From 5b7c4cabbb65f5c469464da6c5f614cbd7f730f2 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Tue, 21 Feb 2023 18:24:12 -0800 Subject: Merge tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next Pull networking updates from Jakub Kicinski: "Core: - Add dedicated kmem_cache for typical/small skb->head, avoid having to access struct page at kfree time, and improve memory use. - Introduce sysctl to set default RPS configuration for new netdevs. - Define Netlink protocol specification format which can be used to describe messages used by each family and auto-generate parsers. Add tools for generating kernel data structures and uAPI headers. - Expose all net/core sysctls inside netns. - Remove 4s sleep in netpoll if carrier is instantly detected on boot. - Add configurable limit of MDB entries per port, and port-vlan. - Continue populating drop reasons throughout the stack. - Retire a handful of legacy Qdiscs and classifiers. Protocols: - Support IPv4 big TCP (TSO frames larger than 64kB). - Add IP_LOCAL_PORT_RANGE socket option, to control local port range on socket by socket basis. - Track and report in procfs number of MPTCP sockets used. - Support mixing IPv4 and IPv6 flows in the in-kernel MPTCP path manager. - IPv6: don't check net.ipv6.route.max_size and rely on garbage collection to free memory (similarly to IPv4). - Support Penultimate Segment Pop (PSP) flavor in SRv6 (RFC8986). - ICMP: add per-rate limit counters. - Add support for user scanning requests in ieee802154. - Remove static WEP support. - Support minimal Wi-Fi 7 Extremely High Throughput (EHT) rate reporting. - WiFi 7 EHT channel puncturing support (client & AP). BPF: - Add a rbtree data structure following the "next-gen data structure" precedent set by recently added linked list, that is, by using kfunc + kptr instead of adding a new BPF map type. - Expose XDP hints via kfuncs with initial support for RX hash and timestamp metadata. - Add BPF_F_NO_TUNNEL_KEY extension to bpf_skb_set_tunnel_key to better support decap on GRE tunnel devices not operating in collect metadata. - Improve x86 JIT's codegen for PROBE_MEM runtime error checks. - Remove the need for trace_printk_lock for bpf_trace_printk and bpf_trace_vprintk helpers. - Extend libbpf's bpf_tracing.h support for tracing arguments of kprobes/uprobes and syscall as a special case. - Significantly reduce the search time for module symbols by livepatch and BPF. - Enable cpumasks to be used as kptrs, which is useful for tracing programs tracking which tasks end up running on which CPUs in different time intervals. - Add support for BPF trampoline on s390x and riscv64. - Add capability to export the XDP features supported by the NIC. - Add __bpf_kfunc tag for marking kernel functions as kfuncs. - Add cgroup.memory=nobpf kernel parameter option to disable BPF memory accounting for container environments. Netfilter: - Remove the CLUSTERIP target. It has been marked as obsolete for years, and we still have WARN splats wrt races of the out-of-band /proc interface installed by this target. - Add 'destroy' commands to nf_tables. They are identical to the existing 'delete' commands, but do not return an error if the referenced object (set, chain, rule...) did not exist. Driver API: - Improve cpumask_local_spread() locality to help NICs set the right IRQ affinity on AMD platforms. - Separate C22 and C45 MDIO bus transactions more clearly. - Introduce new DCB table to control DSCP rewrite on egress. - Support configuration of Physical Layer Collision Avoidance (PLCA) Reconciliation Sublayer (RS) (802.3cg-2019). Modern version of shared medium Ethernet. - Support for MAC Merge layer (IEEE 802.3-2018 clause 99). Allowing preemption of low priority frames by high priority frames. - Add support for controlling MACSec offload using netlink SET. - Rework devlink instance refcounts to allow registration and de-registration under the instance lock. Split the code into multiple files, drop some of the unnecessarily granular locks and factor out common parts of netlink operation handling. - Add TX frame aggregation parameters (for USB drivers). - Add a new attr TCA_EXT_WARN_MSG to report TC (offload) warning messages with notifications for debug. - Allow offloading of UDP NEW connections via act_ct. - Add support for per action HW stats in TC. - Support hardware miss to TC action (continue processing in SW from a specific point in the action chain). - Warn if old Wireless Extension user space interface is used with modern cfg80211/mac80211 drivers. Do not support Wireless Extensions for Wi-Fi 7 devices at all. Everyone should switch to using nl80211 interface instead. - Improve the CAN bit timing configuration. Use extack to return error messages directly to user space, update the SJW handling, including the definition of a new default value that will benefit CAN-FD controllers, by increasing their oscillator tolerance. New hardware / drivers: - Ethernet: - nVidia BlueField-3 support (control traffic driver) - Ethernet support for imx93 SoCs - Motorcomm yt8531 gigabit Ethernet PHY - onsemi NCN26000 10BASE-T1S PHY (with support for PLCA) - Microchip LAN8841 PHY (incl. cable diagnostics and PTP) - Amlogic gxl MDIO mux - WiFi: - RealTek RTL8188EU (rtl8xxxu) - Qualcomm Wi-Fi 7 devices (ath12k) - CAN: - Renesas R-Car V4H Drivers: - Bluetooth: - Set Per Platform Antenna Gain (PPAG) for Intel controllers. - Ethernet NICs: - Intel (1G, igc): - support TSN / Qbv / packet scheduling features of i226 model - Intel (100G, ice): - use GNSS subsystem instead of TTY - multi-buffer XDP support - extend support for GPIO pins to E823 devices - nVidia/Mellanox: - update the shared buffer configuration on PFC commands - implement PTP adjphase function for HW offset control - TC support for Geneve and GRE with VF tunnel offload - more efficient crypto key management method - multi-port eswitch support - Netronome/Corigine: - add DCB IEEE support - support IPsec offloading for NFP3800 - Freescale/NXP (enetc): - support XDP_REDIRECT for XDP non-linear buffers - improve reconfig, avoid link flap and waiting for idle - support MAC Merge layer - Other NICs: - sfc/ef100: add basic devlink support for ef100 - ionic: rx_push mode operation (writing descriptors via MMIO) - bnxt: use the auxiliary bus abstraction for RDMA - r8169: disable ASPM and reset bus in case of tx timeout - cpsw: support QSGMII mode for J721e CPSW9G - cpts: support pulse-per-second output - ngbe: add an mdio bus driver - usbnet: optimize usbnet_bh() by avoiding unnecessary queuing - r8152: handle devices with FW with NCM support - amd-xgbe: support 10Mbps, 2.5GbE speeds and rx-adaptation - virtio-net: support multi buffer XDP - virtio/vsock: replace virtio_vsock_pkt with sk_buff - tsnep: XDP support - Ethernet high-speed switches: - nVidia/Mellanox (mlxsw): - add support for latency TLV (in FW control messages) - Microchip (sparx5): - separate explicit and implicit traffic forwarding rules, make the implicit rules always active - add support for egress DSCP rewrite - IS0 VCAP support (Ingress Classification) - IS2 VCAP filters (protos, L3 addrs, L4 ports, flags, ToS etc.) - ES2 VCAP support (Egress Access Control) - support for Per-Stream Filtering and Policing (802.1Q, 8.6.5.1) - Ethernet embedded switches: - Marvell (mv88e6xxx): - add MAB (port auth) offload support - enable PTP receive for mv88e6390 - NXP (ocelot): - support MAC Merge layer - support for the the vsc7512 internal copper phys - Microchip: - lan9303: convert to PHYLINK - lan966x: support TC flower filter statistics - lan937x: PTP support for KSZ9563/KSZ8563 and LAN937x - lan937x: support Credit Based Shaper configuration - ksz9477: support Energy Efficient Ethernet - other: - qca8k: convert to regmap read/write API, use bulk operations - rswitch: Improve TX timestamp accuracy - Intel WiFi (iwlwifi): - EHT (Wi-Fi 7) rate reporting - STEP equalizer support: transfer some STEP (connection to radio on platforms with integrated wifi) related parameters from the BIOS to the firmware. - Qualcomm 802.11ax WiFi (ath11k): - IPQ5018 support - Fine Timing Measurement (FTM) responder role support - channel 177 support - MediaTek WiFi (mt76): - per-PHY LED support - mt7996: EHT (Wi-Fi 7) support - Wireless Ethernet Dispatch (WED) reset support - switch to using page pool allocator - RealTek WiFi (rtw89): - support new version of Bluetooth co-existance - Mobile: - rmnet: support TX aggregation" * tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1872 commits) page_pool: add a comment explaining the fragment counter usage net: ethtool: fix __ethtool_dev_mm_supported() implementation ethtool: pse-pd: Fix double word in comments xsk: add linux/vmalloc.h to xsk.c sefltests: netdevsim: wait for devlink instance after netns removal selftest: fib_tests: Always cleanup before exit net/mlx5e: Align IPsec ASO result memory to be as required by hardware net/mlx5e: TC, Set CT miss to the specific ct action instance net/mlx5e: Rename CHAIN_TO_REG to MAPPED_OBJ_TO_REG net/mlx5: Refactor tc miss handling to a single function net/mlx5: Kconfig: Make tc offload depend on tc skb extension net/sched: flower: Support hardware miss to tc action net/sched: flower: Move filter handle initialization earlier net/sched: cls_api: Support hardware miss to tc action net/sched: Rename user cookie and act cookie sfc: fix builds without CONFIG_RTC_LIB sfc: clean up some inconsistent indentings net/mlx4_en: Introduce flexible array to silence overflow warning net: lan966x: Fix possible deadlock inside PTP net/ulp: Remove redundant ->clone() test in inet_clone_ulp(). ... --- Documentation/bpf/graph_ds_impl.rst | 267 ++++++++++++++++++++++++++++++++++++ 1 file changed, 267 insertions(+) create mode 100644 Documentation/bpf/graph_ds_impl.rst (limited to 'Documentation/bpf/graph_ds_impl.rst') diff --git a/Documentation/bpf/graph_ds_impl.rst b/Documentation/bpf/graph_ds_impl.rst new file mode 100644 index 000000000..61274622b --- /dev/null +++ b/Documentation/bpf/graph_ds_impl.rst @@ -0,0 +1,267 @@ +========================= +BPF Graph Data Structures +========================= + +This document describes implementation details of new-style "graph" data +structures (linked_list, rbtree), with particular focus on the verifier's +implementation of semantics specific to those data structures. + +Although no specific verifier code is referred to in this document, the document +assumes that the reader has general knowledge of BPF verifier internals, BPF +maps, and BPF program writing. + +Note that the intent of this document is to describe the current state of +these graph data structures. **No guarantees** of stability for either +semantics or APIs are made or implied here. + +.. contents:: + :local: + :depth: 2 + +Introduction +------------ + +The BPF map API has historically been the main way to expose data structures +of various types for use within BPF programs. Some data structures fit naturally +with the map API (HASH, ARRAY), others less so. Consequentially, programs +interacting with the latter group of data structures can be hard to parse +for kernel programmers without previous BPF experience. + +Luckily, some restrictions which necessitated the use of BPF map semantics are +no longer relevant. With the introduction of kfuncs, kptrs, and the any-context +BPF allocator, it is now possible to implement BPF data structures whose API +and semantics more closely match those exposed to the rest of the kernel. + +Two such data structures - linked_list and rbtree - have many verification +details in common. Because both have "root"s ("head" for linked_list) and +"node"s, the verifier code and this document refer to common functionality +as "graph_api", "graph_root", "graph_node", etc. + +Unless otherwise stated, examples and semantics below apply to both graph data +structures. + +Unstable API +------------ + +Data structures implemented using the BPF map API have historically used BPF +helper functions - either standard map API helpers like ``bpf_map_update_elem`` +or map-specific helpers. The new-style graph data structures instead use kfuncs +to define their manipulation helpers. Because there are no stability guarantees +for kfuncs, the API and semantics for these data structures can be evolved in +a way that breaks backwards compatibility if necessary. + +Root and node types for the new data structures are opaquely defined in the +``uapi/linux/bpf.h`` header. + +Locking +------- + +The new-style data structures are intrusive and are defined similarly to their +vanilla kernel counterparts: + +.. code-block:: c + + struct node_data { + long key; + long data; + struct bpf_rb_node node; + }; + + struct bpf_spin_lock glock; + struct bpf_rb_root groot __contains(node_data, node); + +The "root" type for both linked_list and rbtree expects to be in a map_value +which also contains a ``bpf_spin_lock`` - in the above example both global +variables are placed in a single-value arraymap. The verifier considers this +spin_lock to be associated with the ``bpf_rb_root`` by virtue of both being in +the same map_value and will enforce that the correct lock is held when +verifying BPF programs that manipulate the tree. Since this lock checking +happens at verification time, there is no runtime penalty. + +Non-owning references +--------------------- + +**Motivation** + +Consider the following BPF code: + +.. code-block:: c + + struct node_data *n = bpf_obj_new(typeof(*n)); /* ACQUIRED */ + + bpf_spin_lock(&lock); + + bpf_rbtree_add(&tree, n); /* PASSED */ + + bpf_spin_unlock(&lock); + +From the verifier's perspective, the pointer ``n`` returned from ``bpf_obj_new`` +has type ``PTR_TO_BTF_ID | MEM_ALLOC``, with a ``btf_id`` of +``struct node_data`` and a nonzero ``ref_obj_id``. Because it holds ``n``, the +program has ownership of the pointee's (object pointed to by ``n``) lifetime. +The BPF program must pass off ownership before exiting - either via +``bpf_obj_drop``, which ``free``'s the object, or by adding it to ``tree`` with +``bpf_rbtree_add``. + +(``ACQUIRED`` and ``PASSED`` comments in the example denote statements where +"ownership is acquired" and "ownership is passed", respectively) + +What should the verifier do with ``n`` after ownership is passed off? If the +object was ``free``'d with ``bpf_obj_drop`` the answer is obvious: the verifier +should reject programs which attempt to access ``n`` after ``bpf_obj_drop`` as +the object is no longer valid. The underlying memory may have been reused for +some other allocation, unmapped, etc. + +When ownership is passed to ``tree`` via ``bpf_rbtree_add`` the answer is less +obvious. The verifier could enforce the same semantics as for ``bpf_obj_drop``, +but that would result in programs with useful, common coding patterns being +rejected, e.g.: + +.. code-block:: c + + int x; + struct node_data *n = bpf_obj_new(typeof(*n)); /* ACQUIRED */ + + bpf_spin_lock(&lock); + + bpf_rbtree_add(&tree, n); /* PASSED */ + x = n->data; + n->data = 42; + + bpf_spin_unlock(&lock); + +Both the read from and write to ``n->data`` would be rejected. The verifier +can do better, though, by taking advantage of two details: + + * Graph data structure APIs can only be used when the ``bpf_spin_lock`` + associated with the graph root is held + + * Both graph data structures have pointer stability + + * Because graph nodes are allocated with ``bpf_obj_new`` and + adding / removing from the root involves fiddling with the + ``bpf_{list,rb}_node`` field of the node struct, a graph node will + remain at the same address after either operation. + +Because the associated ``bpf_spin_lock`` must be held by any program adding +or removing, if we're in the critical section bounded by that lock, we know +that no other program can add or remove until the end of the critical section. +This combined with pointer stability means that, until the critical section +ends, we can safely access the graph node through ``n`` even after it was used +to pass ownership. + +The verifier considers such a reference a *non-owning reference*. The ref +returned by ``bpf_obj_new`` is accordingly considered an *owning reference*. +Both terms currently only have meaning in the context of graph nodes and API. + +**Details** + +Let's enumerate the properties of both types of references. + +*owning reference* + + * This reference controls the lifetime of the pointee + + * Ownership of pointee must be 'released' by passing it to some graph API + kfunc, or via ``bpf_obj_drop``, which ``free``'s the pointee + + * If not released before program ends, verifier considers program invalid + + * Access to the pointee's memory will not page fault + +*non-owning reference* + + * This reference does not own the pointee + + * It cannot be used to add the graph node to a graph root, nor ``free``'d via + ``bpf_obj_drop`` + + * No explicit control of lifetime, but can infer valid lifetime based on + non-owning ref existence (see explanation below) + + * Access to the pointee's memory will not page fault + +From verifier's perspective non-owning references can only exist +between spin_lock and spin_unlock. Why? After spin_unlock another program +can do arbitrary operations on the data structure like removing and ``free``-ing +via bpf_obj_drop. A non-owning ref to some chunk of memory that was remove'd, +``free``'d, and reused via bpf_obj_new would point to an entirely different thing. +Or the memory could go away. + +To prevent this logic violation all non-owning references are invalidated by the +verifier after a critical section ends. This is necessary to ensure the "will +not page fault" property of non-owning references. So if the verifier hasn't +invalidated a non-owning ref, accessing it will not page fault. + +Currently ``bpf_obj_drop`` is not allowed in the critical section, so +if there's a valid non-owning ref, we must be in a critical section, and can +conclude that the ref's memory hasn't been dropped-and- ``free``'d or +dropped-and-reused. + +Any reference to a node that is in an rbtree _must_ be non-owning, since +the tree has control of the pointee's lifetime. Similarly, any ref to a node +that isn't in rbtree _must_ be owning. This results in a nice property: +graph API add / remove implementations don't need to check if a node +has already been added (or already removed), as the ownership model +allows the verifier to prevent such a state from being valid by simply checking +types. + +However, pointer aliasing poses an issue for the above "nice property". +Consider the following example: + +.. code-block:: c + + struct node_data *n, *m, *o, *p; + n = bpf_obj_new(typeof(*n)); /* 1 */ + + bpf_spin_lock(&lock); + + bpf_rbtree_add(&tree, n); /* 2 */ + m = bpf_rbtree_first(&tree); /* 3 */ + + o = bpf_rbtree_remove(&tree, n); /* 4 */ + p = bpf_rbtree_remove(&tree, m); /* 5 */ + + bpf_spin_unlock(&lock); + + bpf_obj_drop(o); + bpf_obj_drop(p); /* 6 */ + +Assume the tree is empty before this program runs. If we track verifier state +changes here using numbers in above comments: + + 1) n is an owning reference + + 2) n is a non-owning reference, it's been added to the tree + + 3) n and m are non-owning references, they both point to the same node + + 4) o is an owning reference, n and m non-owning, all point to same node + + 5) o and p are owning, n and m non-owning, all point to the same node + + 6) a double-free has occurred, since o and p point to same node and o was + ``free``'d in previous statement + +States 4 and 5 violate our "nice property", as there are non-owning refs to +a node which is not in an rbtree. Statement 5 will try to remove a node which +has already been removed as a result of this violation. State 6 is a dangerous +double-free. + +At a minimum we should prevent state 6 from being possible. If we can't also +prevent state 5 then we must abandon our "nice property" and check whether a +node has already been removed at runtime. + +We prevent both by generalizing the "invalidate non-owning references" behavior +of ``bpf_spin_unlock`` and doing similar invalidation after +``bpf_rbtree_remove``. The logic here being that any graph API kfunc which: + + * takes an arbitrary node argument + + * removes it from the data structure + + * returns an owning reference to the removed node + +May result in a state where some other non-owning reference points to the same +node. So ``remove``-type kfuncs must be considered a non-owning reference +invalidation point as well. -- cgit v1.2.3