| Age | Commit message (Collapse) | Author |
|---|
|
Add property mac-termination-ohms in the device tree bindings for selecting
the resistance value of the builtin series termination resistors of the
PHY. Changing the resistance to an appropriate value can reduce signal
reflections and therefore improve signal quality.
Signed-off-by: Dimitri Fedrau <dimitri.fedrau@liebherr.com>
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://patch.msgid.link/20250416-dp83822-mac-impedance-v3-1-028ac426cddb@liebherr.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Use pci_prepare_to_sleep() like PCI core does in pci_pm_suspend_noirq.
This aligns setting a low-power mode during shutdown with the handling
of the transition to system suspend. Also the transition to runtime
suspend uses pci_target_state() instead of setting D3hot unconditionally.
Note: pci_prepare_to_sleep() uses device_may_wakeup() to check whether
device may generate wakeup events. So we don't lose anything by
not passing tp->saved_wolopts any longer.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://patch.msgid.link/f573fdbd-ba6d-41c1-b68f-311d3c88db2c@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The last use of rvu_npc_enable_bcast_entry() was removed in 2021 by
commit 967db3529eca ("octeontx2-af: add support for multicast/promisc
packet replication feature")
Remove it.
Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://patch.msgid.link/20250420225810.171852-1-linux@treblig.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
When WoL is enabled, we update the software state in phylink to
indicate that the link is down, and disable the resolver from
bringing the link back up.
On resume, we attempt to bring the overall state into consistency
by calling the .mac_link_down() method, but this is wrong if the
link was already down, as phylink strictly orders the .mac_link_up()
and .mac_link_down() methods - and this would break that ordering.
Fixes: f97493657c63 ("net: phylink: add suspend/resume support")
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/E1u55Qf-0016RN-PA@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
- subpage mode fixes:
- access correct object (folio) when looking up bit offset
- fix assertion condition for number of blocks per folio
- fix upper boundary of locking range in hole punch
- zoned fixes:
- fix potential deadlock caught by lockdep when zone reporting and
device freeze run in parallel
- fix zone write pointer mismatch and NULL pointer dereference when
metadata are converted from DUP to RAID1
- fix error handling when reloc inode creation fails
- in tree-checker, unify error code for header level check
- block layer: add helpers to read zone capacity
* tag 'for-6.15-rc3-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: zoned: skip reporting zone for new block group
block: introduce zone capacity helper
btrfs: tree-checker: adjust error code for header level check
btrfs: fix invalid inode pointer after failure to create reloc inode
btrfs: zoned: return EIO on RAID1 block group write pointer mismatch
btrfs: fix the ASSERT() inside GET_SUBPAGE_BITMAP()
btrfs: avoid page_lockend underflow in btrfs_punch_hole_lock_range()
btrfs: subpage: access correct object when reading bitmap start in subpage_calc_start_bit()
|
|
Pull integrity fix from Roberto Sassu:
"One performance fix to avoid unnecessarily taking the inode lock"
* tag 'integrity-6.15-rc3-fix' of https://github.com/linux-integrity/linux:
ima: process_measurement() needlessly takes inode_lock() on MAY_READ
|
|
On IMA policy update, if a measure rule exists in the policy,
IMA_MEASURE is set for ima_policy_flags which makes the violation_check
variable always true. Coupled with a no-action on MAY_READ for a
FILE_CHECK call, we're always taking the inode_lock().
This becomes a performance problem for extremely heavy read-only workloads.
Therefore, prevent this only in the case there's no action to be taken.
Signed-off-by: Frederick Lawler <fred@cloudflare.com>
Acked-by: Roberto Sassu <roberto.sassu@huawei.com>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
|
|
p8022.c defines two external functions, register_8022_client()
and unregister_8022_client(), the last use of which was removed in
2018 by
commit 7a2e838d28cf ("staging: ipx: delete it from the tree")
Remove the p8022.c file, it's corresponding header, and glue
surrounding it. There was one place the header was included in vlan.c
but it didn't use the functions it declared.
There was a comment in net/802/Makefile about checking
against net/core/Makefile, but that's at least 20 years old and
there's no sign of net/core/Makefile mentioning it.
Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Link: https://patch.msgid.link/20250418011519.145320-1-linux@treblig.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
In lwtunnel_{output|xmit}(), dev_xmit_recursion() may be called in
preemptible scope for PREEMPT kernels. This patch disables BHs before
calling dev_xmit_recursion(). BHs are re-enabled only at the end, since
we must ensure the same CPU is used for both dev_xmit_recursion_inc()
and dev_xmit_recursion_dec() (and any other recursion levels in some
cases) in order to maintain valid per-cpu counters.
Reported-by: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Closes: https://lore.kernel.org/netdev/CAADnVQJFWn3dBFJtY+ci6oN1pDFL=TzCmNbRgey7MdYxt_AP2g@mail.gmail.com/
Reported-by: Eduard Zingerman <eddyz87@gmail.com>
Closes: https://lore.kernel.org/netdev/m2h62qwf34.fsf@gmail.com/
Fixes: 986ffb3a57c5 ("net: lwtunnel: fix recursion loops")
Signed-off-by: Justin Iurman <justin.iurman@uliege.be>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250416160716.8823-1-justin.iurman@uliege.be
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Zero-initialize TCP header via memset() to avoid garbage values that
may affect checksum or behavior during test transmission.
Also zero-fill allocated payload and padding regions using memset()
after skb_put(), ensuring deterministic content for all outgoing
test packets.
Fixes: 3e1e58d64c3d ("net: add generic selftest support")
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Cc: stable@vger.kernel.org
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250416160125.2914724-1-o.rempel@pengutronix.de
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Add support for ndo_setup_tc to enable CBS offload functionality as
part of traffic control configuration for network devices, where CBS
is applied from the CPU to the switch. More specifically, CBS is
applied at the GMAC in the topmost architecture diagram.
Signed-off-by: Justin Lai <justinlai0215@realtek.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250416115757.28156-1-justinlai0215@realtek.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Propagate the error code if key_alloc() fails. Don't return
success.
Fixes: 9d1d2b59341f ("rxrpc: rxgk: Implement the yfs-rxgk security class (GSSAPI)")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/Z_-P_1iLDWksH1ik@stanley.mountain
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
With lan88xx based devices the lan78xx driver can get stuck in an
interrupt loop while bringing the device up, flooding the kernel log
with messages like the following:
lan78xx 2-3:1.0 enp1s0u3: kevent 4 may have been dropped
Removing interrupt support from the lan88xx PHY driver forces the
driver to use polling instead, which avoids the problem.
The issue has been observed with Raspberry Pi devices at least since
4.14 (see [1], bug report for their downstream kernel), as well as
with Nvidia devices [2] in 2020, where disabling interrupts was the
vendor-suggested workaround (together with the claim that phylib
changes in 4.9 made the interrupt handling in lan78xx incompatible).
Iperf reports well over 900Mbits/sec per direction with client in
--dualtest mode, so there does not seem to be a significant impact on
throughput (lan88xx device connected via switch to the peer).
[1] https://github.com/raspberrypi/linux/issues/2447
[2] https://forums.developer.nvidia.com/t/jetson-xavier-and-lan7800-problem/142134/11
Link: https://lore.kernel.org/0901d90d-3f20-4a10-b680-9c978e04ddda@lunn.ch
Fixes: 792aec47d59d ("add microchip LAN88xx phy driver")
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Cc: kernel-list@raspberrypi.com
Cc: stable@vger.kernel.org
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20250416102413.30654-1-fiona.klute@gmx.de
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Shannon Nelson says:
====================
ionic: support QSFP CMIS
This patchset sets up support for additional pages and better
handling of the QSFP CMIS data.
v1: https://lore.kernel.org/netdev/20250411182140.63158-1-shannon.nelson@amd.com/
====================
Link: https://patch.msgid.link/20250415231317.40616-1-shannon.nelson@amd.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Make the CMIS module type's page 17 channel data available for
ethtool to request. As done previously, carve space for this
data from the port_info reserved space.
In the future, if additional pages are needed, a new firmware
AdminQ command will be added for accessing random pages.
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250415231317.40616-4-shannon.nelson@amd.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Add support for the newer get_module_eeprom_by_page interface.
Only the upper half of the 256 byte page is available for
reading, and the firmware puts the two sections into the
extended sprom buffer, so a union is used over the extended
sprom buffer to make clear which page is to be accessed.
With get_module_eeprom_by_page implemented there is no need
for the older get_module_info or git_module_eeprom interfaces,
so remove them.
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250415231317.40616-3-shannon.nelson@amd.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Some QSFP modules have more eeprom to be read by ethtool than
the initial high and low page 0 that is currently available
in the DSC's ionic sprom[] buffer. Since the current sprom[]
is baked into the middle of an existing API struct, to make
the high end of page 1 and page 2 available a block is carved
from a reserved space of the existing port_info struct and the
ionic_get_module_eeprom() service is taught how to get there.
Newer firmware writes the additional QSFP page info here,
yet this remains backward compatible because older firmware
sets this space to all 0 and older ionic drivers do not use
the reserved space.
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250415231317.40616-2-shannon.nelson@amd.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Ido Schimmel says:
====================
vxlan: Convert FDB table to rhashtable
The VXLAN driver currently stores FDB entries in a hash table with a
fixed number of buckets (256), resulting in reduced performance as the
number of entries grows. This patchset solves the issue by converting
the driver to use rhashtable which maintains a more or less constant
performance regardless of the number of entries.
Measured transmitted packets per second using a single pktgen thread
with varying number of entries when the transmitted packet always hits
the default entry (worst case):
Number of entries | Improvement
------------------|------------
1k | +1.12%
4k | +9.22%
16k | +55%
64k | +585%
256k | +2460%
The first patches are preparations for the conversion in the last patch.
Specifically, the series is structured as follows:
Patch #1 adds RCU read-side critical sections in the Tx path when
accessing FDB entries. Targeting at net-next as I am not aware of any
issues due to this omission despite the code being structured that way
for a long time. Without it, traces will be generated when converting
FDB lookup to rhashtable_lookup().
Patch #2-#5 simplify the creation of the default FDB entry (all-zeroes).
Current code assumes that insertion into the hash table cannot fail,
which will no longer be true with rhashtable.
Patches #6-#10 add FDB entries to a linked list for entry traversal
instead of traversing over them using the fixed size hash table which is
removed in the last patch.
Patches #11-#12 add wrappers for FDB lookup that make it clear when each
should be used along with lockdep annotations. Needed as a preparation
for rhashtable_lookup() that must be called from an RCU read-side
critical section.
Patch #13 treats dst cache initialization errors as non-fatal. See more
info in the commit message. The current code happens to work because
insertion into the fixed size hash table is slow enough for the per-CPU
allocator to be able to create new chunks of per-CPU memory.
Patch #14 adds an FDB key structure that includes the MAC address and
source VNI. To be used as rhashtable key.
Patch #15 does the conversion to rhashtable.
====================
Link: https://patch.msgid.link/20250415121143.345227-1-idosch@nvidia.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
FDB entries are currently stored in a hash table with a fixed number of
buckets (256), resulting in performance degradation as the number of
entries grows. Solve this by converting the driver to use rhashtable
which maintains more or less constant performance regardless of the
number of entries.
Measured transmitted packets per second using a single pktgen thread
with varying number of entries when the transmitted packet always hits
the default entry (worst case):
Number of entries | Improvement
------------------|------------
1k | +1.12%
4k | +9.22%
16k | +55%
64k | +585%
256k | +2460%
In addition, the change reduces the size of the VXLAN device structure
from 2584 bytes to 672 bytes.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-16-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
In preparation for converting the FDB table to rhashtable, introduce a
key structure that includes the MAC address and source VNI.
No functional changes intended.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-15-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
FDB entries are allocated in an atomic context as they can be added from
the data path when learning is enabled.
After converting the FDB hash table to rhashtable, the insertion rate
will be much higher (*) which will entail a much higher rate of per-CPU
allocations via dst_cache_init().
When adding a large number of entries (e.g., 256k) in a batch, a small
percentage (< 0.02%) of these per-CPU allocations will fail [1]. This
does not happen with the current code since the insertion rate is low
enough to give the per-CPU allocator a chance to asynchronously create
new chunks of per-CPU memory.
Given that:
a. Only a small percentage of these per-CPU allocations fail.
b. The scenario where this happens might not be the most realistic one.
c. The driver can work correctly without dst caches. The dst_cache_*()
APIs first check that the dst cache was properly initialized.
d. The dst caches are not always used (e.g., 'tos inherit').
It seems reasonable to not treat these allocation failures as fatal.
Therefore, do not bail when dst_cache_init() fails and suppress warnings
by specifying '__GFP_NOWARN'.
[1] percpu: allocation failed, size=40 align=8 atomic=1, atomic alloc failed, no space left
(*) 97% reduction in average latency of vxlan_fdb_update() when adding
256k entries in a batch.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-14-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
__vxlan_find_mac() is called from both the data path (e.g., during
learning) and the control path (e.g., when replacing an entry). The
function is missing lockdep annotations to make sure that the FDB hash
lock is held during FDB updates.
Rename __vxlan_find_mac() to vxlan_find_mac_rcu() to reflect the fact
that it should be called from an RCU read-side critical section and call
it from vxlan_find_mac() which checks that the FDB hash lock is held.
Change callers to invoke the appropriate function.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-13-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
vxlan_find_mac() is only expected to be called from the Tx path as it
updates the 'used' timestamp. Rename it to vxlan_find_mac_tx() to
reflect that and to avoid incorrect updates of this timestamp like those
addressed by commit 9722f834fe9a ("vxlan: Avoid unnecessary updates to
FDB 'used' time").
No functional changes intended.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-12-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Instead of holding the FDB hash lock when traversing the FDB linked list
during flushing, use RCU and only acquire the lock for entries that need
to be flushed.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-11-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Instead of holding the FDB hash lock when traversing the FDB linked list
during garbage collection, use RCU and only acquire the lock for entries
that need to be removed (aged out).
Avoid races by using hlist_unhashed() to check that the entry has not
been removed from the list by another thread.
Note that vxlan_fdb_destroy() uses hlist_del_init_rcu() to remove an
entry from the list which should cause list_unhashed() to return true.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-10-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
In preparation for removing the fixed size hash table, convert FDB entry
traversal to use the newly added FDB linked list.
No functional changes intended.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-9-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Currently, FDB entries are stored in a hash table with a fixed number of
buckets. The table is used for both lookups and entry traversal.
Subsequent patches will convert the table to rhashtable which is not
suitable for entry traversal.
In preparation for this conversion, add FDB entries to a linked list.
Subsequent patches will convert the driver to use this list when
traversing entries during dump, flush, etc.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-8-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Currently, the VXLAN driver stores FDB entries in a hash table with a
fixed number of buckets (256). Subsequent patches are going to convert
this table to rhashtable with a linked list for entry traversal, as
rhashtable is more scalable.
In preparation for this conversion, move from a per-bucket spin lock to
a single spin lock that protects the entire FDB table.
The per-bucket spin locks were introduced by commit fe1e0713bbe8
("vxlan: Use FDB_HASH_SIZE hash_locks to reduce contention") citing
"huge contention when inserting/deleting vxlan_fdbs into the fdb_head".
It is not clear from the commit message which code path was holding the
spin lock for long periods of time, but the obvious suspect is the FDB
cleanup routine (vxlan_cleanup()) that periodically traverses the entire
table in order to delete aged-out entries.
This will be solved by subsequent patches that will convert the FDB
cleanup routine to traverse the linked list of FDB entries using RCU,
only acquiring the spin lock when deleting an aged-out entry.
The change reduces the size of the VXLAN device structure from 3600
bytes to 2576 bytes.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-7-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
The default FDB entry can be associated with a net device if a physical
device (i.e., 'dev PHYS_DEV') was specified during the creation of the
VXLAN device.
The assignment of the net device pointer to 'dst->remote_dev' logically
belongs in the if block that resolves the pointer from the specified
ifindex, so move it there.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-6-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Commit 0241b836732f ("vxlan: fix default fdb entry netlink notify
ordering during netdev create") split the creation of the default FDB
entry from its notification to avoid sending a RTM_NEWNEIGH notification
before RTM_NEWLINK.
Previous patches restructured the code so that the default FDB entry is
created after registering the VXLAN device and the notification about
the new entry immediately follows its creation.
Therefore, simplify the code and revert back to vxlan_fdb_update() which
takes care of both creating the FDB entry and notifying user space
about it.
Hold the FDB hash lock when calling vxlan_fdb_update() like it expects.
A subsequent patch will add a lockdep assertion to make sure this is
indeed the case.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-5-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Commit 7c31e54aeee5 ("vxlan: do not destroy fdb if register_netdevice()
is failed") split the insertion of FDB entries into the FDB hash table
from the function where they are created.
This was done in order to work around a problem that is no longer
possible after the previous patch. Simplify the code and move the body
of vxlan_fdb_insert() back into vxlan_fdb_create().
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-4-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
There is asymmetry in how the default FDB entry (all-zeroes) is created
and destroyed in the VXLAN driver. It is created as part of the driver's
newlink() routine, but destroyed as part of its ndo_uninit() routine.
This caused multiple problems in the past. First, commit 0241b836732f
("vxlan: fix default fdb entry netlink notify ordering during netdev
create") split the notification about the entry from its creation so
that it will not be notified to user space before the VXLAN device is
registered.
Then, commit 6db924687139 ("vxlan: Fix error path in
__vxlan_dev_create()") made the error path in __vxlan_dev_create()
asymmetric by destroying the FDB entry before unregistering the net
device. Otherwise, the FDB entry would have been freed twice: By
ndo_uninit() as part of unregister_netdevice() and by
vxlan_fdb_destroy() in the error path.
Finally, commit 7c31e54aeee5 ("vxlan: do not destroy fdb if
register_netdevice() is failed") split the insertion of the FDB entry
into the hash table from its creation, moving the insertion after the
registration of the net device. Otherwise, like before, the FDB entry
would have been freed twice: By ndo_uninit() as part of
register_netdevice()'s error path and by vxlan_fdb_destroy() in the
error path of __vxlan_dev_create().
The end result is that the code is unnecessarily complex. In addition,
the fixed size hash table cannot be converted to rhashtable as
vxlan_fdb_insert() cannot fail, which will no longer be true with
rhashtable.
Solve this by making the addition and deletion of the default FDB entry
completely symmetric. Namely, as part of newlink() routine, create the
entry, insert it into to the hash table and send a notification to user
space after the net device was registered. Note that at this stage the
net device is still administratively down and cannot transmit / receive
packets.
Move the deletion from ndo_uninit() to the dellink routine(): Flush the
default entry together with all the other entries, before unregistering
the net device.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-3-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
The Tx path does not run from an RCU read-side critical section which
makes the current lockless accesses to FDB entries invalid. As far as I
am aware, this has not been a problem in practice, but traces will be
generated once we transition the FDB lookup to rhashtable_lookup().
Add rcu_read_{lock,unlock}() around the handling of FDB entries in the
Tx path. Remove the RCU read-side critical section from vxlan_xmit_nh()
as now the function is always called from an RCU read-side critical
section.
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20250415121143.345227-2-idosch@nvidia.com
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext
Pull sched_ext fixes from Tejun Heo:
- Use kvzalloc() so that large exit_dump buffer allocations don't fail
easily
- Remove cpu.weight / cpu.idle unimplemented warnings which are more
annoying than helpful.
This makes SCX_OPS_HAS_CGROUP_WEIGHT unnecessary. Mark it for
deprecation
* tag 'sched_ext-for-6.15-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext:
sched_ext: Mark SCX_OPS_HAS_CGROUP_WEIGHT for deprecation
sched_ext: Remove cpu.weight / cpu.idle unimplemented warnings
sched_ext: Use kvzalloc for large exit_dump allocation
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup
Pull cgroup fixes from Tejun Heo:
- Fix compilation in CONFIG_LOCKDEP && !CONFIG_PROVE_RCU configurations
- Allow "cpuset_v2_mode" mount option for "cpuset" filesystem type to
make life easier for android
* tag 'cgroup-for-6.15-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
cgroup/cpuset-v1: Add missing support for cpuset_v2_mode
cgroup: Fix compilation issue due to cgroup_mutex not being exported
|
|
Vladimir Oltean says:
====================
ENETC bug fixes for bpf_xdp_adjust_head() and bpf_xdp_adjust_tail()
It has been reported that on the ENETC driver, bpf_xdp_adjust_head()
and bpf_xdp_adjust_tail() are broken in combination with the XDP_PASS
verdict. I have constructed a series a simple XDP programs and tested
with various packet sizes and confirmed that this is the case.
Patch 3/3 fixes the core issue, which is that the sk_buff created on
XDP_PASS is created by the driver as if XDP never ran, but in fact the
geometry needs to be adjusted according to the delta applied by the
program on the original xdp_buff. It depends on commit 539c1fba1ac7
("xdp: add generic xdp_build_skb_from_buff()") which is not available in
"stable" but perhaps should be.
Patch 2/3 is a small refactor necessary for 3/3.
Patch 1/3 fixes a related issue I noticed, which is that
bpf_xdp_adjust_tail() with a positive offset works for linear XDP
buffers, but returns an error for non-linear ones, even if there is
plenty of space in the final page fragment.
====================
Link: https://patch.msgid.link/20250417120005.3288549-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Vlatko Markovikj reported that XDP programs attached to ENETC do not
work well if they use bpf_xdp_adjust_head() or bpf_xdp_adjust_tail(),
combined with the XDP_PASS verdict. A typical use case is to add or
remove a VLAN tag.
The resulting sk_buff passed to the stack is corrupted, because the
algorithm used by the driver for XDP_PASS is to unwind the current
buffer pointer in the RX ring and to re-process the current frame with
enetc_build_skb() as if XDP hadn't run. That is incorrect because XDP
may have modified the geometry of the buffer, which we then are
completely unaware of. We are looking at a modified buffer with the
original geometry.
The initial reaction, both from me and from Vlatko, was to shop around
the kernel for code to steal that would calculate a delta between the
old and the new XDP buffer geometry, and apply that to the sk_buff too.
We noticed that veth and generic xdp have such code.
The headroom adjustment is pretty uncontroversial, but what turned out
severely problematic is the tailroom.
veth has this snippet:
__skb_put(skb, off); /* positive on grow, negative on shrink */
which on first sight looks decent enough, except __skb_put() takes an
"unsigned int" for the second argument, and the arithmetic seems to only
work correctly by coincidence. Second issue, __skb_put() contains a
SKB_LINEAR_ASSERT(). It's not a great pattern to make more widespread.
The skb may still be nonlinear at that point - it only becomes linear
later when resetting skb->data_len to zero.
To avoid the above, bpf_prog_run_generic_xdp() does this instead:
skb_set_tail_pointer(skb, xdp->data_end - xdp->data);
skb->len += off; /* positive on grow, negative on shrink */
which is more open-coded, uses lower-level functions and is in general a
bit too much to spread around in driver code.
Then there is the snippet:
if (xdp_buff_has_frags(xdp))
skb->data_len = skb_shinfo(skb)->xdp_frags_size;
else
skb->data_len = 0;
One would have expected __pskb_trim() to be the function of choice for
this task. But it's not used in veth/xdpgeneric because the extraneous
fragments were _already_ freed by bpf_xdp_adjust_tail() ->
bpf_xdp_frags_shrink_tail() -> ... -> __xdp_return() - the backing
memory for the skb frags and the xdp frags is the same, but they don't
keep individual references.
In fact, that is the biggest reason why this snippet cannot be reused
as-is, because ENETC temporarily constructs an skb with the original len
and the original number of frags. Because the extraneous frags are
already freed by bpf_xdp_adjust_tail() and returned to the page
allocator, it means the entire approach of using enetc_build_skb() is
questionable for XDP_PASS. To avoid that, one would need to elevate the
page refcount of all frags before calling bpf_prog_run_xdp() and drop it
after XDP_PASS.
There are other things that are missing in ENETC's handling of XDP_PASS,
like for example updating skb_shinfo(skb)->meta_len.
These are all handled correctly and cleanly in commit 539c1fba1ac7
("xdp: add generic xdp_build_skb_from_buff()"), added to net-next in
Dec 2024, and in addition might even be quicker that way. I have a very
strong preference towards backporting that commit for "stable", and that
is what is used to fix the handling bugs. It is way too messy to go
this deep into the guts of an sk_buff from the code of a device driver.
Fixes: d1b15102dd16 ("net: enetc: add support for XDP_DROP and XDP_PASS")
Reported-by: Vlatko Markovikj <vlatko.markovikj@etas.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Link: https://patch.msgid.link/20250417120005.3288549-4-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
This small snippet of code ensures that we do something with the array
of RX software buffer descriptor elements after passing the skb to the
stack. In this case, we see if the other half of the page is reusable,
and if so, we "turn around" the buffers, making them directly usable by
enetc_refill_rx_ring() without going to enetc_new_page().
We will need to perform this kind of buffer flipping from a new code
path, i.e. from XDP_PASS. Currently, enetc_build_skb() does it there
buffer by buffer, but in a subsequent change we will stop using
enetc_build_skb() for XDP_PASS.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Link: https://patch.msgid.link/20250417120005.3288549-3-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
At the time when bpf_xdp_adjust_tail() gained support for non-linear
buffers, ENETC was already generating this kind of geometry on RX, due
to its use of 2K half page buffers. Frames larger than 1472 bytes
(without FCS) are stored as multi-buffer, presenting a need for multi
buffer support to work properly even in standard MTU circumstances.
Allow bpf_xdp_frags_increase_tail() to know the allocation size of paged
data, so it can safely permit growing the tailroom of the buffer from
XDP programs.
Fixes: bf25146a5595 ("bpf: add frags support to the bpf_xdp_adjust_tail() API")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Link: https://patch.msgid.link/20250417120005.3288549-2-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The function xdp_convert_buff_to_frame() may return NULL if it fails
to correctly convert the XDP buffer into an XDP frame due to memory
constraints, internal errors, or invalid data. Failing to check for NULL
may lead to a NULL pointer dereference if the result is used later in
processing, potentially causing crashes, data corruption, or undefined
behavior.
On XDP redirect failure, the associated page must be released explicitly
if it was previously retained via get_page(). Failing to do so may result
in a memory leak, as the pages reference count is not decremented.
Cc: stable@vger.kernel.org # v5.9+
Fixes: 6c5aa6fc4def ("xen networking: add basic XDP support for xen-netfront")
Signed-off-by: Alexey Nepomnyashih <sdl@nppct.ru>
Link: https://patch.msgid.link/20250417122118.1009824-1-sdl@nppct.ru
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
Martin KaFai Lau says:
====================
pull-request: bpf-next 2025-04-17
We've added 12 non-merge commits during the last 9 day(s) which contain
a total of 18 files changed, 1748 insertions(+), 19 deletions(-).
The main changes are:
1) bpf qdisc support, from Amery Hung.
A qdisc can be implemented in bpf struct_ops programs and
can be used the same as other existing qdiscs in the
"tc qdisc" command.
2) Add xsk tail adjustment tests, from Tushar Vyavahare.
* tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next:
selftests/bpf: Test attaching bpf qdisc to mq and non root
selftests/bpf: Add a bpf fq qdisc to selftest
selftests/bpf: Add a basic fifo qdisc test
libbpf: Support creating and destroying qdisc
bpf: net_sched: Disable attaching bpf qdisc to non root
bpf: net_sched: Support updating bstats
bpf: net_sched: Add a qdisc watchdog timer
bpf: net_sched: Add basic bpf qdisc kfuncs
bpf: net_sched: Support implementation of Qdisc_ops in bpf
bpf: Prepare to reuse get_ctx_arg_idx
selftests/xsk: Add tail adjustment tests and support check
selftests/xsk: Add packet stream replacement function
====================
Link: https://patch.msgid.link/20250417184338.3152168-1-martin.lau@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Michael Chan says:
====================
bnxt_en: Update for net-next
The first patch changes the FW message timeout threshold for a warning
message. The second patch adjusts the ethtool -w coredump length to
suppress a warning. The last 2 patches are small cleanup patches for
the bnxt_ulp RoCE auxbus code.
v1: https://lore.kernel.org/netdev/20250415174818.1088646-1-michael.chan@broadcom.com/
====================
Link: https://patch.msgid.link/20250417172448.1206107-1-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
BNXT_ROCE_ULP and BNXT_MAX_ULP are no longer used. Remove them to
clean up the code.
Reviewed-by: Shruti Parab <shruti.parab@broadcom.com>
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://patch.msgid.link/20250417172448.1206107-5-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The "ref_count" field in struct bnxt_ulp is unused after
commit a43c26fa2e6c ("RDMA/bnxt_re: Remove the sriov config callback").
So we can just remove it now.
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://patch.msgid.link/20250417172448.1206107-4-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
ethtool first calls .get_dump_flags() to get the dump length. For
coredump, the driver calls the FW to get the coredump length (L1). The
min. of L1 and the user specified length is then passed to
.get_dump_data() (L2) to get the coredump. The actual coredump length
retrieved by the FW (L3) during .get_dump_data() may be smaller than L1.
This length discrepancy will trigger a WARN_ON() in
ethtool_get_dump_data().
ethtool has already vzalloc'ed a buffer with size L1. Just report
the coredump length as L2 even though the actual coredump length L3
may be smaller. The extra zero padding does not matter. This will
prevent the warning that may alarm the user.
For correctness, only do the final length update if there is no error.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Reviewed-by: Damodharam Ammepalli <damodharam.ammepalli@broadcom.com>
Signed-off-by: Shruti Parab <shruti.parab@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://patch.msgid.link/20250417172448.1206107-3-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The firmware advertises a "hwrm_cmd_max_timeout" value to the driver
for NVRAM and coredump related functions that can take tens of seconds
to complete. The driver polls for the operation to complete under
mutex and may trigger hung task watchdog warning if the wait is too long.
To warn the user about this, the driver currently prints a warning if
this advertised value exceeds 40 seconds:
Device requests max timeout of %d seconds, may trigger hung task watchdog
Initially, we chose 40 seconds, well below the kernel's default
CONFIG_DEFAULT_HUNG_TASK_TIMEOUT (120 seconds) to avoid triggering
the hung task watchdog. But 60 seconds is the timeout on most
production FW and cannot be reduced further. Change the driver's warning
threshold to 60 seconds to avoid triggering this warning on all
production devices. We also print the warning if the value exceeds
CONFIG_DEFAULT_HUNG_TASK_TIMEOUT which may be set to architecture
specific defaults as low as 10 seconds.
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://patch.msgid.link/20250417172448.1206107-2-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Russell King says:
====================
net: stmmac: socfpga: fix init ordering and cleanups
This series fixes the init ordering of the socfpga probe function.
The standard rule is to do all setup before publishing any device,
and socfpga violates that. I can see no reason for this, but these
patches have not been tested on hardware.
Address this by moving the initialisation of dwmac->stmmac_rst
along with all the other dwmac initialisers - there's no reason
for this to be late as plat_dat->stmmac_rst has already been
populated.
Next, replace the call to ops->set_phy_mode() with an init function
socfpga_dwmac_init() which will then be linked in to plat_dat->init.
Then, add this to plat_dat->init, and switch to stmmac_pltfr_pm_ops
from the private ops. The runtime suspend/resume socfpga implementations
are identical to the platform ones, but misses the noirq versions
which this will add.
Before we swap the order of socfpga_dwmac_init() and
stmmac_dvr_probe(), we need to change the way the interface is
obtained, as that uses driver data and the struct net_device which
haven't been initialised. Save a pointer to plat_dat in the socfpga
private data, and use that to get the interface mode. We can then swap
the order of the init and probe functions.
Finally, convert to devm_stmmac_pltfr_probe() by moving the call
to ops->set_phy_mode() into an init function appropriately populating
plat_dat->init.
====================
Link: https://patch.msgid.link/aAE2tKlImhwKySq_@shell.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Convert socfpga to use devm_stmmac_pltfr_probe() to further simplify
the probe function, wrapping the call to the set_phy_mode() method
into socfpga_dwmac_init() which can be called from the plat_dat->init()
method. Also call this from socfpga_dwmac_resume() thereby simplifying
that function.
Using the devm variant also means we can remove the call to
stmmac_pltfr_remove().
Unfortunately, we can't also convert to stmmac_pltfr_pm_ops as there is
extra work done in socfpga_dwmac_resume().
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/E1u5Sns-001IJw-OY@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Initialisation/setup after registration is a bug. This is the second
of two patches fixing this in socfpga.
The set_phy_mode() functions do various hardware setup that would
interfere with a netdev that has been published, and thus available to
be opened by the kernel/userspace.
However, set_phy_mode() relies upon the netdev having been initialised
to get at the plat_stmmacenet_data structure, which is probably why it
was placed after stmmac_drv_probe(). We can remove that need by storing
a pointer to struct plat_stmmacenet_data in struct socfpga_dwmac.
Move the call to set_phy_mode() before calling stmmac_dvr_probe().
This also simplifies the probe function as there is no need to
unregister the netdev if set_phy_mode() fails.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/E1u5Snn-001IJq-L0@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Convert socfpga to use the generic stmmac_pltfr_pm_ops, which can be
achieved by adding an appropriate plat_dat->init function to do the
setup.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/E1u5Sni-001IJk-Gi@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
