From 3769478610135e82b262640252d90f6efb05be71 Mon Sep 17 00:00:00 2001 From: Cong Wang Date: Mon, 28 Apr 2025 16:29:54 -0700 Subject: sch_htb: make htb_deactivate() idempotent Alan reported a NULL pointer dereference in htb_next_rb_node() after we made htb_qlen_notify() idempotent. It turns out in the following case it introduced some regression: htb_dequeue_tree(): |-> fq_codel_dequeue() |-> qdisc_tree_reduce_backlog() |-> htb_qlen_notify() |-> htb_deactivate() |-> htb_next_rb_node() |-> htb_deactivate() For htb_next_rb_node(), after calling the 1st htb_deactivate(), the clprio[prio]->ptr could be already set to NULL, which means htb_next_rb_node() is vulnerable here. For htb_deactivate(), although we checked qlen before calling it, in case of qlen==0 after qdisc_tree_reduce_backlog(), we may call it again which triggers the warning inside. To fix the issues here, we need to: 1) Make htb_deactivate() idempotent, that is, simply return if we already call it before. 2) Make htb_next_rb_node() safe against ptr==NULL. Many thanks to Alan for testing and for the reproducer. Fixes: 5ba8b837b522 ("sch_htb: make htb_qlen_notify() idempotent") Reported-by: Alan J. Wylie Signed-off-by: Cong Wang Link: https://patch.msgid.link/20250428232955.1740419-2-xiyou.wangcong@gmail.com Signed-off-by: Jakub Kicinski --- net/sched/sch_htb.c | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 deletions(-) diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c index 4b9a639b642e..14bf71f57057 100644 --- a/net/sched/sch_htb.c +++ b/net/sched/sch_htb.c @@ -348,7 +348,8 @@ static void htb_add_to_wait_tree(struct htb_sched *q, */ static inline void htb_next_rb_node(struct rb_node **n) { - *n = rb_next(*n); + if (*n) + *n = rb_next(*n); } /** @@ -609,8 +610,8 @@ static inline void htb_activate(struct htb_sched *q, struct htb_class *cl) */ static inline void htb_deactivate(struct htb_sched *q, struct htb_class *cl) { - WARN_ON(!cl->prio_activity); - + if (!cl->prio_activity) + return; htb_deactivate_prios(q, cl); cl->prio_activity = 0; } @@ -1485,8 +1486,6 @@ static void htb_qlen_notify(struct Qdisc *sch, unsigned long arg) { struct htb_class *cl = (struct htb_class *)arg; - if (!cl->prio_activity) - return; htb_deactivate(qdisc_priv(sch), cl); } @@ -1740,8 +1739,7 @@ static int htb_delete(struct Qdisc *sch, unsigned long arg, if (cl->parent) cl->parent->children--; - if (cl->prio_activity) - htb_deactivate(q, cl); + htb_deactivate(q, cl); if (cl->cmode != HTB_CAN_SEND) htb_safe_rb_erase(&cl->pq_node, @@ -1949,8 +1947,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, /* turn parent into inner node */ qdisc_purge_queue(parent->leaf.q); parent_qdisc = parent->leaf.q; - if (parent->prio_activity) - htb_deactivate(q, parent); + htb_deactivate(q, parent); /* remove from evt list because of level change */ if (parent->cmode != HTB_CAN_SEND) { -- cgit From 63890286f557aa5c4eed7e90a5a31658de8fdb4d Mon Sep 17 00:00:00 2001 From: Cong Wang Date: Mon, 28 Apr 2025 16:29:55 -0700 Subject: selftests/tc-testing: Add a test case to cover basic HTB+FQ_CODEL case Integrate the reproducer from Alan into TC selftests and use scapy to generate TCP traffic instead of relying on ping command. Cc: Alan J. Wylie Signed-off-by: Cong Wang Link: https://patch.msgid.link/20250428232955.1740419-3-xiyou.wangcong@gmail.com Signed-off-by: Jakub Kicinski --- .../tc-testing/tc-tests/infra/qdiscs.json | 35 ++++++++++++++++++++++ 1 file changed, 35 insertions(+) diff --git a/tools/testing/selftests/tc-testing/tc-tests/infra/qdiscs.json b/tools/testing/selftests/tc-testing/tc-tests/infra/qdiscs.json index 0843f6d37e9c..a951c0d33cd2 100644 --- a/tools/testing/selftests/tc-testing/tc-tests/infra/qdiscs.json +++ b/tools/testing/selftests/tc-testing/tc-tests/infra/qdiscs.json @@ -538,5 +538,40 @@ "$TC qdisc del dev $DUMMY handle 1:0 root", "$IP addr del 10.10.10.10/24 dev $DUMMY || true" ] + }, + { + "id": "62c4", + "name": "Test HTB with FQ_CODEL - basic functionality", + "category": [ + "qdisc", + "htb", + "fq_codel" + ], + "plugins": { + "requires": [ + "nsPlugin", + "scapyPlugin" + ] + }, + "setup": [ + "$TC qdisc add dev $DEV1 root handle 1: htb default 11", + "$TC class add dev $DEV1 parent 1: classid 1:1 htb rate 10kbit", + "$TC class add dev $DEV1 parent 1:1 classid 1:11 htb rate 10kbit prio 0 quantum 1486", + "$TC qdisc add dev $DEV1 parent 1:11 fq_codel quantum 300 noecn", + "sleep 0.5" + ], + "scapy": { + "iface": "$DEV0", + "count": 5, + "packet": "Ether()/IP(dst='10.10.10.1', src='10.10.10.10')/TCP(sport=12345, dport=80)" + }, + "cmdUnderTest": "$TC -s qdisc show dev $DEV1", + "expExitCode": "0", + "verifyCmd": "$TC -s qdisc show dev $DEV1 | grep -A 5 'qdisc fq_codel'", + "matchPattern": "Sent [0-9]+ bytes [0-9]+ pkt", + "matchCount": "1", + "teardown": [ + "$TC qdisc del dev $DEV1 handle 1: root" + ] } ] -- cgit From 1e20324b23f0afba27997434fb978f1e4a1dbcb6 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Wed, 30 Apr 2025 09:37:58 -0700 Subject: virtio-net: don't re-enable refill work too early when NAPI is disabled Commit 4bc12818b363 ("virtio-net: disable delayed refill when pausing rx") fixed a deadlock between reconfig paths and refill work trying to disable the same NAPI instance. The refill work can't run in parallel with reconfig because trying to double-disable a NAPI instance causes a stall under the instance lock, which the reconfig path needs to re-enable the NAPI and therefore unblock the stalled thread. There are two cases where we re-enable refill too early. One is in the virtnet_set_queues() handler. We call it when installing XDP: virtnet_rx_pause_all(vi); ... virtnet_napi_tx_disable(..); ... virtnet_set_queues(..); ... virtnet_rx_resume_all(..); We want the work to be disabled until we call virtnet_rx_resume_all(), but virtnet_set_queues() kicks it before NAPIs were re-enabled. The other case is a more trivial case of mis-ordering in __virtnet_rx_resume() found by code inspection. Taking the spin lock in virtnet_set_queues() (requested during review) may be unnecessary as we are under rtnl_lock and so are all paths writing to ->refill_enabled. Acked-by: Michael S. Tsirkin Reviewed-by: Bui Quang Minh Fixes: 4bc12818b363 ("virtio-net: disable delayed refill when pausing rx") Fixes: 413f0271f396 ("net: protect NAPI enablement with netdev_lock()") Link: https://patch.msgid.link/20250430163758.3029367-1-kuba@kernel.org Signed-off-by: Jakub Kicinski --- drivers/net/virtio_net.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index 848fab51dfa1..b5549d542c02 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -3383,12 +3383,15 @@ static void __virtnet_rx_resume(struct virtnet_info *vi, bool refill) { bool running = netif_running(vi->dev); + bool schedule_refill = false; if (refill && !try_fill_recv(vi, rq, GFP_KERNEL)) - schedule_delayed_work(&vi->refill, 0); - + schedule_refill = true; if (running) virtnet_napi_enable(rq); + + if (schedule_refill) + schedule_delayed_work(&vi->refill, 0); } static void virtnet_rx_resume_all(struct virtnet_info *vi) @@ -3728,8 +3731,10 @@ static int virtnet_set_queues(struct virtnet_info *vi, u16 queue_pairs) succ: vi->curr_queue_pairs = queue_pairs; /* virtnet_open() will refill when device is going to up. */ - if (dev->flags & IFF_UP) + spin_lock_bh(&vi->refill_lock); + if (dev->flags & IFF_UP && vi->refill_enabled) schedule_delayed_work(&vi->refill, 0); + spin_unlock_bh(&vi->refill_lock); return 0; } -- cgit From 4397684a292a71fbc1e815c3e283f7490ddce5ae Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Wed, 30 Apr 2025 09:38:36 -0700 Subject: virtio-net: free xsk_buffs on error in virtnet_xsk_pool_enable() The selftests added to our CI by Bui Quang Minh recently reveals that there is a mem leak on the error path of virtnet_xsk_pool_enable(): unreferenced object 0xffff88800a68a000 (size 2048): comm "xdp_helper", pid 318, jiffies 4294692778 hex dump (first 32 bytes): 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ backtrace (crc 0): __kvmalloc_node_noprof+0x402/0x570 virtnet_xsk_pool_enable+0x293/0x6a0 (drivers/net/virtio_net.c:5882) xp_assign_dev+0x369/0x670 (net/xdp/xsk_buff_pool.c:226) xsk_bind+0x6a5/0x1ae0 __sys_bind+0x15e/0x230 __x64_sys_bind+0x72/0xb0 do_syscall_64+0xc1/0x1d0 entry_SYSCALL_64_after_hwframe+0x77/0x7f Acked-by: Jason Wang Fixes: e9f3962441c0 ("virtio_net: xsk: rx: support fill with xsk buffer") Link: https://patch.msgid.link/20250430163836.3029761-1-kuba@kernel.org Signed-off-by: Jakub Kicinski --- drivers/net/virtio_net.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index b5549d542c02..f9e3e628ec4d 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -5890,8 +5890,10 @@ static int virtnet_xsk_pool_enable(struct net_device *dev, hdr_dma = virtqueue_dma_map_single_attrs(sq->vq, &xsk_hdr, vi->hdr_len, DMA_TO_DEVICE, 0); - if (virtqueue_dma_mapping_error(sq->vq, hdr_dma)) - return -ENOMEM; + if (virtqueue_dma_mapping_error(sq->vq, hdr_dma)) { + err = -ENOMEM; + goto err_free_buffs; + } err = xsk_pool_dma_map(pool, dma_dev, 0); if (err) @@ -5919,6 +5921,8 @@ err_rq: err_xsk_map: virtqueue_dma_unmap_single_attrs(rq->vq, hdr_dma, vi->hdr_len, DMA_TO_DEVICE, 0); +err_free_buffs: + kvfree(rq->xsk_buffs); return err; } -- cgit From c360eb0c3ccb95306704fd221442283ee82f1f58 Mon Sep 17 00:00:00 2001 From: Andrew Lunn Date: Wed, 30 Apr 2025 11:21:35 -0500 Subject: dt-bindings: net: ethernet-controller: Add informative text about RGMII delays Device Tree and Ethernet MAC driver writers often misunderstand RGMII delays. Rewrite the Normative section in terms of the PCB, is the PCB adding the 2ns delay. This meaning was previous implied by the definition, but often wrongly interpreted due to the ambiguous wording and looking at the definition from the wrong perspective. The new definition concentrates clearly on the hardware, and should be less ambiguous. Add an Informative section to the end of the binding describing in detail what the four RGMII delays mean. This expands on just the PCB meaning, adding in the implications for the MAC and PHY. Additionally, when the MAC or PHY needs to add a delay, which is software configuration, describe how Linux does this, in the hope of reducing errors. Make it clear other users of device tree binding may implement the software configuration in other ways while still conforming to the binding. Fixes: 9d3de3c58347 ("dt-bindings: net: Add YAML schemas for the generic Ethernet options") Signed-off-by: Andrew Lunn Acked-by: Conor Dooley Link: https://patch.msgid.link/20250430-v6-15-rc3-net-rgmii-delays-v2-1-099ae651d5e5@lunn.ch Signed-off-by: Jakub Kicinski --- .../bindings/net/ethernet-controller.yaml | 97 ++++++++++++++++++++-- 1 file changed, 90 insertions(+), 7 deletions(-) diff --git a/Documentation/devicetree/bindings/net/ethernet-controller.yaml b/Documentation/devicetree/bindings/net/ethernet-controller.yaml index 45819b235800..a2d4c626f659 100644 --- a/Documentation/devicetree/bindings/net/ethernet-controller.yaml +++ b/Documentation/devicetree/bindings/net/ethernet-controller.yaml @@ -74,19 +74,17 @@ properties: - rev-rmii - moca - # RX and TX delays are added by the MAC when required + # RX and TX delays are provided by the PCB. See below - rgmii - # RGMII with internal RX and TX delays provided by the PHY, - # the MAC should not add the RX or TX delays in this case + # RX and TX delays are not provided by the PCB. This is the most + # frequent case. See below - rgmii-id - # RGMII with internal RX delay provided by the PHY, the MAC - # should not add an RX delay in this case + # TX delay is provided by the PCB. See below - rgmii-rxid - # RGMII with internal TX delay provided by the PHY, the MAC - # should not add an TX delay in this case + # RX delay is provided by the PCB. See below - rgmii-txid - rtbi - smii @@ -286,4 +284,89 @@ allOf: additionalProperties: true +# Informative +# =========== +# +# 'phy-modes' & 'phy-connection-type' properties 'rgmii', 'rgmii-id', +# 'rgmii-rxid', and 'rgmii-txid' are frequently used wrongly by +# developers. This informative section clarifies their usage. +# +# The RGMII specification requires a 2ns delay between the data and +# clock signals on the RGMII bus. How this delay is implemented is not +# specified. +# +# One option is to make the clock traces on the PCB longer than the +# data traces. A sufficiently difference in length can provide the 2ns +# delay. If both the RX and TX delays are implemented in this manner, +# 'rgmii' should be used, so indicating the PCB adds the delays. +# +# If the PCB does not add these delays via extra long traces, +# 'rgmii-id' should be used. Here, 'id' refers to 'internal delay', +# where either the MAC or PHY adds the delay. +# +# If only one of the two delays are implemented via extra long clock +# lines, either 'rgmii-rxid' or 'rgmii-txid' should be used, +# indicating the MAC or PHY should implement one of the delays +# internally, while the PCB implements the other delay. +# +# Device Tree describes hardware, and in this case, it describes the +# PCB between the MAC and the PHY, if the PCB implements delays or +# not. +# +# In practice, very few PCBs make use of extra long clock lines. Hence +# any RGMII phy mode other than 'rgmii-id' is probably wrong, and is +# unlikely to be accepted during review without details provided in +# the commit description and comments in the .dts file. +# +# When the PCB does not implement the delays, the MAC or PHY must. As +# such, this is software configuration, and so not described in Device +# Tree. +# +# The following describes how Linux implements the configuration of +# the MAC and PHY to add these delays when the PCB does not. As stated +# above, developers often get this wrong, and the aim of this section +# is reduce the frequency of these errors by Linux developers. Other +# users of the Device Tree may implement it differently, and still be +# consistent with both the normative and informative description +# above. +# +# By default in Linux, when using phylib/phylink, the MAC is expected +# to read the 'phy-mode' from Device Tree, not implement any delays, +# and pass the value to the PHY. The PHY will then implement delays as +# specified by the 'phy-mode'. The PHY should always be reconfigured +# to implement the needed delays, replacing any setting performed by +# strapping or the bootloader, etc. +# +# Experience to date is that all PHYs which implement RGMII also +# implement the ability to add or not add the needed delays. Hence +# this default is expected to work in all cases. Ignoring this default +# is likely to be questioned by Reviews, and require a strong argument +# to be accepted. +# +# There are a small number of cases where the MAC has hard coded +# delays which cannot be disabled. The 'phy-mode' only describes the +# PCB. The inability to disable the delays in the MAC does not change +# the meaning of 'phy-mode'. It does however mean that a 'phy-mode' of +# 'rgmii' is now invalid, it cannot be supported, since both the PCB +# and the MAC and PHY adding delays cannot result in a functional +# link. Thus the MAC should report a fatal error for any modes which +# cannot be supported. When the MAC implements the delay, it must +# ensure that the PHY does not also implement the same delay. So it +# must modify the phy-mode it passes to the PHY, removing the delay it +# has added. Failure to remove the delay will result in a +# non-functioning link. +# +# Sometimes there is a need to fine tune the delays. Often the MAC or +# PHY can perform this fine tuning. In the MAC node, the Device Tree +# properties 'rx-internal-delay-ps' and 'tx-internal-delay-ps' should +# be used to indicate fine tuning performed by the MAC. The values +# expected here are small. A value of 2000ps, i.e 2ns, and a phy-mode +# of 'rgmii' will not be accepted by Reviewers. +# +# If the PHY is to perform fine tuning, the properties +# 'rx-internal-delay-ps' and 'tx-internal-delay-ps' in the PHY node +# should be used. When the PHY is implementing delays, e.g. 'rgmii-id' +# these properties should have a value near to 2000ps. If the PCB is +# implementing delays, e.g. 'rgmii', a small value can be used to fine +# tune the delay added by the PCB. ... -- cgit From 3e6a0243ff002ddbd7ee18a8974ae61d2e6ed00d Mon Sep 17 00:00:00 2001 From: Guillaume Nault Date: Sat, 3 May 2025 00:57:52 +0200 Subject: gre: Fix again IPv6 link-local address generation. Use addrconf_addr_gen() to generate IPv6 link-local addresses on GRE devices in most cases and fall back to using add_v4_addrs() only in case the GRE configuration is incompatible with addrconf_addr_gen(). GRE used to use addrconf_addr_gen() until commit e5dd729460ca ("ip/ip6_gre: use the same logic as SIT interfaces when computing v6LL address") restricted this use to gretap and ip6gretap devices, and created add_v4_addrs() (borrowed from SIT) for non-Ethernet GRE ones. The original problem came when commit 9af28511be10 ("addrconf: refuse isatap eui64 for INADDR_ANY") made __ipv6_isatap_ifid() fail when its addr parameter was 0. The commit says that this would create an invalid address, however, I couldn't find any RFC saying that the generated interface identifier would be wrong. Anyway, since gre over IPv4 devices pass their local tunnel address to __ipv6_isatap_ifid(), that commit broke their IPv6 link-local address generation when the local address was unspecified. Then commit e5dd729460ca ("ip/ip6_gre: use the same logic as SIT interfaces when computing v6LL address") tried to fix that case by defining add_v4_addrs() and calling it to generate the IPv6 link-local address instead of using addrconf_addr_gen() (apart for gretap and ip6gretap devices, which would still use the regular addrconf_addr_gen(), since they have a MAC address). That broke several use cases because add_v4_addrs() isn't properly integrated into the rest of IPv6 Neighbor Discovery code. Several of these shortcomings have been fixed over time, but add_v4_addrs() remains broken on several aspects. In particular, it doesn't send any Router Sollicitations, so the SLAAC process doesn't start until the interface receives a Router Advertisement. Also, add_v4_addrs() mostly ignores the address generation mode of the interface (/proc/sys/net/ipv6/conf/*/addr_gen_mode), thus breaking the IN6_ADDR_GEN_MODE_RANDOM and IN6_ADDR_GEN_MODE_STABLE_PRIVACY cases. Fix the situation by using add_v4_addrs() only in the specific scenario where the normal method would fail. That is, for interfaces that have all of the following characteristics: * run over IPv4, * transport IP packets directly, not Ethernet (that is, not gretap interfaces), * tunnel endpoint is INADDR_ANY (that is, 0), * device address generation mode is EUI64. In all other cases, revert back to the regular addrconf_addr_gen(). Also, remove the special case for ip6gre interfaces in add_v4_addrs(), since ip6gre devices now always use addrconf_addr_gen() instead. Note: This patch was originally applied as commit 183185a18ff9 ("gre: Fix IPv6 link-local address generation."). However, it was then reverted by commit fc486c2d060f ("Revert "gre: Fix IPv6 link-local address generation."") because it uncovered another bug that ended up breaking net/forwarding/ip6gre_custom_multipath_hash.sh. That other bug has now been fixed by commit 4d0ab3a6885e ("ipv6: Start path selection from the first nexthop"). Therefore we can now revive this GRE patch (no changes since original commit 183185a18ff9 ("gre: Fix IPv6 link-local address generation."). Fixes: e5dd729460ca ("ip/ip6_gre: use the same logic as SIT interfaces when computing v6LL address") Signed-off-by: Guillaume Nault Reviewed-by: Ido Schimmel Link: https://patch.msgid.link/a88cc5c4811af36007645d610c95102dccb360a6.1746225214.git.gnault@redhat.com Signed-off-by: Jakub Kicinski --- net/ipv6/addrconf.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c index 9ba83f0c9928..c6b22170dc49 100644 --- a/net/ipv6/addrconf.c +++ b/net/ipv6/addrconf.c @@ -3214,16 +3214,13 @@ static void add_v4_addrs(struct inet6_dev *idev) struct in6_addr addr; struct net_device *dev; struct net *net = dev_net(idev->dev); - int scope, plen, offset = 0; + int scope, plen; u32 pflags = 0; ASSERT_RTNL(); memset(&addr, 0, sizeof(struct in6_addr)); - /* in case of IP6GRE the dev_addr is an IPv6 and therefore we use only the last 4 bytes */ - if (idev->dev->addr_len == sizeof(struct in6_addr)) - offset = sizeof(struct in6_addr) - 4; - memcpy(&addr.s6_addr32[3], idev->dev->dev_addr + offset, 4); + memcpy(&addr.s6_addr32[3], idev->dev->dev_addr, 4); if (!(idev->dev->flags & IFF_POINTOPOINT) && idev->dev->type == ARPHRD_SIT) { scope = IPV6_ADDR_COMPATv4; @@ -3534,7 +3531,13 @@ static void addrconf_gre_config(struct net_device *dev) return; } - if (dev->type == ARPHRD_ETHER) { + /* Generate the IPv6 link-local address using addrconf_addr_gen(), + * unless we have an IPv4 GRE device not bound to an IP address and + * which is in EUI64 mode (as __ipv6_isatap_ifid() would fail in this + * case). Such devices fall back to add_v4_addrs() instead. + */ + if (!(dev->type == ARPHRD_IPGRE && *(__be32 *)dev->dev_addr == 0 && + idev->cnf.addr_gen_mode == IN6_ADDR_GEN_MODE_EUI64)) { addrconf_addr_gen(idev, true); return; } -- cgit From b6a6006b0e3d10e62c465613f07ff49268baedb1 Mon Sep 17 00:00:00 2001 From: Guillaume Nault Date: Sat, 3 May 2025 00:57:59 +0200 Subject: selftests: Add IPv6 link-local address generation tests for GRE devices. GRE devices have their special code for IPv6 link-local address generation that has been the source of several regressions in the past. Add selftest to check that all gre, ip6gre, gretap and ip6gretap get an IPv6 link-link local address in accordance with the net.ipv6.conf..addr_gen_mode sysctl. Note: This patch was originally applied as commit 6f50175ccad4 ("selftests: Add IPv6 link-local address generation tests for GRE devices."). However, it was then reverted by commit 355d940f4d5a ("Revert "selftests: Add IPv6 link-local address generation tests for GRE devices."") because the commit it depended on was going to be reverted. Now that the situation is resolved, we can add this selftest again (no changes since original patch, appart from context update in tools/testing/selftests/net/Makefile). Signed-off-by: Guillaume Nault Reviewed-by: Ido Schimmel Tested-by: Ido Schimmel Link: https://patch.msgid.link/2c3a5733cb3a6e3119504361a9b9f89fda570a2d.1746225214.git.gnault@redhat.com Signed-off-by: Jakub Kicinski --- tools/testing/selftests/net/Makefile | 1 + tools/testing/selftests/net/gre_ipv6_lladdr.sh | 177 +++++++++++++++++++++++++ 2 files changed, 178 insertions(+) create mode 100755 tools/testing/selftests/net/gre_ipv6_lladdr.sh diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 124078b56fa4..70a38f485d4d 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -31,6 +31,7 @@ TEST_PROGS += veth.sh TEST_PROGS += ioam6.sh TEST_PROGS += gro.sh TEST_PROGS += gre_gso.sh +TEST_PROGS += gre_ipv6_lladdr.sh TEST_PROGS += cmsg_so_mark.sh TEST_PROGS += cmsg_so_priority.sh TEST_PROGS += test_so_rcv.sh diff --git a/tools/testing/selftests/net/gre_ipv6_lladdr.sh b/tools/testing/selftests/net/gre_ipv6_lladdr.sh new file mode 100755 index 000000000000..5b34f6e1f831 --- /dev/null +++ b/tools/testing/selftests/net/gre_ipv6_lladdr.sh @@ -0,0 +1,177 @@ +#!/bin/bash +# SPDX-License-Identifier: GPL-2.0 + +source ./lib.sh + +PAUSE_ON_FAIL="no" + +# The trap function handler +# +exit_cleanup_all() +{ + cleanup_all_ns + + exit "${EXIT_STATUS}" +} + +# Add fake IPv4 and IPv6 networks on the loopback device, to be used as +# underlay by future GRE devices. +# +setup_basenet() +{ + ip -netns "${NS0}" link set dev lo up + ip -netns "${NS0}" address add dev lo 192.0.2.10/24 + ip -netns "${NS0}" address add dev lo 2001:db8::10/64 nodad +} + +# Check if network device has an IPv6 link-local address assigned. +# +# Parameters: +# +# * $1: The network device to test +# * $2: An extra regular expression that should be matched (to verify the +# presence of extra attributes) +# * $3: The expected return code from grep (to allow checking the absence of +# a link-local address) +# * $4: The user visible name for the scenario being tested +# +check_ipv6_ll_addr() +{ + local DEV="$1" + local EXTRA_MATCH="$2" + local XRET="$3" + local MSG="$4" + + RET=0 + set +e + ip -netns "${NS0}" -6 address show dev "${DEV}" scope link | grep "fe80::" | grep -q "${EXTRA_MATCH}" + check_err_fail "${XRET}" $? "" + log_test "${MSG}" + set -e +} + +# Create a GRE device and verify that it gets an IPv6 link-local address as +# expected. +# +# Parameters: +# +# * $1: The device type (gre, ip6gre, gretap or ip6gretap) +# * $2: The local underlay IP address (can be an IPv4, an IPv6 or "any") +# * $3: The remote underlay IP address (can be an IPv4, an IPv6 or "any") +# * $4: The IPv6 interface identifier generation mode to use for the GRE +# device (eui64, none, stable-privacy or random). +# +test_gre_device() +{ + local GRE_TYPE="$1" + local LOCAL_IP="$2" + local REMOTE_IP="$3" + local MODE="$4" + local ADDR_GEN_MODE + local MATCH_REGEXP + local MSG + + ip link add netns "${NS0}" name gretest type "${GRE_TYPE}" local "${LOCAL_IP}" remote "${REMOTE_IP}" + + case "${MODE}" in + "eui64") + ADDR_GEN_MODE=0 + MATCH_REGEXP="" + MSG="${GRE_TYPE}, mode: 0 (EUI64), ${LOCAL_IP} -> ${REMOTE_IP}" + XRET=0 + ;; + "none") + ADDR_GEN_MODE=1 + MATCH_REGEXP="" + MSG="${GRE_TYPE}, mode: 1 (none), ${LOCAL_IP} -> ${REMOTE_IP}" + XRET=1 # No link-local address should be generated + ;; + "stable-privacy") + ADDR_GEN_MODE=2 + MATCH_REGEXP="stable-privacy" + MSG="${GRE_TYPE}, mode: 2 (stable privacy), ${LOCAL_IP} -> ${REMOTE_IP}" + XRET=0 + # Initialise stable_secret (required for stable-privacy mode) + ip netns exec "${NS0}" sysctl -qw net.ipv6.conf.gretest.stable_secret="2001:db8::abcd" + ;; + "random") + ADDR_GEN_MODE=3 + MATCH_REGEXP="stable-privacy" + MSG="${GRE_TYPE}, mode: 3 (random), ${LOCAL_IP} -> ${REMOTE_IP}" + XRET=0 + ;; + esac + + # Check that IPv6 link-local address is generated when device goes up + ip netns exec "${NS0}" sysctl -qw net.ipv6.conf.gretest.addr_gen_mode="${ADDR_GEN_MODE}" + ip -netns "${NS0}" link set dev gretest up + check_ipv6_ll_addr gretest "${MATCH_REGEXP}" "${XRET}" "config: ${MSG}" + + # Now disable link-local address generation + ip -netns "${NS0}" link set dev gretest down + ip netns exec "${NS0}" sysctl -qw net.ipv6.conf.gretest.addr_gen_mode=1 + ip -netns "${NS0}" link set dev gretest up + + # Check that link-local address generation works when re-enabled while + # the device is already up + ip netns exec "${NS0}" sysctl -qw net.ipv6.conf.gretest.addr_gen_mode="${ADDR_GEN_MODE}" + check_ipv6_ll_addr gretest "${MATCH_REGEXP}" "${XRET}" "update: ${MSG}" + + ip -netns "${NS0}" link del dev gretest +} + +test_gre4() +{ + local GRE_TYPE + local MODE + + for GRE_TYPE in "gre" "gretap"; do + printf "\n####\nTesting IPv6 link-local address generation on ${GRE_TYPE} devices\n####\n\n" + + for MODE in "eui64" "none" "stable-privacy" "random"; do + test_gre_device "${GRE_TYPE}" 192.0.2.10 192.0.2.11 "${MODE}" + test_gre_device "${GRE_TYPE}" any 192.0.2.11 "${MODE}" + test_gre_device "${GRE_TYPE}" 192.0.2.10 any "${MODE}" + done + done +} + +test_gre6() +{ + local GRE_TYPE + local MODE + + for GRE_TYPE in "ip6gre" "ip6gretap"; do + printf "\n####\nTesting IPv6 link-local address generation on ${GRE_TYPE} devices\n####\n\n" + + for MODE in "eui64" "none" "stable-privacy" "random"; do + test_gre_device "${GRE_TYPE}" 2001:db8::10 2001:db8::11 "${MODE}" + test_gre_device "${GRE_TYPE}" any 2001:db8::11 "${MODE}" + test_gre_device "${GRE_TYPE}" 2001:db8::10 any "${MODE}" + done + done +} + +usage() +{ + echo "Usage: $0 [-p]" + exit 1 +} + +while getopts :p o +do + case $o in + p) PAUSE_ON_FAIL="yes";; + *) usage;; + esac +done + +setup_ns NS0 + +set -e +trap exit_cleanup_all EXIT + +setup_basenet + +test_gre4 +test_gre6 -- cgit From b344a48cbe5fb24697addc6afbb7358b938a7d31 Mon Sep 17 00:00:00 2001 From: Mohsin Bashir Date: Fri, 2 May 2025 18:35:16 -0700 Subject: selftests: drv: net: fix test failure on ipv6 sys The `get_interface_info` call has ip version hard-coded which leads to failures on an IPV6 system. The NetDrvEnv class already gathers information about remote interface, so instead of fixing the local implementation switch to using cfg.remote_ifname. Before: ./drivers/net/ping.py Traceback (most recent call last): File "/new_tests/./drivers/net/ping.py", line 217, in main() File "/new_tests/./drivers/net/ping.py", line 204, in main get_interface_info(cfg) File "/new_tests/./drivers/net/ping.py", line 128, in get_interface_info raise KsftFailEx('Can not get remote interface') net.lib.py.ksft.KsftFailEx: Can not get remote interface After: ./drivers/net/ping.py TAP version 13 1..6 ok 1 ping.test_default # SKIP Test requires IPv4 connectivity ok 2 ping.test_xdp_generic_sb # SKIP Test requires IPv4 connectivity ok 3 ping.test_xdp_generic_mb # SKIP Test requires IPv4 connectivity ok 4 ping.test_xdp_native_sb # SKIP Test requires IPv4 connectivity ok 5 ping.test_xdp_native_mb # SKIP Test requires IPv4 connectivity ok 6 ping.test_xdp_offload # SKIP device does not support offloaded XDP Totals: pass:0 fail:0 xfail:0 xpass:0 skip:6 error:0 Fixes: 75cc19c8ff89 ("selftests: drv-net: add xdp cases for ping.py") Signed-off-by: Mohsin Bashir Reviewed-by: David Wei Link: https://patch.msgid.link/20250503013518.1722913-2-mohsin.bashr@gmail.com Signed-off-by: Jakub Kicinski --- tools/testing/selftests/drivers/net/ping.py | 22 +++++++++------------- 1 file changed, 9 insertions(+), 13 deletions(-) diff --git a/tools/testing/selftests/drivers/net/ping.py b/tools/testing/selftests/drivers/net/ping.py index 4b6822866066..5272e8b3536d 100755 --- a/tools/testing/selftests/drivers/net/ping.py +++ b/tools/testing/selftests/drivers/net/ping.py @@ -9,7 +9,6 @@ from lib.py import EthtoolFamily, NetDrvEpEnv from lib.py import bkg, cmd, wait_port_listen, rand_port from lib.py import defer, ethtool, ip -remote_ifname="" no_sleep=False def _test_v4(cfg) -> None: @@ -57,7 +56,7 @@ def _set_offload_checksum(cfg, netnl, on) -> None: def _set_xdp_generic_sb_on(cfg) -> None: prog = cfg.net_lib_dir / "xdp_dummy.bpf.o" - cmd(f"ip link set dev {remote_ifname} mtu 1500", shell=True, host=cfg.remote) + cmd(f"ip link set dev {cfg.remote_ifname} mtu 1500", shell=True, host=cfg.remote) cmd(f"ip link set dev {cfg.ifname} mtu 1500 xdpgeneric obj {prog} sec xdp", shell=True) defer(cmd, f"ip link set dev {cfg.ifname} xdpgeneric off") @@ -66,8 +65,8 @@ def _set_xdp_generic_sb_on(cfg) -> None: def _set_xdp_generic_mb_on(cfg) -> None: prog = cfg.net_lib_dir / "xdp_dummy.bpf.o" - cmd(f"ip link set dev {remote_ifname} mtu 9000", shell=True, host=cfg.remote) - defer(ip, f"link set dev {remote_ifname} mtu 1500", host=cfg.remote) + cmd(f"ip link set dev {cfg.remote_ifname} mtu 9000", shell=True, host=cfg.remote) + defer(ip, f"link set dev {cfg.remote_ifname} mtu 1500", host=cfg.remote) ip("link set dev %s mtu 9000 xdpgeneric obj %s sec xdp.frags" % (cfg.ifname, prog)) defer(ip, f"link set dev {cfg.ifname} mtu 1500 xdpgeneric off") @@ -76,7 +75,7 @@ def _set_xdp_generic_mb_on(cfg) -> None: def _set_xdp_native_sb_on(cfg) -> None: prog = cfg.net_lib_dir / "xdp_dummy.bpf.o" - cmd(f"ip link set dev {remote_ifname} mtu 1500", shell=True, host=cfg.remote) + cmd(f"ip link set dev {cfg.remote_ifname} mtu 1500", shell=True, host=cfg.remote) cmd(f"ip -j link set dev {cfg.ifname} mtu 1500 xdp obj {prog} sec xdp", shell=True) defer(ip, f"link set dev {cfg.ifname} mtu 1500 xdp off") xdp_info = ip("-d link show %s" % (cfg.ifname), json=True)[0] @@ -93,8 +92,8 @@ def _set_xdp_native_sb_on(cfg) -> None: def _set_xdp_native_mb_on(cfg) -> None: prog = cfg.net_lib_dir / "xdp_dummy.bpf.o" - cmd(f"ip link set dev {remote_ifname} mtu 9000", shell=True, host=cfg.remote) - defer(ip, f"link set dev {remote_ifname} mtu 1500", host=cfg.remote) + cmd(f"ip link set dev {cfg.remote_ifname} mtu 9000", shell=True, host=cfg.remote) + defer(ip, f"link set dev {cfg.remote_ifname} mtu 1500", host=cfg.remote) try: cmd(f"ip link set dev {cfg.ifname} mtu 9000 xdp obj {prog} sec xdp.frags", shell=True) defer(ip, f"link set dev {cfg.ifname} mtu 1500 xdp off") @@ -112,18 +111,15 @@ def _set_xdp_offload_on(cfg) -> None: except Exception as e: raise KsftSkipEx('device does not support offloaded XDP') defer(ip, f"link set dev {cfg.ifname} xdpoffload off") - cmd(f"ip link set dev {remote_ifname} mtu 1500", shell=True, host=cfg.remote) + cmd(f"ip link set dev {cfg.remote_ifname} mtu 1500", shell=True, host=cfg.remote) if no_sleep != True: time.sleep(10) def get_interface_info(cfg) -> None: - global remote_ifname global no_sleep - remote_info = cmd(f"ip -4 -o addr show to {cfg.remote_addr_v['4']} | awk '{{print $2}}'", shell=True, host=cfg.remote).stdout - remote_ifname = remote_info.rstrip('\n') - if remote_ifname == "": + if cfg.remote_ifname == "": raise KsftFailEx('Can not get remote interface') local_info = ip("-d link show %s" % (cfg.ifname), json=True)[0] if 'parentbus' in local_info and local_info['parentbus'] == "netdevsim": @@ -136,7 +132,7 @@ def set_interface_init(cfg) -> None: cmd(f"ip link set dev {cfg.ifname} xdp off ", shell=True) cmd(f"ip link set dev {cfg.ifname} xdpgeneric off ", shell=True) cmd(f"ip link set dev {cfg.ifname} xdpoffload off", shell=True) - cmd(f"ip link set dev {remote_ifname} mtu 1500", shell=True, host=cfg.remote) + cmd(f"ip link set dev {cfg.remote_ifname} mtu 1500", shell=True, host=cfg.remote) def test_default(cfg, netnl) -> None: _set_offload_checksum(cfg, netnl, "off") -- cgit From 8bb7d8e5cf7f44ea89a445975cbd78888324f234 Mon Sep 17 00:00:00 2001 From: Mohsin Bashir Date: Fri, 2 May 2025 18:35:17 -0700 Subject: selftests: drv: net: avoid skipping tests On a system with either of the ipv4 or ipv6 information missing, tests are currently skipped. Ideally, the test should run as long as at least one address family is present. This patch make test run whenever possible. Before: ./drivers/net/ping.py TAP version 13 1..6 ok 1 ping.test_default # SKIP Test requires IPv4 connectivity ok 2 ping.test_xdp_generic_sb # SKIP Test requires IPv4 connectivity ok 3 ping.test_xdp_generic_mb # SKIP Test requires IPv4 connectivity ok 4 ping.test_xdp_native_sb # SKIP Test requires IPv4 connectivity ok 5 ping.test_xdp_native_mb # SKIP Test requires IPv4 connectivity ok 6 ping.test_xdp_offload # SKIP device does not support offloaded XDP Totals: pass:0 fail:0 xfail:0 xpass:0 skip:6 error:0 After: ./drivers/net/ping.py TAP version 13 1..6 ok 1 ping.test_default ok 2 ping.test_xdp_generic_sb ok 3 ping.test_xdp_generic_mb ok 4 ping.test_xdp_native_sb ok 5 ping.test_xdp_native_mb ok 6 ping.test_xdp_offload # SKIP device does not support offloaded XDP Totals: pass:5 fail:0 xfail:0 xpass:0 skip:1 error:0 Fixes: 75cc19c8ff89 ("selftests: drv-net: add xdp cases for ping.py") Signed-off-by: Mohsin Bashir Link: https://patch.msgid.link/20250503013518.1722913-3-mohsin.bashr@gmail.com Signed-off-by: Jakub Kicinski --- tools/testing/selftests/drivers/net/ping.py | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/tools/testing/selftests/drivers/net/ping.py b/tools/testing/selftests/drivers/net/ping.py index 5272e8b3536d..16b7d3ab0fc8 100755 --- a/tools/testing/selftests/drivers/net/ping.py +++ b/tools/testing/selftests/drivers/net/ping.py @@ -12,7 +12,8 @@ from lib.py import defer, ethtool, ip no_sleep=False def _test_v4(cfg) -> None: - cfg.require_ipver("4") + if not cfg.addr_v["4"]: + return cmd("ping -c 1 -W0.5 " + cfg.remote_addr_v["4"]) cmd("ping -c 1 -W0.5 " + cfg.addr_v["4"], host=cfg.remote) @@ -20,7 +21,8 @@ def _test_v4(cfg) -> None: cmd("ping -s 65000 -c 1 -W0.5 " + cfg.addr_v["4"], host=cfg.remote) def _test_v6(cfg) -> None: - cfg.require_ipver("6") + if not cfg.addr_v["6"]: + return cmd("ping -c 1 -W5 " + cfg.remote_addr_v["6"]) cmd("ping -c 1 -W5 " + cfg.addr_v["6"], host=cfg.remote) -- cgit From 4a9d494ca24b7fd509b3953fcb43d580cb05097b Mon Sep 17 00:00:00 2001 From: Mohsin Bashir Date: Fri, 2 May 2025 18:35:18 -0700 Subject: selftests: drv: net: add version indicator Currently, the test result does not differentiate between the cases when either one of the address families are configured or if both the address families are configured. Ideally, the result should report if a particular case was skipped. ./drivers/net/ping.py TAP version 13 1..7 ok 1 ping.test_default_v4 # SKIP Test requires IPv4 connectivity ok 2 ping.test_default_v6 ok 3 ping.test_xdp_generic_sb ok 4 ping.test_xdp_generic_mb ok 5 ping.test_xdp_native_sb ok 6 ping.test_xdp_native_mb ok 7 ping.test_xdp_offload # SKIP device does not support offloaded XDP Totals: pass:5 fail:0 xfail:0 xpass:0 skip:2 error:0 Fixes: 75cc19c8ff89 ("selftests: drv-net: add xdp cases for ping.py") Signed-off-by: Mohsin Bashir Reviewed-by: David Wei Link: https://patch.msgid.link/20250503013518.1722913-4-mohsin.bashr@gmail.com Signed-off-by: Jakub Kicinski --- tools/testing/selftests/drivers/net/ping.py | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) diff --git a/tools/testing/selftests/drivers/net/ping.py b/tools/testing/selftests/drivers/net/ping.py index 16b7d3ab0fc8..af8df2313a3b 100755 --- a/tools/testing/selftests/drivers/net/ping.py +++ b/tools/testing/selftests/drivers/net/ping.py @@ -136,13 +136,23 @@ def set_interface_init(cfg) -> None: cmd(f"ip link set dev {cfg.ifname} xdpoffload off", shell=True) cmd(f"ip link set dev {cfg.remote_ifname} mtu 1500", shell=True, host=cfg.remote) -def test_default(cfg, netnl) -> None: +def test_default_v4(cfg, netnl) -> None: + cfg.require_ipver("4") + _set_offload_checksum(cfg, netnl, "off") _test_v4(cfg) - _test_v6(cfg) _test_tcp(cfg) _set_offload_checksum(cfg, netnl, "on") _test_v4(cfg) + _test_tcp(cfg) + +def test_default_v6(cfg, netnl) -> None: + cfg.require_ipver("6") + + _set_offload_checksum(cfg, netnl, "off") + _test_v6(cfg) + _test_tcp(cfg) + _set_offload_checksum(cfg, netnl, "on") _test_v6(cfg) _test_tcp(cfg) @@ -200,7 +210,8 @@ def main() -> None: with NetDrvEpEnv(__file__) as cfg: get_interface_info(cfg) set_interface_init(cfg) - ksft_run([test_default, + ksft_run([test_default_v4, + test_default_v6, test_xdp_generic_sb, test_xdp_generic_mb, test_xdp_native_sb, -- cgit From 4720f9707c783f642332dee3d56dccaefa850e42 Mon Sep 17 00:00:00 2001 From: David Wei Date: Fri, 2 May 2025 21:30:50 -0700 Subject: tools: ynl-gen: validate 0 len strings from kernel Strings from the kernel are guaranteed to be null terminated and ynl_attr_validate() checks for this. But it doesn't check if the string has a len of 0, which would cause problems when trying to access data[len - 1]. Fix this by checking that len is positive. Signed-off-by: David Wei Link: https://patch.msgid.link/20250503043050.861238-1-dw@davidwei.uk Signed-off-by: Jakub Kicinski --- tools/net/ynl/lib/ynl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tools/net/ynl/lib/ynl.c b/tools/net/ynl/lib/ynl.c index ce32cb35007d..c4da34048ef8 100644 --- a/tools/net/ynl/lib/ynl.c +++ b/tools/net/ynl/lib/ynl.c @@ -364,7 +364,7 @@ int ynl_attr_validate(struct ynl_parse_arg *yarg, const struct nlattr *attr) "Invalid attribute (binary %s)", policy->name); return -1; case YNL_PT_NUL_STR: - if ((!policy->len || len <= policy->len) && !data[len - 1]) + if (len && (!policy->len || len <= policy->len) && !data[len - 1]) break; yerr(yarg->ys, YNL_ERROR_ATTR_INVALID, "Invalid attribute (string %s)", policy->name); -- cgit From 4db6c75124d871fbabf8243f947d34cc7e0697fc Mon Sep 17 00:00:00 2001 From: Daniel Golle Date: Mon, 5 May 2025 02:07:32 +0100 Subject: net: ethernet: mtk_eth_soc: reset all TX queues on DMA free The purpose of resetting the TX queue is to reset the byte and packet count as well as to clear the software flow control XOFF bit. MediaTek developers pointed out that netdev_reset_queue would only resets queue 0 of the network device. Queues that are not reset may cause unexpected issues. Packets may stop being sent after reset and "transmit timeout" log may be displayed. Import fix from MediaTek's SDK to resolve this issue. Link: https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/319c0d9905579a46dc448579f892f364f1f84818 Fixes: f63959c7eec31 ("net: ethernet: mtk_eth_soc: implement multi-queue support for per-port queues") Signed-off-by: Daniel Golle Link: https://patch.msgid.link/c9ff9adceac4f152239a0f65c397f13547639175.1746406763.git.daniel@makrotopia.org Signed-off-by: Paolo Abeni --- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.c b/drivers/net/ethernet/mediatek/mtk_eth_soc.c index 8fda4ce80d81..53c39561b6d9 100644 --- a/drivers/net/ethernet/mediatek/mtk_eth_soc.c +++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.c @@ -3186,11 +3186,19 @@ static int mtk_dma_init(struct mtk_eth *eth) static void mtk_dma_free(struct mtk_eth *eth) { const struct mtk_soc_data *soc = eth->soc; - int i; + int i, j, txqs = 1; + + if (MTK_HAS_CAPS(eth->soc->caps, MTK_QDMA)) + txqs = MTK_QDMA_NUM_QUEUES; + + for (i = 0; i < MTK_MAX_DEVS; i++) { + if (!eth->netdev[i]) + continue; + + for (j = 0; j < txqs; j++) + netdev_tx_reset_subqueue(eth->netdev[i], j); + } - for (i = 0; i < MTK_MAX_DEVS; i++) - if (eth->netdev[i]) - netdev_reset_queue(eth->netdev[i]); if (!MTK_HAS_CAPS(soc->caps, MTK_SRAM) && eth->scratch_ring) { dma_free_coherent(eth->dma_dev, MTK_QDMA_RING_SIZE * soc->tx.desc_size, -- cgit From e8716b5b0dff1b3d523b4a83fd5e94d57b887c5c Mon Sep 17 00:00:00 2001 From: Frank Wunderlich Date: Mon, 5 May 2025 02:07:58 +0100 Subject: net: ethernet: mtk_eth_soc: do not reset PSE when setting FE Remove redundant PSE reset. When setting FE register there is no need to reset PSE, doing so may cause FE to work abnormal. Link: https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/3a5223473e086a4b54a2b9a44df7d9ddcc2bc75a Fixes: dee4dd10c79aa ("net: ethernet: mtk_eth_soc: ppe: add support for multiple PPEs") Signed-off-by: Frank Wunderlich Link: https://patch.msgid.link/18f0ac7d83f82defa3342c11ef0d1362f6b81e88.1746406763.git.daniel@makrotopia.org Signed-off-by: Paolo Abeni --- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.c b/drivers/net/ethernet/mediatek/mtk_eth_soc.c index 53c39561b6d9..22a532695fb0 100644 --- a/drivers/net/ethernet/mediatek/mtk_eth_soc.c +++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.c @@ -3473,9 +3473,6 @@ static int mtk_open(struct net_device *dev) } mtk_gdm_config(eth, target_mac->id, gdm_config); } - /* Reset and enable PSE */ - mtk_w32(eth, RST_GL_PSE, MTK_RST_GL); - mtk_w32(eth, 0, MTK_RST_GL); napi_enable(ð->tx_napi); napi_enable(ð->rx_napi); -- cgit From dcaeeb8ae84c5506ebc574732838264f3887738c Mon Sep 17 00:00:00 2001 From: Antonios Salios Date: Fri, 25 Apr 2025 13:17:45 +0200 Subject: can: m_can: m_can_class_allocate_dev(): initialize spin lock on device probe The spin lock tx_handling_spinlock in struct m_can_classdev is not being initialized. This leads the following spinlock bad magic complaint from the kernel, eg. when trying to send CAN frames with cansend from can-utils: | BUG: spinlock bad magic on CPU#0, cansend/95 | lock: 0xff60000002ec1010, .magic: 00000000, .owner: /-1, .owner_cpu: 0 | CPU: 0 UID: 0 PID: 95 Comm: cansend Not tainted 6.15.0-rc3-00032-ga79be02bba5c #5 NONE | Hardware name: MachineWare SIM-V (DT) | Call Trace: | [] dump_backtrace+0x1c/0x24 | [] show_stack+0x28/0x34 | [] dump_stack_lvl+0x4a/0x68 | [] dump_stack+0x14/0x1c | [] spin_dump+0x62/0x6e | [] do_raw_spin_lock+0xd0/0x142 | [] _raw_spin_lock_irqsave+0x20/0x2c | [] m_can_start_xmit+0x90/0x34a | [] dev_hard_start_xmit+0xa6/0xee | [] sch_direct_xmit+0x114/0x292 | [] __dev_queue_xmit+0x3b0/0xaa8 | [] can_send+0xc6/0x242 | [] raw_sendmsg+0x1a8/0x36c | [] sock_write_iter+0x9a/0xee | [] vfs_write+0x184/0x3a6 | [] ksys_write+0xa0/0xc0 | [] __riscv_sys_write+0x14/0x1c | [] do_trap_ecall_u+0x168/0x212 | [] handle_exception+0x146/0x152 Initializing the spin lock in m_can_class_allocate_dev solves that problem. Fixes: 1fa80e23c150 ("can: m_can: Introduce a tx_fifo_in_flight counter") Signed-off-by: Antonios Salios Reviewed-by: Vincent Mailhol Link: https://patch.msgid.link/20250425111744.37604-2-antonios@mwa.re Reviewed-by: Markus Schneider-Pargmann Signed-off-by: Marc Kleine-Budde --- drivers/net/can/m_can/m_can.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/net/can/m_can/m_can.c b/drivers/net/can/m_can/m_can.c index 884a6352c42b..326ede9d400f 100644 --- a/drivers/net/can/m_can/m_can.c +++ b/drivers/net/can/m_can/m_can.c @@ -2379,6 +2379,7 @@ struct m_can_classdev *m_can_class_allocate_dev(struct device *dev, SET_NETDEV_DEV(net_dev, dev); m_can_of_parse_mram(class_dev, mram_config_vals); + spin_lock_init(&class_dev->tx_handling_spinlock); out: return class_dev; } -- cgit From 5e1663810e11c64956aa7e280cf74b2f3284d816 Mon Sep 17 00:00:00 2001 From: Kelsey Maes Date: Wed, 30 Apr 2025 09:15:01 -0700 Subject: can: mcp251xfd: fix TDC setting for low data bit rates The TDC is currently hardcoded enabled. This means that even for lower CAN-FD data bitrates (with a DBRP (data bitrate prescaler) > 2) a TDC is configured. This leads to a bus-off condition. ISO 11898-1 section 11.3.3 says "Transmitter delay compensation" (TDC) is only applicable if DBRP is 1 or 2. To fix the problem, switch the driver to use the TDC calculation provided by the CAN driver framework (which respects ISO 11898-1 section 11.3.3). This has the positive side effect that userspace can control TDC as needed. Demonstration of the feature in action: | $ ip link set can0 up type can bitrate 125000 dbitrate 500000 fd on | $ ip -details link show can0 | 3: can0: mtu 72 qdisc pfifo_fast state UP mode DEFAULT group default qlen 10 | link/can promiscuity 0 allmulti 0 minmtu 0 maxmtu 0 | can state ERROR-ACTIVE (berr-counter tx 0 rx 0) restart-ms 0 | bitrate 125000 sample-point 0.875 | tq 50 prop-seg 69 phase-seg1 70 phase-seg2 20 sjw 10 brp 2 | mcp251xfd: tseg1 2..256 tseg2 1..128 sjw 1..128 brp 1..256 brp_inc 1 | dbitrate 500000 dsample-point 0.875 | dtq 125 dprop-seg 6 dphase-seg1 7 dphase-seg2 2 dsjw 1 dbrp 5 | mcp251xfd: dtseg1 1..32 dtseg2 1..16 dsjw 1..16 dbrp 1..256 dbrp_inc 1 | tdcv 0..63 tdco 0..63 | clock 40000000 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535 tso_max_size 65536 tso_max_segs 65535 gro_max_size 65536 parentbus spi parentdev spi0.0 | $ ip link set can0 up type can bitrate 1000000 dbitrate 4000000 fd on | $ ip -details link show can0 | 3: can0: mtu 72 qdisc pfifo_fast state UP mode DEFAULT group default qlen 10 | link/can promiscuity 0 allmulti 0 minmtu 0 maxmtu 0 | can state ERROR-ACTIVE (berr-counter tx 0 rx 0) restart-ms 0 | bitrate 1000000 sample-point 0.750 | tq 25 prop-seg 14 phase-seg1 15 phase-seg2 10 sjw 5 brp 1 | mcp251xfd: tseg1 2..256 tseg2 1..128 sjw 1..128 brp 1..256 brp_inc 1 | dbitrate 4000000 dsample-point 0.700 | dtq 25 dprop-seg 3 dphase-seg1 3 dphase-seg2 3 dsjw 1 dbrp 1 | tdco 7 | mcp251xfd: dtseg1 1..32 dtseg2 1..16 dsjw 1..16 dbrp 1..256 dbrp_inc 1 | tdcv 0..63 tdco 0..63 | clock 40000000 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535 tso_max_size 65536 tso_max_segs 65535 gro_max_size 65536 parentbus spi parentdev spi0.0 There has been some confusion about the MCP2518FD using a relative or absolute TDCO due to the datasheet specifying a range of [-64,63]. I have a custom board with a 40 MHz clock and an estimated loop delay of 100 to 216 ns. During testing at a data bit rate of 4 Mbit/s I found that using can_get_relative_tdco() resulted in bus-off errors. The final TDCO value was 1 which corresponds to a 10% SSP in an absolute configuration. This behavior is expected if the TDCO value is really absolute and not relative. Using priv->can.tdc.tdco instead results in a final TDCO of 8, setting the SSP at exactly 80%. This configuration works. The automatic, manual, and off TDC modes were tested at speeds up to, and including, 8 Mbit/s on real hardware and behave as expected. Fixes: 55e5b97f003e ("can: mcp25xxfd: add driver for Microchip MCP25xxFD SPI CAN") Reported-by: Kelsey Maes Closes: https://lore.kernel.org/all/C2121586-C87F-4B23-A933-845362C29CA1@vpprocess.com Reviewed-by: Vincent Mailhol Signed-off-by: Kelsey Maes Link: https://patch.msgid.link/20250430161501.79370-1-kelsey@vpprocess.com [mkl: add comment] Signed-off-by: Marc Kleine-Budde --- drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c | 40 ++++++++++++++++++++------ 1 file changed, 32 insertions(+), 8 deletions(-) diff --git a/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c b/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c index 3bc56517fe7a..064d81c724f4 100644 --- a/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c +++ b/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c @@ -75,6 +75,24 @@ static const struct can_bittiming_const mcp251xfd_data_bittiming_const = { .brp_inc = 1, }; +/* The datasheet of the mcp2518fd (DS20006027B) specifies a range of + * [-64,63] for TDCO, indicating a relative TDCO. + * + * Manual tests have shown, that using a relative TDCO configuration + * results in bus off, while an absolute configuration works. + * + * For TDCO use the max value (63) from the data sheet, but 0 as the + * minimum. + */ +static const struct can_tdc_const mcp251xfd_tdc_const = { + .tdcv_min = 0, + .tdcv_max = 63, + .tdco_min = 0, + .tdco_max = 63, + .tdcf_min = 0, + .tdcf_max = 0, +}; + static const char *__mcp251xfd_get_model_str(enum mcp251xfd_model model) { switch (model) { @@ -510,8 +528,7 @@ static int mcp251xfd_set_bittiming(const struct mcp251xfd_priv *priv) { const struct can_bittiming *bt = &priv->can.bittiming; const struct can_bittiming *dbt = &priv->can.data_bittiming; - u32 val = 0; - s8 tdco; + u32 tdcmod, val = 0; int err; /* CAN Control Register @@ -575,11 +592,16 @@ static int mcp251xfd_set_bittiming(const struct mcp251xfd_priv *priv) return err; /* Transmitter Delay Compensation */ - tdco = clamp_t(int, dbt->brp * (dbt->prop_seg + dbt->phase_seg1), - -64, 63); - val = FIELD_PREP(MCP251XFD_REG_TDC_TDCMOD_MASK, - MCP251XFD_REG_TDC_TDCMOD_AUTO) | - FIELD_PREP(MCP251XFD_REG_TDC_TDCO_MASK, tdco); + if (priv->can.ctrlmode & CAN_CTRLMODE_TDC_AUTO) + tdcmod = MCP251XFD_REG_TDC_TDCMOD_AUTO; + else if (priv->can.ctrlmode & CAN_CTRLMODE_TDC_MANUAL) + tdcmod = MCP251XFD_REG_TDC_TDCMOD_MANUAL; + else + tdcmod = MCP251XFD_REG_TDC_TDCMOD_DISABLED; + + val = FIELD_PREP(MCP251XFD_REG_TDC_TDCMOD_MASK, tdcmod) | + FIELD_PREP(MCP251XFD_REG_TDC_TDCV_MASK, priv->can.tdc.tdcv) | + FIELD_PREP(MCP251XFD_REG_TDC_TDCO_MASK, priv->can.tdc.tdco); return regmap_write(priv->map_reg, MCP251XFD_REG_TDC, val); } @@ -2083,10 +2105,12 @@ static int mcp251xfd_probe(struct spi_device *spi) priv->can.do_get_berr_counter = mcp251xfd_get_berr_counter; priv->can.bittiming_const = &mcp251xfd_bittiming_const; priv->can.data_bittiming_const = &mcp251xfd_data_bittiming_const; + priv->can.tdc_const = &mcp251xfd_tdc_const; priv->can.ctrlmode_supported = CAN_CTRLMODE_LOOPBACK | CAN_CTRLMODE_LISTENONLY | CAN_CTRLMODE_BERR_REPORTING | CAN_CTRLMODE_FD | CAN_CTRLMODE_FD_NON_ISO | - CAN_CTRLMODE_CC_LEN8_DLC; + CAN_CTRLMODE_CC_LEN8_DLC | CAN_CTRLMODE_TDC_AUTO | + CAN_CTRLMODE_TDC_MANUAL; set_bit(MCP251XFD_FLAGS_DOWN, priv->flags); priv->ndev = ndev; priv->spi = spi; -- cgit From 84f5eb833f53ae192baed4cfb8d9eaab43481fc9 Mon Sep 17 00:00:00 2001 From: Marc Kleine-Budde Date: Fri, 2 May 2025 16:13:44 +0200 Subject: can: mcp251xfd: mcp251xfd_remove(): fix order of unregistration calls If a driver is removed, the driver framework invokes the driver's remove callback. A CAN driver's remove function calls unregister_candev(), which calls net_device_ops::ndo_stop further down in the call stack for interfaces which are in the "up" state. With the mcp251xfd driver the removal of the module causes the following warning: | WARNING: CPU: 0 PID: 352 at net/core/dev.c:7342 __netif_napi_del_locked+0xc8/0xd8 as can_rx_offload_del() deletes the NAPI, while it is still active, because the interface is still up. To fix the warning, first unregister the network interface, which calls net_device_ops::ndo_stop, which disables the NAPI, and then call can_rx_offload_del(). Fixes: 55e5b97f003e ("can: mcp25xxfd: add driver for Microchip MCP25xxFD SPI CAN") Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20250502-can-rx-offload-del-v1-1-59a9b131589d@pengutronix.de Signed-off-by: Marc Kleine-Budde --- drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c b/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c index 064d81c724f4..c30b04f8fc0d 100644 --- a/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c +++ b/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c @@ -2198,8 +2198,8 @@ static void mcp251xfd_remove(struct spi_device *spi) struct mcp251xfd_priv *priv = spi_get_drvdata(spi); struct net_device *ndev = priv->ndev; - can_rx_offload_del(&priv->offload); mcp251xfd_unregister(priv); + can_rx_offload_del(&priv->offload); spi->max_speed_hz = priv->spi_max_speed_hz_orig; free_candev(ndev); } -- cgit From 037ada7a3181300218e4fd78bef6a741cfa7f808 Mon Sep 17 00:00:00 2001 From: Marc Kleine-Budde Date: Fri, 2 May 2025 16:13:45 +0200 Subject: can: rockchip_canfd: rkcanfd_remove(): fix order of unregistration calls If a driver is removed, the driver framework invokes the driver's remove callback. A CAN driver's remove function calls unregister_candev(), which calls net_device_ops::ndo_stop further down in the call stack for interfaces which are in the "up" state. The removal of the module causes a warning, as can_rx_offload_del() deletes the NAPI, while it is still active, because the interface is still up. To fix the warning, first unregister the network interface, which calls net_device_ops::ndo_stop, which disables the NAPI, and then call can_rx_offload_del(). Fixes: ff60bfbaf67f ("can: rockchip_canfd: add driver for Rockchip CAN-FD controller") Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20250502-can-rx-offload-del-v1-2-59a9b131589d@pengutronix.de Reviewed-by: Markus Schneider-Pargmann Signed-off-by: Marc Kleine-Budde --- drivers/net/can/rockchip/rockchip_canfd-core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/can/rockchip/rockchip_canfd-core.c b/drivers/net/can/rockchip/rockchip_canfd-core.c index 7107a37da36c..c3fb3176ce42 100644 --- a/drivers/net/can/rockchip/rockchip_canfd-core.c +++ b/drivers/net/can/rockchip/rockchip_canfd-core.c @@ -937,8 +937,8 @@ static void rkcanfd_remove(struct platform_device *pdev) struct rkcanfd_priv *priv = platform_get_drvdata(pdev); struct net_device *ndev = priv->ndev; - can_rx_offload_del(&priv->offload); rkcanfd_unregister(priv); + can_rx_offload_del(&priv->offload); free_candev(ndev); } -- cgit From 0713a1b3276b98c7dafbeefef00d7bc3a9119a84 Mon Sep 17 00:00:00 2001 From: Marc Kleine-Budde Date: Fri, 2 May 2025 16:13:46 +0200 Subject: can: mcan: m_can_class_unregister(): fix order of unregistration calls If a driver is removed, the driver framework invokes the driver's remove callback. A CAN driver's remove function calls unregister_candev(), which calls net_device_ops::ndo_stop further down in the call stack for interfaces which are in the "up" state. The removal of the module causes a warning, as can_rx_offload_del() deletes the NAPI, while it is still active, because the interface is still up. To fix the warning, first unregister the network interface, which calls net_device_ops::ndo_stop, which disables the NAPI, and then call can_rx_offload_del(). Fixes: 1be37d3b0414 ("can: m_can: fix periph RX path: use rx-offload to ensure skbs are sent from softirq context") Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20250502-can-rx-offload-del-v1-3-59a9b131589d@pengutronix.de Reviewed-by: Markus Schneider-Pargmann Signed-off-by: Marc Kleine-Budde --- drivers/net/can/m_can/m_can.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/can/m_can/m_can.c b/drivers/net/can/m_can/m_can.c index 326ede9d400f..c2c116ce1087 100644 --- a/drivers/net/can/m_can/m_can.c +++ b/drivers/net/can/m_can/m_can.c @@ -2463,9 +2463,9 @@ EXPORT_SYMBOL_GPL(m_can_class_register); void m_can_class_unregister(struct m_can_classdev *cdev) { + unregister_candev(cdev->net); if (cdev->is_peripheral) can_rx_offload_del(&cdev->offload); - unregister_candev(cdev->net); } EXPORT_SYMBOL_GPL(m_can_class_unregister); -- cgit From 511e64e13d8cc72853275832e3f372607466c18c Mon Sep 17 00:00:00 2001 From: Oliver Hartkopp Date: Tue, 29 Apr 2025 09:05:55 +0200 Subject: can: gw: fix RCU/BH usage in cgw_create_job() As reported by Sebastian Andrzej Siewior the use of local_bh_disable() is only feasible in uni processor systems to update the modification rules. The usual use-case to update the modification rules is to update the data of the modifications but not the modification types (AND/OR/XOR/SET) or the checksum functions itself. To omit additional memory allocations to maintain fast modification switching times, the modification description space is doubled at gw-job creation time so that only the reference to the active modification description is changed under rcu protection. Rename cgw_job::mod to cf_mod and make it a RCU pointer. Allocate in cgw_create_job() and free it together with cgw_job in cgw_job_free_rcu(). Update all users to dereference cgw_job::cf_mod with a RCU accessor and if possible once. [bigeasy: Replace mod1/mod2 from the Oliver's original patch with dynamic allocation, use RCU annotation and accessor] Reported-by: Sebastian Andrzej Siewior Closes: https://lore.kernel.org/linux-can/20231031112349.y0aLoBrz@linutronix.de/ Fixes: dd895d7f21b2 ("can: cangw: introduce optional uid to reference created routing jobs") Tested-by: Oliver Hartkopp Signed-off-by: Oliver Hartkopp Signed-off-by: Sebastian Andrzej Siewior Link: https://patch.msgid.link/20250429070555.cs-7b_eZ@linutronix.de Signed-off-by: Marc Kleine-Budde --- net/can/gw.c | 149 ++++++++++++++++++++++++++++++++++++----------------------- 1 file changed, 90 insertions(+), 59 deletions(-) diff --git a/net/can/gw.c b/net/can/gw.c index ef93293c1fae..55eccb1c7620 100644 --- a/net/can/gw.c +++ b/net/can/gw.c @@ -130,7 +130,7 @@ struct cgw_job { u32 handled_frames; u32 dropped_frames; u32 deleted_frames; - struct cf_mod mod; + struct cf_mod __rcu *cf_mod; union { /* CAN frame data source */ struct net_device *dev; @@ -459,6 +459,7 @@ static void can_can_gw_rcv(struct sk_buff *skb, void *data) struct cgw_job *gwj = (struct cgw_job *)data; struct canfd_frame *cf; struct sk_buff *nskb; + struct cf_mod *mod; int modidx = 0; /* process strictly Classic CAN or CAN FD frames */ @@ -506,7 +507,8 @@ static void can_can_gw_rcv(struct sk_buff *skb, void *data) * When there is at least one modification function activated, * we need to copy the skb as we want to modify skb->data. */ - if (gwj->mod.modfunc[0]) + mod = rcu_dereference(gwj->cf_mod); + if (mod->modfunc[0]) nskb = skb_copy(skb, GFP_ATOMIC); else nskb = skb_clone(skb, GFP_ATOMIC); @@ -529,8 +531,8 @@ static void can_can_gw_rcv(struct sk_buff *skb, void *data) cf = (struct canfd_frame *)nskb->data; /* perform preprocessed modification functions if there are any */ - while (modidx < MAX_MODFUNCTIONS && gwj->mod.modfunc[modidx]) - (*gwj->mod.modfunc[modidx++])(cf, &gwj->mod); + while (modidx < MAX_MODFUNCTIONS && mod->modfunc[modidx]) + (*mod->modfunc[modidx++])(cf, mod); /* Has the CAN frame been modified? */ if (modidx) { @@ -546,11 +548,11 @@ static void can_can_gw_rcv(struct sk_buff *skb, void *data) } /* check for checksum updates */ - if (gwj->mod.csumfunc.crc8) - (*gwj->mod.csumfunc.crc8)(cf, &gwj->mod.csum.crc8); + if (mod->csumfunc.crc8) + (*mod->csumfunc.crc8)(cf, &mod->csum.crc8); - if (gwj->mod.csumfunc.xor) - (*gwj->mod.csumfunc.xor)(cf, &gwj->mod.csum.xor); + if (mod->csumfunc.xor) + (*mod->csumfunc.xor)(cf, &mod->csum.xor); } /* clear the skb timestamp if not configured the other way */ @@ -581,9 +583,20 @@ static void cgw_job_free_rcu(struct rcu_head *rcu_head) { struct cgw_job *gwj = container_of(rcu_head, struct cgw_job, rcu); + /* cgw_job::cf_mod is always accessed from the same cgw_job object within + * the same RCU read section. Once cgw_job is scheduled for removal, + * cf_mod can also be removed without mandating an additional grace period. + */ + kfree(rcu_access_pointer(gwj->cf_mod)); kmem_cache_free(cgw_cache, gwj); } +/* Return cgw_job::cf_mod with RTNL protected section */ +static struct cf_mod *cgw_job_cf_mod(struct cgw_job *gwj) +{ + return rcu_dereference_protected(gwj->cf_mod, rtnl_is_locked()); +} + static int cgw_notifier(struct notifier_block *nb, unsigned long msg, void *ptr) { @@ -616,6 +629,7 @@ static int cgw_put_job(struct sk_buff *skb, struct cgw_job *gwj, int type, { struct rtcanmsg *rtcan; struct nlmsghdr *nlh; + struct cf_mod *mod; nlh = nlmsg_put(skb, pid, seq, type, sizeof(*rtcan), flags); if (!nlh) @@ -650,82 +664,83 @@ static int cgw_put_job(struct sk_buff *skb, struct cgw_job *gwj, int type, goto cancel; } + mod = cgw_job_cf_mod(gwj); if (gwj->flags & CGW_FLAGS_CAN_FD) { struct cgw_fdframe_mod mb; - if (gwj->mod.modtype.and) { - memcpy(&mb.cf, &gwj->mod.modframe.and, sizeof(mb.cf)); - mb.modtype = gwj->mod.modtype.and; + if (mod->modtype.and) { + memcpy(&mb.cf, &mod->modframe.and, sizeof(mb.cf)); + mb.modtype = mod->modtype.and; if (nla_put(skb, CGW_FDMOD_AND, sizeof(mb), &mb) < 0) goto cancel; } - if (gwj->mod.modtype.or) { - memcpy(&mb.cf, &gwj->mod.modframe.or, sizeof(mb.cf)); - mb.modtype = gwj->mod.modtype.or; + if (mod->modtype.or) { + memcpy(&mb.cf, &mod->modframe.or, sizeof(mb.cf)); + mb.modtype = mod->modtype.or; if (nla_put(skb, CGW_FDMOD_OR, sizeof(mb), &mb) < 0) goto cancel; } - if (gwj->mod.modtype.xor) { - memcpy(&mb.cf, &gwj->mod.modframe.xor, sizeof(mb.cf)); - mb.modtype = gwj->mod.modtype.xor; + if (mod->modtype.xor) { + memcpy(&mb.cf, &mod->modframe.xor, sizeof(mb.cf)); + mb.modtype = mod->modtype.xor; if (nla_put(skb, CGW_FDMOD_XOR, sizeof(mb), &mb) < 0) goto cancel; } - if (gwj->mod.modtype.set) { - memcpy(&mb.cf, &gwj->mod.modframe.set, sizeof(mb.cf)); - mb.modtype = gwj->mod.modtype.set; + if (mod->modtype.set) { + memcpy(&mb.cf, &mod->modframe.set, sizeof(mb.cf)); + mb.modtype = mod->modtype.set; if (nla_put(skb, CGW_FDMOD_SET, sizeof(mb), &mb) < 0) goto cancel; } } else { struct cgw_frame_mod mb; - if (gwj->mod.modtype.and) { - memcpy(&mb.cf, &gwj->mod.modframe.and, sizeof(mb.cf)); - mb.modtype = gwj->mod.modtype.and; + if (mod->modtype.and) { + memcpy(&mb.cf, &mod->modframe.and, sizeof(mb.cf)); + mb.modtype = mod->modtype.and; if (nla_put(skb, CGW_MOD_AND, sizeof(mb), &mb) < 0) goto cancel; } - if (gwj->mod.modtype.or) { - memcpy(&mb.cf, &gwj->mod.modframe.or, sizeof(mb.cf)); - mb.modtype = gwj->mod.modtype.or; + if (mod->modtype.or) { + memcpy(&mb.cf, &mod->modframe.or, sizeof(mb.cf)); + mb.modtype = mod->modtype.or; if (nla_put(skb, CGW_MOD_OR, sizeof(mb), &mb) < 0) goto cancel; } - if (gwj->mod.modtype.xor) { - memcpy(&mb.cf, &gwj->mod.modframe.xor, sizeof(mb.cf)); - mb.modtype = gwj->mod.modtype.xor; + if (mod->modtype.xor) { + memcpy(&mb.cf, &mod->modframe.xor, sizeof(mb.cf)); + mb.modtype = mod->modtype.xor; if (nla_put(skb, CGW_MOD_XOR, sizeof(mb), &mb) < 0) goto cancel; } - if (gwj->mod.modtype.set) { - memcpy(&mb.cf, &gwj->mod.modframe.set, sizeof(mb.cf)); - mb.modtype = gwj->mod.modtype.set; + if (mod->modtype.set) { + memcpy(&mb.cf, &mod->modframe.set, sizeof(mb.cf)); + mb.modtype = mod->modtype.set; if (nla_put(skb, CGW_MOD_SET, sizeof(mb), &mb) < 0) goto cancel; } } - if (gwj->mod.uid) { - if (nla_put_u32(skb, CGW_MOD_UID, gwj->mod.uid) < 0) + if (mod->uid) { + if (nla_put_u32(skb, CGW_MOD_UID, mod->uid) < 0) goto cancel; } - if (gwj->mod.csumfunc.crc8) { + if (mod->csumfunc.crc8) { if (nla_put(skb, CGW_CS_CRC8, CGW_CS_CRC8_LEN, - &gwj->mod.csum.crc8) < 0) + &mod->csum.crc8) < 0) goto cancel; } - if (gwj->mod.csumfunc.xor) { + if (mod->csumfunc.xor) { if (nla_put(skb, CGW_CS_XOR, CGW_CS_XOR_LEN, - &gwj->mod.csum.xor) < 0) + &mod->csum.xor) < 0) goto cancel; } @@ -1059,7 +1074,7 @@ static int cgw_create_job(struct sk_buff *skb, struct nlmsghdr *nlh, struct net *net = sock_net(skb->sk); struct rtcanmsg *r; struct cgw_job *gwj; - struct cf_mod mod; + struct cf_mod *mod; struct can_can_gw ccgw; u8 limhops = 0; int err = 0; @@ -1078,37 +1093,48 @@ static int cgw_create_job(struct sk_buff *skb, struct nlmsghdr *nlh, if (r->gwtype != CGW_TYPE_CAN_CAN) return -EINVAL; - err = cgw_parse_attr(nlh, &mod, CGW_TYPE_CAN_CAN, &ccgw, &limhops); + mod = kmalloc(sizeof(*mod), GFP_KERNEL); + if (!mod) + return -ENOMEM; + + err = cgw_parse_attr(nlh, mod, CGW_TYPE_CAN_CAN, &ccgw, &limhops); if (err < 0) - return err; + goto out_free_cf; - if (mod.uid) { + if (mod->uid) { ASSERT_RTNL(); /* check for updating an existing job with identical uid */ hlist_for_each_entry(gwj, &net->can.cgw_list, list) { - if (gwj->mod.uid != mod.uid) + struct cf_mod *old_cf; + + old_cf = cgw_job_cf_mod(gwj); + if (old_cf->uid != mod->uid) continue; /* interfaces & filters must be identical */ - if (memcmp(&gwj->ccgw, &ccgw, sizeof(ccgw))) - return -EINVAL; + if (memcmp(&gwj->ccgw, &ccgw, sizeof(ccgw))) { + err = -EINVAL; + goto out_free_cf; + } - /* update modifications with disabled softirq & quit */ - local_bh_disable(); - memcpy(&gwj->mod, &mod, sizeof(mod)); - local_bh_enable(); + rcu_assign_pointer(gwj->cf_mod, mod); + kfree_rcu_mightsleep(old_cf); return 0; } } /* ifindex == 0 is not allowed for job creation */ - if (!ccgw.src_idx || !ccgw.dst_idx) - return -ENODEV; + if (!ccgw.src_idx || !ccgw.dst_idx) { + err = -ENODEV; + goto out_free_cf; + } gwj = kmem_cache_alloc(cgw_cache, GFP_KERNEL); - if (!gwj) - return -ENOMEM; + if (!gwj) { + err = -ENOMEM; + goto out_free_cf; + } gwj->handled_frames = 0; gwj->dropped_frames = 0; @@ -1118,7 +1144,7 @@ static int cgw_create_job(struct sk_buff *skb, struct nlmsghdr *nlh, gwj->limit_hops = limhops; /* insert already parsed information */ - memcpy(&gwj->mod, &mod, sizeof(mod)); + RCU_INIT_POINTER(gwj->cf_mod, mod); memcpy(&gwj->ccgw, &ccgw, sizeof(ccgw)); err = -ENODEV; @@ -1152,9 +1178,11 @@ static int cgw_create_job(struct sk_buff *skb, struct nlmsghdr *nlh, if (!err) hlist_add_head_rcu(&gwj->list, &net->can.cgw_list); out: - if (err) + if (err) { kmem_cache_free(cgw_cache, gwj); - +out_free_cf: + kfree(mod); + } return err; } @@ -1214,19 +1242,22 @@ static int cgw_remove_job(struct sk_buff *skb, struct nlmsghdr *nlh, /* remove only the first matching entry */ hlist_for_each_entry_safe(gwj, nx, &net->can.cgw_list, list) { + struct cf_mod *cf_mod; + if (gwj->flags != r->flags) continue; if (gwj->limit_hops != limhops) continue; + cf_mod = cgw_job_cf_mod(gwj); /* we have a match when uid is enabled and identical */ - if (gwj->mod.uid || mod.uid) { - if (gwj->mod.uid != mod.uid) + if (cf_mod->uid || mod.uid) { + if (cf_mod->uid != mod.uid) continue; } else { /* no uid => check for identical modifications */ - if (memcmp(&gwj->mod, &mod, sizeof(mod))) + if (memcmp(cf_mod, &mod, sizeof(mod))) continue; } -- cgit From 023c1f2f0609218103cbcb48e0104b144d4a16dc Mon Sep 17 00:00:00 2001 From: Veerendranath Jakkam Date: Thu, 24 Apr 2025 18:01:42 +0530 Subject: wifi: cfg80211: fix out-of-bounds access during multi-link element defragmentation Currently during the multi-link element defragmentation process, the multi-link element length added to the total IEs length when calculating the length of remaining IEs after the multi-link element in cfg80211_defrag_mle(). This could lead to out-of-bounds access if the multi-link element or its corresponding fragment elements are the last elements in the IEs buffer. To address this issue, correctly calculate the remaining IEs length by deducting the multi-link element end offset from total IEs end offset. Cc: stable@vger.kernel.org Fixes: 2481b5da9c6b ("wifi: cfg80211: handle BSS data contained in ML probe responses") Signed-off-by: Veerendranath Jakkam Link: https://patch.msgid.link/20250424-fix_mle_defragmentation_oob_access-v1-1-84412a1743fa@quicinc.com Signed-off-by: Johannes Berg --- net/wireless/scan.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/wireless/scan.c b/net/wireless/scan.c index 9865f305275d..ddd3a97f6609 100644 --- a/net/wireless/scan.c +++ b/net/wireless/scan.c @@ -2681,7 +2681,7 @@ cfg80211_defrag_mle(const struct element *mle, const u8 *ie, size_t ielen, /* Required length for first defragmentation */ buf_len = mle->datalen - 1; for_each_element(elem, mle->data + mle->datalen, - ielen - sizeof(*mle) + mle->datalen) { + ie + ielen - mle->data - mle->datalen) { if (elem->id != WLAN_EID_FRAGMENT) break; -- cgit From e12a42f64fc3d74872b349eedd47f90c6676b78a Mon Sep 17 00:00:00 2001 From: Michael-CY Lee Date: Mon, 5 May 2025 16:19:46 +0800 Subject: wifi: mac80211: fix the type of status_code for negotiated TID to Link Mapping The status code should be type of __le16. Fixes: 83e897a961b8 ("wifi: ieee80211: add definitions for negotiated TID to Link map") Fixes: 8f500fbc6c65 ("wifi: mac80211: process and save negotiated TID to Link mapping request") Signed-off-by: Michael-CY Lee Link: https://patch.msgid.link/20250505081946.3927214-1-michael-cy.lee@mediatek.com Signed-off-by: Johannes Berg --- include/linux/ieee80211.h | 2 +- net/mac80211/mlme.c | 12 ++++++------ 2 files changed, 7 insertions(+), 7 deletions(-) diff --git a/include/linux/ieee80211.h b/include/linux/ieee80211.h index 508d466de1cc..457b4fba88bd 100644 --- a/include/linux/ieee80211.h +++ b/include/linux/ieee80211.h @@ -1526,7 +1526,7 @@ struct ieee80211_mgmt { struct { u8 action_code; u8 dialog_token; - u8 status_code; + __le16 status_code; u8 variable[]; } __packed ttlm_res; struct { diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c index 5d1f2d6d09ad..35eaf0812c5b 100644 --- a/net/mac80211/mlme.c +++ b/net/mac80211/mlme.c @@ -7675,6 +7675,7 @@ ieee80211_send_neg_ttlm_res(struct ieee80211_sub_if_data *sdata, int hdr_len = offsetofend(struct ieee80211_mgmt, u.action.u.ttlm_res); int ttlm_max_len = 2 + 1 + sizeof(struct ieee80211_ttlm_elem) + 1 + 2 * 2 * IEEE80211_TTLM_NUM_TIDS; + u16 status_code; skb = dev_alloc_skb(local->tx_headroom + hdr_len + ttlm_max_len); if (!skb) @@ -7697,19 +7698,18 @@ ieee80211_send_neg_ttlm_res(struct ieee80211_sub_if_data *sdata, WARN_ON(1); fallthrough; case NEG_TTLM_RES_REJECT: - mgmt->u.action.u.ttlm_res.status_code = - WLAN_STATUS_DENIED_TID_TO_LINK_MAPPING; + status_code = WLAN_STATUS_DENIED_TID_TO_LINK_MAPPING; break; case NEG_TTLM_RES_ACCEPT: - mgmt->u.action.u.ttlm_res.status_code = WLAN_STATUS_SUCCESS; + status_code = WLAN_STATUS_SUCCESS; break; case NEG_TTLM_RES_SUGGEST_PREFERRED: - mgmt->u.action.u.ttlm_res.status_code = - WLAN_STATUS_PREF_TID_TO_LINK_MAPPING_SUGGESTED; + status_code = WLAN_STATUS_PREF_TID_TO_LINK_MAPPING_SUGGESTED; ieee80211_neg_ttlm_add_suggested_map(skb, neg_ttlm); break; } + mgmt->u.action.u.ttlm_res.status_code = cpu_to_le16(status_code); ieee80211_tx_skb(sdata, skb); } @@ -7875,7 +7875,7 @@ void ieee80211_process_neg_ttlm_res(struct ieee80211_sub_if_data *sdata, * This can be better implemented in the future, to handle request * rejections. */ - if (mgmt->u.action.u.ttlm_res.status_code != WLAN_STATUS_SUCCESS) + if (le16_to_cpu(mgmt->u.action.u.ttlm_res.status_code) != WLAN_STATUS_SUCCESS) __ieee80211_disconnect(sdata); } -- cgit From ebedf8b7f05b9c886d68d63025db8d1b12343157 Mon Sep 17 00:00:00 2001 From: Johannes Berg Date: Tue, 6 May 2025 21:42:59 +0200 Subject: wifi: iwlwifi: add support for Killer on MTL For now, we need another entry for these devices, this will be changed completely for 6.16. Closes: https://bugzilla.kernel.org/show_bug.cgi?id=219926 Link: https://patch.msgid.link/20250506214258.2efbdc9e9a82.I31915ec252bd1c74bd53b89a0e214e42a74b6f2e@changeid Signed-off-by: Johannes Berg --- drivers/net/wireless/intel/iwlwifi/pcie/drv.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/net/wireless/intel/iwlwifi/pcie/drv.c b/drivers/net/wireless/intel/iwlwifi/pcie/drv.c index debeea2b3ae5..00056e76ea3d 100644 --- a/drivers/net/wireless/intel/iwlwifi/pcie/drv.c +++ b/drivers/net/wireless/intel/iwlwifi/pcie/drv.c @@ -588,6 +588,8 @@ VISIBLE_IF_IWLWIFI_KUNIT const struct iwl_dev_info iwl_dev_info_table[] = { IWL_DEV_INFO(0x7A70, 0x1692, iwlax411_2ax_cfg_so_gf4_a0, iwl_ax411_killer_1690i_name), IWL_DEV_INFO(0x7AF0, 0x1691, iwlax411_2ax_cfg_so_gf4_a0, iwl_ax411_killer_1690s_name), IWL_DEV_INFO(0x7AF0, 0x1692, iwlax411_2ax_cfg_so_gf4_a0, iwl_ax411_killer_1690i_name), + IWL_DEV_INFO(0x7F70, 0x1691, iwlax411_2ax_cfg_so_gf4_a0, iwl_ax411_killer_1690s_name), + IWL_DEV_INFO(0x7F70, 0x1692, iwlax411_2ax_cfg_so_gf4_a0, iwl_ax411_killer_1690i_name), IWL_DEV_INFO(0x271C, 0x0214, iwl9260_2ac_cfg, iwl9260_1_name), IWL_DEV_INFO(0x7E40, 0x1691, iwl_cfg_ma, iwl_ax411_killer_1690s_name), -- cgit From 0093cb194a7511d1e68865fa35b763c72e44c2f0 Mon Sep 17 00:00:00 2001 From: Przemek Kitszel Date: Mon, 5 May 2025 09:19:38 -0700 Subject: ice: use DSN instead of PCI BDF for ice_adapter index Use Device Serial Number instead of PCI bus/device/function for the index of struct ice_adapter. Functions on the same physical device should point to the very same ice_adapter instance, but with two PFs, when at least one of them is PCI-e passed-through to a VM, it is no longer the case - PFs will get seemingly random PCI BDF values, and thus indices, what finally leds to each of them being on their own instance of ice_adapter. That causes them to don't attempt any synchronization of the PTP HW clock usage, or any other future resources. DSN works nicely in place of the index, as it is "immutable" in terms of virtualization. Fixes: 0e2bddf9e5f9 ("ice: add ice_adapter for shared data across PFs on the same NIC") Suggested-by: Jacob Keller Suggested-by: Jakub Kicinski Suggested-by: Jiri Pirko Reviewed-by: Aleksandr Loktionov Signed-off-by: Przemek Kitszel Reviewed-by: Simon Horman Tested-by: Rinitha S (A Contingent worker at Intel) Signed-off-by: Tony Nguyen Reviewed-by: Jiri Pirko Link: https://patch.msgid.link/20250505161939.2083581-1-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/intel/ice/ice_adapter.c | 47 +++++++++++----------------- drivers/net/ethernet/intel/ice/ice_adapter.h | 6 ++-- 2 files changed, 22 insertions(+), 31 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_adapter.c b/drivers/net/ethernet/intel/ice/ice_adapter.c index 01a08cfd0090..66e070095d1b 100644 --- a/drivers/net/ethernet/intel/ice/ice_adapter.c +++ b/drivers/net/ethernet/intel/ice/ice_adapter.c @@ -1,7 +1,6 @@ // SPDX-License-Identifier: GPL-2.0-only // SPDX-FileCopyrightText: Copyright Red Hat -#include #include #include #include @@ -14,32 +13,16 @@ static DEFINE_XARRAY(ice_adapters); static DEFINE_MUTEX(ice_adapters_mutex); -/* PCI bus number is 8 bits. Slot is 5 bits. Domain can have the rest. */ -#define INDEX_FIELD_DOMAIN GENMASK(BITS_PER_LONG - 1, 13) -#define INDEX_FIELD_DEV GENMASK(31, 16) -#define INDEX_FIELD_BUS GENMASK(12, 5) -#define INDEX_FIELD_SLOT GENMASK(4, 0) - -static unsigned long ice_adapter_index(const struct pci_dev *pdev) +static unsigned long ice_adapter_index(u64 dsn) { - unsigned int domain = pci_domain_nr(pdev->bus); - - WARN_ON(domain > FIELD_MAX(INDEX_FIELD_DOMAIN)); - - switch (pdev->device) { - case ICE_DEV_ID_E825C_BACKPLANE: - case ICE_DEV_ID_E825C_QSFP: - case ICE_DEV_ID_E825C_SFP: - case ICE_DEV_ID_E825C_SGMII: - return FIELD_PREP(INDEX_FIELD_DEV, pdev->device); - default: - return FIELD_PREP(INDEX_FIELD_DOMAIN, domain) | - FIELD_PREP(INDEX_FIELD_BUS, pdev->bus->number) | - FIELD_PREP(INDEX_FIELD_SLOT, PCI_SLOT(pdev->devfn)); - } +#if BITS_PER_LONG == 64 + return dsn; +#else + return (u32)dsn ^ (u32)(dsn >> 32); +#endif } -static struct ice_adapter *ice_adapter_new(void) +static struct ice_adapter *ice_adapter_new(u64 dsn) { struct ice_adapter *adapter; @@ -47,6 +30,7 @@ static struct ice_adapter *ice_adapter_new(void) if (!adapter) return NULL; + adapter->device_serial_number = dsn; spin_lock_init(&adapter->ptp_gltsyn_time_lock); refcount_set(&adapter->refcount, 1); @@ -77,23 +61,26 @@ static void ice_adapter_free(struct ice_adapter *adapter) * Return: Pointer to ice_adapter on success. * ERR_PTR() on error. -ENOMEM is the only possible error. */ -struct ice_adapter *ice_adapter_get(const struct pci_dev *pdev) +struct ice_adapter *ice_adapter_get(struct pci_dev *pdev) { - unsigned long index = ice_adapter_index(pdev); + u64 dsn = pci_get_dsn(pdev); struct ice_adapter *adapter; + unsigned long index; int err; + index = ice_adapter_index(dsn); scoped_guard(mutex, &ice_adapters_mutex) { err = xa_insert(&ice_adapters, index, NULL, GFP_KERNEL); if (err == -EBUSY) { adapter = xa_load(&ice_adapters, index); refcount_inc(&adapter->refcount); + WARN_ON_ONCE(adapter->device_serial_number != dsn); return adapter; } if (err) return ERR_PTR(err); - adapter = ice_adapter_new(); + adapter = ice_adapter_new(dsn); if (!adapter) return ERR_PTR(-ENOMEM); xa_store(&ice_adapters, index, adapter, GFP_KERNEL); @@ -110,11 +97,13 @@ struct ice_adapter *ice_adapter_get(const struct pci_dev *pdev) * * Context: Process, may sleep. */ -void ice_adapter_put(const struct pci_dev *pdev) +void ice_adapter_put(struct pci_dev *pdev) { - unsigned long index = ice_adapter_index(pdev); + u64 dsn = pci_get_dsn(pdev); struct ice_adapter *adapter; + unsigned long index; + index = ice_adapter_index(dsn); scoped_guard(mutex, &ice_adapters_mutex) { adapter = xa_load(&ice_adapters, index); if (WARN_ON(!adapter)) diff --git a/drivers/net/ethernet/intel/ice/ice_adapter.h b/drivers/net/ethernet/intel/ice/ice_adapter.h index e233225848b3..ac15c0d2bc1a 100644 --- a/drivers/net/ethernet/intel/ice/ice_adapter.h +++ b/drivers/net/ethernet/intel/ice/ice_adapter.h @@ -32,6 +32,7 @@ struct ice_port_list { * @refcount: Reference count. struct ice_pf objects hold the references. * @ctrl_pf: Control PF of the adapter * @ports: Ports list + * @device_serial_number: DSN cached for collision detection on 32bit systems */ struct ice_adapter { refcount_t refcount; @@ -40,9 +41,10 @@ struct ice_adapter { struct ice_pf *ctrl_pf; struct ice_port_list ports; + u64 device_serial_number; }; -struct ice_adapter *ice_adapter_get(const struct pci_dev *pdev); -void ice_adapter_put(const struct pci_dev *pdev); +struct ice_adapter *ice_adapter_get(struct pci_dev *pdev); +void ice_adapter_put(struct pci_dev *pdev); #endif /* _ICE_ADAPTER_H */ -- cgit From 08e9f2d584c4732180edee4cb2dbfa7586d7d5a3 Mon Sep 17 00:00:00 2001 From: Cosmin Ratiu Date: Mon, 5 May 2025 22:47:13 +0300 Subject: net: Lock netdevices during dev_shutdown __qdisc_destroy() calls into various qdiscs .destroy() op, which in turn can call .ndo_setup_tc(), which requires the netdev instance lock. This commit extends the critical section in unregister_netdevice_many_notify() to cover dev_shutdown() (and dev_tcx_uninstall() as a side-effect) and acquires the netdev instance lock in __dev_change_net_namespace() for the other dev_shutdown() call. This should now guarantee that for all qdisc ops, the netdev instance lock is held during .ndo_setup_tc(). Fixes: a0527ee2df3f ("net: hold netdev instance lock during qdisc ndo_setup_tc") Signed-off-by: Cosmin Ratiu Acked-by: Stanislav Fomichev Link: https://patch.msgid.link/20250505194713.1723399-1-cratiu@nvidia.com Signed-off-by: Jakub Kicinski --- net/core/dev.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/net/core/dev.c b/net/core/dev.c index 1be7cb73a602..92e004c354ea 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -11966,9 +11966,9 @@ void unregister_netdevice_many_notify(struct list_head *head, struct sk_buff *skb = NULL; /* Shutdown queueing discipline. */ + netdev_lock_ops(dev); dev_shutdown(dev); dev_tcx_uninstall(dev); - netdev_lock_ops(dev); dev_xdp_uninstall(dev); dev_memory_provider_uninstall(dev); netdev_unlock_ops(dev); @@ -12161,7 +12161,9 @@ int __dev_change_net_namespace(struct net_device *dev, struct net *net, synchronize_net(); /* Shutdown queueing discipline. */ + netdev_lock_ops(dev); dev_shutdown(dev); + netdev_unlock_ops(dev); /* Notify protocols, that we are about to destroy * this device. They should clean all the things. -- cgit From 78cd408356fe3edbac66598772fd347bf3e32c1f Mon Sep 17 00:00:00 2001 From: Stanislav Fomichev Date: Mon, 5 May 2025 18:19:19 -0700 Subject: net: add missing instance lock to dev_set_promiscuity Accidentally spotted while trying to understand what else needs to be renamed to netif_ prefix. Most of the calls to dev_set_promiscuity are adjacent to dev_set_allmulti or dev_disable_lro so it should be safe to add the lock. Note that new netif_set_promiscuity is currently unused, the locked paths call __dev_set_promiscuity directly. Fixes: ad7c7b2172c3 ("net: hold netdev instance lock during sysfs operations") Signed-off-by: Stanislav Fomichev Link: https://patch.msgid.link/20250506011919.2882313-1-sdf@fomichev.me Signed-off-by: Jakub Kicinski --- include/linux/netdevice.h | 1 + net/core/dev.c | 14 +------------- net/core/dev_api.c | 23 +++++++++++++++++++++++ 3 files changed, 25 insertions(+), 13 deletions(-) diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index 2d11d013cabe..7ea022750e4e 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -4972,6 +4972,7 @@ static inline void __dev_mc_unsync(struct net_device *dev, /* Functions used for secondary unicast and multicast support */ void dev_set_rx_mode(struct net_device *dev); +int netif_set_promiscuity(struct net_device *dev, int inc); int dev_set_promiscuity(struct net_device *dev, int inc); int netif_set_allmulti(struct net_device *dev, int inc, bool notify); int dev_set_allmulti(struct net_device *dev, int inc); diff --git a/net/core/dev.c b/net/core/dev.c index 92e004c354ea..11da1e272ec2 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -9193,18 +9193,7 @@ static int __dev_set_promiscuity(struct net_device *dev, int inc, bool notify) return 0; } -/** - * dev_set_promiscuity - update promiscuity count on a device - * @dev: device - * @inc: modifier - * - * Add or remove promiscuity from a device. While the count in the device - * remains above zero the interface remains promiscuous. Once it hits zero - * the device reverts back to normal filtering operation. A negative inc - * value is used to drop promiscuity on the device. - * Return 0 if successful or a negative errno code on error. - */ -int dev_set_promiscuity(struct net_device *dev, int inc) +int netif_set_promiscuity(struct net_device *dev, int inc) { unsigned int old_flags = dev->flags; int err; @@ -9216,7 +9205,6 @@ int dev_set_promiscuity(struct net_device *dev, int inc) dev_set_rx_mode(dev); return err; } -EXPORT_SYMBOL(dev_set_promiscuity); int netif_set_allmulti(struct net_device *dev, int inc, bool notify) { diff --git a/net/core/dev_api.c b/net/core/dev_api.c index 90898cd540ce..f9a160ab596f 100644 --- a/net/core/dev_api.c +++ b/net/core/dev_api.c @@ -267,6 +267,29 @@ void dev_disable_lro(struct net_device *dev) } EXPORT_SYMBOL(dev_disable_lro); +/** + * dev_set_promiscuity() - update promiscuity count on a device + * @dev: device + * @inc: modifier + * + * Add or remove promiscuity from a device. While the count in the device + * remains above zero the interface remains promiscuous. Once it hits zero + * the device reverts back to normal filtering operation. A negative inc + * value is used to drop promiscuity on the device. + * Return 0 if successful or a negative errno code on error. + */ +int dev_set_promiscuity(struct net_device *dev, int inc) +{ + int ret; + + netdev_lock_ops(dev); + ret = netif_set_promiscuity(dev, inc); + netdev_unlock_ops(dev); + + return ret; +} +EXPORT_SYMBOL(dev_set_promiscuity); + /** * dev_set_allmulti() - update allmulti count on a device * @dev: device -- cgit From e34090d7214e0516eb8722aee295cb2507317c07 Mon Sep 17 00:00:00 2001 From: Julian Anastasov Date: Sat, 3 May 2025 01:01:18 +0300 Subject: ipvs: fix uninit-value for saddr in do_output_route4 syzbot reports for uninit-value for the saddr argument [1]. commit 4754957f04f5 ("ipvs: do not use random local source address for tunnels") already implies that the input value of saddr should be ignored but the code is still reading it which can prevent to connect the route. Fix it by changing the argument to ret_saddr. [1] BUG: KMSAN: uninit-value in do_output_route4+0x42c/0x4d0 net/netfilter/ipvs/ip_vs_xmit.c:147 do_output_route4+0x42c/0x4d0 net/netfilter/ipvs/ip_vs_xmit.c:147 __ip_vs_get_out_rt+0x403/0x21d0 net/netfilter/ipvs/ip_vs_xmit.c:330 ip_vs_tunnel_xmit+0x205/0x2380 net/netfilter/ipvs/ip_vs_xmit.c:1136 ip_vs_in_hook+0x1aa5/0x35b0 net/netfilter/ipvs/ip_vs_core.c:2063 nf_hook_entry_hookfn include/linux/netfilter.h:154 [inline] nf_hook_slow+0xf7/0x400 net/netfilter/core.c:626 nf_hook include/linux/netfilter.h:269 [inline] __ip_local_out+0x758/0x7e0 net/ipv4/ip_output.c:118 ip_local_out net/ipv4/ip_output.c:127 [inline] ip_send_skb+0x6a/0x3c0 net/ipv4/ip_output.c:1501 udp_send_skb+0xfda/0x1b70 net/ipv4/udp.c:1195 udp_sendmsg+0x2fe3/0x33c0 net/ipv4/udp.c:1483 inet_sendmsg+0x1fc/0x280 net/ipv4/af_inet.c:851 sock_sendmsg_nosec net/socket.c:712 [inline] __sock_sendmsg+0x267/0x380 net/socket.c:727 ____sys_sendmsg+0x91b/0xda0 net/socket.c:2566 ___sys_sendmsg+0x28d/0x3c0 net/socket.c:2620 __sys_sendmmsg+0x41d/0x880 net/socket.c:2702 __compat_sys_sendmmsg net/compat.c:360 [inline] __do_compat_sys_sendmmsg net/compat.c:367 [inline] __se_compat_sys_sendmmsg net/compat.c:364 [inline] __ia32_compat_sys_sendmmsg+0xc8/0x140 net/compat.c:364 ia32_sys_call+0x3ffa/0x41f0 arch/x86/include/generated/asm/syscalls_32.h:346 do_syscall_32_irqs_on arch/x86/entry/syscall_32.c:83 [inline] __do_fast_syscall_32+0xb0/0x110 arch/x86/entry/syscall_32.c:306 do_fast_syscall_32+0x38/0x80 arch/x86/entry/syscall_32.c:331 do_SYSENTER_32+0x1f/0x30 arch/x86/entry/syscall_32.c:369 entry_SYSENTER_compat_after_hwframe+0x84/0x8e Uninit was created at: slab_post_alloc_hook mm/slub.c:4167 [inline] slab_alloc_node mm/slub.c:4210 [inline] __kmalloc_cache_noprof+0x8fa/0xe00 mm/slub.c:4367 kmalloc_noprof include/linux/slab.h:905 [inline] ip_vs_dest_dst_alloc net/netfilter/ipvs/ip_vs_xmit.c:61 [inline] __ip_vs_get_out_rt+0x35d/0x21d0 net/netfilter/ipvs/ip_vs_xmit.c:323 ip_vs_tunnel_xmit+0x205/0x2380 net/netfilter/ipvs/ip_vs_xmit.c:1136 ip_vs_in_hook+0x1aa5/0x35b0 net/netfilter/ipvs/ip_vs_core.c:2063 nf_hook_entry_hookfn include/linux/netfilter.h:154 [inline] nf_hook_slow+0xf7/0x400 net/netfilter/core.c:626 nf_hook include/linux/netfilter.h:269 [inline] __ip_local_out+0x758/0x7e0 net/ipv4/ip_output.c:118 ip_local_out net/ipv4/ip_output.c:127 [inline] ip_send_skb+0x6a/0x3c0 net/ipv4/ip_output.c:1501 udp_send_skb+0xfda/0x1b70 net/ipv4/udp.c:1195 udp_sendmsg+0x2fe3/0x33c0 net/ipv4/udp.c:1483 inet_sendmsg+0x1fc/0x280 net/ipv4/af_inet.c:851 sock_sendmsg_nosec net/socket.c:712 [inline] __sock_sendmsg+0x267/0x380 net/socket.c:727 ____sys_sendmsg+0x91b/0xda0 net/socket.c:2566 ___sys_sendmsg+0x28d/0x3c0 net/socket.c:2620 __sys_sendmmsg+0x41d/0x880 net/socket.c:2702 __compat_sys_sendmmsg net/compat.c:360 [inline] __do_compat_sys_sendmmsg net/compat.c:367 [inline] __se_compat_sys_sendmmsg net/compat.c:364 [inline] __ia32_compat_sys_sendmmsg+0xc8/0x140 net/compat.c:364 ia32_sys_call+0x3ffa/0x41f0 arch/x86/include/generated/asm/syscalls_32.h:346 do_syscall_32_irqs_on arch/x86/entry/syscall_32.c:83 [inline] __do_fast_syscall_32+0xb0/0x110 arch/x86/entry/syscall_32.c:306 do_fast_syscall_32+0x38/0x80 arch/x86/entry/syscall_32.c:331 do_SYSENTER_32+0x1f/0x30 arch/x86/entry/syscall_32.c:369 entry_SYSENTER_compat_after_hwframe+0x84/0x8e CPU: 0 UID: 0 PID: 22408 Comm: syz.4.5165 Not tainted 6.15.0-rc3-syzkaller-00019-gbc3372351d0c #0 PREEMPT(undef) Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 02/12/2025 Reported-by: syzbot+04b9a82855c8aed20860@syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/68138dfa.050a0220.14dd7d.0017.GAE@google.com/ Fixes: 4754957f04f5 ("ipvs: do not use random local source address for tunnels") Signed-off-by: Julian Anastasov Acked-by: Simon Horman Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipvs/ip_vs_xmit.c | 27 ++++++++------------------- 1 file changed, 8 insertions(+), 19 deletions(-) diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c index 3313bceb6cc9..014f07740369 100644 --- a/net/netfilter/ipvs/ip_vs_xmit.c +++ b/net/netfilter/ipvs/ip_vs_xmit.c @@ -119,13 +119,12 @@ __mtu_check_toobig_v6(const struct sk_buff *skb, u32 mtu) return false; } -/* Get route to daddr, update *saddr, optionally bind route to saddr */ +/* Get route to daddr, optionally bind route to saddr */ static struct rtable *do_output_route4(struct net *net, __be32 daddr, - int rt_mode, __be32 *saddr) + int rt_mode, __be32 *ret_saddr) { struct flowi4 fl4; struct rtable *rt; - bool loop = false; memset(&fl4, 0, sizeof(fl4)); fl4.daddr = daddr; @@ -135,23 +134,17 @@ static struct rtable *do_output_route4(struct net *net, __be32 daddr, retry: rt = ip_route_output_key(net, &fl4); if (IS_ERR(rt)) { - /* Invalid saddr ? */ - if (PTR_ERR(rt) == -EINVAL && *saddr && - rt_mode & IP_VS_RT_MODE_CONNECT && !loop) { - *saddr = 0; - flowi4_update_output(&fl4, 0, daddr, 0); - goto retry; - } IP_VS_DBG_RL("ip_route_output error, dest: %pI4\n", &daddr); return NULL; - } else if (!*saddr && rt_mode & IP_VS_RT_MODE_CONNECT && fl4.saddr) { + } + if (rt_mode & IP_VS_RT_MODE_CONNECT && fl4.saddr) { ip_rt_put(rt); - *saddr = fl4.saddr; flowi4_update_output(&fl4, 0, daddr, fl4.saddr); - loop = true; + rt_mode = 0; goto retry; } - *saddr = fl4.saddr; + if (ret_saddr) + *ret_saddr = fl4.saddr; return rt; } @@ -344,19 +337,15 @@ __ip_vs_get_out_rt(struct netns_ipvs *ipvs, int skb_af, struct sk_buff *skb, if (ret_saddr) *ret_saddr = dest_dst->dst_saddr.ip; } else { - __be32 saddr = htonl(INADDR_ANY); - noref = 0; /* For such unconfigured boxes avoid many route lookups * for performance reasons because we do not remember saddr */ rt_mode &= ~IP_VS_RT_MODE_CONNECT; - rt = do_output_route4(net, daddr, rt_mode, &saddr); + rt = do_output_route4(net, daddr, rt_mode, ret_saddr); if (!rt) goto err_unreach; - if (ret_saddr) - *ret_saddr = saddr; } local = (rt->rt_flags & RTCF_LOCAL) ? 1 : 0; -- cgit From 8478a729c0462273188263136880480729e9efca Mon Sep 17 00:00:00 2001 From: Jozsef Kadlecsik Date: Wed, 7 May 2025 17:01:59 +0200 Subject: netfilter: ipset: fix region locking in hash types Region locking introduced in v5.6-rc4 contained three macros to handle the region locks: ahash_bucket_start(), ahash_bucket_end() which gave back the start and end hash bucket values belonging to a given region lock and ahash_region() which should give back the region lock belonging to a given hash bucket. The latter was incorrect which can lead to a race condition between the garbage collector and adding new elements when a hash type of set is defined with timeouts. Fixes: f66ee0410b1c ("netfilter: ipset: Fix "INFO: rcu detected stall in hash_xxx" reports") Reported-by: Kota Toda Signed-off-by: Jozsef Kadlecsik Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipset/ip_set_hash_gen.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/netfilter/ipset/ip_set_hash_gen.h b/net/netfilter/ipset/ip_set_hash_gen.h index cf3ce72c3de6..5251524b96af 100644 --- a/net/netfilter/ipset/ip_set_hash_gen.h +++ b/net/netfilter/ipset/ip_set_hash_gen.h @@ -64,7 +64,7 @@ struct hbucket { #define ahash_sizeof_regions(htable_bits) \ (ahash_numof_locks(htable_bits) * sizeof(struct ip_set_region)) #define ahash_region(n, htable_bits) \ - ((n) % ahash_numof_locks(htable_bits)) + ((n) / jhash_size(HTABLE_REGION_BITS)) #define ahash_bucket_start(h, htable_bits) \ ((htable_bits) < HTABLE_REGION_BITS ? 0 \ : (h) * jhash_size(HTABLE_REGION_BITS)) -- cgit From 6beb6835c1fbb3f676aebb51a5fee6b77fed9308 Mon Sep 17 00:00:00 2001 From: Eelco Chaudron Date: Tue, 6 May 2025 16:28:54 +0200 Subject: openvswitch: Fix unsafe attribute parsing in output_userspace() This patch replaces the manual Netlink attribute iteration in output_userspace() with nla_for_each_nested(), which ensures that only well-formed attributes are processed. Fixes: ccb1352e76cf ("net: Add Open vSwitch kernel components.") Signed-off-by: Eelco Chaudron Acked-by: Ilya Maximets Acked-by: Aaron Conole Link: https://patch.msgid.link/0bd65949df61591d9171c0dc13e42cea8941da10.1746541734.git.echaudro@redhat.com Signed-off-by: Jakub Kicinski --- net/openvswitch/actions.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c index 61fea7baae5d..2f22ca59586f 100644 --- a/net/openvswitch/actions.c +++ b/net/openvswitch/actions.c @@ -975,8 +975,7 @@ static int output_userspace(struct datapath *dp, struct sk_buff *skb, upcall.cmd = OVS_PACKET_CMD_ACTION; upcall.mru = OVS_CB(skb)->mru; - for (a = nla_data(attr), rem = nla_len(attr); rem > 0; - a = nla_next(a, &rem)) { + nla_for_each_nested(a, attr, rem) { switch (nla_type(a)) { case OVS_USERSPACE_ATTR_USERDATA: upcall.userdata = a; -- cgit From 4a7843cc8a41b9612becccc07715ed017770eb89 Mon Sep 17 00:00:00 2001 From: Lorenzo Bianconi Date: Tue, 6 May 2025 18:56:47 +0200 Subject: net: airoha: Add missing field to ppe_mbox_data struct The official Airoha EN7581 firmware requires adding max_packet field in ppe_mbox_data struct while the unofficial one used to develop the Airoha EN7581 flowtable support does not require this field. This patch does not introduce any real backwards compatible issue since EN7581 fw is not publicly available in linux-firmware or other repositories (e.g. OpenWrt) yet and the official fw version will use this new layout. For this reason this change needs to be backported. Moreover, make explicit the padding added by the compiler introducing the rsv array in init_info struct. At the same time use u32 instead of int for init_info and set_info struct definitions in ppe_mbox_data struct. Fixes: 23290c7bc190d ("net: airoha: Introduce Airoha NPU support") Reviewed-by: Simon Horman Reviewed-by: Jacob Keller Signed-off-by: Lorenzo Bianconi Link: https://patch.msgid.link/20250506-airoha-en7581-fix-ppe_mbox_data-v5-1-29cabed6864d@kernel.org Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/airoha/airoha_npu.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/drivers/net/ethernet/airoha/airoha_npu.c b/drivers/net/ethernet/airoha/airoha_npu.c index 7a5710f9ccf6..ead0625e781f 100644 --- a/drivers/net/ethernet/airoha/airoha_npu.c +++ b/drivers/net/ethernet/airoha/airoha_npu.c @@ -104,12 +104,14 @@ struct ppe_mbox_data { u8 xpon_hal_api; u8 wan_xsi; u8 ct_joyme4; - int ppe_type; - int wan_mode; - int wan_sel; + u8 max_packet; + u8 rsv[3]; + u32 ppe_type; + u32 wan_mode; + u32 wan_sel; } init_info; struct { - int func_id; + u32 func_id; u32 size; u32 data; } set_info; -- cgit From c4327229948879814229b46aa26a750718888503 Mon Sep 17 00:00:00 2001 From: Paul Chaignon Date: Mon, 5 May 2025 21:58:04 +0200 Subject: bpf: Scrub packet on bpf_redirect_peer When bpf_redirect_peer is used to redirect packets to a device in another network namespace, the skb isn't scrubbed. That can lead skb information from one namespace to be "misused" in another namespace. As one example, this is causing Cilium to drop traffic when using bpf_redirect_peer to redirect packets that just went through IPsec decryption to a container namespace. The following pwru trace shows (1) the packet path from the host's XFRM layer to the container's XFRM layer where it's dropped and (2) the number of active skb extensions at each function. NETNS MARK IFACE TUPLE FUNC 4026533547 d00 eth0 10.244.3.124:35473->10.244.2.158:53 xfrm_rcv_cb .active_extensions = (__u8)2, 4026533547 d00 eth0 10.244.3.124:35473->10.244.2.158:53 xfrm4_rcv_cb .active_extensions = (__u8)2, 4026533547 d00 eth0 10.244.3.124:35473->10.244.2.158:53 gro_cells_receive .active_extensions = (__u8)2, [...] 4026533547 0 eth0 10.244.3.124:35473->10.244.2.158:53 skb_do_redirect .active_extensions = (__u8)2, 4026534999 0 eth0 10.244.3.124:35473->10.244.2.158:53 ip_rcv .active_extensions = (__u8)2, 4026534999 0 eth0 10.244.3.124:35473->10.244.2.158:53 ip_rcv_core .active_extensions = (__u8)2, [...] 4026534999 0 eth0 10.244.3.124:35473->10.244.2.158:53 udp_queue_rcv_one_skb .active_extensions = (__u8)2, 4026534999 0 eth0 10.244.3.124:35473->10.244.2.158:53 __xfrm_policy_check .active_extensions = (__u8)2, 4026534999 0 eth0 10.244.3.124:35473->10.244.2.158:53 __xfrm_decode_session .active_extensions = (__u8)2, 4026534999 0 eth0 10.244.3.124:35473->10.244.2.158:53 security_xfrm_decode_session .active_extensions = (__u8)2, 4026534999 0 eth0 10.244.3.124:35473->10.244.2.158:53 kfree_skb_reason(SKB_DROP_REASON_XFRM_POLICY) .active_extensions = (__u8)2, In this case, there are no XFRM policies in the container's network namespace so the drop is unexpected. When we decrypt the IPsec packet, the XFRM state used for decryption is set in the skb extensions. This information is preserved across the netns switch. When we reach the XFRM policy check in the container's netns, __xfrm_policy_check drops the packet with LINUX_MIB_XFRMINNOPOLS because a (container-side) XFRM policy can't be found that matches the (host-side) XFRM state used for decryption. This patch fixes this by scrubbing the packet when using bpf_redirect_peer, as is done on typical netns switches via veth devices except skb->mark and skb->tstamp are not zeroed. Fixes: 9aa1206e8f482 ("bpf: Add redirect_peer helper") Signed-off-by: Paul Chaignon Acked-by: Daniel Borkmann Acked-by: Martin KaFai Lau Link: https://patch.msgid.link/1728ead5e0fe45e7a6542c36bd4e3ca07a73b7d6.1746460653.git.paul.chaignon@gmail.com Signed-off-by: Jakub Kicinski --- net/core/filter.c | 1 + 1 file changed, 1 insertion(+) diff --git a/net/core/filter.c b/net/core/filter.c index 79cab4d78dc3..577a4504e26f 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -2509,6 +2509,7 @@ int skb_do_redirect(struct sk_buff *skb) goto out_drop; skb->dev = dev; dev_sw_netstats_rx_add(dev, skb->len); + skb_scrub_packet(skb, false); return -EAGAIN; } return flags & BPF_F_NEIGH ? -- cgit From f5c79ffdc250bc8c90fd4fdf1e5d7ac4647912d5 Mon Sep 17 00:00:00 2001 From: Paul Chaignon Date: Mon, 5 May 2025 21:58:39 +0200 Subject: bpf: Clarify handling of mark and tstamp by redirect_peer When switching network namespaces with the bpf_redirect_peer helper, the skb->mark and skb->tstamp fields are not zeroed out like they can be on a typical netns switch. This patch clarifies that in the helper description. Signed-off-by: Paul Chaignon Acked-by: Daniel Borkmann Acked-by: Martin KaFai Lau Link: https://patch.msgid.link/ccc86af26d43c5c0b776bcba2601b7479c0d46d0.1746460653.git.paul.chaignon@gmail.com Signed-off-by: Jakub Kicinski --- include/uapi/linux/bpf.h | 3 +++ tools/include/uapi/linux/bpf.h | 3 +++ 2 files changed, 6 insertions(+) diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index 28705ae67784..fd404729b115 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -4968,6 +4968,9 @@ union bpf_attr { * the netns switch takes place from ingress to ingress without * going through the CPU's backlog queue. * + * *skb*\ **->mark** and *skb*\ **->tstamp** are not cleared during + * the netns switch. + * * The *flags* argument is reserved and must be 0. The helper is * currently only supported for tc BPF program types at the * ingress hook and for veth and netkit target device types. The diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index 28705ae67784..fd404729b115 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -4968,6 +4968,9 @@ union bpf_attr { * the netns switch takes place from ingress to ingress without * going through the CPU's backlog queue. * + * *skb*\ **->mark** and *skb*\ **->tstamp** are not cleared during + * the netns switch. + * * The *flags* argument is reserved and must be 0. The helper is * currently only supported for tc BPF program types at the * ingress hook and for veth and netkit target device types. The -- cgit From e5641daa0ea1932e2b0fa7f27843e712244f6538 Mon Sep 17 00:00:00 2001 From: Meghana Malladi Date: Tue, 6 May 2025 16:35:44 +0530 Subject: net: ti: icssg-prueth: Set XDP feature flags for ndev xdp_features demonstrates what all XDP capabilities are supported on a given network device. The driver needs to set these xdp_features flag to let the network stack know what XDP features a given driver is supporting. These flags need to be set for a given ndev irrespective of any XDP program being loaded or not. Fixes: 62aa3246f462 ("net: ti: icssg-prueth: Add XDP support") Signed-off-by: Meghana Malladi Reviewed-by: Jacob Keller Link: https://patch.msgid.link/20250506110546.4065715-2-m-malladi@ti.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/ti/icssg/icssg_prueth.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/drivers/net/ethernet/ti/icssg/icssg_prueth.c b/drivers/net/ethernet/ti/icssg/icssg_prueth.c index 443f90fa6557..ee35fecf61e7 100644 --- a/drivers/net/ethernet/ti/icssg/icssg_prueth.c +++ b/drivers/net/ethernet/ti/icssg/icssg_prueth.c @@ -1109,11 +1109,6 @@ static int emac_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frame static int emac_xdp_setup(struct prueth_emac *emac, struct netdev_bpf *bpf) { struct bpf_prog *prog = bpf->prog; - xdp_features_t val; - - val = NETDEV_XDP_ACT_BASIC | NETDEV_XDP_ACT_REDIRECT | - NETDEV_XDP_ACT_NDO_XMIT; - xdp_set_features_flag(emac->ndev, val); if (!emac->xdpi.prog && !prog) return 0; @@ -1291,6 +1286,10 @@ static int prueth_netdev_init(struct prueth *prueth, ndev->hw_features = NETIF_F_SG; ndev->features = ndev->hw_features | NETIF_F_HW_VLAN_CTAG_FILTER; ndev->hw_features |= NETIF_PRUETH_HSR_OFFLOAD_FEATURES; + xdp_set_features_flag(ndev, + NETDEV_XDP_ACT_BASIC | + NETDEV_XDP_ACT_REDIRECT | + NETDEV_XDP_ACT_NDO_XMIT); netif_napi_add(ndev, &emac->napi_rx, icssg_napi_rx_poll); hrtimer_setup(&emac->rx_hrtimer, &emac_rx_timer_callback, CLOCK_MONOTONIC, -- cgit From 8b3fae3e2376b70b7a76005263bc34d034b9c7bf Mon Sep 17 00:00:00 2001 From: Meghana Malladi Date: Tue, 6 May 2025 16:35:45 +0530 Subject: net: ti: icssg-prueth: Fix kernel panic during concurrent Tx queue access Add __netif_tx_lock() to ensure that only one packet is being transmitted at a time to avoid race conditions in the netif_txq struct and prevent packet data corruption. Failing to do so causes kernel panic with the following error: [ 2184.746764] ------------[ cut here ]------------ [ 2184.751412] kernel BUG at lib/dynamic_queue_limits.c:99! [ 2184.756728] Internal error: Oops - BUG: 00000000f2000800 [#1] PREEMPT SMP logs: https://gist.github.com/MeghanaMalladiTI/9c7aa5fc3b7fb03f87c74aad487956e9 The lock is acquired before calling emac_xmit_xdp_frame() and released after the call returns. This ensures that the TX queue is protected from concurrent access during the transmission of XDP frames. Fixes: 62aa3246f462 ("net: ti: icssg-prueth: Add XDP support") Signed-off-by: Meghana Malladi Reviewed-by: Jacob Keller Link: https://patch.msgid.link/20250506110546.4065715-3-m-malladi@ti.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/ti/icssg/icssg_common.c | 7 ++++++- drivers/net/ethernet/ti/icssg/icssg_prueth.c | 7 ++++++- 2 files changed, 12 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/ti/icssg/icssg_common.c b/drivers/net/ethernet/ti/icssg/icssg_common.c index b4be76e13a2f..38a80e004d1c 100644 --- a/drivers/net/ethernet/ti/icssg/icssg_common.c +++ b/drivers/net/ethernet/ti/icssg/icssg_common.c @@ -650,6 +650,8 @@ static u32 emac_run_xdp(struct prueth_emac *emac, struct xdp_buff *xdp, struct page *page, u32 *len) { struct net_device *ndev = emac->ndev; + struct netdev_queue *netif_txq; + int cpu = smp_processor_id(); struct bpf_prog *xdp_prog; struct xdp_frame *xdpf; u32 pkt_len = *len; @@ -669,8 +671,11 @@ static u32 emac_run_xdp(struct prueth_emac *emac, struct xdp_buff *xdp, goto drop; } - q_idx = smp_processor_id() % emac->tx_ch_num; + q_idx = cpu % emac->tx_ch_num; + netif_txq = netdev_get_tx_queue(ndev, q_idx); + __netif_tx_lock(netif_txq, cpu); result = emac_xmit_xdp_frame(emac, xdpf, page, q_idx); + __netif_tx_unlock(netif_txq); if (result == ICSSG_XDP_CONSUMED) { ndev->stats.tx_dropped++; goto drop; diff --git a/drivers/net/ethernet/ti/icssg/icssg_prueth.c b/drivers/net/ethernet/ti/icssg/icssg_prueth.c index ee35fecf61e7..86fc1278127c 100644 --- a/drivers/net/ethernet/ti/icssg/icssg_prueth.c +++ b/drivers/net/ethernet/ti/icssg/icssg_prueth.c @@ -1075,17 +1075,21 @@ static int emac_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frame { struct prueth_emac *emac = netdev_priv(dev); struct net_device *ndev = emac->ndev; + struct netdev_queue *netif_txq; + int cpu = smp_processor_id(); struct xdp_frame *xdpf; unsigned int q_idx; int nxmit = 0; u32 err; int i; - q_idx = smp_processor_id() % emac->tx_ch_num; + q_idx = cpu % emac->tx_ch_num; + netif_txq = netdev_get_tx_queue(ndev, q_idx); if (unlikely(flags & ~XDP_XMIT_FLAGS_MASK)) return -EINVAL; + __netif_tx_lock(netif_txq, cpu); for (i = 0; i < n; i++) { xdpf = frames[i]; err = emac_xmit_xdp_frame(emac, xdpf, NULL, q_idx); @@ -1095,6 +1099,7 @@ static int emac_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frame } nxmit++; } + __netif_tx_unlock(netif_txq); return nxmit; } -- cgit From 1884fc85ae6ed0652fa324c66b1d89e25174e73b Mon Sep 17 00:00:00 2001 From: Meghana Malladi Date: Tue, 6 May 2025 16:35:46 +0530 Subject: net: ti: icssg-prueth: Report BQL before sending XDP packets When sending out any kind of traffic, it is essential that the driver keeps reporting BQL of the number of bytes that have been sent so that BQL can track the amount of data in the queue and prevents it from overflowing. If BQL is not reported, the driver may continue sending packets even when the queue is full, leading to packet loss, congestion and decreased network performance. Currently this is missing in emac_xmit_xdp_frame() and this patch fixes it. Fixes: 62aa3246f462 ("net: ti: icssg-prueth: Add XDP support") Signed-off-by: Meghana Malladi Reviewed-by: Jacob Keller Link: https://patch.msgid.link/20250506110546.4065715-4-m-malladi@ti.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/ti/icssg/icssg_common.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/ti/icssg/icssg_common.c b/drivers/net/ethernet/ti/icssg/icssg_common.c index 38a80e004d1c..d88a0180294e 100644 --- a/drivers/net/ethernet/ti/icssg/icssg_common.c +++ b/drivers/net/ethernet/ti/icssg/icssg_common.c @@ -187,7 +187,6 @@ int emac_tx_complete_packets(struct prueth_emac *emac, int chn, xdp_return_frame(xdpf); break; default: - netdev_err(ndev, "tx_complete: invalid swdata type %d\n", swdata->type); prueth_xmit_free(tx_chn, desc_tx); ndev->stats.tx_dropped++; continue; @@ -567,6 +566,7 @@ u32 emac_xmit_xdp_frame(struct prueth_emac *emac, { struct cppi5_host_desc_t *first_desc; struct net_device *ndev = emac->ndev; + struct netdev_queue *netif_txq; struct prueth_tx_chn *tx_chn; dma_addr_t desc_dma, buf_dma; struct prueth_swdata *swdata; @@ -620,12 +620,17 @@ u32 emac_xmit_xdp_frame(struct prueth_emac *emac, swdata->data.xdpf = xdpf; } + /* Report BQL before sending the packet */ + netif_txq = netdev_get_tx_queue(ndev, tx_chn->id); + netdev_tx_sent_queue(netif_txq, xdpf->len); + cppi5_hdesc_set_pktlen(first_desc, xdpf->len); desc_dma = k3_cppi_desc_pool_virt2dma(tx_chn->desc_pool, first_desc); ret = k3_udma_glue_push_tx_chn(tx_chn->tx_chn, first_desc, desc_dma); if (ret) { netdev_err(ndev, "xdp tx: push failed: %d\n", ret); + netdev_tx_completed_queue(netif_txq, 1, xdpf->len); goto drop_free_descs; } @@ -984,6 +989,7 @@ enum netdev_tx icssg_ndo_start_xmit(struct sk_buff *skb, struct net_device *ndev ret = k3_udma_glue_push_tx_chn(tx_chn->tx_chn, first_desc, desc_dma); if (ret) { netdev_err(ndev, "tx: push failed: %d\n", ret); + netdev_tx_completed_queue(netif_txq, 1, pkt_len); goto drop_free_descs; } -- cgit From 5f93185a757ff38b36f849c659aeef368db15a68 Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:00 +0200 Subject: net: dsa: b53: allow leaky reserved multicast Allow reserved multicast to ignore VLAN membership so STP and other management protocols work without a PVID VLAN configured when using a vlan aware bridge. Fixes: 967dd82ffc52 ("net: dsa: b53: Add support for Broadcom RoboSwitch") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-2-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index e5ba71897906..62866165ad03 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -373,9 +373,11 @@ static void b53_enable_vlan(struct b53_device *dev, int port, bool enable, b53_read8(dev, B53_VLAN_PAGE, B53_VLAN_CTRL5, &vc5); } + vc1 &= ~VC1_RX_MCST_FWD_EN; + if (enable) { vc0 |= VC0_VLAN_EN | VC0_VID_CHK_EN | VC0_VID_HASH_VID; - vc1 |= VC1_RX_MCST_UNTAG_EN | VC1_RX_MCST_FWD_EN; + vc1 |= VC1_RX_MCST_UNTAG_EN; vc4 &= ~VC4_ING_VID_CHECK_MASK; if (enable_filtering) { vc4 |= VC4_ING_VID_VIO_DROP << VC4_ING_VID_CHECK_S; @@ -393,7 +395,7 @@ static void b53_enable_vlan(struct b53_device *dev, int port, bool enable, } else { vc0 &= ~(VC0_VLAN_EN | VC0_VID_CHK_EN | VC0_VID_HASH_VID); - vc1 &= ~(VC1_RX_MCST_UNTAG_EN | VC1_RX_MCST_FWD_EN); + vc1 &= ~VC1_RX_MCST_UNTAG_EN; vc4 &= ~VC4_ING_VID_CHECK_MASK; vc5 &= ~VC5_DROP_VTABLE_MISS; -- cgit From 425f11d4cc9bd9e97e6825d9abb2c51a068ca7b5 Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:01 +0200 Subject: net: dsa: b53: keep CPU port always tagged again The Broadcom management header does not carry the original VLAN tag state information, just the ingress port, so for untagged frames we do not know from which VLAN they originated. Therefore keep the CPU port always tagged except for VLAN 0. Fixes the following setup: $ ip link add br0 type bridge vlan_filtering 1 $ ip link set sw1p1 master br0 $ bridge vlan add dev br0 pvid untagged self $ ip link add sw1p2.10 link sw1p2 type vlan id 10 Where VID 10 would stay untagged on the CPU port. Fixes: 2c32a3d3c233 ("net: dsa: b53: Do not force CPU to be always tagged") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-3-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index 62866165ad03..9d4fb54b4ced 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -1135,6 +1135,11 @@ static int b53_setup(struct dsa_switch *ds) */ ds->untag_bridge_pvid = dev->tag_protocol == DSA_TAG_PROTO_NONE; + /* The switch does not tell us the original VLAN for untagged + * packets, so keep the CPU port always tagged. + */ + ds->untag_vlan_aware_bridge_pvid = true; + ret = b53_reset_switch(dev); if (ret) { dev_err(ds->dev, "failed to reset switch\n"); @@ -1545,6 +1550,9 @@ int b53_vlan_add(struct dsa_switch *ds, int port, if (vlan->vid == 0 && vlan->vid == b53_default_pvid(dev)) untagged = true; + if (vlan->vid > 0 && dsa_is_cpu_port(ds, port)) + untagged = false; + vl->members |= BIT(port); if (untagged && !b53_vlan_port_needs_forced_tagged(ds, port)) vl->untag |= BIT(port); -- cgit From f480851981043d9bb6447ca9883ade9247b9a0ad Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:02 +0200 Subject: net: dsa: b53: fix clearing PVID of a port Currently the PVID of ports are only set when adding/updating VLANs with PVID set or removing VLANs, but not when clearing the PVID flag of a VLAN. E.g. the following flow $ ip link add br0 type bridge vlan_filtering 1 $ ip link set sw1p1 master bridge $ bridge vlan add dev sw1p1 vid 10 pvid untagged $ bridge vlan add dev sw1p1 vid 10 untagged Would keep the PVID set as 10, despite the flag being cleared. Fix this by checking if we need to unset the PVID on vlan updates. Fixes: a2482d2ce349 ("net: dsa: b53: Plug in VLAN support") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-4-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index 9d4fb54b4ced..65d74c455c57 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -1537,12 +1537,21 @@ int b53_vlan_add(struct dsa_switch *ds, int port, bool untagged = vlan->flags & BRIDGE_VLAN_INFO_UNTAGGED; bool pvid = vlan->flags & BRIDGE_VLAN_INFO_PVID; struct b53_vlan *vl; + u16 old_pvid, new_pvid; int err; err = b53_vlan_prepare(ds, port, vlan); if (err) return err; + b53_read16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), &old_pvid); + if (pvid) + new_pvid = vlan->vid; + else if (!pvid && vlan->vid == old_pvid) + new_pvid = b53_default_pvid(dev); + else + new_pvid = old_pvid; + vl = &dev->vlans[vlan->vid]; b53_get_vlan_entry(dev, vlan->vid, vl); @@ -1562,9 +1571,9 @@ int b53_vlan_add(struct dsa_switch *ds, int port, b53_set_vlan_entry(dev, vlan->vid, vl); b53_fast_age_vlan(dev, vlan->vid); - if (pvid && !dsa_is_cpu_port(ds, port)) { + if (!dsa_is_cpu_port(ds, port) && new_pvid != old_pvid) { b53_write16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), - vlan->vid); + new_pvid); b53_fast_age_vlan(dev, vlan->vid); } -- cgit From 083c6b28c0cbcd83b6af1a10f2c82937129b3438 Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:03 +0200 Subject: net: dsa: b53: fix flushing old pvid VLAN on pvid change Presumably the intention here was to flush the VLAN of the old pvid, not the added VLAN again, which we already flushed before. Fixes: a2482d2ce349 ("net: dsa: b53: Plug in VLAN support") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-5-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index 65d74c455c57..c67c0b5fbc1b 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -1574,7 +1574,7 @@ int b53_vlan_add(struct dsa_switch *ds, int port, if (!dsa_is_cpu_port(ds, port) && new_pvid != old_pvid) { b53_write16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), new_pvid); - b53_fast_age_vlan(dev, vlan->vid); + b53_fast_age_vlan(dev, old_pvid); } return 0; -- cgit From a1c1901c5cc881425cc45992ab6c5418174e9e5a Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:04 +0200 Subject: net: dsa: b53: fix VLAN ID for untagged vlan on bridge leave The untagged default VLAN is added to the default vlan, which may be one, but we modify the VLAN 0 entry on bridge leave. Fix this to use the correct VLAN entry for the default pvid. Fixes: fea83353177a ("net: dsa: b53: Fix default VLAN ID") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-6-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index c67c0b5fbc1b..c60b552b945c 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -1986,7 +1986,7 @@ EXPORT_SYMBOL(b53_br_join); void b53_br_leave(struct dsa_switch *ds, int port, struct dsa_bridge bridge) { struct b53_device *dev = ds->priv; - struct b53_vlan *vl = &dev->vlans[0]; + struct b53_vlan *vl; s8 cpu_port = dsa_to_port(ds, port)->cpu_dp->index; unsigned int i; u16 pvlan, reg, pvid; @@ -2012,6 +2012,7 @@ void b53_br_leave(struct dsa_switch *ds, int port, struct dsa_bridge bridge) dev->ports[port].vlan_ctl_mask = pvlan; pvid = b53_default_pvid(dev); + vl = &dev->vlans[pvid]; /* Make this port join all VLANs without VLAN entries */ if (is58xx(dev)) { -- cgit From 13b152ae40495966501697693f048f47430c50fd Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:05 +0200 Subject: net: dsa: b53: always rejoin default untagged VLAN on bridge leave While JOIN_ALL_VLAN allows to join all VLANs, we still need to keep the default VLAN enabled so that untagged traffic stays untagged. So rejoin the default VLAN even for switches with JOIN_ALL_VLAN support. Fixes: 48aea33a77ab ("net: dsa: b53: Add JOIN_ALL_VLAN support") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-7-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index c60b552b945c..4871e117f5ef 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -2021,12 +2021,12 @@ void b53_br_leave(struct dsa_switch *ds, int port, struct dsa_bridge bridge) if (!(reg & BIT(cpu_port))) reg |= BIT(cpu_port); b53_write16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, reg); - } else { - b53_get_vlan_entry(dev, pvid, vl); - vl->members |= BIT(port) | BIT(cpu_port); - vl->untag |= BIT(port) | BIT(cpu_port); - b53_set_vlan_entry(dev, pvid, vl); } + + b53_get_vlan_entry(dev, pvid, vl); + vl->members |= BIT(port) | BIT(cpu_port); + vl->untag |= BIT(port) | BIT(cpu_port); + b53_set_vlan_entry(dev, pvid, vl); } EXPORT_SYMBOL(b53_br_leave); -- cgit From 45e9d59d39503bb3e6ab4d258caea4ba6496e2dc Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:06 +0200 Subject: net: dsa: b53: do not allow to configure VLAN 0 Since we cannot set forwarding destinations per VLAN, we should not have a VLAN 0 configured, as it would allow untagged traffic to work across ports on VLAN aware bridges regardless if a PVID untagged VLAN exists. So remove the VLAN 0 on join, an re-add it on leave. But only do so if we have a VLAN aware bridge, as without it, untagged traffic would become tagged with VID 0 on a VLAN unaware bridge. Fixes: a2482d2ce349 ("net: dsa: b53: Plug in VLAN support") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-8-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 36 +++++++++++++++++++++++++++--------- 1 file changed, 27 insertions(+), 9 deletions(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index 4871e117f5ef..0b28791cca52 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -1544,6 +1544,9 @@ int b53_vlan_add(struct dsa_switch *ds, int port, if (err) return err; + if (vlan->vid == 0) + return 0; + b53_read16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), &old_pvid); if (pvid) new_pvid = vlan->vid; @@ -1556,10 +1559,7 @@ int b53_vlan_add(struct dsa_switch *ds, int port, b53_get_vlan_entry(dev, vlan->vid, vl); - if (vlan->vid == 0 && vlan->vid == b53_default_pvid(dev)) - untagged = true; - - if (vlan->vid > 0 && dsa_is_cpu_port(ds, port)) + if (dsa_is_cpu_port(ds, port)) untagged = false; vl->members |= BIT(port); @@ -1589,6 +1589,9 @@ int b53_vlan_del(struct dsa_switch *ds, int port, struct b53_vlan *vl; u16 pvid; + if (vlan->vid == 0) + return 0; + b53_read16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), &pvid); vl = &dev->vlans[vlan->vid]; @@ -1935,8 +1938,9 @@ int b53_br_join(struct dsa_switch *ds, int port, struct dsa_bridge bridge, bool *tx_fwd_offload, struct netlink_ext_ack *extack) { struct b53_device *dev = ds->priv; + struct b53_vlan *vl; s8 cpu_port = dsa_to_port(ds, port)->cpu_dp->index; - u16 pvlan, reg; + u16 pvlan, reg, pvid; unsigned int i; /* On 7278, port 7 which connects to the ASP should only receive @@ -1945,6 +1949,9 @@ int b53_br_join(struct dsa_switch *ds, int port, struct dsa_bridge bridge, if (dev->chip_id == BCM7278_DEVICE_ID && port == 7) return -EINVAL; + pvid = b53_default_pvid(dev); + vl = &dev->vlans[pvid]; + /* Make this port leave the all VLANs join since we will have proper * VLAN entries from now on */ @@ -1956,6 +1963,15 @@ int b53_br_join(struct dsa_switch *ds, int port, struct dsa_bridge bridge, b53_write16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, reg); } + if (ds->vlan_filtering) { + b53_get_vlan_entry(dev, pvid, vl); + vl->members &= ~BIT(port); + if (vl->members == BIT(cpu_port)) + vl->members &= ~BIT(cpu_port); + vl->untag = vl->members; + b53_set_vlan_entry(dev, pvid, vl); + } + b53_read16(dev, B53_PVLAN_PAGE, B53_PVLAN_PORT_MASK(port), &pvlan); b53_for_each_port(dev, i) { @@ -2023,10 +2039,12 @@ void b53_br_leave(struct dsa_switch *ds, int port, struct dsa_bridge bridge) b53_write16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, reg); } - b53_get_vlan_entry(dev, pvid, vl); - vl->members |= BIT(port) | BIT(cpu_port); - vl->untag |= BIT(port) | BIT(cpu_port); - b53_set_vlan_entry(dev, pvid, vl); + if (ds->vlan_filtering) { + b53_get_vlan_entry(dev, pvid, vl); + vl->members |= BIT(port) | BIT(cpu_port); + vl->untag |= BIT(port) | BIT(cpu_port); + b53_set_vlan_entry(dev, pvid, vl); + } } EXPORT_SYMBOL(b53_br_leave); -- cgit From f089652b6b16452535dcc5cbaa6e2bb05acd3f93 Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:07 +0200 Subject: net: dsa: b53: do not program vlans when vlan filtering is off Documentation/networking/switchdev.rst says: - with VLAN filtering turned off: the bridge is strictly VLAN unaware and its data path will process all Ethernet frames as if they are VLAN-untagged. The bridge VLAN database can still be modified, but the modifications should have no effect while VLAN filtering is turned off. This breaks if we immediately apply the VLAN configuration, so skip writing it when vlan_filtering is off. Fixes: 0ee2af4ebbe3 ("net: dsa: set configure_vlan_while_not_filtering to true by default") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-9-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 48 +++++++++++++++++++++++----------------- 1 file changed, 28 insertions(+), 20 deletions(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index 0b28791cca52..ee2f1be62618 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -1547,6 +1547,9 @@ int b53_vlan_add(struct dsa_switch *ds, int port, if (vlan->vid == 0) return 0; + if (!ds->vlan_filtering) + return 0; + b53_read16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), &old_pvid); if (pvid) new_pvid = vlan->vid; @@ -1592,6 +1595,9 @@ int b53_vlan_del(struct dsa_switch *ds, int port, if (vlan->vid == 0) return 0; + if (!ds->vlan_filtering) + return 0; + b53_read16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), &pvid); vl = &dev->vlans[vlan->vid]; @@ -1952,18 +1958,20 @@ int b53_br_join(struct dsa_switch *ds, int port, struct dsa_bridge bridge, pvid = b53_default_pvid(dev); vl = &dev->vlans[pvid]; - /* Make this port leave the all VLANs join since we will have proper - * VLAN entries from now on - */ - if (is58xx(dev)) { - b53_read16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, ®); - reg &= ~BIT(port); - if ((reg & BIT(cpu_port)) == BIT(cpu_port)) - reg &= ~BIT(cpu_port); - b53_write16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, reg); - } - if (ds->vlan_filtering) { + /* Make this port leave the all VLANs join since we will have + * proper VLAN entries from now on + */ + if (is58xx(dev)) { + b53_read16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, + ®); + reg &= ~BIT(port); + if ((reg & BIT(cpu_port)) == BIT(cpu_port)) + reg &= ~BIT(cpu_port); + b53_write16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, + reg); + } + b53_get_vlan_entry(dev, pvid, vl); vl->members &= ~BIT(port); if (vl->members == BIT(cpu_port)) @@ -2030,16 +2038,16 @@ void b53_br_leave(struct dsa_switch *ds, int port, struct dsa_bridge bridge) pvid = b53_default_pvid(dev); vl = &dev->vlans[pvid]; - /* Make this port join all VLANs without VLAN entries */ - if (is58xx(dev)) { - b53_read16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, ®); - reg |= BIT(port); - if (!(reg & BIT(cpu_port))) - reg |= BIT(cpu_port); - b53_write16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, reg); - } - if (ds->vlan_filtering) { + /* Make this port join all VLANs without VLAN entries */ + if (is58xx(dev)) { + b53_read16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, ®); + reg |= BIT(port); + if (!(reg & BIT(cpu_port))) + reg |= BIT(cpu_port); + b53_write16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, reg); + } + b53_get_vlan_entry(dev, pvid, vl); vl->members |= BIT(port) | BIT(cpu_port); vl->untag |= BIT(port) | BIT(cpu_port); -- cgit From 2dc2bd57111582895e10f54ea380329c89873f1c Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:08 +0200 Subject: net: dsa: b53: fix toggling vlan_filtering To allow runtime switching between vlan aware and vlan non-aware mode, we need to properly keep track of any bridge VLAN configuration. Likewise, we need to know when we actually switch between both modes, to not have to rewrite the full VLAN table every time we update the VLANs. So keep track of the current vlan_filtering mode, and on changes, apply the appropriate VLAN configuration. Fixes: 0ee2af4ebbe3 ("net: dsa: set configure_vlan_while_not_filtering to true by default") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-10-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 104 +++++++++++++++++++++++++++------------ drivers/net/dsa/b53/b53_priv.h | 2 + 2 files changed, 75 insertions(+), 31 deletions(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index ee2f1be62618..0a7749b416f7 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -763,6 +763,22 @@ static bool b53_vlan_port_needs_forced_tagged(struct dsa_switch *ds, int port) return dev->tag_protocol == DSA_TAG_PROTO_NONE && dsa_is_cpu_port(ds, port); } +static bool b53_vlan_port_may_join_untagged(struct dsa_switch *ds, int port) +{ + struct b53_device *dev = ds->priv; + struct dsa_port *dp; + + if (!dev->vlan_filtering) + return true; + + dp = dsa_to_port(ds, port); + + if (dsa_port_is_cpu(dp)) + return true; + + return dp->bridge == NULL; +} + int b53_configure_vlan(struct dsa_switch *ds) { struct b53_device *dev = ds->priv; @@ -781,7 +797,7 @@ int b53_configure_vlan(struct dsa_switch *ds) b53_do_vlan_op(dev, VTA_CMD_CLEAR); } - b53_enable_vlan(dev, -1, dev->vlan_enabled, ds->vlan_filtering); + b53_enable_vlan(dev, -1, dev->vlan_enabled, dev->vlan_filtering); /* Create an untagged VLAN entry for the default PVID in case * CONFIG_VLAN_8021Q is disabled and there are no calls to @@ -789,26 +805,39 @@ int b53_configure_vlan(struct dsa_switch *ds) * entry. Do this only when the tagging protocol is not * DSA_TAG_PROTO_NONE */ + v = &dev->vlans[def_vid]; b53_for_each_port(dev, i) { - v = &dev->vlans[def_vid]; - v->members |= BIT(i); + if (!b53_vlan_port_may_join_untagged(ds, i)) + continue; + + vl.members |= BIT(i); if (!b53_vlan_port_needs_forced_tagged(ds, i)) - v->untag = v->members; - b53_write16(dev, B53_VLAN_PAGE, - B53_VLAN_PORT_DEF_TAG(i), def_vid); + vl.untag = vl.members; + b53_write16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(i), + def_vid); } + b53_set_vlan_entry(dev, def_vid, &vl); - /* Upon initial call we have not set-up any VLANs, but upon - * system resume, we need to restore all VLAN entries. - */ - for (vid = def_vid; vid < dev->num_vlans; vid++) { - v = &dev->vlans[vid]; + if (dev->vlan_filtering) { + /* Upon initial call we have not set-up any VLANs, but upon + * system resume, we need to restore all VLAN entries. + */ + for (vid = def_vid + 1; vid < dev->num_vlans; vid++) { + v = &dev->vlans[vid]; - if (!v->members) - continue; + if (!v->members) + continue; + + b53_set_vlan_entry(dev, vid, v); + b53_fast_age_vlan(dev, vid); + } - b53_set_vlan_entry(dev, vid, v); - b53_fast_age_vlan(dev, vid); + b53_for_each_port(dev, i) { + if (!dsa_is_cpu_port(ds, i)) + b53_write16(dev, B53_VLAN_PAGE, + B53_VLAN_PORT_DEF_TAG(i), + dev->ports[i].pvid); + } } return 0; @@ -1127,7 +1156,9 @@ EXPORT_SYMBOL(b53_setup_devlink_resources); static int b53_setup(struct dsa_switch *ds) { struct b53_device *dev = ds->priv; + struct b53_vlan *vl; unsigned int port; + u16 pvid; int ret; /* Request bridge PVID untagged when DSA_TAG_PROTO_NONE is set @@ -1146,6 +1177,15 @@ static int b53_setup(struct dsa_switch *ds) return ret; } + /* setup default vlan for filtering mode */ + pvid = b53_default_pvid(dev); + vl = &dev->vlans[pvid]; + b53_for_each_port(dev, port) { + vl->members |= BIT(port); + if (!b53_vlan_port_needs_forced_tagged(ds, port)) + vl->untag |= BIT(port); + } + b53_reset_mib(dev); ret = b53_apply_config(dev); @@ -1499,7 +1539,10 @@ int b53_vlan_filtering(struct dsa_switch *ds, int port, bool vlan_filtering, { struct b53_device *dev = ds->priv; - b53_enable_vlan(dev, port, dev->vlan_enabled, vlan_filtering); + if (dev->vlan_filtering != vlan_filtering) { + dev->vlan_filtering = vlan_filtering; + b53_apply_config(dev); + } return 0; } @@ -1524,7 +1567,7 @@ static int b53_vlan_prepare(struct dsa_switch *ds, int port, if (vlan->vid >= dev->num_vlans) return -ERANGE; - b53_enable_vlan(dev, port, true, ds->vlan_filtering); + b53_enable_vlan(dev, port, true, dev->vlan_filtering); return 0; } @@ -1547,21 +1590,17 @@ int b53_vlan_add(struct dsa_switch *ds, int port, if (vlan->vid == 0) return 0; - if (!ds->vlan_filtering) - return 0; - - b53_read16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), &old_pvid); + old_pvid = dev->ports[port].pvid; if (pvid) new_pvid = vlan->vid; else if (!pvid && vlan->vid == old_pvid) new_pvid = b53_default_pvid(dev); else new_pvid = old_pvid; + dev->ports[port].pvid = new_pvid; vl = &dev->vlans[vlan->vid]; - b53_get_vlan_entry(dev, vlan->vid, vl); - if (dsa_is_cpu_port(ds, port)) untagged = false; @@ -1571,6 +1610,9 @@ int b53_vlan_add(struct dsa_switch *ds, int port, else vl->untag &= ~BIT(port); + if (!dev->vlan_filtering) + return 0; + b53_set_vlan_entry(dev, vlan->vid, vl); b53_fast_age_vlan(dev, vlan->vid); @@ -1595,23 +1637,22 @@ int b53_vlan_del(struct dsa_switch *ds, int port, if (vlan->vid == 0) return 0; - if (!ds->vlan_filtering) - return 0; - - b53_read16(dev, B53_VLAN_PAGE, B53_VLAN_PORT_DEF_TAG(port), &pvid); + pvid = dev->ports[port].pvid; vl = &dev->vlans[vlan->vid]; - b53_get_vlan_entry(dev, vlan->vid, vl); - vl->members &= ~BIT(port); if (pvid == vlan->vid) pvid = b53_default_pvid(dev); + dev->ports[port].pvid = pvid; if (untagged && !b53_vlan_port_needs_forced_tagged(ds, port)) vl->untag &= ~(BIT(port)); + if (!dev->vlan_filtering) + return 0; + b53_set_vlan_entry(dev, vlan->vid, vl); b53_fast_age_vlan(dev, vlan->vid); @@ -1958,7 +1999,7 @@ int b53_br_join(struct dsa_switch *ds, int port, struct dsa_bridge bridge, pvid = b53_default_pvid(dev); vl = &dev->vlans[pvid]; - if (ds->vlan_filtering) { + if (dev->vlan_filtering) { /* Make this port leave the all VLANs join since we will have * proper VLAN entries from now on */ @@ -2038,7 +2079,7 @@ void b53_br_leave(struct dsa_switch *ds, int port, struct dsa_bridge bridge) pvid = b53_default_pvid(dev); vl = &dev->vlans[pvid]; - if (ds->vlan_filtering) { + if (dev->vlan_filtering) { /* Make this port join all VLANs without VLAN entries */ if (is58xx(dev)) { b53_read16(dev, B53_VLAN_PAGE, B53_JOIN_ALL_VLAN_EN, ®); @@ -2803,6 +2844,7 @@ struct b53_device *b53_switch_alloc(struct device *base, ds->ops = &b53_switch_ops; ds->phylink_mac_ops = &b53_phylink_mac_ops; dev->vlan_enabled = true; + dev->vlan_filtering = false; /* Let DSA handle the case were multiple bridges span the same switch * device and different VLAN awareness settings are requested, which * would be breaking filtering semantics for any of the other bridge diff --git a/drivers/net/dsa/b53/b53_priv.h b/drivers/net/dsa/b53/b53_priv.h index 0166c37a13a7..4636e27fd1ee 100644 --- a/drivers/net/dsa/b53/b53_priv.h +++ b/drivers/net/dsa/b53/b53_priv.h @@ -96,6 +96,7 @@ struct b53_pcs { struct b53_port { u16 vlan_ctl_mask; + u16 pvid; struct ethtool_keee eee; }; @@ -147,6 +148,7 @@ struct b53_device { unsigned int num_vlans; struct b53_vlan *vlans; bool vlan_enabled; + bool vlan_filtering; unsigned int num_ports; struct b53_port *ports; -- cgit From 9f34ad89bcf0e6df6f8b01f1bdab211493fc66d1 Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:09 +0200 Subject: net: dsa: b53: fix learning on VLAN unaware bridges When VLAN filtering is off, we configure the switch to forward, but not learn on VLAN table misses. This effectively disables learning while not filtering. Fix this by switching to forward and learn. Setting the learning disable register will still control whether learning actually happens. Fixes: dad8d7c6452b ("net: dsa: b53: Properly account for VLAN filtering") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-11-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index 0a7749b416f7..a2c0b44fc6be 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -383,7 +383,7 @@ static void b53_enable_vlan(struct b53_device *dev, int port, bool enable, vc4 |= VC4_ING_VID_VIO_DROP << VC4_ING_VID_CHECK_S; vc5 |= VC5_DROP_VTABLE_MISS; } else { - vc4 |= VC4_ING_VID_VIO_FWD << VC4_ING_VID_CHECK_S; + vc4 |= VC4_NO_ING_VID_CHK << VC4_ING_VID_CHECK_S; vc5 &= ~VC5_DROP_VTABLE_MISS; } -- cgit From 2e7179c628d3cb9aee75e412473813b099e11ed4 Mon Sep 17 00:00:00 2001 From: Jonas Gorski Date: Tue, 29 Apr 2025 22:17:10 +0200 Subject: net: dsa: b53: do not set learning and unicast/multicast on up When a port gets set up, b53 disables learning and enables the port for flooding. This can undo any bridge configuration on the port. E.g. the following flow would disable learning on a port: $ ip link add br0 type bridge $ ip link set sw1p1 master br0 <- enables learning for sw1p1 $ ip link set br0 up $ ip link set sw1p1 up <- disables learning again Fix this by populating dsa_switch_ops::port_setup(), and set up initial config there. Fixes: f9b3827ee66c ("net: dsa: b53: Support setting learning on port") Signed-off-by: Jonas Gorski Tested-by: Florian Fainelli Reviewed-by: Florian Fainelli Link: https://patch.msgid.link/20250429201710.330937-12-jonas.gorski@gmail.com Signed-off-by: Jakub Kicinski --- drivers/net/dsa/b53/b53_common.c | 21 +++++++++++++-------- drivers/net/dsa/b53/b53_priv.h | 1 + drivers/net/dsa/bcm_sf2.c | 1 + 3 files changed, 15 insertions(+), 8 deletions(-) diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c index a2c0b44fc6be..9eb39cfa5fb2 100644 --- a/drivers/net/dsa/b53/b53_common.c +++ b/drivers/net/dsa/b53/b53_common.c @@ -578,6 +578,18 @@ static void b53_eee_enable_set(struct dsa_switch *ds, int port, bool enable) b53_write16(dev, B53_EEE_PAGE, B53_EEE_EN_CTRL, reg); } +int b53_setup_port(struct dsa_switch *ds, int port) +{ + struct b53_device *dev = ds->priv; + + b53_port_set_ucast_flood(dev, port, true); + b53_port_set_mcast_flood(dev, port, true); + b53_port_set_learning(dev, port, false); + + return 0; +} +EXPORT_SYMBOL(b53_setup_port); + int b53_enable_port(struct dsa_switch *ds, int port, struct phy_device *phy) { struct b53_device *dev = ds->priv; @@ -590,10 +602,6 @@ int b53_enable_port(struct dsa_switch *ds, int port, struct phy_device *phy) cpu_port = dsa_to_port(ds, port)->cpu_dp->index; - b53_port_set_ucast_flood(dev, port, true); - b53_port_set_mcast_flood(dev, port, true); - b53_port_set_learning(dev, port, false); - if (dev->ops->irq_enable) ret = dev->ops->irq_enable(dev, port); if (ret) @@ -724,10 +732,6 @@ static void b53_enable_cpu_port(struct b53_device *dev, int port) b53_write8(dev, B53_CTRL_PAGE, B53_PORT_CTRL(port), port_ctrl); b53_brcm_hdr_setup(dev->ds, port); - - b53_port_set_ucast_flood(dev, port, true); - b53_port_set_mcast_flood(dev, port, true); - b53_port_set_learning(dev, port, false); } static void b53_enable_mib(struct b53_device *dev) @@ -2387,6 +2391,7 @@ static const struct dsa_switch_ops b53_switch_ops = { .phy_read = b53_phy_read16, .phy_write = b53_phy_write16, .phylink_get_caps = b53_phylink_get_caps, + .port_setup = b53_setup_port, .port_enable = b53_enable_port, .port_disable = b53_disable_port, .support_eee = b53_support_eee, diff --git a/drivers/net/dsa/b53/b53_priv.h b/drivers/net/dsa/b53/b53_priv.h index 4636e27fd1ee..2cf3e6a81e37 100644 --- a/drivers/net/dsa/b53/b53_priv.h +++ b/drivers/net/dsa/b53/b53_priv.h @@ -384,6 +384,7 @@ enum dsa_tag_protocol b53_get_tag_protocol(struct dsa_switch *ds, int port, enum dsa_tag_protocol mprot); void b53_mirror_del(struct dsa_switch *ds, int port, struct dsa_mall_mirror_tc_entry *mirror); +int b53_setup_port(struct dsa_switch *ds, int port); int b53_enable_port(struct dsa_switch *ds, int port, struct phy_device *phy); void b53_disable_port(struct dsa_switch *ds, int port); void b53_brcm_hdr_setup(struct dsa_switch *ds, int port); diff --git a/drivers/net/dsa/bcm_sf2.c b/drivers/net/dsa/bcm_sf2.c index fa2bf3fa9019..454a8c7fd7ee 100644 --- a/drivers/net/dsa/bcm_sf2.c +++ b/drivers/net/dsa/bcm_sf2.c @@ -1230,6 +1230,7 @@ static const struct dsa_switch_ops bcm_sf2_ops = { .resume = bcm_sf2_sw_resume, .get_wol = bcm_sf2_sw_get_wol, .set_wol = bcm_sf2_sw_set_wol, + .port_setup = b53_setup_port, .port_enable = bcm_sf2_port_setup, .port_disable = bcm_sf2_port_disable, .support_eee = b53_support_eee, -- cgit From f34343cc11afc7bb1f881c3492bee3484016bf71 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Tue, 6 May 2025 08:59:39 -0700 Subject: fbnic: Fix initialization of mailbox descriptor rings Address to issues with the FW mailbox descriptor initialization. We need to reverse the order of accesses when we invalidate an entry versus writing an entry. When writing an entry we write upper and then lower as the lower 32b contain the valid bit that makes the entire address valid. However for invalidation we should write it in the reverse order so that the upper is marked invalid before we update it. Without this change we may see FW attempt to access pages with the upper 32b of the address set to 0 which will likely result in DMAR faults due to write access failures on mailbox shutdown. Fixes: da3cde08209e ("eth: fbnic: Add FW communication mechanism") Signed-off-by: Alexander Duyck Reviewed-by: Simon Horman Reviewed-by: Jacob Keller Link: https://patch.msgid.link/174654717972.499179.8083789731819297034.stgit@ahduyck-xeon-server.home.arpa Reviewed-by: Jakub Kicinski Signed-off-by: Paolo Abeni --- drivers/net/ethernet/meta/fbnic/fbnic_fw.c | 32 +++++++++++++++++++++--------- 1 file changed, 23 insertions(+), 9 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c index 88db3dacb940..c4956f0a741e 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c @@ -17,11 +17,29 @@ static void __fbnic_mbx_wr_desc(struct fbnic_dev *fbd, int mbx_idx, { u32 desc_offset = FBNIC_IPC_MBX(mbx_idx, desc_idx); + /* Write the upper 32b and then the lower 32b. Doing this the + * FW can then read lower, upper, lower to verify that the state + * of the descriptor wasn't changed mid-transaction. + */ fw_wr32(fbd, desc_offset + 1, upper_32_bits(desc)); fw_wrfl(fbd); fw_wr32(fbd, desc_offset, lower_32_bits(desc)); } +static void __fbnic_mbx_invalidate_desc(struct fbnic_dev *fbd, int mbx_idx, + int desc_idx, u32 desc) +{ + u32 desc_offset = FBNIC_IPC_MBX(mbx_idx, desc_idx); + + /* For initialization we write the lower 32b of the descriptor first. + * This way we can set the state to mark it invalid before we clear the + * upper 32b. + */ + fw_wr32(fbd, desc_offset, desc); + fw_wrfl(fbd); + fw_wr32(fbd, desc_offset + 1, 0); +} + static u64 __fbnic_mbx_rd_desc(struct fbnic_dev *fbd, int mbx_idx, int desc_idx) { u32 desc_offset = FBNIC_IPC_MBX(mbx_idx, desc_idx); @@ -41,21 +59,17 @@ static void fbnic_mbx_init_desc_ring(struct fbnic_dev *fbd, int mbx_idx) * solid stop for the firmware to hit when it is done looping * through the ring. */ - __fbnic_mbx_wr_desc(fbd, mbx_idx, 0, 0); - - fw_wrfl(fbd); + __fbnic_mbx_invalidate_desc(fbd, mbx_idx, 0, 0); /* We then fill the rest of the ring starting at the end and moving * back toward descriptor 0 with skip descriptors that have no * length nor address, and tell the firmware that they can skip * them and just move past them to the one we initialized to 0. */ - for (desc_idx = FBNIC_IPC_MBX_DESC_LEN; --desc_idx;) { - __fbnic_mbx_wr_desc(fbd, mbx_idx, desc_idx, - FBNIC_IPC_MBX_DESC_FW_CMPL | - FBNIC_IPC_MBX_DESC_HOST_CMPL); - fw_wrfl(fbd); - } + for (desc_idx = FBNIC_IPC_MBX_DESC_LEN; --desc_idx;) + __fbnic_mbx_invalidate_desc(fbd, mbx_idx, desc_idx, + FBNIC_IPC_MBX_DESC_FW_CMPL | + FBNIC_IPC_MBX_DESC_HOST_CMPL); } void fbnic_mbx_init(struct fbnic_dev *fbd) -- cgit From 3b12f00ddd08e888273b2ac0488d396d90a836fc Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Tue, 6 May 2025 08:59:46 -0700 Subject: fbnic: Gate AXI read/write enabling on FW mailbox In order to prevent the device from throwing spurious writes and/or reads at us we need to gate the AXI fabric interface to the PCIe until such time as we know the FW is in a known good state. To accomplish this we use the mailbox as a mechanism for us to recognize that the FW has acknowledged our presence and is no longer sending any stale message data to us. We start in fbnic_mbx_init by calling fbnic_mbx_reset_desc_ring function, disabling the DMA in both directions, and then invalidating all the descriptors in each ring. We then poll the mailbox in fbnic_mbx_poll_tx_ready and when the interrupt is set by the FW we pick it up and mark the mailboxes as ready, while also enabling the DMA. Once we have completed all the transactions and need to shut down we call into fbnic_mbx_clean which will in turn call fbnic_mbx_reset_desc_ring for each ring and shut down the DMA and once again invalidate the descriptors. Fixes: 3646153161f1 ("eth: fbnic: Add register init to set PCIe/Ethernet device config") Fixes: da3cde08209e ("eth: fbnic: Add FW communication mechanism") Signed-off-by: Alexander Duyck Reviewed-by: Simon Horman Reviewed-by: Jacob Keller Link: https://patch.msgid.link/174654718623.499179.7445197308109347982.stgit@ahduyck-xeon-server.home.arpa Reviewed-by: Jakub Kicinski Signed-off-by: Paolo Abeni --- drivers/net/ethernet/meta/fbnic/fbnic_csr.h | 2 ++ drivers/net/ethernet/meta/fbnic/fbnic_fw.c | 38 +++++++++++++++++++++++------ drivers/net/ethernet/meta/fbnic/fbnic_mac.c | 6 ----- 3 files changed, 33 insertions(+), 13 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_csr.h b/drivers/net/ethernet/meta/fbnic/fbnic_csr.h index 3b12a0ab5906..51bee8072420 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_csr.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic_csr.h @@ -796,8 +796,10 @@ enum { /* PUL User Registers */ #define FBNIC_CSR_START_PUL_USER 0x31000 /* CSR section delimiter */ #define FBNIC_PUL_OB_TLP_HDR_AW_CFG 0x3103d /* 0xc40f4 */ +#define FBNIC_PUL_OB_TLP_HDR_AW_CFG_FLUSH CSR_BIT(19) #define FBNIC_PUL_OB_TLP_HDR_AW_CFG_BME CSR_BIT(18) #define FBNIC_PUL_OB_TLP_HDR_AR_CFG 0x3103e /* 0xc40f8 */ +#define FBNIC_PUL_OB_TLP_HDR_AR_CFG_FLUSH CSR_BIT(19) #define FBNIC_PUL_OB_TLP_HDR_AR_CFG_BME CSR_BIT(18) #define FBNIC_PUL_USER_OB_RD_TLP_CNT_31_0 \ 0x3106e /* 0xc41b8 */ diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c index c4956f0a741e..f4749bfd840c 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c @@ -51,10 +51,26 @@ static u64 __fbnic_mbx_rd_desc(struct fbnic_dev *fbd, int mbx_idx, int desc_idx) return desc; } -static void fbnic_mbx_init_desc_ring(struct fbnic_dev *fbd, int mbx_idx) +static void fbnic_mbx_reset_desc_ring(struct fbnic_dev *fbd, int mbx_idx) { int desc_idx; + /* Disable DMA transactions from the device, + * and flush any transactions triggered during cleaning + */ + switch (mbx_idx) { + case FBNIC_IPC_MBX_RX_IDX: + wr32(fbd, FBNIC_PUL_OB_TLP_HDR_AW_CFG, + FBNIC_PUL_OB_TLP_HDR_AW_CFG_FLUSH); + break; + case FBNIC_IPC_MBX_TX_IDX: + wr32(fbd, FBNIC_PUL_OB_TLP_HDR_AR_CFG, + FBNIC_PUL_OB_TLP_HDR_AR_CFG_FLUSH); + break; + } + + wrfl(fbd); + /* Initialize first descriptor to all 0s. Doing this gives us a * solid stop for the firmware to hit when it is done looping * through the ring. @@ -90,7 +106,7 @@ void fbnic_mbx_init(struct fbnic_dev *fbd) wr32(fbd, FBNIC_INTR_CLEAR(0), 1u << FBNIC_FW_MSIX_ENTRY); for (i = 0; i < FBNIC_IPC_MBX_INDICES; i++) - fbnic_mbx_init_desc_ring(fbd, i); + fbnic_mbx_reset_desc_ring(fbd, i); } static int fbnic_mbx_map_msg(struct fbnic_dev *fbd, int mbx_idx, @@ -155,7 +171,7 @@ static void fbnic_mbx_clean_desc_ring(struct fbnic_dev *fbd, int mbx_idx) { int i; - fbnic_mbx_init_desc_ring(fbd, mbx_idx); + fbnic_mbx_reset_desc_ring(fbd, mbx_idx); for (i = FBNIC_IPC_MBX_DESC_LEN; i--;) fbnic_mbx_unmap_and_free_msg(fbd, mbx_idx, i); @@ -354,7 +370,7 @@ static int fbnic_fw_xmit_cap_msg(struct fbnic_dev *fbd) return (err == -EOPNOTSUPP) ? 0 : err; } -static void fbnic_mbx_postinit_desc_ring(struct fbnic_dev *fbd, int mbx_idx) +static void fbnic_mbx_init_desc_ring(struct fbnic_dev *fbd, int mbx_idx) { struct fbnic_fw_mbx *mbx = &fbd->mbx[mbx_idx]; @@ -366,10 +382,18 @@ static void fbnic_mbx_postinit_desc_ring(struct fbnic_dev *fbd, int mbx_idx) switch (mbx_idx) { case FBNIC_IPC_MBX_RX_IDX: + /* Enable DMA writes from the device */ + wr32(fbd, FBNIC_PUL_OB_TLP_HDR_AW_CFG, + FBNIC_PUL_OB_TLP_HDR_AW_CFG_BME); + /* Make sure we have a page for the FW to write to */ fbnic_mbx_alloc_rx_msgs(fbd); break; case FBNIC_IPC_MBX_TX_IDX: + /* Enable DMA reads from the device */ + wr32(fbd, FBNIC_PUL_OB_TLP_HDR_AR_CFG, + FBNIC_PUL_OB_TLP_HDR_AR_CFG_BME); + /* Force version to 1 if we successfully requested an update * from the firmware. This should be overwritten once we get * the actual version from the firmware in the capabilities @@ -386,7 +410,7 @@ static void fbnic_mbx_postinit(struct fbnic_dev *fbd) { int i; - /* We only need to do this on the first interrupt following init. + /* We only need to do this on the first interrupt following reset. * this primes the mailbox so that we will have cleared all the * skip descriptors. */ @@ -396,7 +420,7 @@ static void fbnic_mbx_postinit(struct fbnic_dev *fbd) wr32(fbd, FBNIC_INTR_CLEAR(0), 1u << FBNIC_FW_MSIX_ENTRY); for (i = 0; i < FBNIC_IPC_MBX_INDICES; i++) - fbnic_mbx_postinit_desc_ring(fbd, i); + fbnic_mbx_init_desc_ring(fbd, i); } /** @@ -894,7 +918,7 @@ int fbnic_mbx_poll_tx_ready(struct fbnic_dev *fbd) * avoid the mailbox getting stuck closed if the interrupt * is reset. */ - fbnic_mbx_init_desc_ring(fbd, FBNIC_IPC_MBX_TX_IDX); + fbnic_mbx_reset_desc_ring(fbd, FBNIC_IPC_MBX_TX_IDX); msleep(200); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_mac.c b/drivers/net/ethernet/meta/fbnic/fbnic_mac.c index 14291401f463..dde4a37116e2 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_mac.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_mac.c @@ -79,12 +79,6 @@ static void fbnic_mac_init_axi(struct fbnic_dev *fbd) fbnic_init_readrq(fbd, FBNIC_QM_RNI_RBP_CTL, cls, readrq); fbnic_init_mps(fbd, FBNIC_QM_RNI_RDE_CTL, cls, mps); fbnic_init_mps(fbd, FBNIC_QM_RNI_RCM_CTL, cls, mps); - - /* Enable XALI AR/AW outbound */ - wr32(fbd, FBNIC_PUL_OB_TLP_HDR_AW_CFG, - FBNIC_PUL_OB_TLP_HDR_AW_CFG_BME); - wr32(fbd, FBNIC_PUL_OB_TLP_HDR_AR_CFG, - FBNIC_PUL_OB_TLP_HDR_AR_CFG_BME); } static void fbnic_mac_init_qm(struct fbnic_dev *fbd) -- cgit From 682a61281d1036962b68d651994cc407371daef5 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Tue, 6 May 2025 08:59:52 -0700 Subject: fbnic: Add additional handling of IRQs We have two issues that need to be addressed in our IRQ handling. One is the fact that we can end up double-freeing IRQs in the event of an exception handling error such as a PCIe reset/recovery that fails. To prevent that from becoming an issue we can use the msix_vector values to indicate that we have successfully requested/freed the IRQ by only setting or clearing them when we have completed the given action. The other issue is that we have several potential races in our IRQ path due to us manipulating the mask before the vector has been truly disabled. In order to handle that in the case of the FW mailbox we need to not auto-enable the IRQ and instead will be enabling/disabling it separately. In the case of the PCS vector we can mitigate this by unmapping it and synchronizing the IRQ before we clear the mask. The general order of operations after this change is now to request the interrupt, poll the FW mailbox to ready, and then enable the interrupt. For the shutdown we do the reverse where we disable the interrupt, flush any pending Tx, and then free the IRQ. I am renaming the enable/disable to request/free to be equivilent with the IRQ calls being used. We may see additions in the future to enable/disable the IRQs versus request/free them for certain use cases. Fixes: da3cde08209e ("eth: fbnic: Add FW communication mechanism") Fixes: 69684376eed5 ("eth: fbnic: Add link detection") Signed-off-by: Alexander Duyck Reviewed-by: Simon Horman Reviewed-by: Jacob Keller Link: https://patch.msgid.link/174654719271.499179.3634535105127848325.stgit@ahduyck-xeon-server.home.arpa Reviewed-by: Jakub Kicinski Signed-off-by: Paolo Abeni --- drivers/net/ethernet/meta/fbnic/fbnic.h | 8 +- drivers/net/ethernet/meta/fbnic/fbnic_irq.c | 142 +++++++++++++++++-------- drivers/net/ethernet/meta/fbnic/fbnic_netdev.c | 5 +- drivers/net/ethernet/meta/fbnic/fbnic_pci.c | 14 +-- 4 files changed, 110 insertions(+), 59 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic.h b/drivers/net/ethernet/meta/fbnic/fbnic.h index 4ca7b99ef131..de6b1a340f55 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic.h @@ -154,14 +154,14 @@ struct fbnic_dev *fbnic_devlink_alloc(struct pci_dev *pdev); void fbnic_devlink_register(struct fbnic_dev *fbd); void fbnic_devlink_unregister(struct fbnic_dev *fbd); -int fbnic_fw_enable_mbx(struct fbnic_dev *fbd); -void fbnic_fw_disable_mbx(struct fbnic_dev *fbd); +int fbnic_fw_request_mbx(struct fbnic_dev *fbd); +void fbnic_fw_free_mbx(struct fbnic_dev *fbd); void fbnic_hwmon_register(struct fbnic_dev *fbd); void fbnic_hwmon_unregister(struct fbnic_dev *fbd); -int fbnic_pcs_irq_enable(struct fbnic_dev *fbd); -void fbnic_pcs_irq_disable(struct fbnic_dev *fbd); +int fbnic_pcs_request_irq(struct fbnic_dev *fbd); +void fbnic_pcs_free_irq(struct fbnic_dev *fbd); void fbnic_napi_name_irqs(struct fbnic_dev *fbd); int fbnic_napi_request_irq(struct fbnic_dev *fbd, diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_irq.c b/drivers/net/ethernet/meta/fbnic/fbnic_irq.c index 1bbc0e56f3a0..1c88a2bf3a7a 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_irq.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_irq.c @@ -19,69 +19,105 @@ static irqreturn_t fbnic_fw_msix_intr(int __always_unused irq, void *data) return IRQ_HANDLED; } +static int __fbnic_fw_enable_mbx(struct fbnic_dev *fbd, int vector) +{ + int err; + + /* Initialize mailbox and attempt to poll it into ready state */ + fbnic_mbx_init(fbd); + err = fbnic_mbx_poll_tx_ready(fbd); + if (err) { + dev_warn(fbd->dev, "FW mailbox did not enter ready state\n"); + return err; + } + + /* Enable interrupt and unmask the vector */ + enable_irq(vector); + fbnic_wr32(fbd, FBNIC_INTR_MASK_CLEAR(0), 1u << FBNIC_FW_MSIX_ENTRY); + + return 0; +} + /** - * fbnic_fw_enable_mbx - Configure and initialize Firmware Mailbox + * fbnic_fw_request_mbx - Configure and initialize Firmware Mailbox * @fbd: Pointer to device to initialize * - * This function will initialize the firmware mailbox rings, enable the IRQ - * and initialize the communication between the Firmware and the host. The - * firmware is expected to respond to the initialization by sending an - * interrupt essentially notifying the host that it has seen the - * initialization and is now synced up. + * This function will allocate the IRQ and then reinitialize the mailbox + * starting communication between the host and firmware. * * Return: non-zero on failure. **/ -int fbnic_fw_enable_mbx(struct fbnic_dev *fbd) +int fbnic_fw_request_mbx(struct fbnic_dev *fbd) { - u32 vector = fbd->fw_msix_vector; - int err; + struct pci_dev *pdev = to_pci_dev(fbd->dev); + int vector, err; + + WARN_ON(fbd->fw_msix_vector); + + vector = pci_irq_vector(pdev, FBNIC_FW_MSIX_ENTRY); + if (vector < 0) + return vector; /* Request the IRQ for FW Mailbox vector. */ err = request_threaded_irq(vector, NULL, &fbnic_fw_msix_intr, - IRQF_ONESHOT, dev_name(fbd->dev), fbd); + IRQF_ONESHOT | IRQF_NO_AUTOEN, + dev_name(fbd->dev), fbd); if (err) return err; /* Initialize mailbox and attempt to poll it into ready state */ - fbnic_mbx_init(fbd); - err = fbnic_mbx_poll_tx_ready(fbd); - if (err) { - dev_warn(fbd->dev, "FW mailbox did not enter ready state\n"); + err = __fbnic_fw_enable_mbx(fbd, vector); + if (err) free_irq(vector, fbd); - return err; - } - /* Enable interrupts */ - fbnic_wr32(fbd, FBNIC_INTR_MASK_CLEAR(0), 1u << FBNIC_FW_MSIX_ENTRY); + fbd->fw_msix_vector = vector; - return 0; + return err; } /** - * fbnic_fw_disable_mbx - Disable mailbox and place it in standby state - * @fbd: Pointer to device to disable + * fbnic_fw_disable_mbx - Temporarily place mailbox in standby state + * @fbd: Pointer to device * - * This function will disable the mailbox interrupt, free any messages still - * in the mailbox and place it into a standby state. The firmware is - * expected to see the update and assume that the host is in the reset state. + * Shutdown the mailbox by notifying the firmware to stop sending us logs, mask + * and synchronize the IRQ, and then clean up the rings. **/ -void fbnic_fw_disable_mbx(struct fbnic_dev *fbd) +static void fbnic_fw_disable_mbx(struct fbnic_dev *fbd) { - /* Disable interrupt and free vector */ - fbnic_wr32(fbd, FBNIC_INTR_MASK_SET(0), 1u << FBNIC_FW_MSIX_ENTRY); + /* Disable interrupt and synchronize the IRQ */ + disable_irq(fbd->fw_msix_vector); - /* Free the vector */ - free_irq(fbd->fw_msix_vector, fbd); + /* Mask the vector */ + fbnic_wr32(fbd, FBNIC_INTR_MASK_SET(0), 1u << FBNIC_FW_MSIX_ENTRY); /* Make sure disabling logs message is sent, must be done here to * avoid risk of completing without a running interrupt. */ fbnic_mbx_flush_tx(fbd); - - /* Reset the mailboxes to the initialized state */ fbnic_mbx_clean(fbd); } +/** + * fbnic_fw_free_mbx - Disable mailbox and place it in standby state + * @fbd: Pointer to device to disable + * + * This function will disable the mailbox interrupt, free any messages still + * in the mailbox and place it into a disabled state. The firmware is + * expected to see the update and assume that the host is in the reset state. + **/ +void fbnic_fw_free_mbx(struct fbnic_dev *fbd) +{ + /* Vector has already been freed */ + if (!fbd->fw_msix_vector) + return; + + fbnic_fw_disable_mbx(fbd); + + /* Free the vector */ + free_irq(fbd->fw_msix_vector, fbd); + fbd->fw_msix_vector = 0; +} + static irqreturn_t fbnic_pcs_msix_intr(int __always_unused irq, void *data) { struct fbnic_dev *fbd = data; @@ -101,7 +137,7 @@ static irqreturn_t fbnic_pcs_msix_intr(int __always_unused irq, void *data) } /** - * fbnic_pcs_irq_enable - Configure the MAC to enable it to advertise link + * fbnic_pcs_request_irq - Configure the PCS to enable it to advertise link * @fbd: Pointer to device to initialize * * This function provides basic bringup for the MAC/PCS IRQ. For now the IRQ @@ -109,41 +145,61 @@ static irqreturn_t fbnic_pcs_msix_intr(int __always_unused irq, void *data) * * Return: non-zero on failure. **/ -int fbnic_pcs_irq_enable(struct fbnic_dev *fbd) +int fbnic_pcs_request_irq(struct fbnic_dev *fbd) { - u32 vector = fbd->pcs_msix_vector; - int err; + struct pci_dev *pdev = to_pci_dev(fbd->dev); + int vector, err; - /* Request the IRQ for MAC link vector. - * Map MAC cause to it, and unmask it + WARN_ON(fbd->pcs_msix_vector); + + vector = pci_irq_vector(pdev, FBNIC_PCS_MSIX_ENTRY); + if (vector < 0) + return vector; + + /* Request the IRQ for PCS link vector. + * Map PCS cause to it, and unmask it */ err = request_irq(vector, &fbnic_pcs_msix_intr, 0, fbd->netdev->name, fbd); if (err) return err; + /* Map and enable interrupt, unmask vector after link is configured */ fbnic_wr32(fbd, FBNIC_INTR_MSIX_CTRL(FBNIC_INTR_MSIX_CTRL_PCS_IDX), FBNIC_PCS_MSIX_ENTRY | FBNIC_INTR_MSIX_CTRL_ENABLE); + fbd->pcs_msix_vector = vector; + return 0; } /** - * fbnic_pcs_irq_disable - Teardown the MAC IRQ to prepare for stopping + * fbnic_pcs_free_irq - Teardown the PCS IRQ to prepare for stopping * @fbd: Pointer to device that is stopping * - * This function undoes the work done in fbnic_pcs_irq_enable and prepares + * This function undoes the work done in fbnic_pcs_request_irq and prepares * the device to no longer receive traffic on the host interface. **/ -void fbnic_pcs_irq_disable(struct fbnic_dev *fbd) +void fbnic_pcs_free_irq(struct fbnic_dev *fbd) { + /* Vector has already been freed */ + if (!fbd->pcs_msix_vector) + return; + /* Disable interrupt */ fbnic_wr32(fbd, FBNIC_INTR_MSIX_CTRL(FBNIC_INTR_MSIX_CTRL_PCS_IDX), FBNIC_PCS_MSIX_ENTRY); + fbnic_wrfl(fbd); + + /* Synchronize IRQ to prevent race that would unmask vector */ + synchronize_irq(fbd->pcs_msix_vector); + + /* Mask the vector */ fbnic_wr32(fbd, FBNIC_INTR_MASK_SET(0), 1u << FBNIC_PCS_MSIX_ENTRY); /* Free the vector */ free_irq(fbd->pcs_msix_vector, fbd); + fbd->pcs_msix_vector = 0; } void fbnic_synchronize_irq(struct fbnic_dev *fbd, int nr) @@ -226,9 +282,6 @@ void fbnic_free_irqs(struct fbnic_dev *fbd) { struct pci_dev *pdev = to_pci_dev(fbd->dev); - fbd->pcs_msix_vector = 0; - fbd->fw_msix_vector = 0; - fbd->num_irqs = 0; pci_free_irq_vectors(pdev); @@ -254,8 +307,5 @@ int fbnic_alloc_irqs(struct fbnic_dev *fbd) fbd->num_irqs = num_irqs; - fbd->pcs_msix_vector = pci_irq_vector(pdev, FBNIC_PCS_MSIX_ENTRY); - fbd->fw_msix_vector = pci_irq_vector(pdev, FBNIC_FW_MSIX_ENTRY); - return 0; } diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c index 79a01fdd1dd1..2524d9b88d59 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c @@ -44,9 +44,10 @@ int __fbnic_open(struct fbnic_net *fbn) if (err) goto time_stop; - err = fbnic_pcs_irq_enable(fbd); + err = fbnic_pcs_request_irq(fbd); if (err) goto time_stop; + /* Pull the BMC config and initialize the RPC */ fbnic_bmc_rpc_init(fbd); fbnic_rss_reinit(fbd, fbn); @@ -82,7 +83,7 @@ static int fbnic_stop(struct net_device *netdev) struct fbnic_net *fbn = netdev_priv(netdev); fbnic_down(fbn); - fbnic_pcs_irq_disable(fbn->fbd); + fbnic_pcs_free_irq(fbn->fbd); fbnic_time_stop(fbn); fbnic_fw_xmit_ownership_msg(fbn->fbd, false); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_pci.c b/drivers/net/ethernet/meta/fbnic/fbnic_pci.c index 6cbbc2ee3e1f..4e8595239c0f 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_pci.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_pci.c @@ -283,7 +283,7 @@ static int fbnic_probe(struct pci_dev *pdev, const struct pci_device_id *ent) goto free_irqs; } - err = fbnic_fw_enable_mbx(fbd); + err = fbnic_fw_request_mbx(fbd); if (err) { dev_err(&pdev->dev, "Firmware mailbox initialization failure\n"); @@ -363,7 +363,7 @@ static void fbnic_remove(struct pci_dev *pdev) fbnic_hwmon_unregister(fbd); fbnic_dbg_fbd_exit(fbd); fbnic_devlink_unregister(fbd); - fbnic_fw_disable_mbx(fbd); + fbnic_fw_free_mbx(fbd); fbnic_free_irqs(fbd); fbnic_devlink_free(fbd); @@ -387,7 +387,7 @@ static int fbnic_pm_suspend(struct device *dev) rtnl_unlock(); null_uc_addr: - fbnic_fw_disable_mbx(fbd); + fbnic_fw_free_mbx(fbd); /* Free the IRQs so they aren't trying to occupy sleeping CPUs */ fbnic_free_irqs(fbd); @@ -420,7 +420,7 @@ static int __fbnic_pm_resume(struct device *dev) fbd->mac->init_regs(fbd); /* Re-enable mailbox */ - err = fbnic_fw_enable_mbx(fbd); + err = fbnic_fw_request_mbx(fbd); if (err) goto err_free_irqs; @@ -438,15 +438,15 @@ static int __fbnic_pm_resume(struct device *dev) if (netif_running(netdev)) { err = __fbnic_open(fbn); if (err) - goto err_disable_mbx; + goto err_free_mbx; } rtnl_unlock(); return 0; -err_disable_mbx: +err_free_mbx: rtnl_unlock(); - fbnic_fw_disable_mbx(fbd); + fbnic_fw_free_mbx(fbd); err_free_irqs: fbnic_free_irqs(fbd); err_invalidate_uc_addr: -- cgit From 0f9a959a0addd9bbc47e5d16c36b3a7f97981915 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Tue, 6 May 2025 08:59:59 -0700 Subject: fbnic: Actually flush_tx instead of stalling out The fbnic_mbx_flush_tx function had a number of issues. First, we were waiting 200ms for the firmware to process the packets. We can drop this to 20ms and in almost all cases this should be more than enough time. So by changing this we can significantly reduce shutdown time. Second, we were not making sure that the Tx path was actually shut off. As such we could still have packets added while we were flushing the mailbox. To prevent that we can now clear the ready flag for the Tx side and it should stay down since the interrupt is disabled. Third, we kept re-reading the tail due to the second issue. The tail should not move after we have started the flush so we can just read it once while we are holding the mailbox Tx lock. By doing that we are guaranteed that the value should be consistent. Fourth, we were keeping a count of descriptors cleaned due to the second and third issues called out. That count is not a valid reason to be exiting the cleanup, and with the tail only being read once we shouldn't see any cases where the tail moves after the disable so the tracking of count can be dropped. Fifth, we were using attempts * sleep time to determine how long we would wait in our polling loop to flush out the Tx. This can be very imprecise. In order to tighten up the timing we are shifting over to using a jiffies value of jiffies + 10 * HZ + 1 to determine the jiffies value we should stop polling at as this should be accurate within once sleep cycle for the total amount of time spent polling. Fixes: da3cde08209e ("eth: fbnic: Add FW communication mechanism") Signed-off-by: Alexander Duyck Reviewed-by: Simon Horman Reviewed-by: Jacob Keller Link: https://patch.msgid.link/174654719929.499179.16406653096197423749.stgit@ahduyck-xeon-server.home.arpa Reviewed-by: Jakub Kicinski Signed-off-by: Paolo Abeni --- drivers/net/ethernet/meta/fbnic/fbnic_fw.c | 31 +++++++++++++++--------------- 1 file changed, 16 insertions(+), 15 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c index f4749bfd840c..d019191d6ae9 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c @@ -930,35 +930,36 @@ int fbnic_mbx_poll_tx_ready(struct fbnic_dev *fbd) void fbnic_mbx_flush_tx(struct fbnic_dev *fbd) { + unsigned long timeout = jiffies + 10 * HZ + 1; struct fbnic_fw_mbx *tx_mbx; - int attempts = 50; - u8 count = 0; - - /* Nothing to do if there is no mailbox */ - if (!fbnic_fw_present(fbd)) - return; + u8 tail; /* Record current Rx stats */ tx_mbx = &fbd->mbx[FBNIC_IPC_MBX_TX_IDX]; - /* Nothing to do if mailbox never got to ready */ - if (!tx_mbx->ready) - return; + spin_lock_irq(&fbd->fw_tx_lock); + + /* Clear ready to prevent any further attempts to transmit */ + tx_mbx->ready = false; + + /* Read tail to determine the last tail state for the ring */ + tail = tx_mbx->tail; + + spin_unlock_irq(&fbd->fw_tx_lock); /* Give firmware time to process packet, - * we will wait up to 10 seconds which is 50 waits of 200ms. + * we will wait up to 10 seconds which is 500 waits of 20ms. */ do { u8 head = tx_mbx->head; - if (head == tx_mbx->tail) + /* Tx ring is empty once head == tail */ + if (head == tail) break; - msleep(200); + msleep(20); fbnic_mbx_process_tx_msgs(fbd); - - count += (tx_mbx->head - head) % FBNIC_IPC_MBX_DESC_LEN; - } while (count < FBNIC_IPC_MBX_DESC_LEN && --attempts); + } while (time_is_after_jiffies(timeout)); } void fbnic_get_fw_ver_commit_str(struct fbnic_dev *fbd, char *fw_version, -- cgit From cdbb2dc3996a60ed3d7431c1239a8ca98c778e04 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Tue, 6 May 2025 09:00:05 -0700 Subject: fbnic: Cleanup handling of completions There was an issue in that if we were to shutdown we could be left with a completion in flight as the mailbox went away. To address that I have added an fbnic_mbx_evict_all_cmpl function that is meant to essentially create a "broken pipe" type response so that all callers will receive an error indicating that the connection has been broken as a result of us shutting down the mailbox. Fixes: 378e5cc1c6c6 ("eth: fbnic: hwmon: Add completion infrastructure for firmware requests") Signed-off-by: Alexander Duyck Reviewed-by: Jacob Keller Link: https://patch.msgid.link/174654720578.499179.380252598204530873.stgit@ahduyck-xeon-server.home.arpa Reviewed-by: Jakub Kicinski Signed-off-by: Paolo Abeni --- drivers/net/ethernet/meta/fbnic/fbnic_fw.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c index d019191d6ae9..732875aae46c 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c @@ -928,6 +928,20 @@ int fbnic_mbx_poll_tx_ready(struct fbnic_dev *fbd) return attempts ? 0 : -ETIMEDOUT; } +static void __fbnic_fw_evict_cmpl(struct fbnic_fw_completion *cmpl_data) +{ + cmpl_data->result = -EPIPE; + complete(&cmpl_data->done); +} + +static void fbnic_mbx_evict_all_cmpl(struct fbnic_dev *fbd) +{ + if (fbd->cmpl_data) { + __fbnic_fw_evict_cmpl(fbd->cmpl_data); + fbd->cmpl_data = NULL; + } +} + void fbnic_mbx_flush_tx(struct fbnic_dev *fbd) { unsigned long timeout = jiffies + 10 * HZ + 1; @@ -945,6 +959,9 @@ void fbnic_mbx_flush_tx(struct fbnic_dev *fbd) /* Read tail to determine the last tail state for the ring */ tail = tx_mbx->tail; + /* Flush any completions as we are no longer processing Rx */ + fbnic_mbx_evict_all_cmpl(fbd); + spin_unlock_irq(&fbd->fw_tx_lock); /* Give firmware time to process packet, -- cgit From ab064f6005973d456f95ae99cd9ea0d8ab676cce Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Tue, 6 May 2025 09:00:12 -0700 Subject: fbnic: Improve responsiveness of fbnic_mbx_poll_tx_ready There were a couple different issues found in fbnic_mbx_poll_tx_ready. Among them were the fact that we were sleeping much longer than we actually needed to as the actual FW could respond in under 20ms. The other issue was that we would just keep polling the mailbox even if the device itself had gone away. To address the responsiveness issues we can decrease the sleeps to 20ms and use a jiffies based timeout value rather than just counting the number of times we slept and then polled. To address the hardware going away we can move the check for the firmware BAR being present from where it was and place it inside the loop after the mailbox descriptor ring is initialized and before we sleep so that we just abort and return an error if the device went away during initialization. With these two changes we see a significant improvement in boot times for the driver. Fixes: da3cde08209e ("eth: fbnic: Add FW communication mechanism") Signed-off-by: Alexander Duyck Reviewed-by: Jacob Keller Link: https://patch.msgid.link/174654721224.499179.2698616208976624755.stgit@ahduyck-xeon-server.home.arpa Reviewed-by: Jakub Kicinski Signed-off-by: Paolo Abeni --- drivers/net/ethernet/meta/fbnic/fbnic_fw.c | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c index 732875aae46c..d344b454f28b 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c @@ -905,27 +905,30 @@ void fbnic_mbx_poll(struct fbnic_dev *fbd) int fbnic_mbx_poll_tx_ready(struct fbnic_dev *fbd) { + unsigned long timeout = jiffies + 10 * HZ + 1; struct fbnic_fw_mbx *tx_mbx; - int attempts = 50; - - /* Immediate fail if BAR4 isn't there */ - if (!fbnic_fw_present(fbd)) - return -ENODEV; tx_mbx = &fbd->mbx[FBNIC_IPC_MBX_TX_IDX]; - while (!tx_mbx->ready && --attempts) { + while (!tx_mbx->ready) { + if (!time_is_after_jiffies(timeout)) + return -ETIMEDOUT; + /* Force the firmware to trigger an interrupt response to * avoid the mailbox getting stuck closed if the interrupt * is reset. */ fbnic_mbx_reset_desc_ring(fbd, FBNIC_IPC_MBX_TX_IDX); - msleep(200); + /* Immediate fail if BAR4 went away */ + if (!fbnic_fw_present(fbd)) + return -ENODEV; + + msleep(20); fbnic_mbx_poll(fbd); } - return attempts ? 0 : -ETIMEDOUT; + return 0; } static void __fbnic_fw_evict_cmpl(struct fbnic_fw_completion *cmpl_data) -- cgit From 1b34d1c1dc8384884febd83140c9afbc7c4b9eb8 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Tue, 6 May 2025 09:00:18 -0700 Subject: fbnic: Pull fbnic_fw_xmit_cap_msg use out of interrupt context This change pulls the call to fbnic_fw_xmit_cap_msg out of fbnic_mbx_init_desc_ring and instead places it in the polling function for getting the Tx ready. Doing that we can avoid the potential issue with an interrupt coming in later from the firmware that causes it to get fired in interrupt context. Fixes: 20d2e88cc746 ("eth: fbnic: Add initial messaging to notify FW of our presence") Signed-off-by: Alexander Duyck Reviewed-by: Jacob Keller Link: https://patch.msgid.link/174654721876.499179.9839651602256668493.stgit@ahduyck-xeon-server.home.arpa Reviewed-by: Jakub Kicinski Signed-off-by: Paolo Abeni --- drivers/net/ethernet/meta/fbnic/fbnic_fw.c | 43 +++++++++++------------------- 1 file changed, 16 insertions(+), 27 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c index d344b454f28b..90a45c701543 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c @@ -352,24 +352,6 @@ static int fbnic_fw_xmit_simple_msg(struct fbnic_dev *fbd, u32 msg_type) return err; } -/** - * fbnic_fw_xmit_cap_msg - Allocate and populate a FW capabilities message - * @fbd: FBNIC device structure - * - * Return: NULL on failure to allocate, error pointer on error, or pointer - * to new TLV test message. - * - * Sends a single TLV header indicating the host wants the firmware to - * confirm the capabilities and version. - **/ -static int fbnic_fw_xmit_cap_msg(struct fbnic_dev *fbd) -{ - int err = fbnic_fw_xmit_simple_msg(fbd, FBNIC_TLV_MSG_ID_HOST_CAP_REQ); - - /* Return 0 if we are not calling this on ASIC */ - return (err == -EOPNOTSUPP) ? 0 : err; -} - static void fbnic_mbx_init_desc_ring(struct fbnic_dev *fbd, int mbx_idx) { struct fbnic_fw_mbx *mbx = &fbd->mbx[mbx_idx]; @@ -393,15 +375,6 @@ static void fbnic_mbx_init_desc_ring(struct fbnic_dev *fbd, int mbx_idx) /* Enable DMA reads from the device */ wr32(fbd, FBNIC_PUL_OB_TLP_HDR_AR_CFG, FBNIC_PUL_OB_TLP_HDR_AR_CFG_BME); - - /* Force version to 1 if we successfully requested an update - * from the firmware. This should be overwritten once we get - * the actual version from the firmware in the capabilities - * request message. - */ - if (!fbnic_fw_xmit_cap_msg(fbd) && - !fbd->fw_cap.running.mgmt.version) - fbd->fw_cap.running.mgmt.version = 1; break; } } @@ -907,6 +880,7 @@ int fbnic_mbx_poll_tx_ready(struct fbnic_dev *fbd) { unsigned long timeout = jiffies + 10 * HZ + 1; struct fbnic_fw_mbx *tx_mbx; + int err; tx_mbx = &fbd->mbx[FBNIC_IPC_MBX_TX_IDX]; while (!tx_mbx->ready) { @@ -928,7 +902,22 @@ int fbnic_mbx_poll_tx_ready(struct fbnic_dev *fbd) fbnic_mbx_poll(fbd); } + /* Request an update from the firmware. This should overwrite + * mgmt.version once we get the actual version from the firmware + * in the capabilities request message. + */ + err = fbnic_fw_xmit_simple_msg(fbd, FBNIC_TLV_MSG_ID_HOST_CAP_REQ); + if (err) + goto clean_mbx; + + /* Use "1" to indicate we entered the state waiting for a response */ + fbd->fw_cap.running.mgmt.version = 1; + return 0; +clean_mbx: + /* Cleanup Rx buffers and disable mailbox */ + fbnic_mbx_clean(fbd); + return err; } static void __fbnic_fw_evict_cmpl(struct fbnic_fw_completion *cmpl_data) -- cgit From ce2fa1dba204c761582674cf2eb9cbe0b949b5c7 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Tue, 6 May 2025 09:00:25 -0700 Subject: fbnic: Do not allow mailbox to toggle to ready outside fbnic_mbx_poll_tx_ready We had originally thought to have the mailbox go to ready in the background while we were doing other things. One issue with this though is that we can't disable it by clearing the ready state without also blocking interrupts or calls to mbx_poll as it will just pop back to life during an interrupt. In order to prevent that from happening we can pull the code for toggling to ready out of the interrupt path and instead place it in the fbnic_mbx_poll_tx_ready path so that it becomes the only spot where the Rx/Tx can toggle to the ready state. By doing this we can prevent races where we disable the DMA and/or free buffers only to have an interrupt fire and undo what we have done. Fixes: da3cde08209e ("eth: fbnic: Add FW communication mechanism") Signed-off-by: Alexander Duyck Reviewed-by: Jacob Keller Link: https://patch.msgid.link/174654722518.499179.11612865740376848478.stgit@ahduyck-xeon-server.home.arpa Reviewed-by: Jakub Kicinski Signed-off-by: Paolo Abeni --- drivers/net/ethernet/meta/fbnic/fbnic_fw.c | 27 ++++++++++----------------- 1 file changed, 10 insertions(+), 17 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c index 90a45c701543..3d9636a6c968 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_fw.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_fw.c @@ -356,10 +356,6 @@ static void fbnic_mbx_init_desc_ring(struct fbnic_dev *fbd, int mbx_idx) { struct fbnic_fw_mbx *mbx = &fbd->mbx[mbx_idx]; - /* This is a one time init, so just exit if it is completed */ - if (mbx->ready) - return; - mbx->ready = true; switch (mbx_idx) { @@ -379,21 +375,18 @@ static void fbnic_mbx_init_desc_ring(struct fbnic_dev *fbd, int mbx_idx) } } -static void fbnic_mbx_postinit(struct fbnic_dev *fbd) +static bool fbnic_mbx_event(struct fbnic_dev *fbd) { - int i; - /* We only need to do this on the first interrupt following reset. * this primes the mailbox so that we will have cleared all the * skip descriptors. */ if (!(rd32(fbd, FBNIC_INTR_STATUS(0)) & (1u << FBNIC_FW_MSIX_ENTRY))) - return; + return false; wr32(fbd, FBNIC_INTR_CLEAR(0), 1u << FBNIC_FW_MSIX_ENTRY); - for (i = 0; i < FBNIC_IPC_MBX_INDICES; i++) - fbnic_mbx_init_desc_ring(fbd, i); + return true; } /** @@ -870,7 +863,7 @@ next_page: void fbnic_mbx_poll(struct fbnic_dev *fbd) { - fbnic_mbx_postinit(fbd); + fbnic_mbx_event(fbd); fbnic_mbx_process_tx_msgs(fbd); fbnic_mbx_process_rx_msgs(fbd); @@ -879,11 +872,9 @@ void fbnic_mbx_poll(struct fbnic_dev *fbd) int fbnic_mbx_poll_tx_ready(struct fbnic_dev *fbd) { unsigned long timeout = jiffies + 10 * HZ + 1; - struct fbnic_fw_mbx *tx_mbx; - int err; + int err, i; - tx_mbx = &fbd->mbx[FBNIC_IPC_MBX_TX_IDX]; - while (!tx_mbx->ready) { + do { if (!time_is_after_jiffies(timeout)) return -ETIMEDOUT; @@ -898,9 +889,11 @@ int fbnic_mbx_poll_tx_ready(struct fbnic_dev *fbd) return -ENODEV; msleep(20); + } while (!fbnic_mbx_event(fbd)); - fbnic_mbx_poll(fbd); - } + /* FW has shown signs of life. Enable DMA and start Tx/Rx */ + for (i = 0; i < FBNIC_IPC_MBX_INDICES; i++) + fbnic_mbx_init_desc_ring(fbd, i); /* Request an update from the firmware. This should overwrite * mgmt.version once we get the actual version from the firmware -- cgit From 23fa6a23d97182d36ca3c71e43c804fa91e46a03 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Tue, 6 May 2025 17:32:20 -0700 Subject: net: export a helper for adding up queue stats Older drivers and drivers with lower queue counts often have a static array of queues, rather than allocating structs for each queue on demand. Add a helper for adding up qstats from a queue range. Expectation is that driver will pass a queue range [netdev->real_num_*x_queues, MAX). It was tempting to always use num_*x_queues as the end, but virtio seems to clamp its queue count after allocating the netdev. And this way we can trivaly reuse the helper for [0, real_..). Signed-off-by: Jakub Kicinski Link: https://patch.msgid.link/20250507003221.823267-2-kuba@kernel.org Signed-off-by: Paolo Abeni --- include/net/netdev_queues.h | 6 ++++ net/core/netdev-genl.c | 69 ++++++++++++++++++++++++++++++++------------- 2 files changed, 56 insertions(+), 19 deletions(-) diff --git a/include/net/netdev_queues.h b/include/net/netdev_queues.h index 825141d675e5..d01c82983b4d 100644 --- a/include/net/netdev_queues.h +++ b/include/net/netdev_queues.h @@ -103,6 +103,12 @@ struct netdev_stat_ops { struct netdev_queue_stats_tx *tx); }; +void netdev_stat_queue_sum(struct net_device *netdev, + int rx_start, int rx_end, + struct netdev_queue_stats_rx *rx_sum, + int tx_start, int tx_end, + struct netdev_queue_stats_tx *tx_sum); + /** * struct netdev_queue_mgmt_ops - netdev ops for queue management * diff --git a/net/core/netdev-genl.c b/net/core/netdev-genl.c index 230743bdbb14..dae9f0d432fb 100644 --- a/net/core/netdev-genl.c +++ b/net/core/netdev-genl.c @@ -708,25 +708,66 @@ netdev_nl_stats_by_queue(struct net_device *netdev, struct sk_buff *rsp, return 0; } +/** + * netdev_stat_queue_sum() - add up queue stats from range of queues + * @netdev: net_device + * @rx_start: index of the first Rx queue to query + * @rx_end: index after the last Rx queue (first *not* to query) + * @rx_sum: output Rx stats, should be already initialized + * @tx_start: index of the first Tx queue to query + * @tx_end: index after the last Tx queue (first *not* to query) + * @tx_sum: output Tx stats, should be already initialized + * + * Add stats from [start, end) range of queue IDs to *x_sum structs. + * The sum structs must be already initialized. Usually this + * helper is invoked from the .get_base_stats callbacks of drivers + * to account for stats of disabled queues. In that case the ranges + * are usually [netdev->real_num_*x_queues, netdev->num_*x_queues). + */ +void netdev_stat_queue_sum(struct net_device *netdev, + int rx_start, int rx_end, + struct netdev_queue_stats_rx *rx_sum, + int tx_start, int tx_end, + struct netdev_queue_stats_tx *tx_sum) +{ + const struct netdev_stat_ops *ops; + struct netdev_queue_stats_rx rx; + struct netdev_queue_stats_tx tx; + int i; + + ops = netdev->stat_ops; + + for (i = rx_start; i < rx_end; i++) { + memset(&rx, 0xff, sizeof(rx)); + if (ops->get_queue_stats_rx) + ops->get_queue_stats_rx(netdev, i, &rx); + netdev_nl_stats_add(rx_sum, &rx, sizeof(rx)); + } + for (i = tx_start; i < tx_end; i++) { + memset(&tx, 0xff, sizeof(tx)); + if (ops->get_queue_stats_tx) + ops->get_queue_stats_tx(netdev, i, &tx); + netdev_nl_stats_add(tx_sum, &tx, sizeof(tx)); + } +} +EXPORT_SYMBOL(netdev_stat_queue_sum); + static int netdev_nl_stats_by_netdev(struct net_device *netdev, struct sk_buff *rsp, const struct genl_info *info) { - struct netdev_queue_stats_rx rx_sum, rx; - struct netdev_queue_stats_tx tx_sum, tx; - const struct netdev_stat_ops *ops; + struct netdev_queue_stats_rx rx_sum; + struct netdev_queue_stats_tx tx_sum; void *hdr; - int i; - ops = netdev->stat_ops; /* Netdev can't guarantee any complete counters */ - if (!ops->get_base_stats) + if (!netdev->stat_ops->get_base_stats) return 0; memset(&rx_sum, 0xff, sizeof(rx_sum)); memset(&tx_sum, 0xff, sizeof(tx_sum)); - ops->get_base_stats(netdev, &rx_sum, &tx_sum); + netdev->stat_ops->get_base_stats(netdev, &rx_sum, &tx_sum); /* The op was there, but nothing reported, don't bother */ if (!memchr_inv(&rx_sum, 0xff, sizeof(rx_sum)) && @@ -739,18 +780,8 @@ netdev_nl_stats_by_netdev(struct net_device *netdev, struct sk_buff *rsp, if (nla_put_u32(rsp, NETDEV_A_QSTATS_IFINDEX, netdev->ifindex)) goto nla_put_failure; - for (i = 0; i < netdev->real_num_rx_queues; i++) { - memset(&rx, 0xff, sizeof(rx)); - if (ops->get_queue_stats_rx) - ops->get_queue_stats_rx(netdev, i, &rx); - netdev_nl_stats_add(&rx_sum, &rx, sizeof(rx)); - } - for (i = 0; i < netdev->real_num_tx_queues; i++) { - memset(&tx, 0xff, sizeof(tx)); - if (ops->get_queue_stats_tx) - ops->get_queue_stats_tx(netdev, i, &tx); - netdev_nl_stats_add(&tx_sum, &tx, sizeof(tx)); - } + netdev_stat_queue_sum(netdev, 0, netdev->real_num_rx_queues, &rx_sum, + 0, netdev->real_num_tx_queues, &tx_sum); if (netdev_nl_stats_write_rx(rsp, &rx_sum) || netdev_nl_stats_write_tx(rsp, &tx_sum)) -- cgit From 001160ec8c59115efc39e197d40829bdafd4d7f5 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Tue, 6 May 2025 17:32:21 -0700 Subject: virtio-net: fix total qstat values NIPA tests report that the interface statistics reported via qstat are lower than those reported via ip link. Looks like this is because some tests flip the queue count up and down, and we end up with some of the traffic accounted on disabled queues. Add up counters from disabled queues. Fixes: d888f04c09bb ("virtio-net: support queue stat") Signed-off-by: Jakub Kicinski Link: https://patch.msgid.link/20250507003221.823267-3-kuba@kernel.org Signed-off-by: Paolo Abeni --- drivers/net/virtio_net.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index f9e3e628ec4d..e53ba600605a 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -5678,6 +5678,10 @@ static void virtnet_get_base_stats(struct net_device *dev, if (vi->device_stats_cap & VIRTIO_NET_STATS_TYPE_TX_SPEED) tx->hw_drop_ratelimits = 0; + + netdev_stat_queue_sum(dev, + dev->real_num_rx_queues, vi->max_queue_pairs, rx, + dev->real_num_tx_queues, vi->max_queue_pairs, tx); } static const struct netdev_stat_ops virtnet_stat_ops = { -- cgit