- 27 Nov, 2022 1 commit
-
-
Yang Yingliang authored
I got a null-ptr-deref report as following when doing fault injection test: BUG: kernel NULL pointer dereference, address: 0000000000000058 Oops: 0000 [#1] PREEMPT SMP KASAN PTI CPU: 1 PID: 253 Comm: 507-spi-dm9051 Tainted: G B N 6.1.0-rc3+ Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-1ubuntu1.1 04/01/2014 RIP: 0010:klist_put+0x2d/0xd0 Call Trace: <TASK> klist_remove+0xf1/0x1c0 device_release_driver_internal+0x23e/0x2d0 bus_remove_device+0x1bd/0x240 device_del+0x357/0x770 phy_device_remove+0x11/0x30 mdiobus_unregister+0xa5/0x140 release_nodes+0x6a/0xa0 devres_release_all+0xf8/0x150 device_unbind_cleanup+0x19/0xd0 //probe path: phy_device_register() device_add() phy_connect phy_attach_direct() //set device driver probe() //it's failed, driver is not bound device_bind_driver() // probe failed, it's not called //remove path: phy_device_remove() device_del() device_release_driver_internal() __device_release_driver() //dev->drv is not NULL klist_remove() <- knode_driver is not added yet, cause null-ptr-deref In phy_attach_direct(), after setting the 'dev->driver', probe() fails, device_bind_driver() is not called, so the knode_driver->n_klist is not set, then it causes null-ptr-deref in __device_release_driver() while deleting device. Fix this by setting dev->driver to NULL in the error path in phy_attach_direct(). Fixes: e1393456 ("[PATCH] PHY Layer fixup") Signed-off-by:Yang Yingliang <yangyingliang@huawei.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
- 25 Nov, 2022 8 commits
-
-
Rasmus Villemoes authored
When the name_assign_type attribute was introduced (commit 685343fc, "net: add name_assign_type netdev attribute"), the loopback device was explicitly mentioned as one which would make use of NET_NAME_PREDICTABLE: The name_assign_type attribute gives hints where the interface name of a given net-device comes from. These values are currently defined: ... NET_NAME_PREDICTABLE: The ifname has been assigned by the kernel in a predictable way that is guaranteed to avoid reuse and always be the same for a given device. Examples include statically created devices like the loopback device [...] Switch to that so that reading /sys/class/net/lo/name_assign_type produces something sensible instead of returning -EINVAL. Signed-off-by:
Rasmus Villemoes <linux@rasmusvillemoes.dk> Reviewed-by:
Jacob Keller <jacob.e.keller@intel.com> Signed-off-by:
-
Rasmus Villemoes authored
Currently, when a FEC device is brought up, the irq coalesce settings are reset to their default values (1000us, 200 frames). That's unexpected, and breaks for example use of an appropriate .link file to make systemd-udev apply the desired settings (https://www.freedesktop.org/software/systemd/man/systemd.link.html), or any other method that would do a one-time setup during early boot. Refactor the code so that fec_restart() instead uses fec_enet_itr_coal_set(), which simply applies the settings that are stored in the private data, and initialize that private data with the default values. Signed-off-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
Suman Ghosh authored
This patch addresses pfc_alloc_status array overflow occurring for send queue index value greater than PFC priority. Queue index can be greater than supported PFC priority for multiple scenarios (e.g. QoS, during non zero SMQ allocation for a PF/VF). In those scenarios the API should return default tx scheduler '0'. This is causing mbox errors as otx2_get_smq_idx returing invalid smq value. Fixes: 99c969a8 ("octeontx2-pf: Add egress PFC support") Signed-off-by:
Suman Ghosh <sumang@marvell.com> Reviewed-by:
Leon Romanovsky <leonro@nvidia.com> Signed-off-by:
-
Goh, Wei Sheng authored
Currently, pause frame register GMAC_RX_FLOW_CTRL_RFE is not updated correctly when 'ethtool -A <IFACE> autoneg off rx off tx off' command is issued. This fix ensures the flow control change is reflected directly in the GMAC_RX_FLOW_CTRL_RFE register. Fixes: 46f69ded ("net: stmmac: Use resolved link config in mac_link_up()") Cc: <stable@vger.kernel.org> # 5.10.x Signed-off-by:
Goh, Wei Sheng <wei.sheng.goh@intel.com> Signed-off-by:
Noor Azura Ahmad Tarmizi <noor.azura.ahmad.tarmizi@intel.com> Signed-off-by:
-
Izabela Bakollari authored
IPV6 addresses are purged when setting the number of rx/tx rings using ethtool -G. The function aq_set_ringparam calls dev_close, which removes the addresses. As a solution, call an internal function (aq_ndev_close). Fixes: c1af5427 ("net: aquantia: Ethtool based ring size configuration") Signed-off-by:
Izabela Bakollari <ibakolla@redhat.com> Signed-off-by:
-
Lukas Bulwahn authored
Defines prefixed with "CONFIG" should be limited to proper Kconfig options, that are introduced in a Kconfig file. Here, constants for bitmap indices of some configs are defined and these defines begin with the config's name, and are suffixed with BITMAP_IDX. To avoid defines prefixed with "CONFIG", name these constants BITMAP_IDX_FOR_CONFIG_XYZ instead of CONFIG_XYZ_BITMAP_IDX. No functional change. Signed-off-by:
Lukas Bulwahn <lukas.bulwahn@gmail.com> Signed-off-by:
-
Duoming Zhou authored
The watchdog timer is used to monitor whether the process of transmitting data is timeout. If we use qlcnic driver, the dev_watchdog() that is the timer handler of watchdog timer will call qlcnic_tx_timeout() to process the timeout. But the qlcnic_tx_timeout() calls msleep(), as a result, the sleep-in-atomic-context bugs will happen. The processes are shown below: (atomic context) dev_watchdog qlcnic_tx_timeout qlcnic_83xx_idc_request_reset qlcnic_83xx_lock_driver msleep --------------------------- (atomic context) dev_watchdog qlcnic_tx_timeout qlcnic_83xx_idc_request_reset qlcnic_83xx_lock_driver qlcnic_83xx_recover_driver_lock msleep Fix by changing msleep() to mdelay(), the mdelay() is busy-waiting and the bugs could be mitigated. Fixes: 629263ac ("qlcnic: 83xx CNA inter driver communication mechanism") Signed-off-by:Duoming Zhou <duoming@zju.edu.cn> Signed-off-by:
-
David S. Miller authored
Merge tag 'linux-can-fixes-for-6.1-20221124' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can Marc Kleine-Budde says: ==================== linux-can-fixes-for-6.1-20221124 this is a pull request of 8 patches for net/master. Ziyang Xuan contributes a patch for the can327, fixing a potential SKB leak when the netdev is down. Heiko Schocher's patch for the sja1000 driver fixes the width of the definition of the OCR_MODE_MASK. Zhang Changzhong contributes 4 patches. In the sja1000_isa, cc770, and m_can_pci drivers the error path in the probe() function and in case of the etas_es58x a function that is called by probe() are fixed. Jiasheng Jiang add a missing check for the return value of the devm_clk_get() in the m_can driver. Yasushi SHOJI's patch for the mcba_usb fixes setting of the external termination resistor. ==================== Signed-off-by:
-
- 24 Nov, 2022 23 commits
-
-
Yasushi SHOJI authored
Microchip USB Analyzer can activate the internal termination resistors by setting the "termination" option ON, or OFF to to deactivate them. As I've observed, both with my oscilloscope and captured USB packets below, you must send "0" to turn it ON, and "1" to turn it OFF. From the schematics in the user's guide, I can confirm that you must drive the CAN_RES signal LOW "0" to activate the resistors. Reverse the argument value of usb_msg.termination to fix this. These are the two commands sequence, ON then OFF. > No. Time Source Destination Protocol Length Info > 1 0.000000 host 1.3.1 USB 46 URB_BULK out > > Frame 1: 46 bytes on wire (368 bits), 46 bytes captured (368 bits) > USB URB > Leftover Capture Data: a80000000000000000000000000000000000a8 > > No. Time Source Destination Protocol Length Info > 2 4.372547 host 1.3.1 USB 46 URB_BULK out > > Frame 2: 46 bytes on wire (368 bits), 46 bytes captured (368 bits) > USB URB > Leftover Capture Data: a80100000000000000000000000000000000a9 Signed-off-by:
Yasushi SHOJI <yashi@spacecubics.com> Link: https://lore.kernel.org/all/20221124152504.125994-1-yashi@spacecubics.comSigned-off-by:
Marc Kleine-Budde <mkl@pengutronix.de>
-
Jiasheng Jiang authored
Since the devm_clk_get may return error, it should be better to add check for the cdev->hclk, as same as cdev->cclk. Fixes: f524f829 ("can: m_can: Create a m_can platform framework") Signed-off-by:
Jiasheng Jiang <jiasheng@iscas.ac.cn> Link: https://lore.kernel.org/all/20221123063651.26199-1-jiasheng@iscas.ac.cnSigned-off-by:
-
Zhang Changzhong authored
In m_can_pci_remove() and error handling path of m_can_pci_probe(), m_can_class_free_dev() should be called to free resource allocated by m_can_class_allocate_dev(), otherwise there will be memleak. Fixes: cab7ffc0 ("can: m_can: add PCI glue driver for Intel Elkhart Lake") Signed-off-by:
Zhang Changzhong <zhangchangzhong@huawei.com> Reviewed-by:
Jarkko Nikula <jarkko.nikula@linux.intel.com> Link: https://lore.kernel.org/all/1668168684-6390-1-git-send-email-zhangchangzhong@huawei.comSigned-off-by:
-
Zhang Changzhong authored
In case of register_candev() fails, clear es58x_dev->netdev[channel_idx] and add free_candev(). Otherwise es58x_free_netdevs() will unregister the netdev that has never been registered. Fixes: 85372578 ("can: etas_es58x: add core support for ETAS ES58X CAN USB interfaces") Signed-off-by:
Arunachalam Santhanam <Arunachalam.Santhanam@in.bosch.com> Acked-by:
Vincent Mailhol <mailhol.vincent@wanadoo.fr> Link: https://lore.kernel.org/all/1668413685-23354-1-git-send-email-zhangchangzhong@huawei.comSigned-off-by:
-
Zhang Changzhong authored
Add the missing free_cc770dev() before return from cc770_isa_probe() in the register_cc770dev() error handling case. In addition, remove blanks before goto labels. Fixes: 7e02e543 ("can: cc770: legacy CC770 ISA bus driver") Signed-off-by:
-
Zhang Changzhong authored
Add the missing free_sja1000dev() before return from sja1000_isa_probe() in the register_sja1000dev() error handling case. In addition, remove blanks before goto labels. Fixes: 2a6ba39a ("can: sja1000: legacy SJA1000 ISA bus driver") Signed-off-by:
-
Heiko Schocher authored
bitfield mode in ocr register has only 2 bits not 3, so correct the OCR_MODE_MASK define. Signed-off-by:
Heiko Schocher <hs@denx.de> Link: https://lore.kernel.org/all/20221123071636.2407823-1-hs@denx.deSigned-off-by:
-
Ziyang Xuan authored
In can327_feed_frame_to_netdev(), it did not free the skb when netdev is down, and all callers of can327_feed_frame_to_netdev() did not free allocated skb too. That would trigger skb leak. Fix it by adding kfree_skb() in can327_feed_frame_to_netdev() when netdev is down. Not tested, just compiled. Fixes: 43da2f07 ("can: can327: CAN/ldisc driver for ELM327 based OBD-II adapters") Signed-off-by:
Ziyang Xuan <william.xuanziyang@huawei.com> Link: https://lore.kernel.org/all/20221110061437.411525-1-william.xuanziyang@huawei.comReviewed-by:
Max Staudt <max@enpas.org> Cc: stable@vger.kernel.org Signed-off-by:
-
Yu Liao authored
The ACPI buffer memory (string.pointer) should be freed as the buffer is not used after returning from bgx_acpi_match_id(), free it to prevent memory leak. Fixes: 46b903a0 ("net, thunder, bgx: Add support to get MAC address from ACPI.") Signed-off-by:
Yu Liao <liaoyu15@huawei.com> Link: https://lore.kernel.org/r/20221123082237.1220521-1-liaoyu15@huawei.comSigned-off-by:
Paolo Abeni <pabeni@redhat.com>
-
Xiongfeng Wang authored
pci_get_device() will decrease the reference count for the *from* parameter. So we don't need to call put_device() to decrease the reference. Let's remove the put_device() in the loop and only decrease the reference count of the returned 'pdev' for the last loop because it will not be passed to pci_get_device() as input parameter. We don't need to check if 'pdev' is NULL because it is already checked inside pci_dev_put(). Also add pci_dev_put() for the error path. Fixes: fe1939bb ("octeontx2-af: Add SDP interface support") Signed-off-by:
Xiongfeng Wang <wangxiongfeng2@huawei.com> Reviewed-by:
Saeed Mahameed <saeed@kernel.org> Link: https://lore.kernel.org/r/20221123065919.31499-1-wangxiongfeng2@huawei.comSigned-off-by:
-
Liu Jian authored
Call phylink_disconnect_phy() in tse_shutdown() to release the resources occupied by phylink_of_phy_connect() in the tse_open(). Fixes: fef29982 ("net: altera: tse: convert to phylink") Signed-off-by:
Liu Jian <liujian56@huawei.com> Link: https://lore.kernel.org/r/20221123011617.332302-1-liujian56@huawei.comSigned-off-by:
-
Li Zetao authored
When doing the following test steps, an error was found: step 1: modprobe virtio_net succeeded # modprobe virtio_net <-- OK step 2: fault injection in register_netdevice() # modprobe -r virtio_net <-- OK # ... FAULT_INJECTION: forcing a failure. name failslab, interval 1, probability 0, space 0, times 0 CPU: 0 PID: 3521 Comm: modprobe Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), Call Trace: <TASK> ... should_failslab+0xa/0x20 ... dev_set_name+0xc0/0x100 netdev_register_kobject+0xc2/0x340 register_netdevice+0xbb9/0x1320 virtnet_probe+0x1d72/0x2658 [virtio_net] ... </TASK> virtio_net: probe of virtio0 failed with error -22 step 3: modprobe virtio_net failed # modprobe virtio_net <-- failed virtio_net: probe of virtio0 failed with error -2 The root cause of the problem is that the queues are not disable on the error handling path when register_netdevice() fails in virtnet_probe(), resulting in an error "-ENOENT" returned in the next modprobe call in setup_vq(). virtio_pci_modern_device uses virtqueues to send or receive message, and "queue_enable" records whether the queues are available. In vp_modern_find_vqs(), all queues will be selected and activated, but once queues are enabled there is no way to go back except reset. Fix it by reset virtio device on error handling path. This makes error handling follow the same order as normal device cleanup in virtnet_remove() which does: unregister, destroy failover, then reset. And that flow is better tested than error handling so we can be reasonably sure it works well. Fixes: 02465555 ("virtio_net: fix use after free on allocation failure") Signed-off-by:Li Zetao <lizetao1@huawei.com> Acked-by:
Michael S. Tsirkin <mst@redhat.com> Link: https://lore.kernel.org/r/20221122150046.3910638-1-lizetao1@huawei.comSigned-off-by:
-
Hanjun Guo authored
The ACPI buffer memory (buffer.pointer) should be freed as the buffer is not used after acpi_evaluate_object(), free it to prevent memory leak. Fixes: 13e920d9 ("net: wwan: t7xx: Add core components") Signed-off-by:
Hanjun Guo <guohanjun@huawei.com> Link: https://lore.kernel.org/r/1669119580-28977-1-git-send-email-guohanjun@huawei.comSigned-off-by:
-
Jiasheng Jiang authored
As the devm_kcalloc may return NULL pointer, it should be better to add check for the return value, as same as the others. Fixes: e8e095b3 ("octeontx2-af: cn10k: Bandwidth profiles config support") Signed-off-by:
Maciej Fijalkowski <maciej.fijalkowski@intel.com> Link: https://lore.kernel.org/r/20221122055449.31247-1-jiasheng@iscas.ac.cnSigned-off-by:
-
Vladimir Oltean authored
In the blamed commit, a rudimentary reallocation procedure for RX buffer descriptors was implemented, for the situation when their format changes between normal (no PTP) and extended (PTP). enetc_hwtstamp_set() calls enetc_close() and enetc_open() in a sequence, and this sequence loses information which was previously configured in the TX BDR Mode Register, specifically via the enetc_set_bdr_prio() call. The TX ring priority is configured by tc-mqprio and tc-taprio, and affects important things for TSN such as the TX time of packets. The issue manifests itself most visibly by the fact that isochron --txtime reports premature packet transmissions when PTP is first enabled on an enetc interface. Save the TX ring priority in a new field in struct enetc_bdr (occupies a 2 byte hole on arm64) in order to make this survive a ring reconfiguration. Fixes: 434cebab ("enetc: Add dynamic allocation of extended Rx BD rings") Signed-off-by:
Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by:
Alexander Lobakin <alexandr.lobakin@intel.com> Link: https://lore.kernel.org/r/20221122130936.1704151-1-vladimir.oltean@nxp.comSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Zhang Changzhong authored
If prestera_port_sfp_bind() fails, unregister_netdev() should be called in error handling path. Compile tested only. Fixes: 52323ef7 ("net: marvell: prestera: add phylink support") Signed-off-by:
-
Jakub Kicinski authored
Martin Faltesek says: ==================== nfc: st-nci: Restructure validating logic in EVT_TRANSACTION These are the same 3 patches that were applied in st21nfca here: https://lore.kernel.org/netdev/20220607025729.1673212-1-mfaltesek@google.com with a couple minor differences. st-nci has nearly identical code to that of st21nfca for EVT_TRANSACTION, except that there are two extra validation checks that are not present in the st-nci code. The 3/3 patch as coded for st21nfca pulls those checks in, bringing both drivers into parity. ==================== Link: https://lore.kernel.org/r/20221122004246.4186422-1-mfaltesek@google.comSigned-off-by:
-
Martin Faltesek authored
The transaction buffer is allocated by using the size of the packet buf, and subtracting two which seems intended to remove the two tags which are not present in the target structure. This calculation leads to under counting memory because of differences between the packet contents and the target structure. The aid_len field is a u8 in the packet, but a u32 in the structure, resulting in at least 3 bytes always being under counted. Further, the aid data is a variable length field in the packet, but fixed in the structure, so if this field is less than the max, the difference is added to the under counting. To fix, perform validation checks progressively to safely reach the next field, to determine the size of both buffers and verify both tags. Once all validation checks pass, allocate the buffer and copy the data. This eliminates freeing memory on the error path, as validation checks are moved ahead of memory allocation. Reported-by:
Denis Efremov <denis.e.efremov@oracle.com> Reviewed-by:
Guenter Roeck <groeck@google.com> Fixes: 5d1ceb7f ("NFC: st21nfcb: Add HCI transaction event support") Signed-off-by:
Martin Faltesek <mfaltesek@google.com> Reviewed-by:
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Signed-off-by:
-
Martin Faltesek authored
Error path does not free previously allocated memory. Add devm_kfree() to the failure path. Reported-by:
-
Martin Faltesek authored
The first validation check for EVT_TRANSACTION has two different checks tied together with logical AND. One is a check for minimum packet length, and the other is for a valid aid_tag. If either condition is true (fails), then an error should be triggered. The fix is to change && to ||. Reported-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsecJakub Kicinski authored
Steffen Klassert says: ==================== ipsec 2022-11-23 1) Fix "disable_policy" on ipv4 early demuxP Packets after the initial packet in a flow might be incorectly dropped on early demux if there are no matching policies. From Eyal Birger. 2) Fix a kernel warning in case XFRM encap type is not available. From Eyal Birger. 3) Fix ESN wrap around for GSO to avoid a double usage of a sequence number. From Christian Langrock. 4) Fix a send_acquire race with pfkey_register. From Herbert Xu. 5) Fix a list corruption panic in __xfrm_state_delete(). Thomas Jarosch. 6) Fix an unchecked return value in xfrm6_init(). Chen Zhongjin. * 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec: xfrm: Fix ignored return value in xfrm6_init() xfrm: Fix oops in __xfrm_state_delete() af_key: Fix send_acquire race with pfkey_register xfrm: replay: Fix ESN wrap around for GSO xfrm: lwtunnel: squelch kernel warning in case XFRM encap type is not available xfrm: fix "disable_policy" on ipv4 early demux ==================== Link: https://lore.kernel.org/r/20221123093117.434274-1-steffen.klassert@secunet.comSigned-off-by: -
git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nfJakub Kicinski authored
Pablo Neira Ayuso says: ==================== Netfilter fixes for net 1) Fix regression in ipset hash:ip with IPv4 range, from Vishwanath Pai. This is fixing up a bug introduced in the 6.0 release. 2) The "netfilter: ipset: enforce documented limit to prevent allocating huge memory" patch contained a wrong condition which makes impossible to add up to 64 clashing elements to a hash:net,iface type of set while it is the documented feature of the set type. The patch fixes the condition and thus makes possible to add the elements while keeps preventing allocating huge memory, from Jozsef Kadlecsik. This has been broken for several releases. 3) Missing locking when updating the flow block list which might lead a reader to crash. This has been broken since the introduction of the flowtable hardware offload support. * git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf: netfilter: flowtable_offload: add missing locking netfilter: ipset: restore allowing 64 clashing elements in hash:net,iface netfilter: ipset: regression in ip_set_hash_ip.c ==================== Link: https://lore.kernel.org/r/20221122212814.63177-1-pablo@netfilter.orgSigned-off-by:
-
Nir Levy authored
The documentation refers to invalid web page under www.linuxfoundation.org The patch refers to a working URL under wiki.linuxfoundation.org Signed-off-by:
Nir Levy <bhr166@gmail.com> Link: https://lore.kernel.org/all/20221120220630.7443-1-bhr166@gmail.com/Signed-off-by:
-
- 23 Nov, 2022 8 commits
-
-
Santiago Ruano Rincón authored
ZLP for DisplayLink ethernet devices was enabled in 6.0: 266c0190 ("net/cdc_ncm: Enable ZLP for DisplayLink ethernet devices"). The related driver_info should be the "same as cdc_ncm_info, but with FLAG_SEND_ZLP". However, set_rx_mode that enables handling multicast traffic was missing in the new cdc_ncm_zlp_info. usbnet_cdc_update_filter rx mode was introduced in linux 5.9 with: e10dcb1b ("net: cdc_ncm: hook into set_rx_mode to admit multicast traffic") Without this hook, multicast, and then IPv6 SLAAC, is broken. Fixes: 266c0190 ("net/cdc_ncm: Enable ZLP for DisplayLink ethernet devices") Signed-off-by:
Santiago Ruano Rincón <santiago.ruano-rincon@imt-atlantique.fr> Signed-off-by:
-
Davide Tronchin authored
Add RmNet support for LARA-L6. LARA-L6 module can be configured (by AT interface) in three different USB modes: * Default mode (Vendor ID: 0x1546 Product ID: 0x1341) with 4 serial interfaces * RmNet mode (Vendor ID: 0x1546 Product ID: 0x1342) with 4 serial interfaces and 1 RmNet virtual network interface * CDC-ECM mode (Vendor ID: 0x1546 Product ID: 0x1343) with 4 serial interface and 1 CDC-ECM virtual network interface In RmNet mode LARA-L6 exposes the following interfaces: If 0: Diagnostic If 1: AT parser If 2: AT parser If 3: AT parset/alternative functions If 4: RMNET interface Signed-off-by:
Davide Tronchin <davide.tronchin.94@gmail.com> Acked-by:
Bjørn Mork <bjorn@mork.no> Signed-off-by:
-
Jakub Sitnicki authored
When holding a reader-writer spin lock we cannot sleep. Calling setup_udp_tunnel_sock() with write lock held violates this rule, because we end up calling percpu_down_read(), which might sleep, as syzbot reports [1]: __might_resched.cold+0x222/0x26b kernel/sched/core.c:9890 percpu_down_read include/linux/percpu-rwsem.h:49 [inline] cpus_read_lock+0x1b/0x140 kernel/cpu.c:310 static_key_slow_inc+0x12/0x20 kernel/jump_label.c:158 udp_tunnel_encap_enable include/net/udp_tunnel.h:187 [inline] setup_udp_tunnel_sock+0x43d/0x550 net/ipv4/udp_tunnel_core.c:81 l2tp_tunnel_register+0xc51/0x1210 net/l2tp/l2tp_core.c:1509 pppol2tp_connect+0xcdc/0x1a10 net/l2tp/l2tp_ppp.c:723 Trim the writer-side critical section for sk_callback_lock down to the minimum, so that it covers only operations on sk_user_data. Also, when grabbing the sk_callback_lock, we always need to disable BH, as Eric points out. Failing to do so leads to deadlocks because we acquire sk_callback_lock in softirq context, which can get stuck waiting on us if: 1) it runs on the same CPU, or CPU0 ---- lock(clock-AF_INET6); <Interrupt> lock(clock-AF_INET6); 2) lock ordering leads to priority inversion CPU0 CPU1 ---- ---- lock(clock-AF_INET6); local_irq_disable(); lock(&tcp_hashinfo.bhash[i].lock); lock(clock-AF_INET6); <Interrupt> lock(&tcp_hashinfo.bhash[i].lock); ... as syzbot reports [2,3]. Use the _bh variants for write_(un)lock. [1] https://lore.kernel.org/netdev/0000000000004e78ec05eda79749@google.com/ [2] https://lore.kernel.org/netdev/000000000000e38b6605eda76f98@google.com/ [3] https://lore.kernel.org/netdev/000000000000dfa31e05eda76f75@google.com/ v2: - Check and set sk_user_data while holding sk_callback_lock for both L2TP encapsulation types (IP and UDP) (Tetsuo) Cc: Tom Parkin <tparkin@katalix.com> Cc: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp> Fixes: b68777d5 ("l2tp: Serialize access to sk_user_data with sk_callback_lock") Reported-by:Eric Dumazet <edumazet@google.com> Reported-by: syzbot+703d9e154b3b58277261@syzkaller.appspotmail.com Reported-by: syzbot+50680ced9e98a61f7698@syzkaller.appspotmail.com Reported-by: syzbot+de987172bb74a381879b@syzkaller.appspotmail.com Signed-off-by:
Jakub Sitnicki <jakub@cloudflare.com> Signed-off-by:
-
Yuan Can authored
The dm9051_loop_rx() returns without release skb when dm9051_stop_mrcmd() returns error, free the skb to avoid this leak. Fixes: 2dc95a4d ("net: Add dm9051 driver") Signed-off-by:
Yuan Can <yuancan@huawei.com> Reviewed-by:
-
Wang Hai authored
In com20020_probe(), if com20020_config() fails, dev and info will not be freed, which will lead to a memory leak. This patch adds freeing dev and info after com20020_config() fails to fix this bug. Compile tested only. Fixes: 15b99ac1 ("[PATCH] pcmcia: add return value to _config() functions") Signed-off-by:
Wang Hai <wanghai38@huawei.com> Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linuxJakub Kicinski authored
Saeed Mahameed says: ==================== mlx5 fixes 2022-11-21 This series provides bug fixes to mlx5 driver. * tag 'mlx5-fixes-2022-11-21' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux: net/mlx5e: Fix possible race condition in macsec extended packet number update routine net/mlx5e: Fix MACsec update SecY net/mlx5e: Fix MACsec SA initialization routine net/mlx5e: Remove leftovers from old XSK queues enumeration net/mlx5e: Offload rule only when all encaps are valid net/mlx5e: Fix missing alignment in size of MTT/KLM entries net/mlx5: Fix sync reset event handler error flow net/mlx5: E-Switch, Set correctly vport destination net/mlx5: Lag, avoid lockdep warnings net/mlx5: Fix handling of entry refcount when command is not issued to FW net/mlx5: cmdif, Print info on any firmware cmd failure to tracepoint net/mlx5: SF: Fix probing active SFs during driver probe phase net/mlx5: Fix FW tracer timestamp calculation net/mlx5: Do not query pci info while pci disabled ==================== Link: https://lore.kernel.org/r/20221122022559.89459-1-saeed@kernel.orgSigned-off-by:
-
Ziyang Xuan authored
In fib_table_insert(), if the alias was already inserted, but node not exist, the error code should be set before return from error handling path. Fixes: a6c76c17 ("ipv4: Notify route after insertion to the routing table") Signed-off-by:
-
Jakub Kicinski authored
Yan Cangang says: ==================== net: ethernet: mtk_eth_soc: fix memory leak in error path ==================== Link: https://lore.kernel.org/r/20221120055259.224555-1-nalanzeyu@gmail.comSigned-off-by:
-
