This bug was fixed in the package linux-azure - 4.18.0-1006.6
---------------
linux-azure (4.18.0-1006.6) cosmic; urgency=medium
* linux-azure: 4.18.0-1006.6 -proposed tracker (LP: #1805244)
* Accelerated networking (SR-IOV VF) broken in 18.10 daily (LP: #1794477)
- [Packaging] Move pci-hyperv and autofs4 back to linux-modules
linux-azure (4.18.0-1005.5) cosmic; urgency=medium
* linux-azure: 4.18.0-1005.5 -proposed tracker (LP: #1802752)
* [Hyper-V] Fix IRQ spreading on NVMe devices with lower numbers of channels
(LP: #1802358)
- SAUCE: genirq/affinity: Spread IRQs to all available NUMA nodes
- SAUCE: irq/matrix: Split out the CPU selection code into a helper
- SAUCE: irq/matrix: Spread managed interrupts on allocation
- SAUCE: genirq/matrix: Improve target CPU selection for managed interrupts.
[ Ubuntu: 4.18.0-12.13 ]
* linux: 4.18.0-12.13 -proposed tracker (LP: #1802743)
* [FEAT] Guest-dedicated Crypto Adapters (LP: #1787405)
- s390/zcrypt: Add ZAPQ inline function.
- s390/zcrypt: Review inline assembler constraints.
- s390/zcrypt: Integrate ap_asm.h into include/asm/ap.h.
- s390/zcrypt: fix ap_instructions_available() returncodes
- KVM: s390: vsie: simulate VCPU SIE entry/exit
- KVM: s390: introduce and use KVM_REQ_VSIE_RESTART
- KVM: s390: refactor crypto initialization
- s390: vfio-ap: base implementation of VFIO AP device driver
- s390: vfio-ap: register matrix device with VFIO mdev framework
- s390: vfio-ap: sysfs interfaces to configure adapters
- s390: vfio-ap: sysfs interfaces to configure domains
- s390: vfio-ap: sysfs interfaces to configure control domains
- s390: vfio-ap: sysfs interface to view matrix mdev matrix
- KVM: s390: interface to clear CRYCB masks
- s390: vfio-ap: implement mediated device open callback
- s390: vfio-ap: implement VFIO_DEVICE_GET_INFO ioctl
- s390: vfio-ap: zeroize the AP queues
- s390: vfio-ap: implement VFIO_DEVICE_RESET ioctl
- KVM: s390: Clear Crypto Control Block when using vSIE
- KVM: s390: vsie: Do the CRYCB validation first
- KVM: s390: vsie: Make use of CRYCB FORMAT2 clear
- KVM: s390: vsie: Allow CRYCB FORMAT-2
- KVM: s390: vsie: allow CRYCB FORMAT-1
- KVM: s390: vsie: allow CRYCB FORMAT-0
- KVM: s390: vsie: allow guest FORMAT-0 CRYCB on host FORMAT-1
- KVM: s390: vsie: allow guest FORMAT-1 CRYCB on host FORMAT-2
- KVM: s390: vsie: allow guest FORMAT-0 CRYCB on host FORMAT-2
- KVM: s390: device attrs to enable/disable AP interpretation
- KVM: s390: CPU model support for AP virtualization
- s390: doc: detailed specifications for AP virtualization
- KVM: s390: fix locking for crypto setting error path
- KVM: s390: Tracing APCB changes
- s390: vfio-ap: setup APCB mask using KVM dedicated function
- [Config:] Enable CONFIG_S390_AP_IOMMU and set CONFIG_VFIO_AP to module.
* Bypass of mount visibility through userns + mount propagation (LP: #1789161)
- mount: Retest MNT_LOCKED in do_umount
- mount: Don't allow copying MNT_UNBINDABLE|MNT_LOCKED mounts
* CVE-2018-18955: nested user namespaces with more than five extents
incorrectly grant privileges over inode (LP: #1801924) // CVE-2018-18955
- userns: also map extents in the reverse map to kernel IDs
* kdump fail due to an IRQ storm (LP: #1797990)
- SAUCE: x86/PCI: Export find_cap() to be used in early PCI code
- SAUCE: x86/quirks: Add parameter to clear MSIs early on boot
- SAUCE: x86/quirks: Scan all busses for early PCI quirks
* crash in ENA driver on removing an interface (LP: #1802341)
- SAUCE: net: ena: fix crash during ena_remove()
* Ubuntu 18.04.1 - [s390x] Kernel panic while stressing network bonding
(LP: #1797367)
- s390/qeth: reduce hard-coded access to ccw channels
- s390/qeth: sanitize strings in debug messages
* Add checksum offload and TSO support for HiNIC adapters (LP: #1800664)
- net-next/hinic: add checksum offload and TSO support
* smartpqi updates for ubuntu 18.04.2 (LP: #1798208)
- scsi: smartpqi: improve handling for sync requests
- scsi: smartpqi: improve error checking for sync requests
- scsi: smartpqi: add inspur advantech ids
- scsi: smartpqi: fix critical ARM issue reading PQI index registers
- scsi: smartpqi: bump driver version to 1.1.4-130
* [GLK/CLX] Enhanced IBRS (LP: #1786139)
- x86/speculation: Remove SPECTRE_V2_IBRS in enum spectre_v2_mitigation
- x86/speculation: Support Enhanced IBRS on future CPUs
* Enable keyboard wakeup for S2Idle laptops (LP: #1798552)
- Input: i8042 - enable keyboard wakeups by default when s2idle is used
* Overlayfs in user namespace leaks directory content of inaccessible
directories (LP: #1793458) // CVE-2018-6559
- SAUCE: overlayfs: ensure mounter privileges when reading directories
* Update ENA driver to version 2.0.1K (LP: #1798182)
- net: ena: remove ndo_poll_controller
- net: ena: fix auto casting to boolean
- net: ena: minor performance improvement
- net: ena: complete host info to match latest ENA spec
- net: ena: introduce Low Latency Queues data structures according to ENA
spec
- net: ena: add functions for handling Low Latency Queues in ena_com
- net: ena: add functions for handling Low Latency Queues in ena_netdev
- net: ena: use CSUM_CHECKED device indication to report skb's checksum
status
- net: ena: explicit casting and initialization, and clearer error handling
- net: ena: limit refill Rx threshold to 256 to avoid latency issues
- net: ena: change rx copybreak default to reduce kernel memory pressure
- net: ena: remove redundant parameter in ena_com_admin_init()
- net: ena: update driver version to 2.0.1
- net: ena: fix indentations in ena_defs for better readability
- net: ena: Fix Kconfig dependency on X86
- net: ena: enable Low Latency Queues
- net: ena: fix compilation error in xtensa architecture
* Cosmic update: 4.18.17 upstream stable release (LP: #1802119)
- xfrm: Validate address prefix lengths in the xfrm selector.
- xfrm6: call kfree_skb when skb is toobig
- xfrm: reset transport header back to network header after all input
transforms ahave been applied
- xfrm: reset crypto_done when iterating over multiple input xfrms
- mac80211: Always report TX status
- cfg80211: reg: Init wiphy_idx in regulatory_hint_core()
- mac80211: fix pending queue hang due to TX_DROP
- cfg80211: Address some corner cases in scan result channel updating
- mac80211: TDLS: fix skb queue/priority assignment
- mac80211: fix TX status reporting for ieee80211s
- ARM: 8799/1: mm: fix pci_ioremap_io() offset check
- xfrm: validate template mode
- drm/i2c: tda9950: fix timeout counter check
- drm/i2c: tda9950: set MAX_RETRIES for errors only
- netfilter: bridge: Don't sabotage nf_hook calls from an l3mdev
- netfilter: conntrack: get rid of double sizeof
- arm64: hugetlb: Fix handling of young ptes
- ARM: dts: BCM63xx: Fix incorrect interrupt specifiers
- net: macb: Clean 64b dma addresses if they are not detected
- soc: fsl: qbman: qman: avoid allocating from non existing gen_pool
- soc: fsl: qe: Fix copy/paste bug in ucc_get_tdm_sync_shift()
- nl80211: Fix possible Spectre-v1 for NL80211_TXRATE_HT
- mac80211_hwsim: fix locking when iterating radios during ns exit
- mac80211_hwsim: fix race in radio destruction from netlink notifier
- mac80211_hwsim: do not omit multicast announce of first added radio
- Bluetooth: SMP: fix crash in unpairing
- pxa168fb: prepare the clock
- qed: Avoid implicit enum conversion in qed_set_tunn_cls_info
- qed: Fix mask parameter in qed_vf_prep_tunn_req_tlv
- qed: Avoid implicit enum conversion in qed_roce_mode_to_flavor
- qed: Avoid constant logical operation warning in qed_vf_pf_acquire
- qed: Avoid implicit enum conversion in qed_iwarp_parse_rx_pkt
- nl80211: Fix possible Spectre-v1 for CQM RSSI thresholds
- scsi: qedi: Initialize the stats mutex lock
- rxrpc: Fix checks as to whether we should set up a new call
- rxrpc: Fix RTT gathering
- rxrpc: Fix transport sockopts to get IPv4 errors on an IPv6 socket
- rxrpc: Fix error distribution
- netfilter: nft_set_rbtree: add missing rb_erase() in GC routine
- netfilter: avoid erronous array bounds warning
- asix: Check for supported Wake-on-LAN modes
- ax88179_178a: Check for supported Wake-on-LAN modes
- lan78xx: Check for supported Wake-on-LAN modes
- sr9800: Check for supported Wake-on-LAN modes
- r8152: Check for supported Wake-on-LAN Modes
- smsc75xx: Check for Wake-on-LAN modes
- smsc95xx: Check for Wake-on-LAN modes
- cfg80211: fix use-after-free in reg_process_hint()
- KVM: nVMX: Do not expose MPX VMX controls when guest MPX disabled
- KVM: x86: Do not use kvm_x86_ops->mpx_supported() directly
- KVM: nVMX: Fix emulation of VM_ENTRY_LOAD_BNDCFGS
- perf/core: Fix perf_pmu_unregister() locking
- perf/x86/intel/uncore: Use boot_cpu_data.phys_proc_id instead of
hardcorded
physical package ID 0
- perf/ring_buffer: Prevent concurent ring buffer access
- perf/x86/intel/uncore: Fix PCI BDF address of M3UPI on SKX
- perf/x86/amd/uncore: Set ThreadMask and SliceMask for L3 Cache perf events
- thunderbolt: Do not handle ICM events after domain is stopped
- thunderbolt: Initialize after IOMMUs
- net: fec: fix rare tx timeout
- declance: Fix continuation with the adapter identification message
- RISCV: Fix end PFN for low memory
- Revert "serial: 8250_dw: Fix runtime PM handling"
- locking/ww_mutex: Fix runtime warning in the WW mutex selftest
- drm/amd/display: Signal hw_done() after waiting for flip_done()
- be2net: don't flip hw_features when VXLANs are added/deleted
- powerpc/numa: Skip onlining a offline node in kdump path
- net: cxgb3_main: fix a missing-check bug
- yam: fix a missing-check bug
- ocfs2: fix crash in ocfs2_duplicate_clusters_by_page()
- mm/gup_benchmark: fix unsigned comparison to zero in __gup_benchmark_ioctl
- mm/migrate.c: split only transparent huge pages when allocation fails
- x86/paravirt: Fix some warning messages
- clk: mvebu: armada-37xx-periph: Remove unused var num_parents
- libertas: call into generic suspend code before turning off power
- perf report: Don't try to map ip to invalid map
- tls: Fix improper revert in zerocopy_from_iter
- HID: i2c-hid: Remove RESEND_REPORT_DESCR quirk and its handling
- compiler.h: Allow arch-specific asm/compiler.h
- ARM: dts: imx53-qsb: disable 1.2GHz OPP
- perf python: Use -Wno-redundant-decls to build with PYTHON=python3
- perf record: Use unmapped IP for inline callchain cursors
- rxrpc: Don't check RXRPC_CALL_TX_LAST after calling
rxrpc_rotate_tx_window()
- rxrpc: Carry call state out of locked section in rxrpc_rotate_tx_window()
- rxrpc: Only take the rwind and mtu values from latest ACK
- rxrpc: Fix connection-level abort handling
- KVM: x86: support CONFIG_KVM_AMD=y with CONFIG_CRYPTO_DEV_CCP_DD=m
- net: ena: fix warning in rmmod caused by double iounmap
- net: ena: fix rare bug when failed restart/resume is followed by driver
removal
- net: ena: fix NULL dereference due to untimely napi initialization
- gpio: Assign gpio_irq_chip::parents to non-stack pointer
- IB/mlx5: Unmap DMA addr from HCA before IOMMU
- rds: RDS (tcp) hangs on sendto() to unresponding address
- selftests: rtnetlink.sh explicitly requires bash.
- selftests: udpgso_bench.sh explicitly requires bash
- vmlinux.lds.h: Fix incomplete .text.exit discards
- vmlinux.lds.h: Fix linker warnings about orphan .LPBX sections
- afs: Fix cell proc list
- fs/fat/fatent.c: add cond_resched() to fat_count_free_clusters()
- Revert "mm: slowly shrink slabs with a relatively small number of objects"
- Revert "netfilter: ipv6: nf_defrag: drop skb dst before queueing"
- perf tools: Disable parallelism for 'make clean'
- bridge: do not add port to router list when receives query with source
0.0.0.0
- ipv6: mcast: fix a use-after-free in inet6_mc_check
- ipv6/ndisc: Preserve IPv6 control buffer if protocol error handlers are
called
- ipv6: rate-limit probes for neighbourless routes
- llc: set SOCK_RCU_FREE in llc_sap_add_socket()
- net: fec: don't dump RX FIFO register when not available
- net/ipv6: Fix index counter for unicast addresses in in6_dump_addrs
- net/mlx5e: fix csum adjustments caused by RXFCS
- net: sched: gred: pass the right attribute to gred_change_table_def()
- net: socket: fix a missing-check bug
- net: stmmac: Fix stmmac_mdio_reset() when building stmmac as modules
- net: udp: fix handling of CHECKSUM_COMPLETE packets
- r8169: fix NAPI handling under high load
- rtnetlink: Disallow FDB configuration for non-Ethernet device
- sctp: fix race on sctp_id2asoc
- tipc: fix unsafe rcu locking when accessing publication list
- udp6: fix encap return code for resubmitting
- vhost: Fix Spectre V1 vulnerability
- virtio_net: avoid using netif_tx_disable() for serializing tx routine
- ethtool: fix a privilege escalation bug
- bonding: fix length of actor system
- ip6_tunnel: Fix encapsulation layout
- openvswitch: Fix push/pop ethernet validation
- net: ipmr: fix unresolved entry dumps
- net/mlx5: Take only bit 24-26 of wqe.pftype_wq for page fault type
- net: bcmgenet: Poll internal PHY for GENETv5
- net: sched: Fix for duplicate class dump
- net/sched: cls_api: add missing validation of netlink attributes
- net/ipv6: Allow onlink routes to have a device mismatch if it is the
default
route
- sctp: fix the data size calculation in sctp_data_size
- sctp: not free the new asoc when sctp_wait_for_connect returns err
- net/mlx5: Fix memory leak when setting fpga ipsec caps
- net/smc: fix smc_buf_unuse to use the lgr pointer
- mlxsw: spectrum_switchdev: Don't ignore deletions of learned MACs
- net: bpfilter: use get_pid_task instead of pid_task
- net: drop skb on failure in ip_check_defrag()
- net: fix pskb_trim_rcsum_slow() with odd trim offset
- mlxsw: core: Fix devlink unregister flow
- sparc64: Export __node_distance.
- sparc64: Make corrupted user stacks more debuggable.
- sparc64: Make proc_id signed.
- sparc64: Set %l4 properly on trap return after handling signals.
- sparc64: Wire up compat getpeername and getsockname.
- sparc: Fix single-pcr perf event counter management.
- sparc: Fix syscall fallback bugs in VDSO.
- sparc: Throttle perf events properly.
- net: bridge: remove ipv6 zero address check in mcast queries
- Linux 4.18.17
* Cosmic update: 4.18.16 upstream stable release (LP: #1802100)
- soundwire: Fix duplicate stream state assignment
- soundwire: Fix incorrect exit after configuring stream
- soundwire: Fix acquiring bus lock twice during master release
- media: af9035: prevent buffer overflow on write
- spi: gpio: Fix copy-and-paste error
- batman-adv: Avoid probe ELP information leak
- batman-adv: Fix segfault when writing to throughput_override
- batman-adv: Fix segfault when writing to sysfs elp_interval
- batman-adv: Prevent duplicated gateway_node entry
- batman-adv: Prevent duplicated nc_node entry
- batman-adv: Prevent duplicated softif_vlan entry
- batman-adv: Prevent duplicated global TT entry
- batman-adv: Prevent duplicated tvlv handler
- batman-adv: fix backbone_gw refcount on queue_work() failure
- batman-adv: fix hardif_neigh refcount on queue_work() failure
- cxgb4: fix abort_req_rss6 struct
- clocksource/drivers/ti-32k: Add CLOCK_SOURCE_SUSPEND_NONSTOP flag for non-
am43 SoCs
- scsi: ibmvscsis: Fix a stringop-overflow warning
- scsi: ibmvscsis: Ensure partition name is properly NUL terminated
- intel_th: pci: Add Ice Lake PCH support
- Input: atakbd - fix Atari keymap
- Input: atakbd - fix Atari CapsLock behaviour
- selftests: pmtu: properly redirect stderr to /dev/null
- net: emac: fix fixed-link setup for the RTL8363SB switch
- ravb: do not write 1 to reserved bits
- net/smc: fix non-blocking connect problem
- net/smc: fix sizeof to int comparison
- qed: Fix populating the invalid stag value in multi function mode.
- qed: Do not add VLAN 0 tag to untagged frames in multi-function mode.
- PCI: dwc: Fix scheduling while atomic issues
- RDMA/uverbs: Fix validity check for modify QP
- scsi: lpfc: Synchronize access to remoteport via rport
- drm: mali-dp: Call drm_crtc_vblank_reset on device init
- scsi: ipr: System hung while dlpar adding primary ipr adapter back
- scsi: sd: don't crash the host on invalid commands
- bpf: sockmap only allow ESTABLISHED sock state
- bpf: sockmap, fix transition through disconnect without close
- bpf: test_maps, only support ESTABLISHED socks
- net/mlx4: Use cpumask_available for eq->affinity_mask
- clocksource/drivers/fttmr010: Fix set_next_event handler
- RDMA/bnxt_re: Fix system crash during RDMA resource initialization
- RISC-V: include linux/ftrace.h in asm-prototypes.h
- iommu/rockchip: Free irqs in shutdown handler
- pinctrl/amd: poll InterruptEnable bits in amd_gpio_irq_set_type
- powerpc/tm: Fix userspace r13 corruption
- powerpc/tm: Avoid possible userspace r1 corruption on reclaim
- powerpc/numa: Use associativity if VPHN hcall is successful
- iommu/amd: Return devid as alias for ACPI HID devices
- x86/boot: Fix kexec booting failure in the SEV bit detection code
- Revert "vfs: fix freeze protection in mnt_want_write_file() for overlayfs"
- mremap: properly flush TLB before releasing the page
- ARC: build: Get rid of toolchain check
- ARC: build: Don't set CROSS_COMPILE in arch's Makefile
- Linux 4.18.16
* Cosmic update: 4.18.15 upstream stable release (LP: #1802082)
- bnxt_en: Fix TX timeout during netpoll.
- bnxt_en: free hwrm resources, if driver probe fails.
- bonding: avoid possible dead-lock
- ip6_tunnel: be careful when accessing the inner header
- ip_tunnel: be careful when accessing the inner header
- ipv4: fix use-after-free in ip_cmsg_recv_dstaddr()
- ipv6: take rcu lock in rawv6_send_hdrinc()
- net: dsa: bcm_sf2: Call setup during switch resume
- net: hns: fix for unmapping problem when SMMU is on
- net: ipv4: update fnhe_pmtu when first hop's MTU changes
- net/ipv6: Display all addresses in output of /proc/net/if_inet6
- netlabel: check for IPV4MASK in addrinfo_get
- net: mvpp2: Extract the correct ethtype from the skb for tx csum offload
- net: mvpp2: fix a txq_done race condition
- net: sched: Add policy validation for tc attributes
- net: sched: cls_u32: fix hnode refcounting
- net: systemport: Fix wake-up interrupt race during resume
- net/usb: cancel pending work when unbinding smsc75xx
- qlcnic: fix Tx descriptor corruption on 82xx devices
- qmi_wwan: Added support for Gemalto's Cinterion ALASxx WWAN interface
- rtnl: limit IFLA_NUM_TX_QUEUES and IFLA_NUM_RX_QUEUES to 4096
- sctp: update dst pmtu with the correct daddr
- team: Forbid enslaving team device to itself
- tipc: fix flow control accounting for implicit connect
- udp: Unbreak modules that rely on external __skb_recv_udp() availability
- net: qualcomm: rmnet: Skip processing loopback packets
- net: qualcomm: rmnet: Fix incorrect allocation flag in transmit
- net: qualcomm: rmnet: Fix incorrect allocation flag in receive path
- tun: remove unused parameters
- tun: initialize napi_mutex unconditionally
- tun: napi flags belong to tfile
- net: stmmac: Fixup the tail addr setting in xmit path
- net/packet: fix packet drop as of virtio gso
- net: dsa: bcm_sf2: Fix unbind ordering
- net/mlx5e: Set vlan masks for all offloaded TC rules
- net: aquantia: memory corruption on jumbo frames
- net/mlx5: E-Switch, Fix out of bound access when setting vport rate
- bonding: pass link-local packets to bonding master also.
- bonding: fix warning message
- net: stmmac: Rework coalesce timer and fix multi-queue races
- nfp: avoid soft lockups under control message storm
- bnxt_en: don't try to offload VLAN 'modify' action
- net-ethtool: ETHTOOL_GUFO did not and should not require CAP_NET_ADMIN
- net: phy: phylink: fix SFP interface autodetection
- sfp: fix oops with ethtool -m
- tcp/dccp: fix lockdep issue when SYN is backlogged
- inet: make sure to grab rcu_read_lock before using ireq->ireq_opt
- net: dsa: b53: Keep CPU port as tagged in all VLANs
- rtnetlink: Fail dump if target netnsid is invalid
- bnxt_en: Fix VNIC reservations on the PF.
- net: ipv4: don't let PMTU updates increase route MTU
- net/mlx5: Check for SQ and not RQ state when modifying hairpin SQ
- bnxt_en: Fix enables field in HWRM_QUEUE_COS2BW_CFG request
- bnxt_en: get the reduced max_irqs by the ones used by RDMA
- net/ipv6: Remove extra call to ip6_convert_metrics for multipath case
- net/ipv6: stop leaking percpu memory in fib6 info
- net: mscc: fix the frame extraction into the skb
- qed: Fix shmem structure inconsistency between driver and the mfw.
- r8169: fix network stalls due to missing bit TXCFG_AUTO_FIFO
- r8169: set RX_MULTI_EN bit in RxConfig for 8168F-family chips
- vxlan: fill ttl inherit info
- ASoC: dapm: Fix NULL pointer deference on CODEC to CODEC DAIs
- ASoC: max98373: Added speaker FS gain cotnrol register to volatile.
- ASoC: rt5514: Fix the issue of the delay volume applied again
- selftests: android: move config up a level
- selftests: kselftest: Remove outdated comment
- ASoC: max98373: Added 10ms sleep after amp software reset
- ASoC: wm8804: Add ACPI support
- ASoC: sigmadsp: safeload should not have lower byte limit
- ASoC: q6routing: initialize data correctly
- selftests: add headers_install to lib.mk
- selftests/efivarfs: add required kernel configs
- selftests: memory-hotplug: add required configs
- ASoC: rsnd: adg: care clock-frequency size
- ASoC: rsnd: don't fallback to PIO mode when -EPROBE_DEFER
- hwmon: (nct6775) Fix access to fan pulse registers
- Fix cg_read_strcmp()
- ASoC: AMD: Ensure reset bit is cleared before configuring
- drm/pl111: Make sure of_device_id tables are NULL terminated
- Bluetooth: SMP: Fix trying to use non-existent local OOB data
- Bluetooth: Use correct tfm to generate OOB data
- Bluetooth: hci_ldisc: Free rw_semaphore on close
- mfd: omap-usb-host: Fix dts probe of children
- KVM: PPC: Book3S HV: Don't use compound_order to determine host mapping
size
- scsi: iscsi: target: Don't use stack buffer for scatterlist
- scsi: qla2xxx: Fix an endian bug in fcpcmd_is_corrupted()
- sound: enable interrupt after dma buffer initialization
- sound: don't call skl_init_chip() to reset intel skl soc
- bpf: btf: Fix end boundary calculation for type section
- bpf: use __GFP_COMP while allocating page
- hwmon: (nct6775) Fix virtual temperature sources for NCT6796D
- hwmon: (nct6775) Fix RPM output for fan7 on NCT6796D
- stmmac: fix valid numbers of unicast filter entries
- hwmon: (nct6775) Use different register to get fan RPM for fan7
- net: ethernet: ti: add missing GENERIC_ALLOCATOR dependency
- net: macb: disable scatter-gather for macb on sama5d3
- ARM: dts: at91: add new compatibility string for macb on sama5d3
- PCI: hv: support reporting serial number as slot information
- clk: x86: add "ether_clk" alias for Bay Trail / Cherry Trail
- clk: x86: Stop marking clocks as CLK_IS_CRITICAL
- pinctrl: cannonlake: Fix gpio base for GPP-E
- x86/kvm/lapic: always disable MMIO interface in x2APIC mode
- drm/amdgpu: Fix SDMA HQD destroy error on gfx_v7
- drm/amdkfd: Change the control stack MTYPE from UC to NC on GFX9
- drm/amdkfd: Fix ATS capablity was not reported correctly on some APUs
- mm: slowly shrink slabs with a relatively small number of objects
- mm/vmstat.c: fix outdated vmstat_text
- afs: Fix afs_server struct leak
- afs: Fix clearance of reply
- MIPS: Fix CONFIG_CMDLINE handling
- MIPS: VDSO: Always map near top of user memory
- mach64: detect the dot clock divider correctly on sparc
- vsprintf: Fix off-by-one bug in bstr_printf() processing dereferenced
pointers
- percpu: stop leaking bitmap metadata blocks
- perf script python: Fix export-to-postgresql.py occasional failure
- perf script python: Fix export-to-sqlite.py sample columns
- s390/cio: Fix how vfio-ccw checks pinned pages
- dm cache: destroy migration_cache if cache target registration failed
- dm: fix report zone remapping to account for partition offset
- dm linear: eliminate linear_end_io call if CONFIG_DM_ZONED disabled
- dm linear: fix linear_end_io conditional definition
- cgroup: Fix dom_cgrp propagation when enabling threaded mode
- Input: xpad - add support for Xbox1 PDP Camo series gamepad
- drm/nouveau/drm/nouveau: Grab runtime PM ref in nv50_mstc_detect()
- mmc: block: avoid multiblock reads for the last sector in SPI mode
- pinctrl: mcp23s08: fix irq and irqchip setup order
- arm64: perf: Reject stand-alone CHAIN events for PMUv3
- mm/mmap.c: don't clobber partially overlapping VMA with
MAP_FIXED_NOREPLACE
- mm/thp: fix call to mmu_notifier in set_pmd_migration_entry() v2
- filesystem-dax: Fix dax_layout_busy_page() livelock
- mm: Preserve _PAGE_DEVMAP across mprotect() calls
- i2c: i2c-scmi: fix for i2c_smbus_write_block_data
- KVM: PPC: Book3S HV: Avoid crash from THP collapse during radix page fault
- Linux 4.18.15
* Cosmic update: 4.18.14 upstream stable release (LP: #1801986)
- perf/core: Add sanity check to deal with pinned event failure
- mm: migration: fix migration of huge PMD shared pages
- mm, thp: fix mlocking THP page with migration enabled
- mm/vmstat.c: skip NR_TLB_REMOTE_FLUSH* properly
- KVM: VMX: check for existence of secondary exec controls before accessing
- blk-mq: I/O and timer unplugs are inverted in blktrace
- pstore/ram: Fix failure-path memory leak in ramoops_init
- clocksource/drivers/timer-atmel-pit: Properly handle error cases
- fbdev/omapfb: fix omapfb_memory_read infoleak
- mmc: core: Fix debounce time to use microseconds
- mmc: slot-gpio: Fix debounce time to use miliseconds again
- mac80211: allocate TXQs for active monitor interfaces
- drm/amdgpu: Fix vce work queue was not cancelled when suspend
- drm: fix use-after-free read in drm_mode_create_lease_ioctl()
- x86/vdso: Fix asm constraints on vDSO syscall fallbacks
- selftests/x86: Add clock_gettime() tests to test_vdso
- x86/vdso: Only enable vDSO retpolines when enabled and supported
- x86/vdso: Fix vDSO syscall fallback asm constraint regression
- Revert "UBUNTU: SAUCE: PCI: Reprogram bridge prefetch registers on resume"
- PCI: Reprogram bridge prefetch registers on resume
- mac80211: fix setting IEEE80211_KEY_FLAG_RX_MGMT for AP mode keys
- PM / core: Clear the direct_complete flag on errors
- dm mpath: fix attached_handler_name leak and dangling hw_handler_name
pointer
- dm cache metadata: ignore hints array being too small during resize
- dm cache: fix resize crash if user doesn't reload cache table
- xhci: Add missing CAS workaround for Intel Sunrise Point xHCI
- usb: xhci-mtk: resume USB3 roothub first
- USB: serial: simple: add Motorola Tetra MTP6550 id
- USB: serial: option: improve Quectel EP06 detection
- USB: serial: option: add two-endpoints device-id flag
- usb: cdc_acm: Do not leak URB buffers
- tty: Drop tty->count on tty_reopen() failure
- of: unittest: Disable interrupt node tests for old world MAC systems
- powerpc: Avoid code patching freed init sections
- powerpc/lib: fix book3s/32 boot failure due to code patching
- ARC: clone syscall to setp r25 as thread pointer
- f2fs: fix invalid memory access
- tipc: call start and done ops directly in __tipc_nl_compat_dumpit()
- ucma: fix a use-after-free in ucma_resolve_ip()
- ubifs: Check for name being NULL while mounting
- rds: rds_ib_recv_alloc_cache() should call alloc_percpu_gfp() instead
- ath10k: fix scan crash due to incorrect length calculation
- Linux 4.18.14
* Cosmic update: 4.18.13 upstream stable release (LP: #1801931)
- rseq/selftests: fix parametrized test with -fpie
- mac80211: Run TXQ teardown code before de-registering interfaces
- mac80211_hwsim: require at least one channel
- Btrfs: fix unexpected failure of nocow buffered writes after snapshotting
when low on space
- KVM: PPC: Book3S HV: Don't truncate HPTE index in xlate function
- cfg80211: remove division by size of sizeof(struct ieee80211_wmm_rule)
- btrfs: btrfs_shrink_device should call commit transaction at the end
- scsi: csiostor: add a check for NULL pointer after kmalloc()
- scsi: csiostor: fix incorrect port capabilities
- scsi: libata: Add missing newline at end of file
- scsi: aacraid: fix a signedness bug
- bpf, sockmap: fix potential use after free in bpf_tcp_close
- bpf, sockmap: fix psock refcount leak in bpf_tcp_recvmsg
- bpf: sockmap, decrement copied count correctly in redirect error case
- mac80211: correct use of IEEE80211_VHT_CAP_RXSTBC_X
- mac80211_hwsim: correct use of IEEE80211_VHT_CAP_RXSTBC_X
- cfg80211: make wmm_rule part of the reg_rule structure
- mac80211_hwsim: Fix possible Spectre-v1 for hwsim_world_regdom_custom
- nl80211: Fix nla_put_u8 to u16 for NL80211_WMMR_TXOP
- nl80211: Pass center frequency in kHz instead of MHz
- bpf: fix several offset tests in bpf_msg_pull_data
- gpio: adp5588: Fix sleep-in-atomic-context bug
- mac80211: mesh: fix HWMP sequence numbering to follow standard
- mac80211: avoid kernel panic when building AMSDU from non-linear SKB
- gpiolib: acpi: Switch to cansleep version of GPIO library call
- gpiolib-acpi: Register GpioInt ACPI event handlers from a late_initcall
- gpio: dwapb: Fix error handling in dwapb_gpio_probe()
- bpf: fix msg->data/data_end after sg shift repair in bpf_msg_pull_data
- bpf: fix shift upon scatterlist ring wrap-around in bpf_msg_pull_data
- bpf: fix sg shift repair start offset in bpf_msg_pull_data
- tipc: switch to rhashtable iterator
- sh_eth: Add R7S9210 support
- net: mvpp2: initialize port of_node pointer
- tc-testing: add test-cases for numeric and invalid control action
- cfg80211: nl80211_update_ft_ies() to validate NL80211_ATTR_IE
- mac80211: do not convert to A-MSDU if frag/subframe limited
- mac80211: always account for A-MSDU header changes
- tools/kvm_stat: fix python3 issues
- tools/kvm_stat: fix handling of invalid paths in debugfs provider
- tools/kvm_stat: fix updates for dead guests
- gpio: Fix crash due to registration race
- ARC: atomics: unbork atomic_fetch_##op()
- Revert "blk-throttle: fix race between blkcg_bio_issue_check() and
cgroup_rmdir()"
- md/raid5-cache: disable reshape completely
- RAID10 BUG_ON in raise_barrier when force is true and conf->barrier is 0
- selftests: pmtu: maximum MTU for vti4 is 2^16-1-20
- selftests: pmtu: detect correct binary to ping ipv6 addresses
- ibmvnic: Include missing return code checks in reset function
- bpf: Fix bpf_msg_pull_data()
- bpf: avoid misuse of psock when TCP_ULP_BPF collides with another ULP
- i2c: uniphier: issue STOP only for last message or I2C_M_STOP
- i2c: uniphier-f: issue STOP only for last message or I2C_M_STOP
- net: cadence: Fix a sleep-in-atomic-context bug in macb_halt_tx()
- fs/cifs: don't translate SFM_SLASH (U+F026) to backslash
- mac80211: fix an off-by-one issue in A-MSDU max_subframe computation
- cfg80211: fix a type issue in ieee80211_chandef_to_operating_class()
- mac80211: fix WMM TXOP calculation
- mac80211: fix a race between restart and CSA flows
- mac80211: Fix station bandwidth setting after channel switch
- mac80211: don't Tx a deauth frame if the AP forbade Tx
- mac80211: shorten the IBSS debug messages
- fsnotify: fix ignore mask logic in fsnotify()
- net/ibm/emac: wrong emac_calc_base call was used by typo
- nds32: fix logic for module
- nds32: add NULL entry to the end of_device_id array
- nds32: Fix empty call trace
- nds32: Fix get_user/put_user macro expand pointer problem
- nds32: fix build error because of wrong semicolon
- tools/vm/slabinfo.c: fix sign-compare warning
- tools/vm/page-types.c: fix "defined but not used" warning
- nds32: linker script: GCOV kernel may refers data in __exit
- ceph: avoid a use-after-free in ceph_destroy_options()
- firmware: arm_scmi: fix divide by zero when sustained_perf_level is zero
- afs: Fix cell specification to permit an empty address list
- mm: madvise(MADV_DODUMP): allow hugetlbfs pages
- bpf: 32-bit RSH verification must truncate input before the ALU op
- netfilter: xt_cluster: add dependency on conntrack module
- netfilter: xt_checksum: ignore gso skbs
- HID: intel-ish-hid: Enable Sunrise Point-H ish driver
- HID: add support for Apple Magic Keyboards
- usb: gadget: fotg210-udc: Fix memory leak of fotg210->ep[i]
- HID: hid-saitek: Add device ID for RAT 7 Contagion
- scsi: iscsi: target: Set conn->sess to NULL when
iscsi_login_set_conn_values
fails
- scsi: iscsi: target: Fix conn_ops double free
- scsi: qedi: Add the CRC size within iSCSI NVM image
- perf annotate: Properly interpret indirect call
- perf evsel: Fix potential null pointer dereference in
perf_evsel__new_idx()
- perf util: Fix bad memory access in trace info.
- perf probe powerpc: Ignore SyS symbols irrespective of endianness
- perf annotate: Fix parsing aarch64 branch instructions after objdump
update
- netfilter: kconfig: nat related expression depend on nftables core
- netfilter: nf_tables: release chain in flushing set
- Revert "iio: temperature: maxim_thermocouple: add MAX31856 part"
- iio: imu: st_lsm6dsx: take into account ts samples in wm configuration
- RDMA/ucma: check fd type in ucma_migrate_id()
- riscv: Do not overwrite initrd_start and initrd_end
- HID: sensor-hub: Restore fixup for Lenovo ThinkPad Helix 2 sensor hub
report
- usb: host: xhci-plat: Iterate over parent nodes for finding quirks
- USB: yurex: Check for truncation in yurex_read()
- nvmet-rdma: fix possible bogus dereference under heavy load
- bnxt_re: Fix couple of memory leaks that could lead to IOMMU call traces
- net/mlx5: Consider PCI domain in search for next dev
- dm raid: fix reshape race on small devices
- drm/nouveau: fix oops in client init failure path
- drm/nouveau/mmu: don't attempt to dereference vmm without valid instance
pointer
- drm/nouveau/TBDdevinit: don't fail when PMU/PRE_OS is missing from VBIOS
- drm/nouveau/disp: fix DP disable race
- drm/nouveau/disp/gm200-: enforce identity-mapped SOR assignment for
LVDS/eDP
panels
- dm raid: fix stripe adding reshape deadlock
- dm raid: fix rebuild of specific devices by updating superblock
- dm raid: fix RAID leg rebuild errors
- r8169: set TxConfig register after TX / RX is enabled, just like RxConfig
- fs/cifs: suppress a string overflow warning
- perf/x86/intel: Add support/quirk for the MISPREDICT bit on Knights
Landing
CPUs
- sched/topology: Set correct NUMA topology type
- dm thin metadata: try to avoid ever aborting transactions
- netfilter: nfnetlink_queue: Solve the NFQUEUE/conntrack clash for
NF_REPEAT
- netfilter: xt_hashlimit: use s->file instead of s->private
- arch/hexagon: fix kernel/dma.c build warning
- hexagon: modify ffs() and fls() to return int
- drm/amdgpu: Fix SDMA hang in prt mode v2
- arm64: jump_label.h: use asm_volatile_goto macro instead of "asm goto"
- drm/amdgpu: fix error handling in amdgpu_cs_user_fence_chunk
- r8169: Clear RTL_FLAG_TASK_*_PENDING when clearing RTL_FLAG_TASK_ENABLED
- s390/qeth: don't dump past end of unknown HW header
- cifs: read overflow in is_valid_oplock_break()
- asm-generic: io: Fix ioport_map() for !CONFIG_GENERIC_IOMAP &&
CONFIG_INDIRECT_PIO
- xen/manage: don't complain about an empty value in control/sysrq node
- xen: avoid crash in disable_hotplug_cpu
- xen: fix GCC warning and remove duplicate EVTCHN_ROW/EVTCHN_COL usage
- x86/APM: Fix build warning when PROC_FS is not enabled
- new primitive: discard_new_inode()
- vfs: don't evict uninitialized inode
- ovl: set I_CREATING on inode being created
- ovl: fix access beyond unterminated strings
- ovl: fix memory leak on unlink of indexed file
- ovl: fix format of setxattr debug
- sysfs: Do not return POSIX ACL xattrs via listxattr
- b43: fix DMA error related regression with proprietary firmware
- firmware: Fix security issue with request_firmware_into_buf()
- firmware: Always initialize the fw_priv list object
- cpufreq: qcom-kryo: Fix section annotations
- smb2: fix missing files in root share directory listing
- iommu/amd: Clear memory encryption mask from physical address
- crypto: qat - Fix KASAN stack-out-of-bounds bug in adf_probe()
- crypto: chelsio - Fix memory corruption in DMA Mapped buffers.
- crypto: mxs-dcp - Fix wait logic on chan threads
- crypto: caam/jr - fix ablkcipher_edesc pointer arithmetic
- gpiolib: Free the last requested descriptor
- Drivers: hv: vmbus: Use get/put_cpu() in vmbus_connect()
- tools: hv: fcopy: set 'error' in case an unknown operation was requested
- proc: restrict kernel stack dumps to root
- ocfs2: fix locking for res->tracking and dlm->tracking_list
- HID: i2c-hid: disable runtime PM operations on hantick touchpad
- ixgbe: check return value of napi_complete_done()
- dm thin metadata: fix __udivdi3 undefined on 32-bit
- Revert "drm/amd/pp: Send khz clock values to DC for smu7/8"
- Linux 4.18.13
* Volume control not working Dell XPS 27 (7760) (LP: #1775068) // Cosmic
update: 4.18.13 upstream stable release (LP: #1801931)
- ALSA: hda/realtek - Cannot adjust speaker's volume on Dell XPS 27 7760
* [Bionic][Cosmic] ipmi: Fix timer race with module unload (LP: #1799281)
- ipmi: Fix timer race with module unload
* [Bionic][Cosmic] Fix to ipmi to support vendor specific messages greater
than 255 bytes (LP: #1799794)
- ipmi:ssif: Add support for multi-part transmit messages > 2 parts
* 18.10 kernel does not appear to validate kernel module signatures correctly
(LP: #1798863) // CVE-2018-18653
- SAUCE: (efi-lockdown) module: remove support for deferring module
signature
verification to IMA
* 18.10 kernel does not appear to validate kernel module signatures correctly
(LP: #1798863)
- SAUCE: (efi-lockdown) module: trust keys from secondary keyring for module
signing
* [Ubuntu] net/af_iucv: fix skb leaks for HiperTransport (LP: #1800639)
- net/af_iucv: drop inbound packets with invalid flags
- net/af_iucv: fix skb handling on HiperTransport xmit error
* Power consumption during s2idle is higher than long idle(sk hynix)
(LP: #1801875)
- SAUCE: pci: prevent sk hynix nvme from entering D3
- SAUCE: nvme: add quirk to not call disable function when suspending
* NULL pointer dereference at 0000000000000020 when access
dst_orig->ops->family in function xfrm_lookup_with_ifid() (LP: #1801878)
- xfrm: Fix NULL pointer dereference when skb_dst_force clears the
dst_entry.
* hns3: map tx ring to tc (LP: #1802023)
- net: hns3: Set tx ring' tc info when netdev is up
* [Ubuntu] qeth: Fix potential array overrun in cmd/rc lookup (LP: #1800641)
- s390: qeth_core_mpc: Use ARRAY_SIZE instead of reimplementing its function
- s390: qeth: Fix potential array overrun in cmd/rc lookup
* Mellanox CX5 stops pinging with rx_wqe_err (mlx5_core) (LP: #1799393)
- net/mlx5: WQ, fixes for fragmented WQ buffers API
* Vulkan applications cause permanent memory leak with Intel GPU
(LP: #1798165)
- drm/syncobj: Don't leak fences when WAIT_FOR_SUBMIT is set
* Packaging resync (LP: #1786013)
- [Package] add support for specifying the primary makefile
linux-azure (4.18.0-1004.4) cosmic; urgency=medium
* linux-azure: 4.18.0-1004.4 -proposed tracker (LP: #1799451)
* Shared folders cannot be mounted in ubuntu/cosmic64 due to missing vbox
modules (LP: #1796647)
- [Config] azure: CONFIG_VBOXGUEST=n
* netfilter: nf_conntrack: resolve clash for matching conntracks
(LP: #1795493)
- SAUCE: netfilter: nf_conntrack: resolve clash for matching conntracks
- SAUCE: netfilter: nf_nat: return the same reply tuple for matching CTs
* [Hyper-V] Enable NVME devices (LP: #1793386)
- [Config] linux-azure: CONFIG_BLK_DEV_NVME=y
* linux-azure: fix getabis information (LP: #1793462)
- [Packaging] linux-azure: fix getabis
[ Ubuntu: 4.18.0-11.12 ]
* linux: 4.18.0-11.12 -proposed tracker (LP: #1799445)
* arm64: snapdragon: WARNING: CPU: 0 PID: 1 arch/arm64/kernel/setup.c:271
reserve_memblock_reserved_regions (LP: #1797139)
- SAUCE: arm64: Fix /proc/iomem for reserved but not memory regions
* arm64: snapdragon: WARNING: CPU: 0 PID: 1 at drivers/irqchip/irq-gic.c:1016
gic_irq_domain_translate (LP: #1797143)
- SAUCE: arm64: dts: msm8916: camms: fix gic_irq_domain_translate warnings
* The front MIC can't work on the Lenovo M715 (LP: #1797292)
- ALSA: hda/realtek - Fix the problem of the front MIC on the Lenovo M715
* Provide mode where all vCPUs on a core must be the same VM (LP: #1792957)
- KVM: PPC: Book3S HV: Provide mode where all vCPUs on a core must be the
same
VM
* fscache: bad refcounting in fscache_op_complete leads to OOPS (LP: #1797314)
- SAUCE: fscache: Fix race in decrementing refcount of op->npages
* hns3: autoneg settings get lost on down/up (LP: #1797654)
- net: hns3: Fix for information of phydev lost problem when down/up
* not able to unwind the stack from within __kernel_clock_gettime in the Linux
vDSO (LP: #1797963)
- powerpc/vdso: Correct call frame information
* Signal 7 error when running GPFS tracing in cluster (LP: #1792195)
- powerpc/mm/books3s: Add new pte bit to mark pte temporarily invalid.
- powerpc/mm/radix: Only need the Nest MMU workaround for R -> RW transition
* Support Edge Gateway's WIFI LED (LP: #1798330)
- SAUCE: mwifiex: Switch WiFi LED state according to the device status
* Support Edge Gateway's Bluetooth LED (LP: #1798332)
- SAUCE: Bluetooth: Support for LED on Edge Gateways
* kvm doesn't work on 36 physical bits systems (LP: #1798427)
- KVM: x86: fix L1TF's MMIO GFN calculation
* CVE-2018-15471
- xen-netback: fix input validation in xenvif_set_hash_mapping()
* regression in 'ip --family bridge neigh' since linux v4.12 (LP: #1796748)
- rtnetlink: fix rtnl_fdb_dump() for ndmsg header
[ Ubuntu: 4.18.0-10.11 ]
* linux: 4.18.0-10.11 -proposed tracker (LP: #1797379)
* the machine of lenovo M715 with the AMD GPU (Radeon Vega 8 Mobile, rev ca,
1002:15dd) often hangs randomly (LP: #1796789)
- drm/amd: Add missing fields in atom_integrated_system_info_v1_11
* Miscellaneous Ubuntu changes
- [Config] CONFIG_VBOXGUEST=n
- ubuntu: vbox -- update to 5.2.18-dfsg-2
- ubuntu: enable vbox build
-- Marcelo Henrique Cerri <marcelo.ce...@canonical.com> Mon, 26 Nov
2018 21:48:04 +0000
** Changed in: linux-azure (Ubuntu Cosmic)
Status: New => Fix Released
** CVE added: https://cve.mitre.org/cgi-bin/cvename.cgi?name=2018-15471
** CVE added: https://cve.mitre.org/cgi-bin/cvename.cgi?name=2018-18653
** CVE added: https://cve.mitre.org/cgi-bin/cvename.cgi?name=2018-18955
** CVE added: https://cve.mitre.org/cgi-bin/cvename.cgi?name=2018-6559
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux-azure in Ubuntu.
https://bugs.launchpad.net/bugs/1802358
Title:
[Hyper-V] Fix IRQ spreading on NVMe devices with lower numbers of
channels
Status in linux-azure package in Ubuntu:
Fix Released
Status in linux-azure source package in Bionic:
Fix Released
Status in linux-azure source package in Cosmic:
Fix Released
Bug description:
1. Patch to kernel/irq/affinity.c:
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-
next.git/commit/kernel/irq/affinity.c?h=next-20181108&id=b82592199032bf7c778f861b936287e37ebc9f62.
2. Patches to kernel/irq/matrix.c. There are three patches for this. The
first two from Fujitsu, and then there is the patch from Long that actually
makes the previous two work correctly.
a.
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/kernel/irq/matrix.c?h=next-20181108&id=8ffe4e61c06a48324cfd97f1199bb9838acce2f2
b.
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/kernel/irq/matrix.c?h=next-20181108&id=76f99ae5b54d48430d1f0c5512a84da0ff9761e0
c.
https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git/commit/?h=irq/core&id=e8da8794a7fd9eef1ec9a07f0d4897c68581c72b
We expect the tip patch to be applied to Linux-next soon.
genirq/affinity: Spread IRQs to all available NUMA nodes
If the number of NUMA nodes exceeds the number of MSI/MSI-X interrupts
which are allocated for a device, the interrupt affinity spreading code
fails to spread them across all nodes.
The reason is, that the spreading code starts from node 0 and continues up
to the number of interrupts requested for allocation. This leaves the nodes
past the last interrupt unused.
This results in interrupt concentration on the first nodes which violates
the assumption of the block layer that all nodes are covered evenly. As a
consequence the NUMA nodes above the number of interrupts are all assigned
to hardware queue 0 and therefore NUMA node 0, which results in bad
performance and has CPU hotplug implications, because queue 0 gets shut
down when the last CPU of node 0 is offlined.
Go over all NUMA nodes and assign them round-robin to all requested
interrupts to solve this.
irq/matrix: Split out the CPU selection code into a helper
Linux finds the CPU which has the lowest vector allocation count to spread
out the non managed interrupts across the possible target CPUs, but does
not do so for managed interrupts.
Split out the CPU selection code into a helper function for reuse. No
functional change.
irq/matrix: Spread managed interrupts on allocation
Linux spreads out the non managed interrupt across the possible target CPUs
to avoid vector space exhaustion.
Managed interrupts are treated differently, as for them the vectors are
reserved (with guarantee) when the interrupt descriptors are initialized.
When the interrupt is requested a real vector is assigned. The assignment
logic uses the first CPU in the affinity mask for assignment. If the
interrupt has more than one CPU in the affinity mask, which happens when a
multi queue device has less queues than CPUs, then doing the same search as
for non managed interrupts makes sense as it puts the interrupt on the
least interrupt plagued CPU. For single CPU affine vectors that's obviously
a NOOP.
Restructre the matrix allocation code so it does the 'best CPU' search, add
the sanity check for an empty affinity mask and adapt the call site in the
x86 vector management code.
genirq/matrix: Improve target CPU selection for managed interrupts.irq/core
On large systems with multiple devices of the same class (e.g. NVMe disks,
using managed interrupts), the kernel can affinitize these interrupts to a
small subset of CPUs instead of spreading them out evenly.
irq_matrix_alloc_managed() tries to select the CPU in the supplied cpumask
of possible target CPUs which has the lowest number of interrupt vectors
allocated.
This is done by searching the CPU with the highest number of available
vectors. While this is correct for non-managed CPUs it can select the wrong
CPU for managed interrupts. Under certain constellations this results in
affinitizing the managed interrupts of several devices to a single CPU in
a set.
The book keeping of available vectors works the following way:
1) Non-managed interrupts:
available is decremented when the interrupt is actually requested by
the device driver and a vector is assigned. It's incremented when the
interrupt and the vector are freed.
2) Managed interrupts:
Managed interrupts guarantee vector reservation when the MSI/MSI-X
functionality of a device is enabled, which is achieved by reserving
vectors in the bitmaps of the possible target CPUs. This reservation
decrements the available count on each possible target CPU.
When the interrupt is requested by the device driver then a vector is
allocated from the reserved region. The operation is reversed when the
interrupt is freed by the device driver. Neither of these operations
affect the available count.
The reservation persist up to the point where the MSI/MSI-X
functionality is disabled and only this operation increments the
available count again.
For non-managed interrupts the available count is the correct selection
criterion because the guaranteed reservations need to be taken into
account. Using the allocated counter could lead to a failing allocation in
the following situation (total vector space of 10 assumed):
CPU0 CPU1
available: 2 0
allocated: 5 3 <--- CPU1 is selected, but available space = 0
managed reserved: 3 7
while available yields the correct result.
For managed interrupts the available count is not the appropriate
selection criterion because as explained above the available count is not
affected by the actual vector allocation.
The following example illustrates that. Total vector space of 10
assumed. The starting point is:
CPU0 CPU1
available: 5 4
allocated: 2 3
managed reserved: 3 3
Allocating vectors for three non-managed interrupts will result in
affinitizing the first two to CPU0 and the third one to CPU1 because the
available count is adjusted with each allocation:
CPU0 CPU1
available: 5 4 <- Select CPU0 for 1st allocation
--> allocated: 3 3
available: 4 4 <- Select CPU0 for 2nd allocation
--> allocated: 4 3
available: 3 4 <- Select CPU1 for 3rd allocation
--> allocated: 4 4
But the allocation of three managed interrupts starting from the same
point will affinitize all of them to CPU0 because the available count is
not affected by the allocation (see above). So the end result is:
CPU0 CPU1
available: 5 4
allocated: 5 3
Introduce a "managed_allocated" field in struct cpumap to track the vector
allocation for managed interrupts separately. Use this information to
select the target CPU when a vector is allocated for a managed interrupt,
which results in more evenly distributed vector assignments. The above
example results in the following allocations:
CPU0 CPU1
managed_allocated: 0 0 <- Select CPU0 for 1st allocation
--> allocated: 3 3
managed_allocated: 1 0 <- Select CPU1 for 2nd allocation
--> allocated: 3 4
managed_allocated: 1 1 <- Select CPU0 for 3rd allocation
--> allocated: 4 4
The allocation of non-managed interrupts is not affected by this change and
is still evaluating the available count.
The overall distribution of interrupt vectors for both types of interrupts
might still not be perfectly even depending on the number of non-managed
and managed interrupts in a system, but due to the reservation guarantee
for managed interrupts this cannot be avoided.
Expose the new field in debugfs as well.
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux-azure/+bug/1802358/+subscriptions
--
Mailing list: https://launchpad.net/~kernel-packages
Post to : kernel-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~kernel-packages
More help : https://help.launchpad.net/ListHelp
