:-1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x97fcd0
[ 2084.035462] flags: 0x17ffffc0000000(node=0|zone=2|lastcpupid=0x1fffff)
[ 2084.041990] raw: 0017ffffc0000000 dead000000000100 dead000000000122 0000000000000000
[ 2084.049730] raw: 0000000000000000 0000000000000000 ffffffffffffffff 0000000000000000
[ 2084.057468] page dumped because: nonzero _refcount
[ 2084.062260] Modules linked in: bonding tls sunrpc intel_rapl_msr intel_rapl_common intel_uncore_frequency intel_uncore_frequency_common i10nm_edac nfit libnvdimm x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel kvm mgag200 irqd
[ 2084.137829] CPU: 34 PID: 0 Comm: swapper/34 Kdump: loaded Not tainted 5.14.0-427.37.1.el9_4.x86_64 #1
[ 2084.147039] Hardware name: Dell Inc. PowerEdge R750/0216NK, BIOS 1.13.2 12/19/2023
[ 2084.154604] Call Trace:
[ 2084.157058] <IRQ>
[ 2084.159080] dump_stack_lvl+0x34/0x48
[ 2084.162752] bad_page.cold+0x63/0x94
[ 2084.166333] check_new_pages+0xb3/0xe0
[ 2084.170083] rmqueue_bulk+0x2d2/0x9e0
[ 2084.173749] ? ktime_get+0x35/0xa0
[ 2084.177159] rmqueue_pcplist+0x13b/0x210
[ 2084.181081] rmqueue+0x7d3/0xd40
[ 2084.184316] ? xas_load+0x9/0xa0
[ 2084.187547] ? xas_find+0x183/0x1d0
[ 2084.191041] ? xa_find_after+0xd0/0x130
[ 2084.194879] ? intel_iommu_iotlb_sync_map+0x89/0xe0
[ 2084.199759] get_page_from_freelist+0x11f/0x530
[ 2084.204291] __alloc_pages+0xf2/0x250
[ 2084.207958] ice_alloc_rx_bufs+0xcc/0x1c0 [ice]
[ 2084.212543] ice_clean_rx_irq+0x631/0xa20 [ice]
[ 2084.217111] ice_napi_poll+0xdf/0x2a0 [ice]
[ 2084.221330] __napi_poll+0x27/0x170
[ 2084.224824] net_rx_action+0x233/0x2f0
[ 2084.228575] __do_softirq+0xc7/0x2ac
[ 2084.232155] __irq_exit_rcu+0xa1/0xc0
[ 2084.235821] common_interrupt+0x80/0xa0
[ 2084.239662] </IRQ>
[ 2084.241768] <TASK>
The fix is mostly about reverting what was done in commit 1dc1a7e7f4
("ice: Centrallize Rx buffer recycling") followed by proper timing on
page_count() storage and then removing the ice_rx_buf::act related logic
(which was mostly introduced for purposes from cited commit).
Special thanks to Xu Du for providing reproducer and Jacob Keller for
initial extensive analysis.
* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
ice: stop storing XDP verdict within ice_rx_buf
ice: gather page_count()'s of each frag right before XDP prog call
ice: put Rx buffers after being done with current frame
====================
Link: https://patch.msgid.link/20250131185415.3741532-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Disable PCIe AER on the tg3 device on system reboot on a limited
list of Dell PowerEdge systems. This prevents a fatal PCIe AER event
on the tg3 device during the ACPI _PTS (prepare to sleep) method for
S5 on those systems. The _PTS is invoked by acpi_enter_sleep_state_prep()
as part of the kernel's reboot sequence as a result of commit
38f34dba80 ("PM: ACPI: reboot: Reinstate S5 for reboot").
There was an earlier fix for this problem by commit 2ca1c94ce0
("tg3: Disable tg3 device on system reboot to avoid triggering AER").
But it was discovered that this earlier fix caused a reboot hang
when some Dell PowerEdge servers were booted via ipxe. To address
this reboot hang, the earlier fix was essentially reverted by commit
9fc3bc7643 ("tg3: power down device only on SYSTEM_POWER_OFF").
This re-exposed the tg3 PCIe AER on reboot problem.
This fix is not an ideal solution because the root cause of the AER
is in system firmware. Instead, it's a targeted work-around in the
tg3 driver.
Note also that the PCIe AER must be disabled on the tg3 device even
if the system is configured to use "firmware first" error handling.
V3:
- Fix sparse warning on improper comparison of pdev->current_state
- Adhere to netdev comment style
Fixes: 9fc3bc7643 ("tg3: power down device only on SYSTEM_POWER_OFF")
Signed-off-by: Lenny Szubowicz <lszubowi@redhat.com>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some Wake-on-LAN modes such as WAKE_FILTER may only be supported by the MAC,
while others might be only supported by the PHY. Make sure that the .get_wol()
returns the union of both rather than only that of the PHY if the PHY supports
Wake-on-LAN.
When disabling Wake-on-LAN, make sure that this is done at both the PHY
and MAC level, rather than doing an early return from the PHY driver.
Fixes: 7e400ff35c ("net: bcmgenet: Add support for PHY-based Wake-on-LAN")
Fixes: 9ee09edc05 ("net: bcmgenet: Properly overlay PHY and MAC Wake-on-LAN capabilities")
Signed-off-by: Florian Fainelli <florian.fainelli@broadcom.com>
Link: https://patch.msgid.link/20250129231342.35013-1-florian.fainelli@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Idea behind having ice_rx_buf::act was to simplify and speed up the Rx
data path by walking through buffers that were representing cleaned HW
Rx descriptors. Since it caused us a major headache recently and we
rolled back to old approach that 'puts' Rx buffers right after running
XDP prog/creating skb, this is useless now and should be removed.
Get rid of ice_rx_buf::act and related logic. We still need to take care
of a corner case where XDP program releases a particular fragment.
Make ice_run_xdp() to return its result and use it within
ice_put_rx_mbuf().
Fixes: 2fba7dc515 ("ice: Add support for XDP multi-buffer on Rx side")
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Chandan Kumar Rout <chandanx.rout@intel.com> (A Contingent Worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
If we store the pgcnt on few fragments while being in the middle of
gathering the whole frame and we stumbled upon DD bit not being set, we
terminate the NAPI Rx processing loop and come back later on. Then on
next NAPI execution we work on previously stored pgcnt.
Imagine that second half of page was used actively by networking stack
and by the time we came back, stack is not busy with this page anymore
and decremented the refcnt. The page reuse algorithm in this case should
be good to reuse the page but given the old refcnt it will not do so and
attempt to release the page via page_frag_cache_drain() with
pagecnt_bias used as an arg. This in turn will result in negative refcnt
on struct page, which was initially observed by Xu Du.
Therefore, move the page count storage from ice_get_rx_buf() to a place
where we are sure that whole frame has been collected, but before
calling XDP program as it internally can also change the page count of
fragments belonging to xdp_buff.
Fixes: ac07533911 ("ice: Store page count inside ice_rx_buf")
Reported-and-tested-by: Xu Du <xudu@redhat.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Co-developed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Chandan Kumar Rout <chandanx.rout@intel.com> (A Contingent Worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Introduce a new helper ice_put_rx_mbuf() that will go through gathered
frags from current frame and will call ice_put_rx_buf() on them. Current
logic that was supposed to simplify and optimize the driver where we go
through a batch of all buffers processed in current NAPI instance turned
out to be broken for jumbo frames and very heavy load that was coming
from both multi-thread iperf and nginx/wrk pair between server and
client. The delay introduced by approach that we are dropping is simply
too big and we need to take the decision regarding page
recycling/releasing as quick as we can.
While at it, address an error path of ice_add_xdp_frag() - we were
missing buffer putting from day 1 there.
As a nice side effect we get rid of annoying and repetitive three-liner:
xdp->data = NULL;
rx_ring->first_desc = ntc;
rx_ring->nr_frags = 0;
by embedding it within introduced routine.
Fixes: 1dc1a7e7f4 ("ice: Centrallize Rx buffer recycling")
Reported-and-tested-by: Xu Du <xudu@redhat.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Co-developed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Chandan Kumar Rout <chandanx.rout@intel.com> (A Contingent Worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Pull networking fixes from Jakub Kicinski:
"Including fixes from IPSec, netfilter and Bluetooth.
Nothing really stands out, but as usual there's a slight concentration
of fixes for issues added in the last two weeks before the merge
window, and driver bugs from 6.13 which tend to get discovered upon
wider distribution.
Current release - regressions:
- net: revert RTNL changes in unregister_netdevice_many_notify()
- Bluetooth: fix possible infinite recursion of btusb_reset
- eth: adjust locking in some old drivers which protect their state
with spinlocks to avoid sleeping in atomic; core protects netdev
state with a mutex now
Previous releases - regressions:
- eth:
- mlx5e: make sure we pass node ID, not CPU ID to kvzalloc_node()
- bgmac: reduce max frame size to support just 1500 bytes; the
jumbo frame support would previously cause OOB writes, but now
fails outright
- mptcp: blackhole only if 1st SYN retrans w/o MPC is accepted, avoid
false detection of MPTCP blackholing
Previous releases - always broken:
- mptcp: handle fastopen disconnect correctly
- xfrm:
- make sure skb->sk is a full sock before accessing its fields
- fix taking a lock with preempt disabled for RT kernels
- usb: ipheth: improve safety of packet metadata parsing; prevent
potential OOB accesses
- eth: renesas: fix missing rtnl lock in suspend/resume path"
* tag 'net-6.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (88 commits)
MAINTAINERS: add Neal to TCP maintainers
net: revert RTNL changes in unregister_netdevice_many_notify()
net: hsr: fix fill_frame_info() regression vs VLAN packets
doc: mptcp: sysctl: blackhole_timeout is per-netns
mptcp: blackhole only if 1st SYN retrans w/o MPC is accepted
netfilter: nf_tables: reject mismatching sum of field_len with set key length
net: sh_eth: Fix missing rtnl lock in suspend/resume path
net: ravb: Fix missing rtnl lock in suspend/resume path
selftests/net: Add test for loading devbound XDP program in generic mode
net: xdp: Disallow attaching device-bound programs in generic mode
tcp: correct handling of extreme memory squeeze
bgmac: reduce max frame size to support just MTU 1500
vsock/test: Add test for connect() retries
vsock/test: Add test for UAF due to socket unbinding
vsock/test: Introduce vsock_connect_fd()
vsock/test: Introduce vsock_bind()
vsock: Allow retrying on connect() failure
vsock: Keep the binding until socket destruction
Bluetooth: L2CAP: accept zero as a special value for MTU auto-selection
Bluetooth: btnxpuart: Fix glitches seen in dual A2DP streaming
...
Fix the suspend/resume path by ensuring the rtnl lock is held where
required. Calls to sh_eth_close, sh_eth_open and wol operations must be
performed under the rtnl lock to prevent conflicts with ongoing ndo
operations.
Fixes: b71af04676 ("sh_eth: add more PM methods")
Tested-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
