- 26 Aug, 2024 40 commits
-
-
Jason Xing authored
We found that one close-wait socket was reset by the other side due to a new connection reusing the same port which is beyond our expectation, so we have to investigate the underlying reason. The following experiment is conducted in the test environment. We limit the port range from 40000 to 40010 and delay the time to close() after receiving a fin from the active close side, which can help us easily reproduce like what happened in production. Here are three connections captured by tcpdump: 127.0.0.1.40002 > 127.0.0.1.9999: Flags [S], seq 2965525191 127.0.0.1.9999 > 127.0.0.1.40002: Flags [S.], seq 2769915070 127.0.0.1.40002 > 127.0.0.1.9999: Flags [.], ack 1 127.0.0.1.40002 > 127.0.0.1.9999: Flags [F.], seq 1, ack 1 // a few seconds later, within 60 seconds 127.0.0.1.40002 > 127.0.0.1.9999: Flags [S], seq 2965590730 127.0.0.1.9999 > 127.0.0.1.40002: Flags [.], ack 2 127.0.0.1.40002 > 127.0.0.1.9999: Flags [R], seq 2965525193 // later, very quickly 127.0.0.1.40002 > 127.0.0.1.9999: Flags [S], seq 2965590730 127.0.0.1.9999 > 127.0.0.1.40002: Flags [S.], seq 3120990805 127.0.0.1.40002 > 127.0.0.1.9999: Flags [.], ack 1 As we can see, the first flow is reset because: 1) client starts a new connection, I mean, the second one 2) client tries to find a suitable port which is a timewait socket (its state is timewait, substate is fin_wait2) 3) client occupies that timewait port to send a SYN 4) server finds a corresponding close-wait socket in ehash table, then replies with a challenge ack 5) client sends an RST to terminate this old close-wait socket. I don't think the port selection algo can choose a FIN_WAIT2 socket when we turn on tcp_tw_reuse because on the server side there remain unread data. In some cases, if one side haven't call close() yet, we should not consider it as expendable and treat it at will. Even though, sometimes, the server isn't able to call close() as soon as possible like what we expect, it can not be terminated easily, especially due to a second unrelated connection happening. After this patch, we can see the expected failure if we start a connection when all the ports are occupied in fin_wait2 state: "Ncat: Cannot assign requested address." Reported-by:
Jade Dong <jadedong@tencent.com> Signed-off-by:
Jason Xing <kernelxing@tencent.com> Reviewed-by:
Eric Dumazet <edumazet@google.com> Link: https://patch.msgid.link/20240823001152.31004-1-kerneljasonxing@gmail.comSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Jakub Kicinski authored
Kyle Swenson says: ==================== net: pse-pd: tps23881: Reset GPIO support On some boards, the TPS2388x's reset line (active low) is pulled low to keep the chip in reset until the SoC pulls the device out of reset. This series updates the device-tree binding for the tps23881 and then adds support for the reset gpio handling in the tps23881 driver. v1: https://lore.kernel.org/20240819190151.93253-1-kyle.swenson@est.tech ==================== Link: https://patch.msgid.link/20240822220100.3030184-1-kyle.swenson@est.techSigned-off-by:
-
Kyle Swenson authored
The TPS23880/1 has an active-low reset pin that some boards connect to the SoC to control when the TPS23880 is pulled out of reset. Add support for this via a reset-gpios property in the DTS. Signed-off-by:
Kyle Swenson <kyle.swenson@est.tech> Acked-by:
Oleksij Rempel <o.rempel@pengutronix.de> Reviewed-by:
Kory Maincent <kory.maincent@bootlin.com> Link: https://patch.msgid.link/20240822220100.3030184-3-kyle.swenson@est.techSigned-off-by:
-
Kyle Swenson authored
The TPS23881 has an active-low reset pin that can be connected to an SoC. Document this with the device-tree binding. Signed-off-by:
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Reviewed-by:
-
Rosen Penev authored
It's only used in one place. It doesn't need to be in the struct. Signed-off-by:
Rosen Penev <rosenp@gmail.com> Link: https://patch.msgid.link/20240822192758.141201-1-rosenp@gmail.comSigned-off-by:
-
Cong Wang authored
Although commit 4a4cd703 ("l2tp: don't set sk_user_data in tunnel socket") removed sk->sk_user_data usage, setup_udp_tunnel_sock() still touches sk->sk_user_data, this conflicts with sockmap which also leverages sk->sk_user_data to save psock. Restore this sk->sk_user_data check to avoid such conflicts. Fixes: 4a4cd703 ("l2tp: don't set sk_user_data in tunnel socket") Reported-by: syzbot+8dbe3133b840c470da0e@syzkaller.appspotmail.com Cc: Tom Parkin <tparkin@katalix.com> Signed-off-by:
Cong Wang <cong.wang@bytedance.com> Tested-by:
James Chapman <jchapman@katalix.com> Reviewed-by:
-
Jakub Kicinski authored
Sean Anderson says: ==================== net: xilinx: axienet: Multicast fixes and improvements This series has a few small patches improving the handling of multicast addresses. In particular, it makes the driver a whole lot less spammy, and adjusts things so we aren't in promiscuous mode when we have more than four multicast addresses (a common occurance on modern systems). As the hardware has a 4-entry CAM, the ideal method would be to "pack" multiple addresses into one CAM entry. Something like: entry.address = address[0] | address[1]; entry.mask = ~(address[0] ^ address[1]); Which would make the entry match both addresses (along with some others that would need to be filtered in software). Mapping addresses to entries in an efficient way is a bit tricky. If anyone knows of an in-tree example of something like this, I'd be glad to hear about it. ==================== Link: https://patch.msgid.link/20240822154059.1066595-1-sean.anderson@linux.devSigned-off-by:
-
Sean Anderson authored
Add support for IFF_ALLMULTI by configuring a single filter to match the multicast address bit. This allows us to keep promiscuous mode disabled, even when we have more than four multicast addresses. An even better solution would be to "pack" addresses into the available CAM registers, but that can wait for a future series. Signed-off-by:
Sean Anderson <sean.anderson@linux.dev> Reviewed-by:
Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20240822154059.1066595-6-sean.anderson@linux.devSigned-off-by:
-
Sean Anderson authored
Contrary to the comment, we don't have to inform the net subsystem. Signed-off-by:
-
Sean Anderson authored
A message about being in promiscuous mode is printed every time each additional multicast address beyond four is added. Suppress this message like is done in other drivers. Signed-off-by:
-
Jakub Kicinski authored
Li Zetao says: ==================== Some modifications to optimize code readability This patchset is mainly optimized for readability in contexts where size needs to be determined. By using min() or max(), or even directly removing redundant judgments (such as the 5th patch), the code is more consistent with the context. ==================== Link: https://patch.msgid.link/20240822133908.1042240-1-lizetao1@huawei.comSigned-off-by:
-
Li Zetao authored
When calculating size of own domain based on number of peers, the result should be less than MAX_MON_DOMAIN, so using min() here is very semantic. Signed-off-by:
Li Zetao <lizetao1@huawei.com> Reviewed-by:
-
Li Zetao authored
When coping sockaddr in ip6_mc_msfget(), the time of copies depends on the minimum value between sl_count and gf_numsrc. Using min() here is very semantic. Signed-off-by:
-
Li Zetao authored
When processing the tail append of sk buffer, the final length needs to be determined based on expectlen and addlen. Using max() here can increase the readability of the code. Signed-off-by:
-
Jakub Kicinski authored
Simon Horman says: ==================== net: header and core spelling corrections This patchset addresses a number of spelling errors in comments in Networking files under include/, and files in net/core/. Spelling problems are as flagged by codespell. It aims to provide patches that can be accepted directly into net-next. And splits patches up based on maintainer boundaries: many things feed directly into net-next. This is a complex process and I apologise for any errors. I also plan to address, via separate patches, spelling errors in other files in the same directories, for files whose changes typically go through trees other than net-next (which feed into net-next). ==================== Link: https://patch.msgid.link/20240822-net-spell-v1-0-3a98971ce2d2@kernel.orgSigned-off-by:
-
Simon Horman authored
Correct spelling in net/core. As reported by codespell. Signed-off-by:
-
Simon Horman authored
Correct spelling in Networking headers. As reported by codespell. Signed-off-by:
-
Simon Horman authored
Correct spelling in x25.h As reported by codespell. Signed-off-by:
Martin Schiller <ms@dev.tdt.de> Link: https://patch.msgid.link/20240822-net-spell-v1-11-3a98971ce2d2@kernel.orgSigned-off-by:
-
Simon Horman authored
Correct spelling in sctp.h and structs.h. As reported by codespell. Cc: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Signed-off-by:
Xin Long <lucien.xin@gmail.com> Link: https://patch.msgid.link/20240822-net-spell-v1-10-3a98971ce2d2@kernel.orgSigned-off-by:
-
Simon Horman authored
Correct spelling in pkt_cls.h and red.h. As reported by codespell. Cc: Krzysztof Kozlowski <krzk@kernel.org> Signed-off-by:
-
Simon Horman authored
Correct spelling in NFC headers. As reported by codespell. Signed-off-by:
-
Simon Horman authored
Correct spelling in netlabel.h. As reported by codespell. Cc: Paul Moore <paul@paul-moore.com> Signed-off-by:
-
Simon Horman authored
Correct spelling in if_rmnet.h As reported by codespell. Cc: Sean Tranchetti <quic_stranche@quicinc.com> Signed-off-by:
Subash Abhinov Kasiviswanathan <quic_subashab@quicinc.com> Link: https://patch.msgid.link/20240822-net-spell-v1-6-3a98971ce2d2@kernel.orgSigned-off-by:
-
Simon Horman authored
Correct spelling in bond_3ad.h and bond_alb.h. As reported by codespell. Cc: Jay Vosburgh <jv@jvosburgh.net> Cc: Andy Gospodarek <andy@greyhouse.net> Signed-off-by:
-
Simon Horman authored
Correct spelling in ip_tunnels.h As reported by codespell. Cc: David Ahern <dsahern@kernel.org> Signed-off-by:
-
Simon Horman authored
Correct spelling in ip_tunnels.h As reported by codespell. Cc: David Ahern <dsahern@kernel.org> Signed-off-by:
-
Simon Horman authored
Correct spelling in iucv.h As reported by codespell. Cc: Alexandra Winter <wintera@linux.ibm.com> Cc: Thorsten Winkler <twinkler@linux.ibm.com> Signed-off-by:
-
Simon Horman authored
Correct spelling in if_packet.h As reported by codespell. Signed-off-by:
Willem de Bruijn <willemb@google.com> Link: https://patch.msgid.link/20240822-net-spell-v1-1-3a98971ce2d2@kernel.orgSigned-off-by:
-
Jakub Kicinski authored
MD Danish Anwar says: ==================== Add support for ICSSG PA_STATS This series adds support for PA_STATS. Previously this series was a standalone patch adding documentation for PA_STATS in dt-bindings file ti,pruss.yaml. v1 https://lore.kernel.org/all/20240430121915.1561359-1-danishanwar@ti.com/ v2 https://lore.kernel.org/all/20240529115149.630273-1-danishanwar@ti.com/ v3 https://lore.kernel.org/all/20240625153319.795665-1-danishanwar@ti.com/ v4 https://lore.kernel.org/all/20240729113226.2905928-1-danishanwar@ti.com/ v5 https://lore.kernel.org/all/20240814092033.2984734-1-danishanwar@ti.com/ v6 https://lore.kernel.org/all/20240820091657.4068304-1-danishanwar@ti.com/ ==================== Link: https://patch.msgid.link/20240822122652.1071801-1-danishanwar@ti.comSigned-off-by:
-
MD Danish Anwar authored
Add support for dumping PA stats registers via ethtool. Firmware maintained stats are stored at PA Stats registers. Also modify emac_get_strings() API to use ethtool_puts(). This commit also maintains consistency between miig_stats and pa_stats by - renaming the array icssg_all_stats to icssg_all_miig_stats - renaming the structure icssg_stats to icssg_miig_stats - renaming ICSSG_STATS() to ICSSG_MIIG_STATS() - changing order of stats related data structures and arrays so that data structures of a certain stats type is clubbed together. Signed-off-by:
MD Danish Anwar <danishanwar@ti.com> Link: https://patch.msgid.link/20240822122652.1071801-3-danishanwar@ti.comSigned-off-by:
-
MD Danish Anwar authored
Add documentation for pa-stats node which is syscon regmap for PA_STATS registers. This will be used to dump statistics maintained by ICSSG firmware. Reviewed-by:
Roger Quadros <rogerq@kernel.org> Acked-by:
Nishanth Menon <nm@ti.com> Signed-off-by:
-
Jakub Kicinski authored
Oleksij Rempel says: ==================== Add ALCD Support to Cable Testing Interface This patch series introduces support for Active Link Cable Diagnostics (ALCD) in the ethtool cable testing interface and the DP83TD510 PHY driver. Why ALCD? On a 10BaseT1L interface, TDR (Time Domain Reflectometry) is not possible if the link partner is active - TDR will fail in these cases because it requires interrupting the link. Since the link is active, we already know the cable is functioning, so instead of using TDR, we can use ALCD. ALCD lets us measure cable length without disrupting the active link, which is crucial in environments where network uptime is important. It provides a way to gather diagnostic data without the need for downtime. What's in this series: - Extended the ethtool cable testing interface to specify the source of diagnostic results (TDR or ALCD). - Updated the DP83TD510 PHY driver to use ALCD when the link is active, ensuring we can still get cable length info without dropping the connection. ==================== Link: https://patch.msgid.link/20240822120703.1393130-1-o.rempel@pengutronix.deSigned-off-by:
-
Oleksij Rempel authored
In industrial environments where 10BaseT1L PHYs are replacing existing field bus systems like CAN, it's often essential to retain the existing cable infrastructure. After installation, collecting metrics such as cable length is crucial for assessing the quality of the infrastructure. Traditionally, TDR (Time Domain Reflectometry) is used for this purpose. However, TDR requires interrupting the link, and if the link partner remains active, the TDR measurement will fail. Unlike multi-pair systems, where TDR can be attempted during the MDI-X switching window, 10BaseT1L systems face greater challenges. The TDR sequence on 10BaseT1L is longer and coincides with uninterrupted autonegotiation pulses, making TDR impossible when the link partner is active. The DP83TD510 PHY provides an alternative through ALCD (Active Link Cable Diagnostics), which allows for cable length measurement without disrupting an active link. Since a live link indicates no short or open cable states, ALCD can be used effectively to gather cable length information. Enhance the dp83td510 driver by: - Leveraging ALCD to measure cable length when the link is active. - Bypassing TDR when a link is detected, as ALCD provides the required information without disruption. Signed-off-by:
Andrew Lunn <andrew@lunn.ch> Link: https://patch.msgid.link/20240822120703.1393130-4-o.rempel@pengutronix.deSigned-off-by:
-
Oleksij Rempel authored
Enhance the ethtool cable test interface by introducing the ability to specify the source of the diagnostic information for cable test results. This is particularly useful for PHYs that offer multiple diagnostic methods, such as Time Domain Reflectometry (TDR) and Active Link Cable Diagnostic (ALCD). Key changes: - Added `ethnl_cable_test_result_with_src` and `ethnl_cable_test_fault_length_with_src` functions to allow specifying the information source when reporting cable test results. - Updated existing `ethnl_cable_test_result` and `ethnl_cable_test_fault_length` functions to use TDR as the default source, ensuring backward compatibility. - Modified the UAPI to support these new attributes, enabling drivers to provide more detailed diagnostic information. Signed-off-by:
-
Oleksij Rempel authored
Extend the ethtool netlink cable testing interface by adding support for specifying the source of cable testing results. This allows users to differentiate between results obtained through different diagnostic methods. For example, some TI 10BaseT1L PHYs provide two variants of cable diagnostics: Time Domain Reflectometry (TDR) and Active Link Cable Diagnostic (ALCD). By introducing `ETHTOOL_A_CABLE_RESULT_SRC` and `ETHTOOL_A_CABLE_FAULT_LENGTH_SRC` attributes, this update enables drivers to indicate whether the result was derived from TDR or ALCD, improving the clarity and utility of diagnostic information. Signed-off-by:
-
Breno Leitao authored
Adds a selftest that creates two virtual interfaces, assigns one to a new namespace, and assigns IP addresses to both. It listens on the destination interface using socat and configures a dynamic target on netconsole, pointing to the destination IP address. The test then checks if the message was received properly on the destination interface. Signed-off-by:
Breno Leitao <leitao@debian.org> Acked-by:
Matthieu Baerts (NGI0) <matttbe@kernel.org> Link: https://patch.msgid.link/20240822095652.3806208-1-leitao@debian.orgSigned-off-by:
-
Jakub Kicinski authored
Breno Leitao says: ==================== netconsole: Populate dynamic entry even if netpoll fails The current implementation of netconsole removes the entry and fails entirely if netpoll fails to initialize. This approach is suboptimal, as it prevents reconfiguration or re-enabling of the target through configfs. While this issue might seem minor if it were rare, it actually occurs frequently when the network module is configured as a loadable module. In such cases, the network is unavailable when netconsole initializes, causing netpoll to fail. This failure forces users to reconfigure the target from scratch, discarding any settings provided via the command line. The proposed change would keep the target available in configfs, albeit in a disabled state. This modification allows users to adjust settings or simply re-enable the target once the network module has loaded, providing a more flexible and user-friendly solution. v2: https://lore.kernel.org/20240819103616.2260006-1-leitao@debian.org v1: https://lore.kernel.org/20240809161935.3129104-1-leitao@debian.org ==================== Link: https://patch.msgid.link/20240822111051.179850-1-leitao@debian.orgSigned-off-by:
-
Breno Leitao authored
Currently, netconsole discards targets that fail during initialization, causing two issues: 1) Inconsistency between target list and configfs entries * user pass cmdline0, cmdline1. If cmdline0 fails, then cmdline1 becomes cmdline0 in configfs. 2) Inability to manage failed targets from userspace * If user pass a target that fails with netpoll (interface not loaded at netcons initialization time, such as interface is a module), then the target will not exist in the configfs, so, user cannot re-enable or modify it from userspace. Failed targets are now added to the target list and configfs, but remain disabled until manually enabled or reconfigured. This change does not change the behaviour if CONFIG_NETCONSOLE_DYNAMIC is not set. CC: Aijay Adams <aijay@meta.com> Signed-off-by: -
Breno Leitao authored
Modify netpoll_setup() and __netpoll_setup() to ensure that the netpoll structure (np) is left in a clean state if setup fails for any reason. This prevents carrying over misconfigured fields in case of partial setup success. Key changes: - np->dev is now set only after successful setup, ensuring it's always NULL if netpoll is not configured or if netpoll_setup() fails. - np->local_ip is zeroed if netpoll setup doesn't complete successfully. - Added DEBUG_NET_WARN_ON_ONCE() checks to catch unexpected states. - Reordered some operations in __netpoll_setup() for better logical flow. These changes improve the reliability of netpoll configuration, since it assures that the structure is fully initialized or totally unset. Suggested-by:
Paolo Abeni <pabeni@redhat.com> Signed-off-by:
-
Jakub Kicinski authored
Divya Koppera says: ==================== Adds support for lan887x phy Adds support for lan887x phy and accept autoneg configuration in phy driver only when feature is enabled in supported list. v2: https://lore.kernel.org/20240813181515.863208-1-divya.koppera@microchip.com v1: https://lore.kernel.org/20240808145916.26006-1-Divya.Koppera@microchip.com ==================== Link: https://patch.msgid.link/20240821055906.27717-1-Divya.Koppera@microchip.comSigned-off-by:
-
