- 30 Jan, 2024 27 commits
-
-
Paolo Abeni authored
Petr Machata says: ==================== mlxsw: Refactor reference counting code Amit Cohen writes: This set converts all reference counters defined as 'unsigned int' to refcount_t type. The reference counting of LAGs can be simplified, so first refactor the related code and then change the type of the reference counter. Patch set overview: Patches #1-#4 are preparations for LAG refactor Patch #5 refactors LAG code and change the type of reference counter Patch #6 converts the remaining reference counters in mlxsw driver ==================== Link: https://lore.kernel.org/r/cover.1706293430.git.petrm@nvidia.comSigned-off-by:
Paolo Abeni <pabeni@redhat.com>
-
Amit Cohen authored
mlxsw driver uses 'unsigned int' for reference counters in several structures. Instead, use refcount_t type which allows us to catch overflow and underflow issues. Change the type of the counters and use the appropriate API. Signed-off-by:
Amit Cohen <amcohen@nvidia.com> Reviewed-by:
Ido Schimmel <idosch@nvidia.com> Signed-off-by:
Petr Machata <petrm@nvidia.com> Reviewed-by:
Simon Horman <horms@kernel.org> Signed-off-by:
-
Amit Cohen authored
mlxsw_sp stores an array of LAGs. When a port joins a LAG, in case that this LAG is already in use, we only have to increase the reference counter. Otherwise, we have to search for an unused LAG ID and configure it in hardware. When a port leaves a LAG, we have to destroy it only for the last user. This code can be simplified, for such requirements we usually add get() and put() functions which create and destroy the object. Add mlxsw_sp_lag_{get,put}() and use them. These functions take care of the reference counter and hardware configuration if needed. Change the reference counter to refcount_t type which catches overflow and underflow issues. Signed-off-by: -
Amit Cohen authored
Currently, the function mlxsw_sp_lag_index_get() is called twice - first as part of NETDEV_PRECHANGEUPPER event and later as part of NETDEV_CHANGEUPPER. This function will be changed in the next patch. To simplify the code, call it only once as part of NETDEV_CHANGEUPPER event and set an error message using 'extack' in case of failure. Signed-off-by:
-
Amit Cohen authored
The maximum number of LAGs is queried from core several times. It is used to allocate LAG array, and then to iterate over it. In addition, it is used for PGT initialization. To simplify the code, instead of querying it several times, store the value as part of 'mlxsw_sp' and use it. Signed-off-by:
-
Amit Cohen authored
A next patch will add mlxsw_sp_lag_{get,put}() functions to handle LAG reference counting and create/destroy it only for first user/last user. Remove mlxsw_sp_lag_get() function and access LAG array directly. Signed-off-by: -
Amit Cohen authored
The structure mlxsw_sp_upper is used only as LAG. Rename it to mlxsw_sp_lag and move it to spectrum.c file, as it is used only there. Move the function mlxsw_sp_lag_get() with the structure. Signed-off-by:
-
Petr Machata authored
The config file contains a partial kernel configuration to be used by `virtme-configkernel --custom'. The presumption is that the config file contains all Kconfig options needed by the selftests from the directory. In net/forwarding/config, many are missing, which manifests as spurious failures when running the selftests, with messages about unknown device types, qdisc kinds or classifier actions. Add the missing configurations. Tested the resulting configuration using virtme-ng as follows: # vng -b -f tools/testing/selftests/net/forwarding/config # vng --user root (within the VM:) # make -C tools/testing/selftests TARGETS=net/forwarding run_tests Signed-off-by:
-
Horatiu Vultur authored
When setting or getting PHC time, the higher bits of the second time (>32 bits) they were ignored. Meaning that setting some time in the future like year 2150, it was failing to set this. The issue can be reproduced like this: # phc_ctl /dev/ptp1 set 10000000000 phc_ctl[12.290]: set clock time to 10000000000.000000000 or Sat Nov 20 17:46:40 2286 # phc_ctl /dev/ptp1 get phc_ctl[15.309]: clock time is 1410065411.018055420 or Sun Sep 7 04:50:11 2014 Signed-off-by:
Horatiu Vultur <horatiu.vultur@microchip.com> Reviewed-by:
Maxime Chevallier <maxime.chevallier@bootlin.com> Reviewed-by:
Divya Koppera <divya.koppera@microchip.com> Reviewed-by:
Andrew Lunn <andrew@lunn.ch> Link: https://lore.kernel.org/r/20240126073042.1845153-1-horatiu.vultur@microchip.comSigned-off-by:
-
Stephen Hemminger authored
Use the inline function reciprocal_scale rather than open coding the scale optimization. Also, remove unnecessary initializations. Resulting compiled code is unchanged (according to godbolt). Signed-off-by:
Stephen Hemminger <stephen@networkplumber.org> Reviewed-by:
Willem de Bruijn <willemb@google.com> Acked-by:
Jason Wang <jasowang@redhat.com> Link: https://lore.kernel.org/r/20240126002550.169608-1-stephen@networkplumber.orgSigned-off-by:
-
Paolo Abeni authored
Tony Nguyen says: ==================== ice: fix timestamping in reset process Karol Kolacinski says: PTP reset process has multiple places where timestamping can end up in an incorrect state. This series introduces a proper state machine for PTP and refactors a large part of the code to ensure that timestamping does not break. The following are changes since commit 91374ba5: net: dsa: mt7530: support OF-based registration of switch MDIO bus and are available in the git repository at: git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue 100GbE ==================== Link: https://lore.kernel.org/r/20240125215757.2601799-1-anthony.l.nguyen@intel.comSigned-off-by:
-
Jacob Keller authored
The ice driver currently attempts to destroy and re-initialize the Tx timestamp tracker during the reset flow. The release of the Tx tracker only happened during CORE reset or GLOBAL reset. The ice_ptp_rebuild() function always calls the ice_ptp_init_tx function which will allocate a new tracker data structure, resulting in memory leaks during PF reset. Certainly the driver should not be allocating a new tracker without removing the old tracker data, as this results in a memory leak. Additionally, there's no reason to remove the tracker memory during a reset. Remove this logic from the reset and rebuild flow. Instead of releasing the Tx tracker, flush outstanding timestamps just before we reset the PHY timestamp block in ice_ptp_cfg_phy_interrupt(). Signed-off-by:
Jacob Keller <jacob.e.keller@intel.com> Signed-off-by:
Karol Kolacinski <karol.kolacinski@intel.com> Reviewed-by:
Tony Nguyen <anthony.l.nguyen@intel.com> Signed-off-by:
-
Jacob Keller authored
The ice_ptp_reset() function uses a goto to skip past clock owner operations if performing a PF reset or if the device is not the clock owner. This is a bit confusing. Factor this out into ice_ptp_rebuild_owner() instead. The ice_ptp_reset() function is called by ice_rebuild() to restore PTP functionality after a device reset. Follow the convention set by the ice_main.c file and rename this function to ice_ptp_rebuild(), in the same way that we have ice_prepare_for_reset() and ice_ptp_prepare_for_reset(). Signed-off-by:
-
Jacob Keller authored
The ice_ptp_tx_cfg_intr() function sends a control queue message to configure the PHY timestamp interrupt block. This is a very similar name to a function which is used to configure the MAC Other Interrupt Cause Enable register. Rename this function to ice_ptp_cfg_phy_interrupt in order to make it more obvious to the reader what action it performs, and distinguish it from other similarly named functions. Signed-off-by:
-
Jacob Keller authored
E810 hardware does not have a Tx timestamp ready bitmap. Don't check has_ready_bitmap in E810-specific functions. Add has_ready_bitmap check in ice_ptp_process_tx_tstamp() to stop relying on the fact that ice_get_phy_tx_tstamp_ready() returns all 1s. Signed-off-by:
-
Jacob Keller authored
The tx->verify_cached flag is used to inform the Tx timestamp tracking code whether it needs to verify the cached Tx timestamp value against a previous captured value. This is necessary on E810 hardware which does not have a Tx timestamp ready bitmap. In addition, we currently rely on the fact that the ice_get_phy_tx_tstamp_ready() function returns all 1s for E810 hardware. Instead of introducing a brand new flag, rename and verify_cached to has_ready_bitmap, inverting the relevant checks. Signed-off-by:
-
Jacob Keller authored
The ice_ptp_prepare_for_reset() and ice_ptp_reset() functions currently check the pf->flags ICE_FLAG_PFR_REQ bit to determine if the current reset is a PF reset or not. This is problematic, because it is possible that a PF reset and a higher level reset (CORE reset, GLOBAL reset, EMP reset) are requested simultaneously. In that case, the driver performs the highest level reset requested. However, the ICE_FLAG_PFR_REQ flag will still be set. The main driver reset functions take an enum ice_reset_req indicating which reset is actually being performed. Pass this data into the PTP functions and rely on this instead of relying on the driver flags. This ensures that the PTP code performs the proper level of reset that the driver is actually undergoing. Signed-off-by:
-
Jacob Keller authored
Add PTP state machine so that the driver can correctly identify PTP state around resets. When the driver got information about ungraceful reset, PTP was not prepared for reset and it returned error. When this situation occurs, prepare PTP before rebuilding its structures. Signed-off-by:
-
Menglong Dong authored
For now, the packet with an old ack is not accepted if we are in FIN_WAIT1 state, which can cause retransmission. Taking the following case as an example: Client Server | | FIN_WAIT1(Send FIN, seq=10) FIN_WAIT1(Send FIN, seq=20, ack=10) | | | Send ACK(seq=21, ack=11) Recv ACK(seq=21, ack=11) | Recv FIN(seq=20, ack=10) In the case above, simultaneous close is happening, and the FIN and ACK packet that send from the server is out of order. Then, the FIN will be dropped by the client, as it has an old ack. Then, the server has to retransmit the FIN, which can cause delay if the server has set the SO_LINGER on the socket. Old ack is accepted in the ESTABLISHED and TIME_WAIT state, and I think it should be better to keep the same logic. In this commit, we accept old ack in FIN_WAIT1/FIN_WAIT2/CLOSING/LAST_ACK states. Maybe we should limit it to FIN_WAIT1 for now? Signed-off-by:Menglong Dong <menglong8.dong@gmail.com> Reviewed-by:
Eric Dumazet <edumazet@google.com> Link: https://lore.kernel.org/r/20240126040519.1846345-1-menglong8.dong@gmail.comSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Jakub Kicinski authored
Arınç ÜNAL says: ==================== MT7530 DSA Subdriver Improvements Act I This patch series simplifies the MT7530 DSA subdriver and improves the logic of the support for MT7530, MT7531, and the switch on the MT7988 SoC. I have done a simple ping test to confirm basic communication on all switch ports on MCM and standalone MT7530, and MT7531 switch with this patch series applied. MT7621 Unielec, MCM MT7530: rgmii-only-gmac0-mt7621-unielec-u7621-06-16m.dtb gmac0-and-gmac1-mt7621-unielec-u7621-06-16m.dtb tftpboot 0x80008000 mips-uzImage.bin; tftpboot 0x83000000 mips-rootfs.cpio.uboot; tftpboot 0x83f00000 $dtb; bootm 0x80008000 0x83000000 0x83f00000 MT7622 Bananapi, MT7531: gmac0-and-gmac1-mt7622-bananapi-bpi-r64.dtb tftpboot 0x40000000 arm64-Image; tftpboot 0x45000000 arm64-rootfs.cpio.uboot; tftpboot 0x4a000000 $dtb; booti 0x40000000 0x45000000 0x4a000000 MT7623 Bananapi, standalone MT7530: rgmii-only-gmac0-mt7623n-bananapi-bpi-r2.dtb gmac0-and-gmac1-mt7623n-bananapi-bpi-r2.dtb tftpboot 0x80008000 arm-zImage; tftpboot 0x83000000 arm-rootfs.cpio.uboot; tftpboot 0x83f00000 $dtb; bootz 0x80008000 0x83000000 0x83f00000 This patch series is the continuation of the patch series linked below. https://lore.kernel.org/r/20230522121532.86610-1-arinc.unal@arinc9.com v2: https://lore.kernel.org/r/20231227044347.107291-1-arinc.unal@arinc9.com v1: https://lore.kernel.org/r/20231118123205.266819-1-arinc.unal@arinc9.comSigned-off-by:
Arınç ÜNAL <arinc.unal@arinc9.com> ==================== Link: https://lore.kernel.org/r/20240122-for-netnext-mt7530-improvements-1-v3-0-042401f2b279@arinc9.comSigned-off-by:
-
Arınç ÜNAL authored
There's no need to run all the code on mt7530_setup_port5() if port 5 is disabled. The only case for calling mt7530_setup_port5() from mt7530_setup() is when PHY muxing is enabled. That is because port 5 is not defined as a port on the devicetree, therefore, it cannot be controlled by phylink. Because of this, run mt7530_setup_port5() if priv->p5_intf_sel is P5_INTF_SEL_PHY_P0 or P5_INTF_SEL_PHY_P4. Remove the P5_DISABLED case from mt7530_setup_port5(). Stop initialising the interface variable as the remaining cases will always call mt7530_setup_port5() with it initialised. Signed-off-by:
Vladimir Oltean <olteanv@gmail.com> Link: https://lore.kernel.org/r/20240122-for-netnext-mt7530-improvements-1-v3-7-042401f2b279@arinc9.comSigned-off-by:
-
Arınç ÜNAL authored
Running mt7530_setup_port5() from mt7530_setup() used to handle all cases of configuring port 5, including phylink. Setting priv->p5_interface under mt7530_setup_port5() makes sure that mt7530_setup_port5() from mt753x_phylink_mac_config() won't run. The commit ("net: dsa: mt7530: improve code path for setting up port 5") makes so that mt7530_setup_port5() from mt7530_setup() runs only on non-phylink cases. Get rid of unnecessarily setting priv->p5_interface under mt7530_setup_port5() as port 5 phylink configuration will be done by running mt7530_setup_port5() from mt753x_phylink_mac_config() now. Signed-off-by: -
Arınç ÜNAL authored
There're two code paths for setting up port 5: mt7530_setup() -> mt7530_setup_port5() mt753x_phylink_mac_config() -> mt753x_mac_config() -> mt7530_mac_config() -> mt7530_setup_port5() Currently mt7530_setup_port5() from mt7530_setup() always runs. If port 5 is used as a CPU, DSA, or user port, mt7530_setup_port5() from mt753x_phylink_mac_config() won't run. That is because priv->p5_interface set on mt7530_setup_port5() will match state->interface on mt753x_phylink_mac_config() which will stop running mt7530_setup_port5() again. Therefore, mt7530_setup_port5() will never run from mt753x_phylink_mac_config(). Address this by not running mt7530_setup_port5() from mt7530_setup() if port 5 is used as a CPU, DSA, or user port. This driver isn't in the dsa_switches_apply_workarounds[] array so phylink will always be present. To keep the cases where port 5 isn't controlled by phylink working as before, preserve the mt7530_setup_port5() call from mt7530_setup(). Do not set priv->p5_intf_sel to P5_DISABLED. It is already set to that when "priv" is allocated. Move setting the interface to a more specific location. It's supposed to be overwritten if PHY muxing is detected. Improve the comment which explains the process. Signed-off-by: -
Arınç ÜNAL authored
There's no logic to numerically order the CPU ports. Just state the port number instead. Remove the irrelevant PHY muxing information from mt7530_mac_port_get_caps(). Explain the supported MII modes instead. Remove the out of place PHY muxing information from mt753x_phylink_mac_config(). The function is for MT7530, MT7531, and the switch on the MT7988 SoC but there's no PHY muxing on MT7531 or the switch on the MT7988 SoC. These comments were gradually introduced with the commits below. commit ca366d6c ("net: dsa: mt7530: Convert to PHYLINK API") commit 38f790a8 ("net: dsa: mt7530: Add support for port 5") commit 88bdef8b ("net: dsa: mt7530: Extend device data ready for adding a new hardware") commit c288575f ("net: dsa: mt7530: Add the support of MT7531 switch") Signed-off-by:
Daniel Golle <daniel@makrotopia.org> Reviewed-by:
-
Arınç ÜNAL authored
Introduce the p5_sgmii field to store the information for whether port 5 has got SGMII or not. Instead of reading the MT7531_TOP_SIG_SR register multiple times, the register will be read once and the value will be stored on the p5_sgmii field. This saves unnecessary reads of the register. Move the comment about MT7531AE and MT7531BE to mt7531_setup(), where the switch is identified. Get rid of mt7531_dual_sgmii_supported() now that priv->p5_sgmii stores the information. Address the code where mt7531_dual_sgmii_supported() is used. Get rid of mt7531_is_rgmii_port() which just prints the opposite of priv->p5_sgmii. Instead of calling mt7531_pll_setup() then returning, do not call it if port 5 is SGMII. Remove P5_INTF_SEL_GMAC5_SGMII. The p5_interface_select enum is supposed to represent the mode that port 5 is being used in, not the hardware information of port 5. Set p5_intf_sel to P5_INTF_SEL_GMAC5 instead, if port 5 is not dsa_is_unused_port(). Signed-off-by:
-
Arınç ÜNAL authored
Use the p5_interface_select enumeration as the data type for the p5_intf_sel field. This ensures p5_intf_sel can only take the values defined in the p5_interface_select enumeration. Remove the explicit assignment of 0 to P5_DISABLED as the first enum item is automatically assigned 0. Signed-off-by:
-
Arınç ÜNAL authored
On the MT7530 switch, the CPU_PORT field indicates which CPU port to trap frames to, regardless of the affinity of the inbound user port. When multiple CPU ports are in use, if the DSA conduit interface is down, trapped frames won't be passed to the conduit interface. To make trapping frames work including this case, implement ds->ops->conduit_state_change() on this subdriver and set the CPU_PORT field to the numerically smallest CPU port whose conduit interface is up. Introduce the active_cpu_ports field to store the information of the active CPU ports. Correct the macros, CPU_PORT is bits 4 through 6 of the register. Add a comment to explain frame trapping for this switch. Currently, the driver doesn't support the use of multiple CPU ports so this is not necessarily a bug fix. Suggested-by:
Russell King (Oracle) <linux@armlinux.org.uk> Signed-off-by:
-
- 29 Jan, 2024 13 commits
-
-
Jakub Kicinski authored
We had to add another synchronize_rcu() in recent fix. Bite the bullet and add an rcu_head to netdev_name_node, free from RCU. Note that name_node does not hold any reference on dev to which it points, but there must be a synchronize_rcu() on device removal path, so we should be fine. Signed-off-by:
Jiri Pirko <jiri@nvidia.com> Reviewed-by:
David S. Miller <davem@davemloft.net>
-
Tobias Schramm authored
The TRF7970A is a SPI device, not I2C. Signed-off-by:
Tobias Schramm <t.schramm@manjaro.org> Reviewed-by:
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Signed-off-by:
-
Min Li authored
The RENESAS FemtoClock3 Wireless is a high-performance jitter attenuator, frequency translator, and clock synthesizer. The device is comprised of 3 digital PLLs (DPLL) to track CLKIN inputs and three independent low phase noise fractional output dividers (FOD) that output low phase noise clocks. FemtoClock3 supports one Time Synchronization (Time Sync) channel to enable an external processor to control the phase and frequency of the Time Sync channel and to take phase measurements using the TDC. Intended applications are synchronization using the precision time protocol (PTP) and synchronization with 0.5 Hz and 1 Hz signals from GNSS. Signed-off-by:
Min Li <min.li.xe@renesas.com> Acked-by:
Lee Jones <lee@kernel.org> Signed-off-by:
-
Min Li authored
This change is for the PHC devices that can measure the phase offset between PHC signal and the external signal, such as the 1PPS signal of GNSS. Reporting PTP_CLOCK_EXTOFF to user space will be piggy-backed to the existing ptp_extts_event so that application such as ts2phc can poll the external offset the same way as extts. Hence, ts2phc can use the offset to achieve the alignment between PHC and the external signal by the help of either SW or HW filters. Signed-off-by:
Richard Cochran <richardcochran@gmail.com> Signed-off-by:
-
David S. Miller authored
Breno Leitao says: ==================== Fix MODULE_DESCRIPTION() for net (p3) There are hundreds of network modules that misses MODULE_DESCRIPTION(), causing a warning when compiling with W=1. Example: WARNING: modpost: missing MODULE_DESCRIPTION() in drivers/net/arcnet/com90io.o WARNING: modpost: missing MODULE_DESCRIPTION() in drivers/net/arcnet/arc-rimi.o WARNING: modpost: missing MODULE_DESCRIPTION() in drivers/net/arcnet/com20020.o This part3 of the patchset focus on the missing ethernet drivers, which is now warning free. This also fixes net/pcs and ieee802154. ==================== Signed-off-by: -
Breno Leitao authored
W=1 builds now warn if module is built without a MODULE_DESCRIPTION(). Add descriptions to arcnet module. Signed-off-by:
Breno Leitao <leitao@debian.org> Signed-off-by:
-
Breno Leitao authored
W=1 builds now warn if module is built without a MODULE_DESCRIPTION(). Add descriptions to ieee802154 modules. Signed-off-by:
Miquel Raynal <miquel.raynal@bootlin.com> Acked-by:
Stefan Schmidt <stefan@datenfreihafen.org> Signed-off-by:
-
Breno Leitao authored
W=1 builds now warn if module is built without a MODULE_DESCRIPTION(). Add descriptions to the Lynx, XPCS and LynxI PCS drivers. Signed-off-by:
-
Breno Leitao authored
W=1 builds now warn if module is built without a MODULE_DESCRIPTION(). Add descriptions to the Beckhoff CX5020 EtherCAT Ethernet driver. Signed-off-by:
-
Breno Leitao authored
W=1 builds now warn if module is built without a MODULE_DESCRIPTION(). Add descriptions to the TI CPSW switch module. Signed-off-by:
Roger Quadros <rogerq@kernel.org> Signed-off-by:
-
Breno Leitao authored
W=1 builds now warn if module is built without a MODULE_DESCRIPTION(). Add descriptions to the STMicro DWMAC for Altera SOCs. Signed-off-by:
-
Breno Leitao authored
W=1 builds now warn if module is built without a MODULE_DESCRIPTION(). Add descriptions to the Qualcom rmnet and emac drivers. Signed-off-by:
Subash Abhinov Kasiviswanathan <quic_subashab@quicinc.com> Signed-off-by:
-
Breno Leitao authored
W=1 builds now warn if module is built without a MODULE_DESCRIPTION(). Add descriptions to the SMSC 91x/911x/9420 Ethernet drivers. Signed-off-by:
-
