Hangbin Liu [Wed, 13 Dec 2023 06:08:44 +0000 (14:08 +0800)]
selftests/net: add variable NS_LIST for lib.sh
Add a global variable NS_LIST to store all the namespaces that setup_ns
created, so the caller could call cleanup_all_ns() instead of remember
all the netns names when using cleanup_ns().
Justin Chen [Wed, 13 Dec 2023 22:27:43 +0000 (14:27 -0800)]
net: mdio: mdio-bcm-unimac: Delay before first poll
With a clock interval of 400 nsec and a 64 bit transactions (32 bit
preamble & 16 bit control & 16 bit data), it is reasonable to assume
the mdio transaction will take 25.6 usec. Add a 30 usec delay before
the first poll to reduce the chance of a 1000-2000 usec sleep.
Reduce the timeout from 1000ms to 100ms as it is unlikely for the bus
to take this long.
====================
tools: ynl-gen: fill in the gaps in support of legacy families
Fill in the gaps in YNL C code gen so that we can generate user
space code for all genetlink families for which we have specs.
The two major changes we need are support for fixed headers and
support for recursive nests.
For fixed header support - place the struct for the fixed header
directly in the request struct (and don't bother generating access
helpers). The member of a fixed header can't be too complex, and
also are by definition not optional so the user has to fill them in.
The YNL core needs a bit of a tweak to understand that the attrs
may now start at a fixed offset, which is not necessarily equal
to sizeof(struct genlmsghdr).
Dealing with nested sets is much harder. Previously we'd gen
the nested structs as:
struct outer {
struct inner inner;
};
If structs are recursive (e.g. inner contains outer again)
we must break this chain and allocate one of the structs
dynamically (store a pointer rather than full struct).
====================
Jakub Kicinski [Wed, 13 Dec 2023 23:14:32 +0000 (15:14 -0800)]
tools: ynl-gen: print prototypes for recursive stuff
We avoid printing forward declarations and prototypes for most
types by sorting things topologically. But if structs nest we
do need the forward declarations, there's no other way.
Jakub Kicinski [Wed, 13 Dec 2023 23:14:31 +0000 (15:14 -0800)]
tools: ynl-gen: store recursive nests by a pointer
To avoid infinite nesting store recursive structs by pointer.
If recursive struct is placed in the op directly - the first
instance can be stored by value. That makes the code much
less of a pain for majority of practical uses.
Jakub Kicinski [Wed, 13 Dec 2023 23:14:30 +0000 (15:14 -0800)]
tools: ynl-gen: re-sort ignoring recursive nests
We try to keep the structures and helpers "topologically sorted",
to avoid forward declarations. When recursive nests are at play
we need to sort twice, because structs which end up being marked
as recursive will get a full set of forward declarations, so we
should ignore them for the purpose of sorting.
Jakub Kicinski [Wed, 13 Dec 2023 23:14:29 +0000 (15:14 -0800)]
tools: ynl-gen: record information about recursive nests
Track which nests are recursive. Non-recursive nesting gets
rendered in C as directly nested structs. For recursive
ones we need to put a pointer in, rather than full struct.
Track this information, no change to generated code, yet.
Jakub Kicinski [Wed, 13 Dec 2023 23:14:26 +0000 (15:14 -0800)]
tools: ynl-gen: use enum user type for members and args
Commit 30c902001534 ("tools: ynl-gen: use enum name from the spec")
added pre-cooked user type for enums. Use it to fix ignoring
enum-name provided in the spec.
This changes a type in struct ethtool_tunnel_udp_entry but is
generally inconsequential for current families.
drivers/net/ethernet/intel/iavf/iavf_ethtool.c 3a0b5a2929fd ("iavf: Introduce new state machines for flow director") 95260816b489 ("iavf: use iavf_schedule_aq_request() helper")
https://lore.kernel.org/all/84e12519-04dc-bd80-bc34-8cf50d7898ce@intel.com/
drivers/net/ethernet/broadcom/bnxt/bnxt.c c13e268c0768 ("bnxt_en: Fix HWTSTAMP_FILTER_ALL packet timestamp logic") c2f8063309da ("bnxt_en: Refactor RX VLAN acceleration logic.") a7445d69809f ("bnxt_en: Add support for new RX and TPA_START completion types for P7") 1c7fd6ee2fe4 ("bnxt_en: Rename some macros for the P5 chips")
https://lore.kernel.org/all/20231211110022.27926ad9@canb.auug.org.au/
drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c bd6781c18cb5 ("bnxt_en: Fix wrong return value check in bnxt_close_nic()") 84793a499578 ("bnxt_en: Skip nic close/open when configuring tstamp filters")
https://lore.kernel.org/all/20231214113041.3a0c003c@canb.auug.org.au/
drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c 3d7a3f2612d7 ("net/mlx5: Nack sync reset request when HotPlug is enabled") cecf44ea1a1f ("net/mlx5: Allow sync reset flow when BF MGT interface device is present")
https://lore.kernel.org/all/20231211110328.76c925af@canb.auug.org.au/
Linus Torvalds [Thu, 14 Dec 2023 21:11:49 +0000 (13:11 -0800)]
Merge tag 'net-6.7-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Pull networking fixes from Paolo Abeni:
"Current release - regressions:
- tcp: fix tcp_disordered_ack() vs usec TS resolution
Current release - new code bugs:
- dpll: sanitize possible null pointer dereference in
dpll_pin_parent_pin_set()
- eth: octeon_ep: initialise control mbox tasks before using APIs
Previous releases - regressions:
- io_uring/af_unix: disable sending io_uring over sockets
- eth: mlx5e:
- TC, don't offload post action rule if not supported
- fix possible deadlock on mlx5e_tx_timeout_work
- eth: iavf: fix iavf_shutdown to call iavf_remove instead iavf_close
- eth: bnxt_en: fix skb recycling logic in bnxt_deliver_skb()
- eth: ena: fix DMA syncing in XDP path when SWIOTLB is on
- eth: team: fix use-after-free when an option instance allocation
fails
Previous releases - always broken:
- neighbour: don't let neigh_forced_gc() disable preemption for long
- net: prevent mss overflow in skb_segment()
- ipv6: support reporting otherwise unknown prefix flags in
RTM_NEWPREFIX
- tcp: remove acked SYN flag from packet in the transmit queue
correctly
- eth: octeontx2-af:
- fix a use-after-free in rvu_nix_register_reporters
- fix promisc mcam entry action
- eth: dwmac-loongson: make sure MDIO is initialized before use
- eth: atlantic: fix double free in ring reinit logic"
* tag 'net-6.7-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (62 commits)
net: atlantic: fix double free in ring reinit logic
appletalk: Fix Use-After-Free in atalk_ioctl
net: stmmac: Handle disabled MDIO busses from devicetree
net: stmmac: dwmac-qcom-ethqos: Fix drops in 10M SGMII RX
dpaa2-switch: do not ask for MDB, VLAN and FDB replay
dpaa2-switch: fix size of the dma_unmap
net: prevent mss overflow in skb_segment()
vsock/virtio: Fix unsigned integer wrap around in virtio_transport_has_space()
Revert "tcp: disable tcp_autocorking for socket when TCP_NODELAY flag is set"
MIPS: dts: loongson: drop incorrect dwmac fallback compatible
stmmac: dwmac-loongson: drop useless check for compatible fallback
stmmac: dwmac-loongson: Make sure MDIO is initialized before use
tcp: disable tcp_autocorking for socket when TCP_NODELAY flag is set
dpll: sanitize possible null pointer dereference in dpll_pin_parent_pin_set()
net: ena: Fix XDP redirection error
net: ena: Fix DMA syncing in XDP path when SWIOTLB is on
net: ena: Fix xdp drops handling due to multibuf packets
net: ena: Destroy correct number of xdp queues upon failure
net: Remove acked SYN flag from packet in the transmit queue correctly
qed: Fix a potential use-after-free in qed_cxt_tables_alloc
...
Linus Torvalds [Thu, 14 Dec 2023 19:53:00 +0000 (11:53 -0800)]
Merge tag 'for-6.7-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
"Some fixes to quota accounting code, mostly around error handling and
correctness:
- free reserves on various error paths, after IO errors or
transaction abort
- don't clear reserved range at the folio release time, it'll be
properly cleared after final write
- fix integer overflow due to int used when passing around size of
freed reservations
- fix a regression in squota accounting that missed some cases with
delayed refs"
* tag 'for-6.7-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: ensure releasing squota reserve on head refs
btrfs: don't clear qgroup reserved bit in release_folio
btrfs: free qgroup pertrans reserve on transaction abort
btrfs: fix qgroup_free_reserved_data int overflow
btrfs: free qgroup reserve when ORDERED_IOERR is set
Stefan Eichenberger [Tue, 12 Dec 2023 14:12:00 +0000 (15:12 +0100)]
net: mvpp2: add support for mii
Currently, mvpp2 only supports RGMII. This commit adds support for MII.
The description in Marvell's functional specification seems to be wrong.
To enable MII, we need to set GENCONF_CTRL0_PORT3_RGMII, while for RGMII
we need to clear it. This is also how U-Boot handles it.
Signed-off-by: Stefan Eichenberger <eichest@gmail.com> Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com> Reviewed-by: Marcin Wojtas <mw@semihalf.com> Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Link: https://lore.kernel.org/r/20231212141200.62579-1-eichest@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Igor Russkikh [Wed, 13 Dec 2023 09:40:44 +0000 (10:40 +0100)]
net: atlantic: fix double free in ring reinit logic
Driver has a logic leak in ring data allocation/free,
where double free may happen in aq_ring_free if system is under
stress and driver init/deinit is happening.
The probability is higher to get this during suspend/resume cycle.
Verification was done simulating same conditions with
Hyunwoo Kim [Wed, 13 Dec 2023 04:10:56 +0000 (23:10 -0500)]
appletalk: Fix Use-After-Free in atalk_ioctl
Because atalk_ioctl() accesses sk->sk_receive_queue
without holding a sk->sk_receive_queue.lock, it can
cause a race with atalk_recvmsg().
A use-after-free for skb occurs with the following flow.
```
atalk_ioctl() -> skb_peek()
atalk_recvmsg() -> skb_recv_datagram() -> skb_free_datagram()
```
Add sk->sk_receive_queue.lock to atalk_ioctl() to fix this issue.
Sneh Shah [Tue, 12 Dec 2023 09:22:08 +0000 (14:52 +0530)]
net: stmmac: dwmac-qcom-ethqos: Fix drops in 10M SGMII RX
In 10M SGMII mode all the packets are being dropped due to wrong Rx clock.
SGMII 10MBPS mode needs RX clock divider programmed to avoid drops in Rx.
Update configure SGMII function with Rx clk divider programming.
Fixes: 463120c31c58 ("net: stmmac: dwmac-qcom-ethqos: add support for SGMII") Tested-by: Andrew Halaney <ahalaney@redhat.com> Signed-off-by: Sneh Shah <quic_snehshah@quicinc.com> Reviewed-by: Bjorn Andersson <quic_bjorande@quicinc.com> Link: https://lore.kernel.org/r/20231212092208.22393-1-quic_snehshah@quicinc.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Jakub Kicinski [Thu, 14 Dec 2023 06:08:58 +0000 (22:08 -0800)]
Merge branch '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2023-12-12 (igb, e1000e)
This series contains updates to igb and e1000e drivers.
Ilpo Järvinen does some cleanups to both drivers: utilizing FIELD_GET()
helpers and using standard kernel defines over driver created ones.
* '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
e1000e: Use pcie_capability_read_word() for reading LNKSTA
e1000e: Use PCI_EXP_LNKSTA_NLW & FIELD_GET() instead of custom defines/code
igb: Use FIELD_GET() to extract Link Width
====================
Jakub Kicinski [Thu, 14 Dec 2023 06:07:19 +0000 (22:07 -0800)]
Merge branch 'support-symmetric-xor-rss-hash'
Ahmed Zaki says:
====================
Support symmetric-xor RSS hash
Patches 1 and 2 modify the get/set_rxh ethtool API to take a pointer to
struct of parameters instead of individual params. This will allow future
changes to the uAPI-shared struct ethtool_rxfh without changing the
drivers' API.
Patch 3 adds the support at the Kernel level, allowing the user to set a
symmetric-xor RSS hash for a netdevice via:
# ethtool -X eth0 hfunc toeplitz symmetric-xor
and clears the flag via:
# ethtool -X eth0 hfunc toeplitz
The "symmetric-xor" is set in a new "input_xfrm" field in struct
ethtool_rxfh. Support for the new "symmetric-xor" flag will be later sent
to the "ethtool" user-space tool.
Patch 4 fixes a long standing bug with the ice hash function register
values. The bug has been benign for now since only (asymmetric) Toeplitz
hash (Zero) has been used.
Patches 5 and 6 lay some groundwork refactoring. While the first is
mainly cosmetic, the second is needed since there is no more room in the
previous 64-bit RSS profile ID for the symmetric attribute introduced in
the next patch.
Finally, patches 7 and 8 add the symmetric-xor support for the ice
(E800 PFs) and the iAVF drivers.
====================
Jeff Guo [Wed, 13 Dec 2023 00:33:20 +0000 (17:33 -0700)]
ice: enable symmetric-xor RSS for Toeplitz hash function
Allow the user to set the symmetric Toeplitz hash function via:
# ethtool -X eth0 hfunc toeplitz symmetric-xor
All existing RSS configurations will be converted to symmetric unless they
have a non-symmetric field (other than IP src/dst and L4 src/dst ports)
used for hashing. The driver will reject a new RSS configuration if such
a field is requested.
The hash function in the E800 NICs is set per-VSI and a specific AQ
command is needed to modify the hash function. Use the AQ command to
enable setting the symmetric Toeplitz RSS hash function for any VSI
in the new ice_set_rss_hfunc().
When the Symmetric Toeplitz hash function is used, the hardware sets the
input set of the RSS (Toeplitz) algorithm to be the XOR of the fields
index by HSYMM and the fields index by the INSET registers. We use this
to create a symmetric hash by setting the HSYMM registers to point to
their counterparts in the INSET registers:
HSYMM [src_fv] = dst_fv;
HSYMM [dst_fv] = src_fv;
where src_fv and dst_fv are the indexes of the protocol's src and dst
fields.
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com> Signed-off-by: Jeff Guo <jia.guo@intel.com> Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Co-developed-by: Ahmed Zaki <ahmed.zaki@intel.com> Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com> Link: https://lore.kernel.org/r/20231213003321.605376-8-ahmed.zaki@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Ahmed Zaki [Wed, 13 Dec 2023 00:33:19 +0000 (17:33 -0700)]
ice: refactor the FD and RSS flow ID generation
The flow director and RSS blocks use separate methods to generate a
unique 64 bit ID for the flow. This is not extendable, especially for
the RSS that already uses all 64 bit space.
Refactor the flow generation API so that the ID is generated within
ice_flow_add_prof(). The FD and RSS blocks caches the generated ID for
later use.
Qi Zhang [Wed, 13 Dec 2023 00:33:18 +0000 (17:33 -0700)]
ice: refactor RSS configuration
Refactor the driver to use a communication data structure for RSS
config. To do so we introduce the new ice_rss_hash_cfg struct, and then
pass it as an argument to several functions.
Also introduce enum ice_rss_cfg_hdr_type to specify a more granular and
flexible RSS configuration:
ICE_RSS_OUTER_HEADERS - take outer layer as RSS input set
ICE_RSS_INNER_HEADERS - take inner layer as RSS input set
ICE_RSS_INNER_HEADERS_W_OUTER_IPV4 - take inner layer as RSS input set for
packet with outer IPV4
ICE_RSS_INNER_HEADERS_W_OUTER_IPV6 - take inner layer as RSS input set for
packet with outer IPV6
ICE_RSS_ANY_HEADERS - try with outer first then inner (same as the
behaviour without this change)
Finally, move the virtchnl_rss_algorithm enum to be with the other RSS
related structures in the virtchnl.h file.
There should be no functional change due to this patch.
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com> Signed-off-by: Qi Zhang <qi.z.zhang@intel.com> Co-developed-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Co-developed-by: Ahmed Zaki <ahmed.zaki@intel.com> Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com> Link: https://lore.kernel.org/r/20231213003321.605376-6-ahmed.zaki@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Ahmed Zaki [Wed, 13 Dec 2023 00:33:17 +0000 (17:33 -0700)]
ice: fix ICE_AQ_VSI_Q_OPT_RSS_* register values
Fix the values of the ICE_AQ_VSI_Q_OPT_RSS_* registers. Shifting is
already done when the values are used, no need to double shift. Bug was
not discovered earlier since only ICE_AQ_VSI_Q_OPT_RSS_TPLZ (Zero) is
currently used.
Also, rename ICE_AQ_VSI_Q_OPT_RSS_XXX to ICE_AQ_VSI_Q_OPT_RSS_HASH_XXX
for consistency.
Co-developed-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com> Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com> Link: https://lore.kernel.org/r/20231213003321.605376-5-ahmed.zaki@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Ahmed Zaki [Wed, 13 Dec 2023 00:33:16 +0000 (17:33 -0700)]
net: ethtool: add support for symmetric-xor RSS hash
Symmetric RSS hash functions are beneficial in applications that monitor
both Tx and Rx packets of the same flow (IDS, software firewalls, ..etc).
Getting all traffic of the same flow on the same RX queue results in
higher CPU cache efficiency.
A NIC that supports "symmetric-xor" can achieve this RSS hash symmetry
by XORing the source and destination fields and pass the values to the
RSS hash algorithm.
The user may request RSS hash symmetry for a specific algorithm, via:
# ethtool -X eth0 hfunc <hash_alg> symmetric-xor
or turn symmetry off (asymmetric) by:
# ethtool -X eth0 hfunc <hash_alg>
The specific fields for each flow type should then be specified as usual
via:
# ethtool -N|-U eth0 rx-flow-hash <flow_type> s|d|f|n
Ahmed Zaki [Wed, 13 Dec 2023 00:33:15 +0000 (17:33 -0700)]
net: ethtool: get rid of get/set_rxfh_context functions
Add the RSS context parameters to struct ethtool_rxfh_param and use the
get/set_rxfh to handle the RSS contexts as well.
This is part 2/2 of the fix suggested in [1]:
- Add a rss_context member to the argument struct and a capability
like cap_link_lanes_supported to indicate whether driver supports
rss contexts, then you can remove *et_rxfh_context functions,
and instead call *et_rxfh() with a non-zero rss_context.
Link: https://lore.kernel.org/netdev/20231121152906.2dd5f487@kernel.org/ CC: Jesse Brandeburg <jesse.brandeburg@intel.com> CC: Tony Nguyen <anthony.l.nguyen@intel.com> CC: Marcin Wojtas <mw@semihalf.com> CC: Russell King <linux@armlinux.org.uk> CC: Sunil Goutham <sgoutham@marvell.com> CC: Geetha sowjanya <gakula@marvell.com> CC: Subbaraya Sundeep <sbhatta@marvell.com> CC: hariprasad <hkelam@marvell.com> CC: Saeed Mahameed <saeedm@nvidia.com> CC: Leon Romanovsky <leon@kernel.org> CC: Edward Cree <ecree.xilinx@gmail.com> CC: Martin Habets <habetsm.xilinx@gmail.com> Suggested-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com> Link: https://lore.kernel.org/r/20231213003321.605376-3-ahmed.zaki@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Ahmed Zaki [Wed, 13 Dec 2023 00:33:14 +0000 (17:33 -0700)]
net: ethtool: pass a pointer to parameters to get/set_rxfh ethtool ops
The get/set_rxfh ethtool ops currently takes the rxfh (RSS) parameters
as direct function arguments. This will force us to change the API (and
all drivers' functions) every time some new parameters are added.
This is part 1/2 of the fix, as suggested in [1]:
- First simplify the code by always providing a pointer to all params
(indir, key and func); the fact that some of them may be NULL seems
like a weird historic thing or a premature optimization.
It will simplify the drivers if all pointers are always present.
- Then make the functions take a dev pointer, and a pointer to a
single struct wrapping all arguments. The set_* should also take
an extack.
Jakub Kicinski [Thu, 14 Dec 2023 06:03:01 +0000 (22:03 -0800)]
Merge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2023-12-12 (iavf)
This series contains updates to iavf driver only.
Piotr reworks Flow Director states to deal with issues in restoring
filters.
Slawomir fixes shutdown processing as it was missing needed calls.
* '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
iavf: Fix iavf_shutdown to call iavf_remove instead iavf_close
iavf: Handle ntuple on/off based on new state machines for flow director
iavf: Introduce new state machines for flow director
====================
Jakub Kicinski [Thu, 14 Dec 2023 02:38:56 +0000 (18:38 -0800)]
Merge branch 'dpaa2-switch-various-fixes'
Ioana Ciornei says:
====================
dpaa2-switch: various fixes
The first patch fixes the size passed to two dma_unmap_single() calls
which was wrongly put as the size of the pointer.
The second patch is new to this series and reverts the behavior of the
dpaa2-switch driver to not ask for object replay upon offloading so that
we avoid the errors encountered when a VLAN is installed multiple times
on the same port.
====================
Ioana Ciornei [Tue, 12 Dec 2023 16:43:26 +0000 (18:43 +0200)]
dpaa2-switch: do not ask for MDB, VLAN and FDB replay
Starting with commit 4e51bf44a03a ("net: bridge: move the switchdev
object replay helpers to "push" mode") the switchdev_bridge_port_offload()
helper was extended with the intention to provide switchdev drivers easy
access to object addition and deletion replays. This works by calling
the replay helpers with non-NULL notifier blocks.
In the same commit, the dpaa2-switch driver was updated so that it
passes valid notifier blocks to the helper. At that moment, no
regression was identified through testing.
In the meantime, the blamed commit changed the behavior in terms of
which ports get hit by the replay. Before this commit, only the initial
port which identified itself as offloaded through
switchdev_bridge_port_offload() got a replay of all port objects and
FDBs. After this, the newly joining port will trigger a replay of
objects on all bridge ports and on the bridge itself.
This behavior leads to errors in dpaa2_switch_port_vlans_add() when a
VLAN gets installed on the same interface multiple times.
The intended mechanism to address this is to pass a non-NULL ctx to the
switchdev_bridge_port_offload() helper and then check it against the
port's private structure. But since the driver does not have any use for
the replayed port objects and FDBs until it gains support for LAG
offload, it's better to fix the issue by reverting the dpaa2-switch
driver to not ask for replay. The pointers will be added back when we
are prepared to ignore replays on unrelated ports.
Ioana Ciornei [Tue, 12 Dec 2023 16:43:25 +0000 (18:43 +0200)]
dpaa2-switch: fix size of the dma_unmap
The size of the DMA unmap was wrongly put as a sizeof of a pointer.
Change the value of the DMA unmap to be the actual macro used for the
allocation and the DMA map.
Liang Chen [Tue, 12 Dec 2023 04:46:11 +0000 (12:46 +0800)]
page_pool: transition to reference count management after page draining
To support multiple users referencing the same fragment,
'pp_frag_count' is renamed to 'pp_ref_count', transitioning pp pages
from fragment management to reference count management after draining
based on the suggestion from [1].
The idea is that the concept of fragmenting exists before the page is
drained, and all related functions retain their current names.
However, once the page is drained, its management shifts to being
governed by 'pp_ref_count'. Therefore, all functions associated with
that lifecycle stage of a pp page are renamed.
Kees Cook [Tue, 12 Dec 2023 22:09:57 +0000 (14:09 -0800)]
cxgb3: Avoid potential string truncation in desc
Builds with W=1 were warning about potential string truncations:
drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c: In function 'cxgb_up':
drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c:394:38: warning: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size between 5 and 20 [-Wformat-truncation=]
394 | "%s-%d", d->name, pi->first_qset + i);
| ^~
In function 'name_msix_vecs',
inlined from 'cxgb_up' at drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c:1264:3: drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c:394:34: note: directive argument in the range [-2147483641, 509]
394 | "%s-%d", d->name, pi->first_qset + i);
| ^~~~~~~
drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c:393:25: note: 'snprintf' output between 3 and 28 bytes into a destination of size 21
393 | snprintf(adap->msix_info[msi_idx].desc, n,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
394 | "%s-%d", d->name, pi->first_qset + i);
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Avoid open-coded %NUL-termination (this code was assuming snprintf
wasn't %NUL terminating when it does -- likely thinking of strncpy),
and grow the size of the string to handle a maximal value.
Kees Cook [Tue, 12 Dec 2023 22:13:12 +0000 (14:13 -0800)]
amd-xgbe: Avoid potential string truncation in name
Build with W=1 were warning about a potential string truncation:
drivers/net/ethernet/amd/xgbe/xgbe-drv.c: In function 'xgbe_alloc_channels':
drivers/net/ethernet/amd/xgbe/xgbe-drv.c:211:73: warning: '%u' directive output may be truncated writing between 1 and 10 bytes into a region of size 8 [-Wformat-truncation=]
211 | snprintf(channel->name, sizeof(channel->name), "channel-%u", i);
| ^~
drivers/net/ethernet/amd/xgbe/xgbe-drv.c:211:64: note: directive argument in the range [0, 4294967294]
211 | snprintf(channel->name, sizeof(channel->name), "channel-%u", i);
| ^~~~~~~~~~~~
drivers/net/ethernet/amd/xgbe/xgbe-drv.c:211:17: note: 'snprintf' output between 10 and 19 bytes into a destination of size 16
211 | snprintf(channel->name, sizeof(channel->name), "channel-%u", i);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Increase the size of the "name" buffer to handle the full format range.
Reported-by: kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202312100937.ZPZCARhB-lkp@intel.com/ Cc: Shyam Sundar S K <Shyam-sundar.S-k@amd.com> Signed-off-by: Kees Cook <keescook@chromium.org> Link: https://lore.kernel.org/r/20231212221312.work.830-kees@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Eric Dumazet [Tue, 12 Dec 2023 14:55:50 +0000 (14:55 +0000)]
sctp: support MSG_ERRQUEUE flag in recvmsg()
For some reason sctp_poll() generates EPOLLERR if sk->sk_error_queue
is not empty but recvmsg() can not drain the error queue yet.
This is needed to better support timestamping.
I had to export inet_recv_error(), since sctp
can be compiled as a module.
Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Cc: Willem de Bruijn <willemb@google.com> Acked-by: Xin Long <lucien.xin@gmail.com> Link: https://lore.kernel.org/r/20231212145550.3872051-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Fixes: 3953c46c3ac7 ("sk_buff: allow segmenting based on frag sizes") Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Link: https://lore.kernel.org/r/20231212164621.4131800-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
====================
idpf: add get/set for Ethtool's header split ringparam
Currently, the header split feature (putting headers in one smaller
buffer and then the data in a separate bigger one) is always enabled
in idpf when supported.
One may want to not have fragmented frames per each packet, for example,
to avoid XDP frags. To better optimize setups for particular workloads,
add ability to switch the header split state on and off via Ethtool's
ringparams, as well as to query the current status.
There's currently only GET in the Ethtool Netlink interface for now,
so add SET first. I suspect idpf is not the only one supporting this.
====================
Michal Kubiak [Tue, 12 Dec 2023 14:27:52 +0000 (15:27 +0100)]
idpf: add get/set for Ethtool's header split ringparam
idpf supports the header split feature and that feature is always
enabled by default.
However, for flexibility reasons and to simplify some scenarios, it
would be useful to have the support for switching the header split
off (and on) from the userspace.
Address that need by adding the user config parameter, the functions
for disabling (or enabling) the header split feature, and calls to
them from the Ethtool ringparam callbacks.
It still is enabled by default if supported by the hardware.
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com> Signed-off-by: Michal Kubiak <michal.kubiak@intel.com> Co-developed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://lore.kernel.org/r/20231212142752.935000-3-aleksander.lobakin@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Alexander Lobakin [Tue, 12 Dec 2023 14:27:51 +0000 (15:27 +0100)]
ethtool: add SET for TCP_DATA_SPLIT ringparam
Follow up commit 9690ae604290 ("ethtool: add header/data split
indication") and add the set part of Ethtool's header split, i.e.
ability to enable/disable header split via the Ethtool Netlink
interface. This might be helpful to optimize the setup for particular
workloads, for example, to avoid XDP frags, and so on.
A driver should advertise ``ETHTOOL_RING_USE_TCP_DATA_SPLIT`` in its
ops->supported_ring_params to allow doing that. "Unknown" passed from
the userspace when the header split is supported means the driver is
free to choose the preferred state.
Oleksij Rempel [Tue, 12 Dec 2023 05:41:44 +0000 (06:41 +0100)]
net: phy: Add support for the DP83TG720S Ethernet PHY
The DP83TG720S-Q1 device is an IEEE 802.3bp and Open Alliance compliant
automotive Ethernet physical layer transceiver.
This driver was tested with i.MX8MP EQOS (stmmac) on the MAC side and
same TI PHY on other side.
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Link: https://lore.kernel.org/r/20231212054144.87527-3-o.rempel@pengutronix.de Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Oleksij Rempel [Tue, 12 Dec 2023 05:41:43 +0000 (06:41 +0100)]
net: phy: c45: add genphy_c45_pma_read_ext_abilities() function
Move part of the genphy_c45_pma_read_abilities() code to a separate
function.
Some PHYs do not implement PMA/PMD status 2 register (Register 1.8) but
do implement PMA/PMD extended ability register (Register 1.11). To make
use of it, we need to be able to access this part of code separately.
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Link: https://lore.kernel.org/r/20231212054144.87527-2-o.rempel@pengutronix.de Signed-off-by: Jakub Kicinski <kuba@kernel.org>
====================
net/sched: optimizations around action binding and init
Scaling optimizations for action binding in rtnl-less filters.
We saw a noticeable lock contention around idrinfo->lock when
testing in a 56 core system, which disappeared after the patches.
====================
Pedro Tammela [Mon, 11 Dec 2023 18:18:07 +0000 (15:18 -0300)]
net/sched: act_api: skip idr replace on bound actions
tcf_idr_insert_many will replace the allocated -EBUSY pointer in
tcf_idr_check_alloc with the real action pointer, exposing it
to all operations. This operation is only needed when the action pointer
is created (ACT_P_CREATED). For actions which are bound to (returned 0),
the pointer already resides in the idr making such operation a nop.
Even though it's a nop, it's still not a cheap operation as internally
the idr code walks the idr and then does a replace on the appropriate slot.
So if the action was bound, better skip the idr replace entirely.
Pedro Tammela [Mon, 11 Dec 2023 18:18:06 +0000 (15:18 -0300)]
net/sched: act_api: rely on rcu in tcf_idr_check_alloc
Instead of relying only on the idrinfo->lock mutex for
bind/alloc logic, rely on a combination of rcu + mutex + atomics
to better scale the case where multiple rtnl-less filters are
binding to the same action object.
Action binding happens when an action index is specified explicitly and
an action exists which such index exists. Example:
tc actions add action drop index 1
tc filter add ... matchall action drop index 1
tc filter add ... matchall action drop index 1
tc filter add ... matchall action drop index 1
tc filter ls ...
filter protocol all pref 49150 matchall chain 0 filter protocol all pref 49150 matchall chain 0 handle 0x1
not_in_hw
action order 1: gact action drop
random type none pass val 0
index 1 ref 4 bind 3
filter protocol all pref 49151 matchall chain 0 filter protocol all pref 49151 matchall chain 0 handle 0x1
not_in_hw
action order 1: gact action drop
random type none pass val 0
index 1 ref 4 bind 3
filter protocol all pref 49152 matchall chain 0 filter protocol all pref 49152 matchall chain 0 handle 0x1
not_in_hw
action order 1: gact action drop
random type none pass val 0
index 1 ref 4 bind 3
When no index is specified, as before, grab the mutex and allocate
in the idr the next available id. In this version, as opposed to before,
it's simplified to store the -EBUSY pointer instead of the previous
alloc + replace combination.
When an index is specified, rely on rcu to find if there's an object in
such index. If there's none, fallback to the above, serializing on the
mutex and reserving the specified id. If there's one, it can be an -EBUSY
pointer, in which case we just try again until it's an action, or an action.
Given the rcu guarantees, the action found could be dead and therefore
we need to bump the refcount if it's not 0, handling the case it's
in fact 0.
As bind and the action refcount are already atomics, these increments can
happen without the mutex protection while many tcf_idr_check_alloc race
to bind to the same action instance.
In case binding encounters a parallel delete or add, it will return
-EAGAIN in order to try again. Both filter and action apis already
have the retry machinery in-place. In case it's an unlocked filter it
retries under the rtnl lock.
Yusong Gao [Wed, 13 Dec 2023 10:31:10 +0000 (10:31 +0000)]
sign-file: Fix incorrect return values check
There are some wrong return values check in sign-file when call OpenSSL
API. The ERR() check cond is wrong because of the program only check the
return value is < 0 which ignored the return val is 0. For example:
1. CMS_final() return 1 for success or 0 for failure.
2. i2d_CMS_bio_stream() returns 1 for success or 0 for failure.
3. i2d_TYPEbio() return 1 for success and 0 for failure.
4. BIO_free() return 1 for success and 0 for failure.
Linus Torvalds [Wed, 13 Dec 2023 19:09:58 +0000 (11:09 -0800)]
Merge tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull ufs fix from Al Viro:
"ufs got broken this merge window on folio conversion - calling
conventions for filemap_lock_folio() are not the same as for
find_lock_page()"
* tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
fix ufs_get_locked_folio() breakage
Linus Torvalds [Wed, 13 Dec 2023 18:54:50 +0000 (10:54 -0800)]
Merge tag 'efi-urgent-for-v6.7-2' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi
Pull EFI fixes from Ard Biesheuvel:
- Deal with a regression in the recently refactored x86 EFI stub code
on older Dell systems by disabling randomization of the physical load
address
- Use the correct load address for relocatable Loongarch kernels
* tag 'efi-urgent-for-v6.7-2' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi:
efi/x86: Avoid physical KASLR on older Dell systems
efi/loongarch: Use load address to calculate kernel entry address
====================
virtio-net: support dynamic coalescing moderation
Now, virtio-net already supports per-queue moderation parameter
setting. Based on this, we use the linux dimlib to support
dynamic coalescing moderation for virtio-net.
Due to some scheduling issues, we only support and test the rx dim.
Some test results:
I. Sockperf UDP
=================================================
1. Env
rxq_0 with affinity to cpu_0.
3. Result
dim off: 1143277.00 rxpps, throughput 17.844 MBps, cpu is 100%.
dim on: 1124161.00 rxpps, throughput 17.610 MBps, cpu is 83.5%.
=================================================
II. Redis
=================================================
1. Env
There are 8 rxqs, and rxq_i with affinity to cpu_i.
2. Result
When all cpus are 100%, ops/sec of memtier_benchmark client is
dim off: 978437.23
dim on: 1143638.28
=================================================
III. Nginx
=================================================
1. Env
There are 8 rxqs and rxq_i with affinity to cpu_i.
2. Result
When all cpus are 100%, requests/sec of wrk client is
dim off: 877931.67
dim on: 1019160.31
=================================================
IV. Latency of sockperf udp
=================================================
1. Rx cmd
taskset -c 0 sockperf sr -p 8989
After running this cmd 5 times and averaging the results,
3. Result
dim off: 17.7735 usec
dim on: 18.0110 usec
=================================================
Changelog:
v7->v8:
- Add select DIMLIB.
v6->v7:
- Drop the patch titled "spin lock for ctrl cmd access"
- Use rtnl_trylock to avoid the deadlock.
v5->v6:
- Add patch(4/5): spin lock for ctrl cmd access
- Patch(5/5):
- Use spin lock and cancel_work_sync to synchronize
v4->v5:
- Patch(4/4):
- Fix possible synchronization issues with cancel_work_sync.
- Reduce if/else nesting levels
v3->v4:
- Patch(5/5): drop.
v2->v3:
- Patch(4/5): some minor modifications.
v1->v2:
- Patch(2/5): a minor fix.
- Patch(4/5):
- improve the judgment of dim switch conditions.
- Cancel the work when vq reset.
- Patch(5/5): drop the tx dim implementation.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Heng Qi [Mon, 11 Dec 2023 10:36:07 +0000 (18:36 +0800)]
virtio-net: support rx netdim
By comparing the traffic information in the complete napi processes,
let the virtio-net driver automatically adjust the coalescing
moderation parameters of each receive queue.
Signed-off-by: Heng Qi <hengqi@linux.alibaba.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Heng Qi [Mon, 11 Dec 2023 10:36:04 +0000 (18:36 +0800)]
virtio-net: returns whether napi is complete
rx netdim needs to count the traffic during a complete napi process,
and start updating and comparing samples to make decisions after
the napi ends. Let virtqueue_napi_complete() return true if napi is done,
otherwise vice versa.
Signed-off-by: Heng Qi <hengqi@linux.alibaba.com> Acked-by: Jason Wang <jasowang@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Shannon Nelson [Mon, 11 Dec 2023 18:58:04 +0000 (10:58 -0800)]
ionic: fill out pci error handlers
Set up the pci_error_handlers error_detected and resume to be useful in
handling AER events. If the error detected is pci_channel_io_frozen we
set up to do an FLR at the end of the AER handling - this tends to clear
things up well enough that traffic can continue. Else, let the AER/PCI
machinery do what is needed for the less serious errors seen.
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Shannon Nelson [Mon, 11 Dec 2023 18:58:03 +0000 (10:58 -0800)]
ionic: lif debugfs refresh on reset
Remove and restore the lif's debugfs pointers on a reset,
and make sure to check for the dentry before removing it
in case an earlier reset failed to rebuild the lif.
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Shannon Nelson [Mon, 11 Dec 2023 18:58:02 +0000 (10:58 -0800)]
ionic: use timer_shutdown_sync
When stopping the watchdog timer at remove time we should
be using the new timer_shutdown_sync to assure the timer
doesn't ever get rearmed.
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Shannon Nelson [Mon, 11 Dec 2023 18:58:01 +0000 (10:58 -0800)]
ionic: no fw read when PCI reset failed
If there was a failed attempt to reset the PCI connection,
don't later try to read from PCI as the space is unmapped
and will cause a paging request crash. When clearing the PCI
setup we can clear the dev_info register pointer, and check
it before using it in the fw_running test.
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Shannon Nelson [Mon, 11 Dec 2023 18:58:00 +0000 (10:58 -0800)]
ionic: prevent pci disable of already disabled device
If a reset fails, the PCI device is left in a disabled
state, so don't try to disable it again on driver remove.
This prevents a scary looking WARN trace in the kernel log.
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Shannon Nelson [Mon, 11 Dec 2023 18:57:59 +0000 (10:57 -0800)]
ionic: bypass firmware cmds when stuck in reset
If the driver or firmware is stuck in reset state, don't bother
trying to use adminq commands. This speeds up shutdown and
prevents unnecessary timeouts and error messages.
This includes a bit of rework on ionic_adminq_post_wait()
and ionic_adminq_post_wait_nomsg() to both use
__ionic_adminq_post_wait() which can do the checks needed in
both cases.
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Shannon Nelson [Mon, 11 Dec 2023 18:57:58 +0000 (10:57 -0800)]
ionic: keep filters across FLR
Make sure we keep and replay the filters and RSS config across
an FLR by using our FW_RESET flag. This gets checked on the
way down and on the way back up to help determine how much LIF
state to keep and restore across a reset action.
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Shannon Nelson [Mon, 11 Dec 2023 18:57:57 +0000 (10:57 -0800)]
ionic: pass opcode to devcmd_wait
Don't rely on the PCI memory for the devcmd opcode because we
read a 0xff value if the PCI bus is broken, which can cause us
to report a bogus dev_cmd opcode later.
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Wed, 13 Dec 2023 10:57:01 +0000 (10:57 +0000)]
Merge branch 'stmmac-bug-fixes'
Yanteng Si says:
====================
stmmac: Some bug fixes
* Put Krzysztof's patch into my thread, pick Conor's Reviewed-by
tag and Jiaxun's Acked-by tag.(prev version is RFC patch)
* I fixed an Oops related to mdio, mainly to ensure that
mdio is initialized before use, because it will be used
in a series of patches I am working on.
see <https://lore.kernel.org/loongarch/cover.1699533745.git.siyanteng@loongson.cn/T/#t>
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Krzysztof Kozlowski [Mon, 11 Dec 2023 10:33:54 +0000 (18:33 +0800)]
MIPS: dts: loongson: drop incorrect dwmac fallback compatible
Device binds to proper PCI ID (LOONGSON, 0x7a03), already listed in DTS,
so checking for some other compatible does not make sense. It cannot be
bound to unsupported platform.
Drop useless, incorrect (space in between) and undocumented compatible.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Signed-off-by: Yanteng Si <siyanteng@loongson.cn> Reviewed-by: Conor Dooley <conor.dooley@microchip.com> Acked-by: Jiaxun Yang <jiaxun.yang@flygoat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Krzysztof Kozlowski [Mon, 11 Dec 2023 10:33:53 +0000 (18:33 +0800)]
stmmac: dwmac-loongson: drop useless check for compatible fallback
Device binds to proper PCI ID (LOONGSON, 0x7a03), already listed in DTS,
so checking for some other compatible does not make sense. It cannot be
bound to unsupported platform.
Drop useless, incorrect (space in between) and undocumented compatible.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Signed-off-by: Yanteng Si <siyanteng@loongson.cn> Reviewed-by: Conor Dooley <conor.dooley@microchip.com> Acked-by: Jiaxun Yang <jiaxun.yang@flygoat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Yanteng Si [Mon, 11 Dec 2023 10:33:11 +0000 (18:33 +0800)]
stmmac: dwmac-loongson: Make sure MDIO is initialized before use
Generic code will use mdio. If it is not initialized before use,
the kernel will Oops.
Fixes: 30bba69d7db4 ("stmmac: pci: Add dwmac support for Loongson") Signed-off-by: Yanteng Si <siyanteng@loongson.cn> Signed-off-by: Feiyang Chen <chenfeiyang@loongson.cn> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
Salvatore Dipietro [Fri, 8 Dec 2023 18:20:49 +0000 (10:20 -0800)]
tcp: disable tcp_autocorking for socket when TCP_NODELAY flag is set
Based on the tcp man page, if TCP_NODELAY is set, it disables Nagle's algorithm
and packets are sent as soon as possible. However in the `tcp_push` function
where autocorking is evaluated the `nonagle` value set by TCP_NODELAY is not
considered which can trigger unexpected corking of packets and induce delays.
For example, if two packets are generated as part of a server's reply, if the
first one is not transmitted on the wire quickly enough, the second packet can
trigger the autocorking in `tcp_push` and be delayed instead of sent as soon as
possible. It will either wait for additional packets to be coalesced or an ACK
from the client before transmitting the corked packet. This can interact badly
if the receiver has tcp delayed acks enabled, introducing 40ms extra delay in
completion times. It is not always possible to control who has delayed acks
set, but it is possible to adjust when and how autocorking is triggered.
Patch prevents autocorking if the TCP_NODELAY flag is set on the socket.
Patch has been tested using an AWS c7g.2xlarge instance with Ubuntu 22.04 and
Apache Tomcat 9.0.83 running the basic servlet below:
public class HelloWorldServlet extends HttpServlet {
@Override
protected void doGet(HttpServletRequest request, HttpServletResponse response)
throws ServletException, IOException {
response.setContentType("text/html;charset=utf-8");
OutputStreamWriter osw = new OutputStreamWriter(response.getOutputStream(),"UTF-8");
String s = "a".repeat(3096);
osw.write(s,0,s.length());
osw.flush();
}
}
Load was applied using wrk2 (https://github.com/kinvolk/wrk2) from an AWS
c6i.8xlarge instance. With the current auto-corking behavior and TCP_NODELAY
set an additional 40ms latency from P99.99+ values are observed. With the
patch applied we see no occurrences of 40ms latencies. The patch has also been
tested with iperf and uperf benchmarks and no regression was observed.
# No patch with tcp_autocorking=1 and TCP_NODELAY set on all sockets
./wrk -t32 -c128 -d40s --latency -R10000 http://172.31.49.177:8080/hello/hello'
...
50.000% 0.91ms
75.000% 1.12ms
90.000% 1.46ms
99.000% 1.73ms
99.900% 1.96ms
99.990% 43.62ms <<< 40+ ms extra latency
99.999% 48.32ms
100.000% 49.34ms
# With patch
./wrk -t32 -c128 -d40s --latency -R10000 http://172.31.49.177:8080/hello/hello'
...
50.000% 0.89ms
75.000% 1.13ms
90.000% 1.44ms
99.000% 1.67ms
99.900% 1.78ms
99.990% 2.27ms <<< no 40+ ms extra latency
99.999% 3.71ms
100.000% 4.57ms
Fixes: f54b311142a9 ("tcp: auto corking") Signed-off-by: Salvatore Dipietro <dipiets@amazon.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Wed, 13 Dec 2023 10:34:28 +0000 (10:34 +0000)]
Merge branch 'net-at803x-cleanups'
Christian Marangi says:
====================
net: phy: at803x: cleanup
The intention of this big series is to try to cleanup the big
at803x PHY driver.
It currently have 3 different family of PHY in it. at803x, qca83xx
and qca808x.
The current codebase required lots of cleanup and reworking to
make the split possible as currently there is a greater use of
adding special function matching the phy_id.
This has been reworked to make the function actually generic
and make the change only in more specific one. The result
is the addition of micro additional function but that is for good
as it massively simplify splitting the driver later.
Consider that this is all in preparation for the addition of
qca807x PHY driver that will also uso some of the functions of
at803x.
Subsequent series will come with the actual PHY split and other
required cleanup. This is only to start the process with minor
changes.
Changes v4:
- Improve at8031_probe function
Changes v3:
- Add Reviewed-by tag from Andrew
- Split patch 10 (at8031 rename) to rename and move
Changes v2:
- Drop split part due to series too big
- Split changes even more
- Fix problem pointed out by Russell (flawed reworked function logic)
- Add Reviewed-by tag from Andrew
- Minor rework to prevent further code duplication for cdt
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Christian Marangi [Fri, 8 Dec 2023 14:52:00 +0000 (15:52 +0100)]
net: phy: at803x: drop specific PHY ID check from cable test functions
Drop specific PHY ID check for cable test functions for at803x. This is
done to make functions more generic. While at it better describe what
the functions does by using more symbolic function names.
PHYs that requires to set additional reg are moved to specific function
calling the more generic one.
cdt_start and cdt_wait_for_completion are changed to take an additional
arg to pass specific values specific to the PHY.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Christian Marangi [Fri, 8 Dec 2023 14:51:55 +0000 (15:51 +0100)]
net: phy: at803x: move specific at8031 WOL bits to dedicated function
Move specific at8031 WOL enable/disable to dedicated function to make
at803x_set_wol more generic.
This is needed in preparation for PHY driver split as qca8081 share the
same function to toggle WOL settings.
In this new implementation WOL module in at8031 is enabled after the
generic interrupt is setup. This should not cause any problem as the
WOL_INT has a separate implementation and only relay on MAC bits.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Christian Marangi [Fri, 8 Dec 2023 14:51:50 +0000 (15:51 +0100)]
net: phy: at803x: raname hw_stats functions to qca83xx specific name
The function and the struct related to hw_stats were specific to qca83xx
PHY but were called following the convention in the driver of calling
everything with at803x prefix.
To better organize the code, rename these function a more specific name
to better describe that they are specific to 83xx PHY family.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
Jiri Pirko [Thu, 7 Dec 2023 15:12:04 +0000 (16:12 +0100)]
dpll: remove leftover mode_supported() op and use mode_get() instead
Mode supported is currently reported to the user exactly the same, as
the current mode. That's because mode changing is not implemented.
Remove the leftover mode_supported() op and use mode_get() to fill up
the supported mode exposed to user.
One, if even, mode changing is going to be introduced, this could be
very easily taken back. In the meantime, prevent drivers form
implementing this in wrong way (as for example recent netdevsim
implementation attempt intended to do).
Signed-off-by: Jiri Pirko <jiri@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
* tag 'hid-for-linus-2023121201' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid:
HID: apple: Add "hfd.cn" and "WKB603" to the list of non-apple keyboards
HID: lenovo: Restrict detection of patched firmware only to USB cptkbd
HID: Add quirk for Labtec/ODDOR/aikeec handbrake
HID: i2c-hid: Add IDEA5002 to i2c_hid_acpi_blacklist[]
mailmap: add address mapping for Jiri Kosina
Jiri Pirko [Mon, 11 Dec 2023 08:37:58 +0000 (09:37 +0100)]
dpll: sanitize possible null pointer dereference in dpll_pin_parent_pin_set()
User may not pass DPLL_A_PIN_STATE attribute in the pin set operation
message. Sanitize that by checking if the attr pointer is not null
and process the passed state attribute value only in that case.
Reported-by: Xingyuan Mo <hdthky0@gmail.com> Fixes: 9d71b54b65b1 ("dpll: netlink: Add DPLL framework base functions") Signed-off-by: Jiri Pirko <jiri@nvidia.com> Acked-by: Vadim Fedorenko <vadim.fedorenko@linux.dev> Link: https://lore.kernel.org/r/20231211083758.1082853-1-jiri@resnulli.us Signed-off-by: Jakub Kicinski <kuba@kernel.org>
David Arinzon [Mon, 11 Dec 2023 06:28:00 +0000 (06:28 +0000)]
net: ena: Fix DMA syncing in XDP path when SWIOTLB is on
This patch fixes two issues:
Issue 1
-------
Description
```````````
Current code does not call dma_sync_single_for_cpu() to sync data from
the device side memory to the CPU side memory before the XDP code path
uses the CPU side data.
This causes the XDP code path to read the unset garbage data in the CPU
side memory, resulting in incorrect handling of the packet by XDP.
Solution
````````
1. Add a call to dma_sync_single_for_cpu() before the XDP code starts to
use the data in the CPU side memory.
2. The XDP code verdict can be XDP_PASS, in which case there is a
fallback to the non-XDP code, which also calls
dma_sync_single_for_cpu().
To avoid calling dma_sync_single_for_cpu() twice:
2.1. Put the dma_sync_single_for_cpu() in the code in such a place where
it happens before XDP and non-XDP code.
2.2. Remove the calls to dma_sync_single_for_cpu() in the non-XDP code
for the first buffer only (rx_copybreak and non-rx_copybreak
cases), since the new call that was added covers these cases.
The call to dma_sync_single_for_cpu() for the second buffer and on
stays because only the first buffer is handled by the newly added
dma_sync_single_for_cpu(). And there is no need for special
handling of the second buffer and on for the XDP path since
currently the driver supports only single buffer packets.
Issue 2
-------
Description
```````````
In case the XDP code forwarded the packet (ENA_XDP_FORWARDED),
ena_unmap_rx_buff_attrs() is called with attrs set to 0.
This means that before unmapping the buffer, the internal function
dma_unmap_page_attrs() will also call dma_sync_single_for_cpu() on
the whole buffer (not only on the data part of it).
This sync is both wasteful (since a sync was already explicitly
called before) and also causes a bug, which will be explained
using the below diagram.
The following diagram shows the flow of events causing the bug.
The order of events is (1)-(4) as shown in the diagram.
CPU side memory area
(3)convert_to_xdp_frame() initializes the
headroom with xdpf metadata
||
\/
___________________________________
| |
0 | V 4K
---------------------------------------------------------------------
| xdpf->data | other xdpf | < data > | tailroom ||...|
| | fields | | GARBAGE || |
---------------------------------------------------------------------
/\ /\
|| ||
(4)ena_unmap_rx_buff_attrs() calls (2)dma_sync_single_for_cpu()
dma_sync_single_for_cpu() on the copies data from device
whole buffer page, overwriting side to CPU side memory
the xdpf->data with GARBAGE. ||
0 4K
---------------------------------------------------------------------
| headroom | < data > | tailroom ||...|
| GARBAGE | | GARBAGE || |
---------------------------------------------------------------------
Device side memory area /\
||
(1) device writes RX packet data
After the call to ena_unmap_rx_buff_attrs() in (4), the xdpf->data
becomes corrupted, and so when it is later accessed in
ena_clean_xdp_irq()->xdp_return_frame(), it causes a page fault,
crashing the kernel.
Solution
````````
Explicitly tell ena_unmap_rx_buff_attrs() not to call
dma_sync_single_for_cpu() by passing it the ENA_DMA_ATTR_SKIP_CPU_SYNC
flag.
Fixes: f7d625adeb7b ("net: ena: Add dynamic recycling mechanism for rx buffers") Signed-off-by: Arthur Kiyanovski <akiyano@amazon.com> Signed-off-by: David Arinzon <darinzon@amazon.com> Link: https://lore.kernel.org/r/20231211062801.27891-4-darinzon@amazon.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
projects / users / jedix / linux-maple.git / log