Linux 5.17 changelog

Summary:

1. Prominent features

1.1. Support recursive id-mapped mounts

Linux 5.12 added support for id-mapped mounts, that is, creating a mount of an existing file system with the user and group IDs mapped to different ones. This is very useful for containers, where you might want to mount a directory with some random user ID as the user ID expected by the container. This feature, however, required that the base file system was a file system that was not id-mapped itself. This release supports recursive id-mapped mounts.

1.2. BPF CO-RE support and other improvements

The nature of BPF programs makes hard for them to be portable: The position of a member in a data structure can change depending from version to version and even on the same kernel depending on the compiler options. This requires compiling BPF programs in the target computer. In order to make BPF programs more portable, a few features are required; some have already been merged, like BTF, which was introduced in Linux 4.18 and provides type information of all kernel data structures. This release introduces "CO-RE" -Compile Once, Run Everywhere- support. Along with other pieces in the userspace BPF infrastructure, this helps to create BPF programs that can be run in different kernels without requiring compilation.

Recommended article: BPF CO-RE reference guide.

This release also adds the bpf_loop helper, which helps to move loop logic of BPF programs into the kernel and thereby guarantee that the loop will always terminate

Recommended article: A different approach to BPF loops

2. Core (various)

• (FEATURED) vfs: Extend the mapping infrastructure in order to support mapped mounts of mapped filesystems commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• fanotify: Add FAN_RENAME event to report extra info records about new and old parent+name details. This new event was designed as a replacement for the "inotify way" of joining the MOVED_FROM/MOVED_TO events using a cookie commit, commit, commit, commit, commit, commit, commit, commit, commit

• devtmpfs: default mount with noexec and nosuid commit

• fuse,virtiofs: support per-file DAX commit, commit, commit, commit, commit, commit, commit

• Task scheduler

  • Adds accounting for "forced idle" time, which is time where a cookie'd task forces its SMT sibling to idle, despite the presence of runnable tasks commit

  • delayacct: support swapin delay accounting for swapping without blkio commit

  • delayacct: track delays from memory compact commit

• cgroup

  • cpuacct: Include guest time in user time in cpuacct.stat commit

  • cpuacct: Make user/system times in cpuacct.stat more precise commit

• signals: requeuing undeliverable signals commit, commit, commit

• topology: only export used sysfs attributes commit, commit, commit

• io_uring

  • Allow to skip CQE posting commit, commit, commit, commit

  • Task work optimization commit, commit, commit, commit, commit

  • Reworkg io_uring's poll and internal poll commit, commit, commit, commit, commit, commit, commit

• RCU

  • Tasks updates commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • RCU no-CB CPUs updates, most notably the ability to offload CPUs that are in de-offloaded state at boot time commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • Remove CONFIG_RCU_FAST_NO_HZ support commit, commit, commit, commit

• Reference counting tracking infrastructure. Recommended LWN article: A reference-count tracking infrastructure.commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• module: add in-kernel support for decompressing for security purposes commit

• Build: introduce debug.config for CI-like setup commit

• fscache, cachefiles: Rewrite, significantly simplifying the code compared to what's upstream, removing the complex operation scheduling and object state machine in favour of something much smaller and simpler commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• irq: [PCI]MSI refactoring which aims to provide the ability of expanding MSI-X vectors after enabling MSI-X commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

3. File systems

• BTRFS

  • Speedup directory logging/fsync by copying index keys only. This reduces the amount of logged metadata by about half, and therefore the time spent logging or fsyncing large directories commit, commit

  • Enable adding of a device when balance is paused (i.e an fs is mounted with skip_balance options) commit, commit, commit

  • Allow defrag to be interruptible commit

  • Index free space entries on size commit, commit, commit

  • Make send work with concurrent block group relocation commit

  • btrfs: remove reada infrastructure commit

• CEPH

  • New mount device syntax commit, commit, commit, commit, commit

  • Mount syntax module parameter commit

  • Add new "nopagecache" option commit

• CIFS

  • Port to the new fscache I/O commit, commit, commit, commit, commit

  • smb3: send NTLMSSP version information commit

• XFS

  • Support large folios commit

  • Kill dead ioctls commit, commit, commit

• EXT4

  • Convert to new mount API commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • Remove lazytime/nolazytime mount options handled by MS_LAZYTIME commit

  • Implement support for get/set fs label commit

• DLM

  • Add debugfs rawmsg send functionality commit

  • Add lkb debugfs functionality commit

• EROFS

  • Support tail-packing inline compressed data commit, commit, commit, commit, commit

  • Add sysfs interface commit

  • Add sysfs node to control sync decompression strategy commit

• F2FS

  • Add gc_urgent_high_remaining sysfs node commit

  • Show number of pending discard commands commit

  • Use iomap for Direct I/O commit, commit

  • Support POSIX_FADV_DONTNEED drop compressed page cache commit

• FUSE

  • fuse,virtiofs: support per-file DAX commit, commit, commit, commit, commit, commit, commit

  • Send file/inode security context during creation commit

  • Extend init flags to avoid running out of flags commit

• KSMBD

  • Add support for key exchange commit

  • Set 445 port to smbdirect port by default commit

  • Add smb-direct shutdown commit

  • Add support for smb2 max credit parameter commit

  • Add reserved room in ipc request/response commit

• NFS

  • NFSv4.1+ support for session trunking discovery commit, commit, commit, commit, commit, commit, commit

  • NFSv4: Add some support for case insensitive filesystems commit

• UBIFS

  • Export filesystem error counters commit

4. Memory management

• Batch of memory folio patches commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• Segregate a struct slab from struct page commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• Extend vmalloc support for constrained allocations commit, commit, commit, commit

• Support for anonymous VMA naming. It adds a field to /proc/pid/maps and /proc/pid/smaps to show a userspace-provided name for anonymous vmas. The names of named anonymous vmas are shown as [anon:<name>]. Userspace can set the name for a region of memory by calling prctl(PR_SET_VMA, PR_SET_VMA_ANON_NAME, start, len, (unsigned long)name); commit, commit, commit

• cgroup

  • memcg: add oom_group_kill memory event when memory.oom.group triggers to allow userspace to cleanly identify when an entire cgroup is oom killed commit * memcg: add per-memcg vmalloc stat commit

  • memcg: better bounds on the memcg stats updates commit

  • hugetlb: add hugetlb.*.numa_stat file commit

• kcsan: Support detecting a subset of missing memory barriers commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• damon

  • schemes: Extend stats for better online analysis and tuning commit, commit, commit, commit, commit, commit

  • Add access checking for hugetlb pages commit

• Adds set_mempolicy_home_node syscall to set a home node for the MPOL_BIND and MPOL_PREFERRED_MANY memory policy. Users should use this syscall after setting up a memory policy for the specified range. The syscall allows specifying a home node/preferred node from which kernel will fulfill memory allocation requests first. This helps applications to hint at a memory allocation preference node and fallback to _only_ a set of nodes if the memory is not available on the preferred node. Fallback allocation is attempted from the node which is nearest to the preferred node commit, commit, commit

• device-dax: Introduce compound pages in devmap in order to minimize 'struct page' overhead commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• sysctl: change watermark_scale_factor max limit to 30% commit

• vmstat: add events for THP max_ptes_* commit

• Introduce PAGE_TABLE_CHECK debug option to check user page table entries at the time they are added and removed, in order to catch memory corruption issues related to double mapping commit, commit, commit, commit

• Improve the migration stats commit, commit, commit

• Migration: support multiple target nodes demotion commit

5. Block layer

• Memory folio support in block + iomap layers commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• md: drop queue limitation for RAID1 and RAID10 commit

• Remove the rsxx driver commit

6. Tracing, perf and BPF

• Add rtla(1) tool, a meta-tool including a set of commands that aims to analyze the real-time properties of Linux. But instead of testing Linux as a black box, rtla leverages kernel tracing capabilities to provide precise information about the properties and root causes of unexpected results. In this release it presents an interface to the osnoise and timerlat tracers. In the future, it will also serve as home to the rtsl and other latency/noise tracers commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• BPF

  • (FEATURED) BPF CO-RE (COmpile once - Run Everywhere) support commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • (FEATURED) Add bpf_loop helper. One of the complexities of using for loops in bpf programs is that the verifier needs to ensure that in every possibility of the loop logic, the loop will always terminate. As such, there is a limit on how many iterations the loop can do. The bpf_loop helper moves the loop logic into the kernel and can thereby guarantee that the loop will always terminate. The bpf_loop helper simplifies a lot of the complexity the verifier needs to check, as well as removes the constraint on the number of loops able to be run commit, commit, commit, commit

  • Introduce composable BPF types commit, commit, commit, commit, commit, commit, commit, commit, commit

  • Adds unstable conntrack lookup helpers using BPF kfunc support

  • Sleepable local storage commit

  • Add helpers to access traced function arguments commit, commit, commit, commit, commit

  • Support BTF_KIND_TYPE_TAG for btf_type_tag attributes. The main motivation for btf_type_tag is to bring kernel annotations __user, __rcu etc. to btf. With such information available in btf, bpf verifier can detect mis-usages and reject the program. For example, for __user tagged pointer, developers can then use proper helper like bpf_probe_read_kernel() etc. to read the data commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• perf

  • Support event alias in form foo-bar-baz commit, commit, commit

  • perf stat: Support --cputype option for hybrid events commit

  • perf ftrace: Implement function latency histogram commit, commit, commit, commit, commit

  • perf evlist: Allow setting arbitrary leader commit, commit

• libbpf

  • Deprecate bpf_prog_load_xattr() API commit, commit, commit, commit, commit, commit, commit, commit, commit

  • Enhance and rework logging controls commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • Add runtime APIs to query libbpf version commit

  • Revamp and fix libbpf's feature-probing APIs commit, commit, commit

  • Use probe_name for legacy kprobe commit, commit

  • bpftool: Probes for bounded loops and instruction set extensions commit, commit, commit

  • bpftool: Enable libbpf's strict mode by default commit

  • bpftool: Add current libbpf_strict mode to version output commit

• tracing

  • Add ustring operation to filtering string pointers commit

  • 'rel_loc' relative dynamic array attribute support commit, commit, commit, commit, commit

7. Virtualization

• Allow for configuring max number of virtqueue pairs commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• Add XEN guest pvUSB support commit, commit, commit

• vdpa: add driver_override support and sysfs ABI documentation commit

8. Cryptography

• Add SP800-108 KDF implementation to crypto API commit, commit, commit, commit

• DRBG - improve 'nopr' reseeding commit, commit, commit, commit, commit, commit

• jitter: add oversampling of noise source commit

• keys: X.509 public key issuer lookup without AKID commit

9. Networking

• Bluetooth

  • aosp: Support AOSP Bluetooth Quality Report commit, commit

  • Set Privacy Mode when updating the resolving list commit, commit

  • hci_sync: Add support for waiting specific LE subevents commit

• sysctl:

  • Enable max_dgram_qlen unix sysctl to be configurable by non-init user namespaces commit

  • Enable neighbor sysctls that is save for userns root commit

  • Namespaceify min_pmtu sysctl commit

  • Namespaceify mtu_expires sysctl commit

• Allow CAP_NET_RAW to setsockopt SO_PRIORITY commit

• Allow SO_MARK with CAP_NET_RAW commit

• batman-adv: allow netlink usage in unprivileged containers commit

• Batching iter for AF_UNIX sockets

• Expose SO_RCVBUF/SO_SNDBUF through bpf_getsockopt() commit

• Get ingress_ifindex in BPF_SK_LOOKUP prog type commit, commit

• bridge: Allow base 16 inputs in sysfs commit

• Wireless

  • Add support for notifying association comeback commit, commit

  • Allow continuous radar monitoring on offchannel chain commit

  • Implement APIs for dedicated radar detection HW commit

  • Allow drivers to provide a destination device + info for flow offload. Only supported in combination with 802.3 encap offload commit

  • Indicate SA Query procedures offload for AP SME device commit, commit

• ethtool

  • Add support to set/get rx buf len via ethtool commit

  • Add support to set/get tx copybreak buf size via ethtool commit

• Allow user to offload tc action to net device commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• GRO: add ability to control gro max packet size commit

• IPv4: support binding to nonlocal addresses commit

• IPv6: ioam: Support for Queue depth data field commit

• mctp: emit RTM_NEWADDR and RTM_DELADDR netlink messages for MTCP address changes commit

• MPTCP

  • More socket option support: IP_TOS, IP_FREEBIND, IP_TRANSPARENT, IPV6_FREEBIND, and IPV6_TRANSPARENT commit, commit, commit, commit

  • Adds socket support for a few socket options, ioctls, and one ancillary data type commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • Add mibs counter for ignored incoming options commit

  • Enforce HoL-blocking estimation commit

  • improve accept() and disconnect() commit, commit, commit, commit, commit, commit

• Netfilter

  • bridge: add support for pppoe filtering commit

  • Add register tracking infrastructure to skip redundant store-to-register operations, this includes support for payload, meta and bitwise expresssions commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • nft_fwd_netdev: Support egress hook commit

• sit: allow encapsulated IPv6 traffic to be delivered locally commit

• smc

  • Introduce TCP ULP support commit

  • ntroduce net namespace support for linkgroup commit

• xfrm: Add support for SM3 and SM4 secure hashes commit, commit

10. Architectures

10.1. ARM

• Device Tree Sources

  • New SoC: Qualcomm gets support for two newly announced platforms, both of hich can now work in production environments: the SDX65 5G modem that can run a minimal Linux on its Cortex-A7 core, and the Snapdragon 8 Gen 1, their latest high-end phone SoC commit

  • New SoC: Renesas adds support for R-Car S4-8, the most recent automotive Server/Communication SoC commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • New SoC: TI adds support for J721s2, a new automotive SoC in the K3 family commit, commit, commit, commit, commit

  • New SoC: Mediatek MT7986a/b is a SoC used in Wifi routers, the latest generation following their popular MT76xx series. Only basic support is added for now commit, commit

  • New SoC: NXP i.MX8 ULP8 is a new low-power variant of the widespread i.MX8 series commit, commit

  • New SoC: TI SPEAr320s is a minor variant of the old SPEAr320 SoC that we have supported for a long time commit

  • Aspeed AST2500/AST2600 BMCs in TYAN, Facebook and Yadro servers commit

  • AT91/SAMA5 based evaluation board commit

  • Intel IXP4xx now supports the final two machines in device tree that were previously only supported in old style board files commit

  • Mediatek MT6589, used in the Fairphone FP1 phone from 2013, while MT8183 is used in the Acer Chromebook 314 commit, commit

  • Qualcomm gains support for the reference machines using the two new SoCs, plus a number of Chromebook variants and phones based on the Snapdragon 7c, 845 and 888 SoCs, including Sony Xperia devices (Xperia XZ2 / XZ2C / XZ3 (Tama platform), Xperia 1 III / 5 III), Samsung J5, and the Microsoft Surface Duo 2 commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • ST STM32 now supports the Engicam i.Core STM32MP1 carrier board commit

  • Tegra now boots various older Android devices based on 32-bit chips out of the box, including a number of ASUS Transformer tablets (Prime TF201, Pad TF701T, Pad TF300T, Infinity TF700T, EeePad TF101, Pad TF300TG). There is also a new Jetson AGX Orin developer kit commit, commit, commit, commit, commit, commit, commit, commit

  • Apple support adds the missing device trees for all the remaining M1 Macbook and iMac variants, though not yet the M1 Pro/Max versions commit

  • Allwinner now supports another version of the Tanix TX6 set-top box based on the H6 SoC commit

  • Broadcom gains support for the Netgear RAXE500 Wireless router based on BCM4908 commit

  • Adds support for the following modules: TQMa8Mx, TQMa8MxML, TQMa8MxNL. Each of the modules is available with different i.MX8M variants commit, commit, commit

  • tegra: Add device-tree for 1080p version of Nyan Big commit

  • Add Goramo MultiLink device tree commit

  • Add JOZ Access Point commit

  • aspeed: Adding Facebook Bletchley BMC commit

  • aspeed: add device tree for YADRO VEGMAN BMC commit

  • imx6: phytec: Add PEB-WLBT-05 support commit

  • imx6dl-yapp4: Add Y Soft IOTA Crux/Crux+ board commit

  • imx6qdl: add TQ-Systems MBa6x device trees commit

  • imx6qdl: phytec: Add support for optional PEB-EVAL-01 board commit

  • apple: Add t6000/t6001 MacBook Pro 14/16" compatibles commit

• Enable KCSAN commit

• Support KFENCE commit, commit, commit

• Add support for Cortex-M55 processor commit

• Add support for Cortex-M33 processor commit

• Report Spectre v2 status through sysfs commit

• arm64: Add two HWCAPs for Arm v8.7 FP behaviour commit, commit, commit

• Xen: Add support of extended regions (safe ranges) on Arm commit, commit, commit, commit, commit, commit

• Adds runtime PM support to Tegra drivers and enables core voltage scaling for Tegra20/30 SoCs, resolving overheating troubles commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

• perf

  • Userspace counter access support commit, commit, commit, commit, commit

  • vendor events: Arm Neoverse N2 commit

  • arm-spe: Add SPE total latency as PERF_SAMPLE_WEIGHT commit

  • arm-spe: Synthesize SPE instruction events commit

  • Inject missing frames when using 'perf record --call-graph=fp' commit

  • hisi: Add driver for HiSilicon PCIe PMU commit, commit

  • Add LLC-TAD perf counter support commit

  • Arm CMN updates commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit

  • smmuv3: Add devicetree support commit

• Add Xilinx Event Management Driver commit, commit, commit

• coresight: syscfg: dynamic load commit, commit, commit, commit, commit, commit

• coresight: trbe: Workaround Cortex-A510 erratas commit, commit, commit, commit

• Apple SoC PMGR device power states driver commit, commit, commit, commit, commit, commit

• samsung: Add USI driver commit

• samsung: exynos-chipid: add Exynos7885 SoC support commit

• KVM: arm64: series implements an unshare hypercall at EL2 in nVHE protected mode, and makes use of it to unmmap guest-specific data-structures from EL2 stage-1 during guest tear-down commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit, commit