- 13 Aug, 2023 3 commits
-
-
Menglong Dong authored
In tcp_retransmit_timer(), a window shrunk connection will be regarded as timeout if 'tcp_jiffies32 - tp->rcv_tstamp > TCP_RTO_MAX'. This is not right all the time. The retransmits will become zero-window probes in tcp_retransmit_timer() if the 'snd_wnd==0'. Therefore, the icsk->icsk_rto will come up to TCP_RTO_MAX sooner or later. However, the timer can be delayed and be triggered after 122877ms, not TCP_RTO_MAX, as I tested. Therefore, 'tcp_jiffies32 - tp->rcv_tstamp > TCP_RTO_MAX' is always true once the RTO come up to TCP_RTO_MAX, and the socket will die. Fix this by replacing the 'tcp_jiffies32' with '(u32)icsk->icsk_timeout', which is exact the timestamp of the timeout. However, "tp->rcv_tstamp" can restart from idle, then tp->rcv_tstamp could already be a long time (minutes or hours) in the past even on the first RTO. So we double check the timeout with the duration of the retransmission. Meanwhile, making "2 * TCP_RTO_MAX" as the timeout to avoid the socket dying too soon. Fixes: 1da177e4 ("Linux-2.6.12-rc2") Link: https://lore.kernel.org/netdev/CADxym3YyMiO+zMD4zj03YPM3FBi-1LHi6gSD2XT8pyAMM096pg@mail.gmail.com/Signed-off-by:
Menglong Dong <imagedong@tencent.com> Reviewed-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Menglong Dong authored
Fow now, an ACK can update the window in following case, according to the tcp_may_update_window(): 1. the ACK acknowledged new data 2. the ACK has new data 3. the ACK expand the window and the seq of it is valid Now, we allow the ACK update the window if the window is 0, and the seq/ack of it is valid. This is for the case that the receiver replies an zero-window ACK when it is under memory stress and can't queue the new data. Signed-off-by:
-
Menglong Dong authored
For now, skb will be dropped when no memory, which makes client keep retrans util timeout and it's not friendly to the users. In this patch, we reply an ACK with zero-window in this case to update the snd_wnd of the sender to 0. Therefore, the sender won't timeout the connection and will probe the zero-window with the retransmits. Signed-off-by:
-
- 12 Aug, 2023 3 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queueJakub Kicinski authored
Tony Nguyen says: ==================== i40e: Replace one-element arrays with flexible-array members Replace one-element arrays with flexible-array members in multiple structures. This results in no differences in binary output. * '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue: i40e: Replace one-element array with flex-array member in struct i40e_profile_aq_section i40e: Replace one-element array with flex-array member in struct i40e_section_table i40e: Replace one-element array with flex-array member in struct i40e_profile_segment i40e: Replace one-element array with flex-array member in struct i40e_package_header ==================== Link: https://lore.kernel.org/r/20230810175302.1964182-1-anthony.l.nguyen@intel.comSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Arnd Bergmann authored
The init function is only referenced locally, so it should be static to avoid this warning: drivers/net/ethernet/amd/atarilance.c:370:28: error: no previous prototype for 'atarilance_probe' [-Werror=missing-prototypes] Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Reviewed-by:
Yang Yingliang <yangyingliang@huawei.com> Link: https://lore.kernel.org/r/20230810122528.1220434-2-arnd@kernel.orgSigned-off-by:
-
Arnd Bergmann authored
The function is exported for no reason and should just be static: drivers/net/ethernet/sun/ldmvsw.c:127:5: error: no previous prototype for 'ldmvsw_open' [-Werror=missing-prototypes] Signed-off-by:
Simon Horman <horms@kernel.org> Tested-by: Simon Horman <horms@kernel.org> # build-tested Link: https://lore.kernel.org/r/20230810122528.1220434-1-arnd@kernel.orgSigned-off-by:
-
- 11 Aug, 2023 33 commits
-
-
David S. Miller authored
Alexis Lothoré says: ==================== net: dsa: rzn1-a5psw: add support for vlan and .port_bridge_flags this series enables vlan support in Renesas RZN1 internal ethernet switch, and is a follow up to the work initiated by Clement Leger a few months ago, who handed me over the topic. This new revision aims to iron the last few points raised by Vladimir to ensure that the driver is in line with switch drivers expectations, and is based on the lengthy discussion in [1] (thanks Vladimir for the valuable explanations) [1] https://lore.kernel.org/netdev/20230314163651.242259-1-clement.leger@bootlin.com/ ---- V5: - ensure that flooding can be enabled only if port belongs to a bridge - enable learning in a5psw_port_stp_state_set() only if port has learning enabled - toggle vlan tagging on vlan filtering in - removed reviewed-by on second patch since its modified RESEND V4: - Resent due to net-next being closed V4: - Fix missing CPU port bit in a5psw->bridged_ports - Use unsigned int for vlan_res_id parameters - Rename a5psw_get_vlan_res_entry() to a5psw_new_vlan_res_entry() - In a5psw_port_vlan_add(), return -ENOSPC when no VLAN entry is found - In a5psw_port_vlan_filtering(), compute "val" from "mask" V3: - Target net-next tree and correct version... V2: - Fixed a few formatting errors - Add .port_bridge_flags implementation ==================== Reviewed-by:
-
Clément Léger authored
Add support for vlan operation (add, del, filtering) on the RZN1 driver. The a5psw switch supports up to 32 VLAN IDs with filtering, tagged/untagged VLANs and PVID for each ports. Signed-off-by:
Clément Léger <clement.leger@bootlin.com> Signed-off-by:
Alexis Lothoré <alexis.lothore@bootlin.com> Signed-off-by:
-
Clément Léger authored
When running vlan test (bridge_vlan_aware/unaware.sh), there were some failure due to the lack .port_bridge_flag function to disable port flooding. Implement this operation for BR_LEARNING, BR_FLOOD, BR_MCAST_FLOOD and BR_BCAST_FLOOD. Since .port_bridge_flags affects the bits disabling learning for a port, ensure that any other modification on the same register done by a5psw_port_stp_state_set is in sync by using the port learning state to enable/disable learning on the port. Signed-off-by:
-
Clément Léger authored
.port_bridge_flags will be added and allows to modify the flood mask independently for each port. Keeping the existing bridged_ports write in a5psw_flooding_set_resolution() would potentially messed up this. Use a read-modify-write to set that value and move bridged_ports handling in bridge_port_join/leave. Signed-off-by:
Florian Fainelli <f.fainelli@gmail.com> Signed-off-by:
-
Hariprasad Kelam authored
The current implementation does not allow the user to enable both hw-tc-offload and ntuple features on the interface. These checks are added as TC flower offload and ntuple features internally configures the same hardware resource MCAM. But TC HTB offload configures the transmit scheduler which can be safely enabled on the interface with ntuple feature. This patch adds the same and ensures only TC flower offload and ntuple features are mutually exclusive. Signed-off-by:
Hariprasad Kelam <hkelam@marvell.com> Signed-off-by:
Sunil Goutham <sgoutham@marvell.com> Signed-off-by:
-
David S. Miller authored
Zhengchao Shao says: ==================== bonding: do some cleanups in bond driver Do some cleanups in bond driver. --- v2: use IS_ERR instead of NULL check in patch 2/5, update commit information in patch 3/5, remove inline modifier in patch 4/5 ==================== Reviewed-by:Hangbin Liu <liuhangbin@gmail.com> Signed-off-by:
-
Zhengchao Shao authored
The free_percpu function also could check whether "rr_tx_counter" parameter is NULL. Therefore, remove NULL check in bond_destructor. Signed-off-by:
Zhengchao Shao <shaozhengchao@huawei.com> Signed-off-by:
-
Zhengchao Shao authored
In bond_reset_slave_arr(), values are assigned and memory is released only when the variables "usable" and "all" are not NULL. But even if the "usable" and "all" variables are NULL, they can still work, because value will be checked in kfree_rcu. Therefore, use bond_set_slave_arr() and set the input parameters "usable_slaves" and "all_slaves" to NULL to simplify the code in bond_reset_slave_arr(). And the same to bond_uninit(). Signed-off-by:
-
Zhengchao Shao authored
Because debugfs_create_dir returns ERR_PTR, so bonding_debug_root will never be NULL. Remove redundant NULL check for bonding_debug_root in debugfs function. The later debugfs_create_dir/debugfs_remove_recursive /debugfs_remove_recursive functions will check the dentry with IS_ERR(). Signed-off-by:
-
Zhengchao Shao authored
Because debugfs_create_dir returns ERR_PTR, so IS_ERR should be used to check whether the directory is successfully created. Signed-off-by:
-
Zhengchao Shao authored
Some functions are only used in initialization and exit functions, so add the __init/__net_init and __net_exit modifiers to these functions. Signed-off-by:
-
Krzysztof Kozlowski authored
'type' is an enum, thus cast of pointer on 64-bit compile test with W=1 causes: mvmdio.c:272:9: error: cast to smaller integer type 'enum orion_mdio_bus_type' from 'const void *' [-Werror,-Wvoid-pointer-to-enum-cast] Signed-off-by:
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Reviewed-by:
-
Krzysztof Kozlowski authored
'enet_id' is an enum, thus cast of pointer on 64-bit compile test with W=1 causes: xgene_enet_main.c:2044:20: error: cast to smaller integer type 'enum xgene_enet_id' from 'const void *' [-Werror,-Wvoid-pointer-to-enum-cast] Signed-off-by:
-
Shradha Gupta authored
Extended performance counter stats in 'ethtool -S <interface>' for MANA VF to include GDMA tx LSO packets and bytes count. Tested-on: Ubuntu22 Testcases: 1. LISA testcase: PERF-NETWORK-TCP-THROUGHPUT-MULTICONNECTION-NTTTCP-Synthetic 2. LISA testcase: PERF-NETWORK-TCP-THROUGHPUT-MULTICONNECTION-NTTTCP-SRIOV 3. Validated the GDMA stat packets and byte counters Signed-off-by:
Shradha Gupta <shradhagupta@linux.microsoft.com> Reviewed-by:
Pavan Chebbi <pavan.chebbi@broadcom.com> Signed-off-by:
-
Sathesh Edara authored
Implement control plane mailbox versions for host and firmware. Versions are published in info area of control mailbox bar4 memory structure.Firmware will publish minimum and maximum supported versions.Control plane mailbox apis will check for firmware version before sending any control commands to firmware. Notifications from firmware will similarly be checked for host version compatibility. Signed-off-by:
Sathesh Edara <sedara@marvell.com> Signed-off-by:
-
Ratheesh Kannoth authored
Accept TC offload classifier rule only if SPI field can be extracted by HW. Signed-off-by:
Ratheesh Kannoth <rkannoth@marvell.com> Reviewed-by:
-
Sergei Antonov authored
If netdev_mc_count() is not zero and not IFF_ALLMULTI, filter incoming multicast packets. The chip has a Multicast Address Hash Table for allowed multicast addresses, so we fill it. Implement .ndo_set_rx_mode and recalculate multicast hash table. Also observe change of IFF_PROMISC and IFF_ALLMULTI netdev flags. Signed-off-by:
Sergei Antonov <saproj@gmail.com> Reviewed-by:
Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by:
-
Tahsin Erdogan authored
When gso.hdr_len is zero and a packet is transmitted via write() or writev(), all payload is treated as header which requires a contiguous memory allocation. This allocation request is harder to satisfy, and may even fail if there is enough fragmentation. Note that sendmsg() code path limits the linear copy length, so this change makes write()/writev() and sendmsg() paths more consistent. Signed-off-by:
Tahsin Erdogan <trdgn@amazon.com> Acked-by:
Jason Wang <jasowang@redhat.com> Reviewed-by:
Willem de Bruijn <willemb@google.com> Link: https://lore.kernel.org/r/20230809164753.2247594-1-trdgn@amazon.comSigned-off-by:
-
Yang Yingliang authored
The driver init/exit() function don't do anything special, it can use the module_pci_driver() macro to eliminate boilerplate code. Signed-off-by:
-
Yue Haibing authored
Commit 61c9fed4 ("[SCTP]: A better solution to fix the race between sctp_peeloff() and sctp_rcv().") removed the implementation but left declaration in place. Remove it. Signed-off-by:
Yue Haibing <yuehaibing@huawei.com> Acked-by:
Xin Long <lucien.xin@gmail.com> Link: https://lore.kernel.org/r/20230809142323.9428-1-yuehaibing@huawei.comSigned-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/netdev/netJakub Kicinski authored
Merge net again, after pulling in x86/bugs fixes to clang linking errors. Signed-off-by: -
Yue Haibing authored
Commit 43e36921 ("caif: Move refcount from service layer to sock and dev.") declared but never implemented this. Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tipJakub Kicinski authored
Cross merge x86 fixes to fix clang linking errors: ld.lld: error: ./arch/x86/kernel/vmlinux.lds:221: at least one side of the expression must be absolute These will hopefully be downstream by the time we ship the next batch of fixes. * 'x86/bugs' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: x86: Move gds_ucode_mitigated() declaration to header x86/speculation: Add cpu_show_gds() prototype driver core: cpu: Make cpu_show_not_affected() static x86/srso: Fix build breakage with the LLVM linker Documentation/srso: Document IBPB aspect and fix formatting driver core: cpu: Unify redundant silly stubs Documentation/hw-vuln: Unify filename specification in index Link: https://lore.kernel.org/all/CAHk-=wj_b+FGTnevQSBAtCWuhCk=0oQ_THvthBW2hzqpOTLFmg@mail.gmail.com/Signed-off-by:
-
Furong Xu authored
Commit abe80fdc ("net: stmmac: RX queue routing configuration") introduced RX queue routing to DWMAC4 core. This patch extend the support to XGMAC2 core. Signed-off-by:
Furong Xu <0x1207@gmail.com> Reviewed-by:
-
Jakub Kicinski authored
Andrew Lunn says: ==================== Support offload LED blinking to PHY. Allow offloading of the LED trigger netdev to PHY drivers and implement it for the Marvell PHY driver. Additionally, correct the handling of when the initial state of the LED cannot be represented by the trigger, and so an error is returned. As with ledtrig-timer, disable offload when the trigger is deactivate, or replaced by another trigger. ==================== Link: https://lore.kernel.org/r/20230808210436.838995-1-andrew@lunn.chSigned-off-by:
-
Andrew Lunn authored
Ensure that the offloading of blinking is stopped when the trigger is deactivated. Calling led_set_brightness() is documented as stopping offload and setting the LED to a constant brightness. Suggested-by:
Daniel Golle <daniel@makrotopia.org> Signed-off-by:
Andrew Lunn <andrew@lunn.ch> Reviewed-by:
Simon Horman <simon.horman@corigine.com> Tested-by:
-
Andrew Lunn authored
Add the code needed to indicate if a given blinking pattern can be offloaded, to offload a pattern and to try to return the current pattern. Reviewed-by:
-
Andrew Lunn authored
Linux LEDs can be requested to perform hardware accelerated blinking to indicate link, RX, TX etc. Pass the rules for blinking to the PHY driver, if it implements the ops needed to determine if a given pattern can be offloaded, to offload it, and what the current offload is. Additionally implement the op needed to get what device the LED is for. Reviewed-by:
-
Andrew Lunn authored
When the netdev trigger is activates, it tries to determine what device the LED blinks for, and what the current blink mode is. The documentation for hw_control_get() says: * Return 0 on success, a negative error number on failing parsing the * initial mode. Error from this function is NOT FATAL as the device * may be in a not supported initial state by the attached LED trigger. */ For the Marvell PHY and the Armada 370-rd board, the initial LED blink mode is not supported by the trigger, so it returns an error. This resulted in not getting the device the LED is blinking for. As a result, the device is unknown and offloaded is never performed. Change to condition to always get the device if offloading is supported, and reduce the scope of testing for an error from hw_control_get() to skip setting trigger internal state if there is an error. Reviewed-by:
-
Li Zetao authored
The module_mhi_driver() will set "THIS_MODULE" to driver.owner when register a mhi_driver driver, so it is redundant initialization to set driver.owner in the statement. Remove it for clean code. Signed-off-by:
Li Zetao <lizetao1@huawei.com> Reviewed-by:
-
Jakub Kicinski authored
Shenwei Wang says: ==================== update stmmac fix_mac_speed ==================== Link: https://lore.kernel.org/r/20230807160716.259072-1-shenwei.wang@nxp.comSigned-off-by:
-
Shenwei Wang authored
When using a fixed-link setup, certain devices like the SJA1105 require a small pause in the TXC clock line to enable their internal tunable delay line (TDL). To satisfy this requirement, this patch temporarily disables the TX clock, and restarts it after a required period. This provides the required silent interval on the clock line for SJA1105 to complete the frequency transition and enable the internal TDLs. This action occurs before the link is built up, so it does not impact a normal device too. There is no need to identify if the connected device is an SJA1105 alike or not during the implementation. So far we have only enabled this feature on the i.MX93 platform. Signed-off-by:
Shenwei Wang <shenwei.wang@nxp.com> Reviewed-by:
Frank Li <frank.li@nxp.com> Link: https://lore.kernel.org/r/20230807160716.259072-3-shenwei.wang@nxp.comSigned-off-by:
-
Shenwei Wang authored
A mode parameter has been added to the callback function of fix_mac_speed to indicate the physical layer type. The mode can be one the following: MLO_AN_PHY - Conventional PHY MLO_AN_FIXED - Fixed-link mode MLO_AN_INBAND - In-band protocol Signed-off-by:
-
- 10 Aug, 2023 1 commit
-
-
https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-nextJakub Kicinski authored
Martin KaFai Lau says: ==================== pull-request: bpf-next 2023-08-09 We've added 19 non-merge commits during the last 6 day(s) which contain a total of 25 files changed, 369 insertions(+), 141 deletions(-). The main changes are: 1) Fix array-index-out-of-bounds access when detaching from an already empty mprog entry from Daniel Borkmann. 2) Adjust bpf selftest because of a recent llvm change related to the cpu-v4 ISA from Eduard Zingerman. 3) Add uprobe support for the bpf_get_func_ip helper from Jiri Olsa. 4) Fix a KASAN splat due to the kernel incorrectly accepted an invalid program using the recent cpu-v4 instruction from Yonghong Song. * tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: bpf: btf: Remove two unused function declarations bpf: lru: Remove unused declaration bpf_lru_promote() selftests/bpf: relax expected log messages to allow emitting BPF_ST selftests/bpf: remove duplicated functions bpf, docs: Fix small typo and define semantics of sign extension selftests/bpf: Add bpf_get_func_ip test for uprobe inside function selftests/bpf: Add bpf_get_func_ip tests for uprobe on function entry bpf: Add support for bpf_get_func_ip helper for uprobe program selftests/bpf: Add a movsx selftest for sign-extension of R10 bpf: Fix an incorrect verification success with movsx insn bpf, docs: Formalize type notation and function semantics in ISA standard bpf: change bpf_alu_sign_string and bpf_movsx_string to static libbpf: Use local includes inside the library bpf: fix bpf_dynptr_slice() to stop return an ERR_PTR. bpf: fix inconsistent return types of bpf_xdp_copy_buf(). selftests/bpf: fix the incorrect verification of port numbers. selftests/bpf: Add test for detachment on empty mprog entry bpf: Fix mprog detachment for empty mprog entry bpf: bpf_struct_ops: Remove unnecessary initial values of variables ==================== Link: https://lore.kernel.org/r/20230810055123.109578-1-martin.lau@linux.devSigned-off-by:
-
