linux-2.6-microblaze.git
6 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 14 Aug 2018 00:01:03 +0000 (17:01 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull misc x86 fixes from Thomas Gleixner:
 "Two fixes for x86:

   - Provide a declaration for native_save_fl() which unbreaks the
     wreckage caused by making it 'extern inline'.

   - Fix the failing paravirt patching which is supposed to replace
     indirect with direct calls. The wreckage is caused by an incorrect
     clobber test"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/paravirt: Fix spectre-v2 mitigations for paravirt guests
  x86/irqflags: Provide a declaration for native_save_fl

6 years agokconfig: remove unused sym_get_env_prop() function
Masahiro Yamada [Mon, 13 Aug 2018 16:48:38 +0000 (01:48 +0900)]
kconfig: remove unused sym_get_env_prop() function

This function is unused since commit 104daea149c4 ("kconfig: reference
environment variables directly and remove 'option env='").

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
6 years agococcicheck: return proper error code on fail
Denis Efremov [Fri, 10 Aug 2018 20:25:55 +0000 (23:25 +0300)]
coccicheck: return proper error code on fail

If coccicheck fails, it should return an error code distinct from zero
to signal about an internal problem. Current code instead of exiting with
the tool's error code returns the error code of 'echo "coccicheck failed"'
which is almost always equals to zero, thus failing the original intention
of alerting about a problem. This patch fixes the code.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Denis Efremov <efremov@linux.com>
Acked-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
6 years agoCoccinelle: doubletest: reduce side effect false positives
Julia Lawall [Fri, 3 Aug 2018 16:32:11 +0000 (18:32 +0200)]
Coccinelle: doubletest: reduce side effect false positives

Ensure that the cited expression is not a function call or an
assignment to reduce the chance of false positives.

Slightly modify the warning message to indicate another source
of false positves.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
6 years agoMerge branch 'x86-mm-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 23:29:35 +0000 (16:29 -0700)]
Merge branch 'x86-mm-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 mm updates from Thomas Gleixner:

 - Make lazy TLB mode even lazier to avoid pointless switch_mm()
   operations, which reduces CPU load by 1-2% for memcache workloads

 - Small cleanups and improvements all over the place

* 'x86-mm-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/mm: Remove redundant check for kmem_cache_create()
  arm/asm/tlb.h: Fix build error implicit func declaration
  x86/mm/tlb: Make clear_asid_other() static
  x86/mm/tlb: Skip atomic operations for 'init_mm' in switch_mm_irqs_off()
  x86/mm/tlb: Always use lazy TLB mode
  x86/mm/tlb: Only send page table free TLB flush to lazy TLB CPUs
  x86/mm/tlb: Make lazy TLB mode lazier
  x86/mm/tlb: Restructure switch_mm_irqs_off()
  x86/mm/tlb: Leave lazy TLB mode at page table free time
  mm: Allocate the mm_cpumask (mm->cpu_bitmap[]) dynamically based on nr_cpu_ids
  x86/mm: Add TLB purge to free pmd/pte page interfaces
  ioremap: Update pgtable free interfaces with addr
  x86/mm: Disable ioremap free page handling on x86-PAE

6 years agoMerge branch 'x86-platform-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 Aug 2018 23:08:26 +0000 (16:08 -0700)]
Merge branch 'x86-platform-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 platform updates from Thomas Gleixner:
 "Trivial cleanups and improvements"

* 'x86-platform-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/platform/UV: Remove redundant check of p == q
  x86/platform/olpc: Use PTR_ERR_OR_ZERO()
  x86/platform/UV: Mark memblock related init code and data correctly

6 years agoMerge tag 'mlx5e-updates-2018-08-10' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Mon, 13 Aug 2018 23:04:18 +0000 (16:04 -0700)]
Merge tag 'mlx5e-updates-2018-08-10' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5e-updates-2018-08-10

This series provides the following updates to mlx5e netdevice driver.

1) First 4 patches extends the support for ethtool rxnfc flow steering
   - Added ipv6 support
   - l4 proto ip field for both ip6 and ip4

2) Next 4 patches, reorganizing flow steering structures and declaration into
one header file, and add two Kconfig flags to allow disabling/enabling mlx5
netdevice rx flow steering at compile time:
CONFIG_MLX5_EN_ARFS for en_arfs.c
CONFIG_MLX5_EN_RXNFC for en_fs_ehtool.c

3) More kconfig flags dependencies
- vxlan.c depends on CONFIG_VXLAN
- clock.c depends on CONFIG_PTP_1588_CLOCK

4) Reorganize the Makefile
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'x86-cache-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 23:01:46 +0000 (16:01 -0700)]
Merge branch 'x86-cache-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 cache QoS (RDT/CAR) updates from Thomas Gleixner:
 "Add support for pseudo-locked cache regions.

  Cache Allocation Technology (CAT) allows on certain CPUs to isolate a
  region of cache and 'lock' it. Cache pseudo-locking builds on the fact
  that a CPU can still read and write data pre-allocated outside its
  current allocated area on cache hit. With cache pseudo-locking data
  can be preloaded into a reserved portion of cache that no application
  can fill, and from that point on will only serve cache hits. The cache
  pseudo-locked memory is made accessible to user space where an
  application can map it into its virtual address space and thus have a
  region of memory with reduced average read latency.

  The locking is not perfect and gets totally screwed by WBINDV and
  similar mechanisms, but it provides a reasonable enhancement for
  certain types of latency sensitive applications.

  The implementation extends the current CAT mechanism and provides a
  generally useful exclusive CAT mode on which it builds the extra
  pseude-locked regions"

* 'x86-cache-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (45 commits)
  x86/intel_rdt: Disable PMU access
  x86/intel_rdt: Fix possible circular lock dependency
  x86/intel_rdt: Make CPU information accessible for pseudo-locked regions
  x86/intel_rdt: Support restoration of subset of permissions
  x86/intel_rdt: Fix cleanup of plr structure on error
  x86/intel_rdt: Move pseudo_lock_region_clear()
  x86/intel_rdt: Limit C-states dynamically when pseudo-locking active
  x86/intel_rdt: Support L3 cache performance event of Broadwell
  x86/intel_rdt: More precise L2 hit/miss measurements
  x86/intel_rdt: Create character device exposing pseudo-locked region
  x86/intel_rdt: Create debugfs files for pseudo-locking testing
  x86/intel_rdt: Create resctrl debug area
  x86/intel_rdt: Ensure RDT cleanup on exit
  x86/intel_rdt: Resctrl files reflect pseudo-locked information
  x86/intel_rdt: Support creation/removal of pseudo-locked region
  x86/intel_rdt: Pseudo-lock region creation/removal core
  x86/intel_rdt: Discover supported platforms via prefetch disable bits
  x86/intel_rdt: Add utilities to test pseudo-locked region possibility
  x86/intel_rdt: Split resource group removal in two
  x86/intel_rdt: Enable entering of pseudo-locksetup mode
  ...

6 years agoMerge branch 'x86-hyperv-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 Aug 2018 22:49:04 +0000 (15:49 -0700)]
Merge branch 'x86-hyperv-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86/hyper-v update from Thomas Gleixner:
 "Add fast hypercall support for guest running on the Microsoft HyperV(isor)"

* 'x86-hyperv-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/hyper-v: Fix wrong merge conflict resolution
  x86/hyper-v: Check for VP_INVAL in hyperv_flush_tlb_others()
  x86/hyper-v: Check cpumask_to_vpset() return value in hyperv_flush_tlb_others_ex()
  x86/hyper-v: Trace PV IPI send
  x86/hyper-v: Use cheaper HVCALL_SEND_IPI hypercall when possible
  x86/hyper-v: Use 'fast' hypercall for HVCALL_SEND_IPI
  x86/hyper-v: Implement hv_do_fast_hypercall16
  x86/hyper-v: Use cheaper HVCALL_FLUSH_VIRTUAL_ADDRESS_{LIST,SPACE} hypercalls when possible

6 years agoMerge branch 'x86-debug-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 22:21:12 +0000 (15:21 -0700)]
Merge branch 'x86-debug-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 dump printing cleanup from Thomas Gleixner:
 "Clean up the show_opcodes() printout so nested dumps can be properly
  differentiated"

* 'x86-debug-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86: Avoid pr_cont() in show_opcodes()

6 years agoMerge branch 'x86-cpu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 21:41:53 +0000 (14:41 -0700)]
Merge branch 'x86-cpu-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 cpu updates from Thomas Gleixner:
 "Two small updates for the CPU code:

   - Improve NUMA emulation

   - Add the EPT_AD CPU feature bit"

* 'x86-cpu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/cpufeatures: Add EPT_AD feature bit
  x86/numa_emulation: Introduce uniform split capability
  x86/numa_emulation: Fix emulated-to-physical node mapping

6 years agoMerge branch 'x86-cleanups-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 Aug 2018 21:13:53 +0000 (14:13 -0700)]
Merge branch 'x86-cleanups-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 cleanups from Thomas Gleixner:
 "Trival cleanups"

* 'x86-cleanups-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/iommu: Use NULL instead of 0
  x86/platform/pcspeaker: Use PTR_ERR_OR_ZERO() to fix ptr_ret.cocci warning

6 years agoMerge branch 'x86-build-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 21:12:24 +0000 (14:12 -0700)]
Merge branch 'x86-build-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 build cleanup from Thomas Gleixner:
 "Remove a stale quirk for a no longer supported GCC version"

* 'x86-build-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/build: Remove old -funit-at-a-time GCC quirk

6 years agoMerge branch 'x86-asm-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 20:35:26 +0000 (13:35 -0700)]
Merge branch 'x86-asm-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 asm updates from Thomas Gleixner:
 "The lowlevel and ASM code updates for x86:

   - Make stack trace unwinding more reliable

   - ASM instruction updates for better code generation

   - Various cleanups"

* 'x86-asm-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/entry/64: Add two more instruction suffixes
  x86/asm/64: Use 32-bit XOR to zero registers
  x86/build/vdso: Simplify 'cmd_vdso2c'
  x86/build/vdso: Remove unused vdso-syms.lds
  x86/stacktrace: Enable HAVE_RELIABLE_STACKTRACE for the ORC unwinder
  x86/unwind/orc: Detect the end of the stack
  x86/stacktrace: Do not fail for ORC with regs on stack
  x86/stacktrace: Clarify the reliable success paths
  x86/stacktrace: Remove STACKTRACE_DUMP_ONCE
  x86/stacktrace: Do not unwind after user regs
  x86/asm: Use CC_SET/CC_OUT in percpu_cmpxchg8b_double() to micro-optimize code generation

6 years agoMerge branch 'x86-boot-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 20:32:42 +0000 (13:32 -0700)]
Merge branch 'x86-boot-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 boot updates from Thomas Gleixner:
 "Boot code updates for x86:

   - Allow to skip a given amount of huge pages for address layout
     randomization on the kernel command line to prevent regressions in
     the huge page allocation with small memory sizes

   - Various cleanups"

* 'x86-boot-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/boot: Use CC_SET()/CC_OUT() instead of open coding it
  x86/boot/KASLR: Make local variable mem_limit static
  x86/boot/KASLR: Skip specified number of 1GB huge pages when doing physical randomization (KASLR)
  x86/boot/KASLR: Add two new functions for 1GB huge pages handling

6 years agoMerge branch 'x86-apic-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 20:31:08 +0000 (13:31 -0700)]
Merge branch 'x86-apic-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 apic update from Thomas Gleixner:
 "Trivial cleanups of the APIC related code"

* 'x86-apic-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/apic: Trivial coding style fixes
  x86/vector: Merge allocate_vector() into assign_vector_locked()

6 years agoMerge branch 'timers-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 Aug 2018 20:02:31 +0000 (13:02 -0700)]
Merge branch 'timers-core-for-linus' of git://git./linux/kernel/git/tip/tip

Pull timer updates from Thomas Gleixner:
 "The timers departement more or less proudly presents:

   - More Y2038 timekeeping work mostly in the core code. The work is
     slowly, but steadily targeting the actuall syscalls.

   - Enhanced timekeeping suspend/resume support by utilizing
     clocksources which do not stop during suspend, but are otherwise
     not the main timekeeping clocksources.

   - Make NTP adjustmets more accurate and immediate when the frequency
     is set directly and not incrementally.

   - Sanitize the overrung handing of posix timers

   - A new timer driver for Mediatek SoCs

   - The usual pile of fixes and updates all over the place"

* 'timers-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (32 commits)
  clockevents: Warn if cpu_all_mask is used as cpumask
  tick/broadcast-hrtimer: Use cpu_possible_mask for ce_broadcast_hrtimer
  clocksource/drivers/arm_arch_timer: Fix bogus cpu_all_mask usage
  clocksource: ti-32k: Remove CLOCK_SOURCE_SUSPEND_NONSTOP flag
  timers: Clear timer_base::must_forward_clk with timer_base::lock held
  clocksource/drivers/sprd: Register one always-on timer to compensate suspend time
  clocksource/drivers/timer-mediatek: Add support for system timer
  clocksource/drivers/timer-mediatek: Convert the driver to timer-of
  clocksource/drivers/timer-mediatek: Use specific prefix for GPT
  clocksource/drivers/timer-mediatek: Rename mtk_timer to timer-mediatek
  clocksource/drivers/timer-mediatek: Add system timer bindings
  clocksource/drivers: Set clockevent device cpumask to cpu_possible_mask
  time: Introduce one suspend clocksource to compensate the suspend time
  time: Fix extra sleeptime injection when suspend fails
  timekeeping/ntp: Constify some function arguments
  ntp: Use kstrtos64 for s64 variable
  ntp: Remove redundant arguments
  timer: Fix coding style
  ktime: Provide typesafe ktime_to_ns()
  hrtimer: Improve kernel message printing
  ...

6 years agoMerge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 19:55:49 +0000 (12:55 -0700)]
Merge branch 'perf-core-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf update from Thomas Gleixner:
 "The perf crowd presents:

  Kernel updates:

   - Removal of jprobes

   - Cleanup and consolidatation the handling of kprobes

   - Cleanup and consolidation of hardware breakpoints

   - The usual pile of fixes and updates to PMUs and event descriptors

  Tooling updates:

   - Updates and improvements all over the place. Nothing outstanding,
     just the (good) boring incremental grump work"

* 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (103 commits)
  perf trace: Do not require --no-syscalls to suppress strace like output
  perf bpf: Include uapi/linux/bpf.h from the 'perf trace' script's bpf.h
  perf tools: Allow overriding MAX_NR_CPUS at compile time
  perf bpf: Show better message when failing to load an object
  perf list: Unify metric group description format with PMU event description
  perf vendor events arm64: Update ThunderX2 implementation defined pmu core events
  perf cs-etm: Generate branch sample for CS_ETM_TRACE_ON packet
  perf cs-etm: Generate branch sample when receiving a CS_ETM_TRACE_ON packet
  perf cs-etm: Support dummy address value for CS_ETM_TRACE_ON packet
  perf cs-etm: Fix start tracing packet handling
  perf build: Fix installation directory for eBPF
  perf c2c report: Fix crash for empty browser
  perf tests: Fix indexing when invoking subtests
  perf trace: Beautify the AF_INET & AF_INET6 'socket' syscall 'protocol' args
  perf trace beauty: Add beautifiers for 'socket''s 'protocol' arg
  perf trace beauty: Do not print NULL strarray entries
  perf beauty: Add a generator for IPPROTO_ socket's protocol constants
  tools include uapi: Grab a copy of linux/in.h
  perf tests: Fix complex event name parsing
  perf evlist: Fix error out while applying initial delay and LBR
  ...

6 years agonet/mlx5: Improve argument name for add flow API
Eli Cohen [Sun, 29 Apr 2018 14:17:34 +0000 (09:17 -0500)]
net/mlx5: Improve argument name for add flow API

The last argument to mlx5_add_flow_rules passes the number of
destinations in the struct pointed to by the dest arg. Change the name
to better reflect this fact.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
6 years agonet/mlx5: Reorganize the makefile
Saeed Mahameed [Tue, 31 Jul 2018 21:44:00 +0000 (14:44 -0700)]
net/mlx5: Reorganize the makefile

Reorganize the Makefile and group files together according to their
functionality and importance.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
6 years agonet/mlx5e: clock.c depends on CONFIG_PTP_1588_CLOCK
Moshe Shemesh [Sun, 29 Jul 2018 10:29:45 +0000 (13:29 +0300)]
net/mlx5e: clock.c depends on CONFIG_PTP_1588_CLOCK

lib/clock.c includes clock related functions which require ptp support.
Thus compile out lib/clock.c and add the needed function stubs in case
kconfig CONFIG_PTP_1588_CLOCK is off.

Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
6 years agonet/mlx5e: vxlan.c depends on CONFIG_VXLAN
Saeed Mahameed [Fri, 13 Jul 2018 17:58:49 +0000 (10:58 -0700)]
net/mlx5e: vxlan.c depends on CONFIG_VXLAN

When vxlan is not enabled by kernel, no need to enable it in mlx5.
Compile out lib/vxlan.c if CONFIG_VXLAN is not selected.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
6 years agonet/mlx5e: Move flow steering declarations into en/fs.h
Saeed Mahameed [Thu, 12 Jul 2018 10:09:19 +0000 (03:09 -0700)]
net/mlx5e: Move flow steering declarations into en/fs.h

Move flow steering declarations and definitions into the dedicated
en/fs.h header file

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
6 years agonet/mlx5e: Add CONFIG_MLX5_EN_ARFS for accelerated flow steering support
Saeed Mahameed [Thu, 12 Jul 2018 10:01:26 +0000 (03:01 -0700)]
net/mlx5e: Add CONFIG_MLX5_EN_ARFS for accelerated flow steering support

Add new mlx5 Kconfig flag to allow selecting accelerated flow steering
support, and compile out en_arfs.c if not selected.

Move arfs declarations and definitions to en/fs.h header file.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
6 years agonet/mlx5e: Add CONFIG_MLX5_EN_RXNFC for ethtool rx nfc
Saeed Mahameed [Wed, 11 Jul 2018 19:02:42 +0000 (12:02 -0700)]
net/mlx5e: Add CONFIG_MLX5_EN_RXNFC for ethtool rx nfc

Add new mlx5 Kconfig flag to allow selecting ethtool rx nfc support,
and compile out en_fs_ehtool.c if not selected.

Add en/fs.h header file to host all steering declarations and
definitions.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
6 years agonet/mlx5e: Ethtool steering, move ethtool callbacks
Saeed Mahameed [Wed, 11 Jul 2018 00:04:49 +0000 (17:04 -0700)]
net/mlx5e: Ethtool steering, move ethtool callbacks

Move ethool rxnfc callback into en_fs_etthool file where they belong.
This will allow us to make many ethtool fs related helper functions
static.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
6 years agonet/mlx5e: Ethtool steering, l4 proto support
Saeed Mahameed [Sun, 8 Jul 2018 07:24:21 +0000 (00:24 -0700)]
net/mlx5e: Ethtool steering, l4 proto support

Add support for l4 proto ip field in ethtool flow steering.

Example: Redirect icmpv6 to rx queue #2

ethtool -U eth0 flow-type ip6 l4proto 58 action 2

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
6 years agonet/mlx5e: Ethtool steering, ip6 support
Saeed Mahameed [Fri, 6 Jul 2018 00:11:09 +0000 (17:11 -0700)]
net/mlx5e: Ethtool steering, ip6 support

Add ip6 support for ethtool flow steering.

New supported flow types: ip6|tcp6|udp6|
Supported fields: src-ip|dst-ip|src-port|dst-port

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
6 years agoMerge branch 'locking-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 Aug 2018 19:23:39 +0000 (12:23 -0700)]
Merge branch 'locking-core-for-linus' of git://git./linux/kernel/git/tip/tip

Pull locking/atomics update from Thomas Gleixner:
 "The locking, atomics and memory model brains delivered:

   - A larger update to the atomics code which reworks the ordering
     barriers, consolidates the atomic primitives, provides the new
     atomic64_fetch_add_unless() primitive and cleans up the include
     hell.

   - Simplify cmpxchg() instrumentation and add instrumentation for
     xchg() and cmpxchg_double().

   - Updates to the memory model and documentation"

* 'locking-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (48 commits)
  locking/atomics: Rework ordering barriers
  locking/atomics: Instrument cmpxchg_double*()
  locking/atomics: Instrument xchg()
  locking/atomics: Simplify cmpxchg() instrumentation
  locking/atomics/x86: Reduce arch_cmpxchg64*() instrumentation
  tools/memory-model: Rename litmus tests to comply to norm7
  tools/memory-model/Documentation: Fix typo, smb->smp
  sched/Documentation: Update wake_up() & co. memory-barrier guarantees
  locking/spinlock, sched/core: Clarify requirements for smp_mb__after_spinlock()
  sched/core: Use smp_mb() in wake_woken_function()
  tools/memory-model: Add informal LKMM documentation to MAINTAINERS
  locking/atomics/Documentation: Describe atomic_set() as a write operation
  tools/memory-model: Make scripts executable
  tools/memory-model: Remove ACCESS_ONCE() from model
  tools/memory-model: Remove ACCESS_ONCE() from recipes
  locking/memory-barriers.txt/kokr: Update Korean translation to fix broken DMA vs. MMIO ordering example
  MAINTAINERS: Add Daniel Lustig as an LKMM reviewer
  tools/memory-model: Fix ISA2+pooncelock+pooncelock+pombonce name
  tools/memory-model: Add litmus test for full multicopy atomicity
  locking/refcount: Always allow checked forms
  ...

6 years agoMerge branch 'smp-hotplug-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 Aug 2018 19:21:46 +0000 (12:21 -0700)]
Merge branch 'smp-hotplug-for-linus' of git://git./linux/kernel/git/tip/tip

Pull CPU hotplug update from Thomas Gleixner:
 "A trivial name fix for the hotplug state machine"

* 'smp-hotplug-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  cpu/hotplug: Clarify CPU hotplug step name for timers

6 years agonet/mlx5e: Ethtool steering flow parsing refactoring
Saeed Mahameed [Fri, 6 Jul 2018 00:31:14 +0000 (17:31 -0700)]
net/mlx5e: Ethtool steering flow parsing refactoring

Have a parsing function per flow type, that converts from ethtool rx flow
spec to mlx5 flow spec.

Will be useful to add support for ip6 ethtool flow steering in the
next patch.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
6 years agonet/mlx5e: Ethtool steering flow validation refactoring
Saeed Mahameed [Fri, 6 Jul 2018 00:30:34 +0000 (17:30 -0700)]
net/mlx5e: Ethtool steering flow validation refactoring

Have a ethtool rx flow spec validation helper function per flow type.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
6 years agonet: sched: act_ife: disable bh when taking ife_mod_lock
Vlad Buslov [Mon, 13 Aug 2018 17:20:11 +0000 (20:20 +0300)]
net: sched: act_ife: disable bh when taking ife_mod_lock

Lockdep reports deadlock for following locking scenario in ife action:

Task one:
1) Executes ife action update.
2) Takes tcfa_lock.
3) Waits on ife_mod_lock which is already taken by task two.

Task two:

1) Executes any path that obtains ife_mod_lock without disabling bh (any
path that takes ife_mod_lock while holding tcfa_lock has bh disabled) like
loading a meta module, or creating new action.
2) Takes ife_mod_lock.
3) Task is preempted by rate estimator timer.
4) Timer callback waits on tcfa_lock which is taken by task one.

In described case tasks deadlock because they take same two locks in
different order. To prevent potential deadlock reported by lockdep, always
disable bh when obtaining ife_mod_lock.

Lockdep warning:

[  508.101192] =====================================================
[  508.107708] WARNING: SOFTIRQ-safe -> SOFTIRQ-unsafe lock order detected
[  508.114728] 4.18.0-rc8+ #646 Not tainted
[  508.119050] -----------------------------------------------------
[  508.125559] tc/5460 [HC0[0]:SC0[2]:HE1:SE0] is trying to acquire:
[  508.132025] 000000005a938c68 (ife_mod_lock){++++}, at: find_ife_oplist+0x1e/0xc0 [act_ife]
[  508.140996]
               and this task is already holding:
[  508.147548] 00000000d46f6c56 (&(&p->tcfa_lock)->rlock){+.-.}, at: tcf_ife_init+0x6ae/0xf40 [act_ife]
[  508.157371] which would create a new lock dependency:
[  508.162828]  (&(&p->tcfa_lock)->rlock){+.-.} -> (ife_mod_lock){++++}
[  508.169572]
               but this new dependency connects a SOFTIRQ-irq-safe lock:
[  508.178197]  (&(&p->tcfa_lock)->rlock){+.-.}
[  508.178201]
               ... which became SOFTIRQ-irq-safe at:
[  508.189771]   _raw_spin_lock+0x2c/0x40
[  508.193906]   est_fetch_counters+0x41/0xb0
[  508.198391]   est_timer+0x83/0x3c0
[  508.202180]   call_timer_fn+0x16a/0x5d0
[  508.206400]   run_timer_softirq+0x399/0x920
[  508.210967]   __do_softirq+0x157/0x97d
[  508.215102]   irq_exit+0x152/0x1c0
[  508.218888]   smp_apic_timer_interrupt+0xc0/0x4e0
[  508.223976]   apic_timer_interrupt+0xf/0x20
[  508.228540]   cpuidle_enter_state+0xf8/0x5d0
[  508.233198]   do_idle+0x28a/0x350
[  508.236881]   cpu_startup_entry+0xc7/0xe0
[  508.241296]   start_secondary+0x2e8/0x3f0
[  508.245678]   secondary_startup_64+0xa5/0xb0
[  508.250347]
               to a SOFTIRQ-irq-unsafe lock:  (ife_mod_lock){++++}
[  508.256531]
               ... which became SOFTIRQ-irq-unsafe at:
[  508.267279] ...
[  508.267283]   _raw_write_lock+0x2c/0x40
[  508.273653]   register_ife_op+0x118/0x2c0 [act_ife]
[  508.278926]   do_one_initcall+0xf7/0x4d9
[  508.283214]   do_init_module+0x18b/0x44e
[  508.287521]   load_module+0x4167/0x5730
[  508.291739]   __do_sys_finit_module+0x16d/0x1a0
[  508.296654]   do_syscall_64+0x7a/0x3f0
[  508.300788]   entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  508.306302]
               other info that might help us debug this:

[  508.315286]  Possible interrupt unsafe locking scenario:

[  508.322771]        CPU0                    CPU1
[  508.327681]        ----                    ----
[  508.332604]   lock(ife_mod_lock);
[  508.336300]                                local_irq_disable();
[  508.342608]                                lock(&(&p->tcfa_lock)->rlock);
[  508.349793]                                lock(ife_mod_lock);
[  508.355990]   <Interrupt>
[  508.358974]     lock(&(&p->tcfa_lock)->rlock);
[  508.363803]
                *** DEADLOCK ***

[  508.370715] 2 locks held by tc/5460:
[  508.374680]  #0: 00000000e27e4fa4 (rtnl_mutex){+.+.}, at: rtnetlink_rcv_msg+0x583/0x7b0
[  508.383366]  #1: 00000000d46f6c56 (&(&p->tcfa_lock)->rlock){+.-.}, at: tcf_ife_init+0x6ae/0xf40 [act_ife]
[  508.393648]
               the dependencies between SOFTIRQ-irq-safe lock and the holding lock:
[  508.403505] -> (&(&p->tcfa_lock)->rlock){+.-.} ops: 1001553 {
[  508.409646]    HARDIRQ-ON-W at:
[  508.413136]                     _raw_spin_lock_bh+0x34/0x40
[  508.419059]                     gnet_stats_start_copy_compat+0xa2/0x230
[  508.426021]                     gnet_stats_start_copy+0x16/0x20
[  508.432333]                     tcf_action_copy_stats+0x95/0x1d0
[  508.438735]                     tcf_action_dump_1+0xb0/0x4e0
[  508.444795]                     tcf_action_dump+0xca/0x200
[  508.450673]                     tcf_exts_dump+0xd9/0x320
[  508.456392]                     fl_dump+0x1b7/0x4a0 [cls_flower]
[  508.462798]                     tcf_fill_node+0x380/0x530
[  508.468601]                     tfilter_notify+0xdf/0x1c0
[  508.474404]                     tc_new_tfilter+0x84a/0xc90
[  508.480270]                     rtnetlink_rcv_msg+0x5bd/0x7b0
[  508.486419]                     netlink_rcv_skb+0x184/0x220
[  508.492394]                     netlink_unicast+0x31b/0x460
[  508.507411]                     netlink_sendmsg+0x3fb/0x840
[  508.513390]                     sock_sendmsg+0x7b/0xd0
[  508.518907]                     ___sys_sendmsg+0x4c6/0x610
[  508.524797]                     __sys_sendmsg+0xd7/0x150
[  508.530510]                     do_syscall_64+0x7a/0x3f0
[  508.536201]                     entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  508.543301]    IN-SOFTIRQ-W at:
[  508.546834]                     _raw_spin_lock+0x2c/0x40
[  508.552522]                     est_fetch_counters+0x41/0xb0
[  508.558571]                     est_timer+0x83/0x3c0
[  508.563912]                     call_timer_fn+0x16a/0x5d0
[  508.569699]                     run_timer_softirq+0x399/0x920
[  508.575840]                     __do_softirq+0x157/0x97d
[  508.581538]                     irq_exit+0x152/0x1c0
[  508.586882]                     smp_apic_timer_interrupt+0xc0/0x4e0
[  508.593533]                     apic_timer_interrupt+0xf/0x20
[  508.599686]                     cpuidle_enter_state+0xf8/0x5d0
[  508.605895]                     do_idle+0x28a/0x350
[  508.611147]                     cpu_startup_entry+0xc7/0xe0
[  508.617097]                     start_secondary+0x2e8/0x3f0
[  508.623029]                     secondary_startup_64+0xa5/0xb0
[  508.629245]    INITIAL USE at:
[  508.632686]                    _raw_spin_lock_bh+0x34/0x40
[  508.638557]                    gnet_stats_start_copy_compat+0xa2/0x230
[  508.645491]                    gnet_stats_start_copy+0x16/0x20
[  508.651719]                    tcf_action_copy_stats+0x95/0x1d0
[  508.657992]                    tcf_action_dump_1+0xb0/0x4e0
[  508.663937]                    tcf_action_dump+0xca/0x200
[  508.669716]                    tcf_exts_dump+0xd9/0x320
[  508.675337]                    fl_dump+0x1b7/0x4a0 [cls_flower]
[  508.681650]                    tcf_fill_node+0x380/0x530
[  508.687366]                    tfilter_notify+0xdf/0x1c0
[  508.693031]                    tc_new_tfilter+0x84a/0xc90
[  508.698820]                    rtnetlink_rcv_msg+0x5bd/0x7b0
[  508.704869]                    netlink_rcv_skb+0x184/0x220
[  508.710758]                    netlink_unicast+0x31b/0x460
[  508.716627]                    netlink_sendmsg+0x3fb/0x840
[  508.722510]                    sock_sendmsg+0x7b/0xd0
[  508.727931]                    ___sys_sendmsg+0x4c6/0x610
[  508.733729]                    __sys_sendmsg+0xd7/0x150
[  508.739346]                    do_syscall_64 +0x7a/0x3f0
[  508.744943]                    entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  508.751930]  }
[  508.753964]  ... key      at: [<ffffffff916b3e20>] __key.61145+0x0/0x40
[  508.760946]  ... acquired at:
[  508.764294]    _raw_read_lock+0x2f/0x40
[  508.768513]    find_ife_oplist+0x1e/0xc0 [act_ife]
[  508.773692]    tcf_ife_init+0x82f/0xf40 [act_ife]
[  508.778785]    tcf_action_init_1+0x510/0x750
[  508.783468]    tcf_action_init+0x1e8/0x340
[  508.787938]    tcf_action_add+0xc5/0x240
[  508.792241]    tc_ctl_action+0x203/0x2a0
[  508.796550]    rtnetlink_rcv_msg+0x5bd/0x7b0
[  508.801200]    netlink_rcv_skb+0x184/0x220
[  508.805674]    netlink_unicast+0x31b/0x460
[  508.810129]    netlink_sendmsg+0x3fb/0x840
[  508.814611]    sock_sendmsg+0x7b/0xd0
[  508.818665]    ___sys_sendmsg+0x4c6/0x610
[  508.823029]    __sys_sendmsg+0xd7/0x150
[  508.827246]    do_syscall_64+0x7a/0x3f0
[  508.831483]    entry_SYSCALL_64_after_hwframe+0x49/0xbe

               the dependencies between the lock to be acquired
[  508.838945]  and SOFTIRQ-irq-unsafe lock:
[  508.851177] -> (ife_mod_lock){++++} ops: 95 {
[  508.855920]    HARDIRQ-ON-W at:
[  508.859478]                     _raw_write_lock+0x2c/0x40
[  508.865264]                     register_ife_op+0x118/0x2c0 [act_ife]
[  508.872071]                     do_one_initcall+0xf7/0x4d9
[  508.877947]                     do_init_module+0x18b/0x44e
[  508.883819]                     load_module+0x4167/0x5730
[  508.889595]                     __do_sys_finit_module+0x16d/0x1a0
[  508.896043]                     do_syscall_64+0x7a/0x3f0
[  508.901734]                     entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  508.908827]    HARDIRQ-ON-R at:
[  508.912359]                     _raw_read_lock+0x2f/0x40
[  508.918043]                     find_ife_oplist+0x1e/0xc0 [act_ife]
[  508.924692]                     tcf_ife_init+0x82f/0xf40 [act_ife]
[  508.931252]                     tcf_action_init_1+0x510/0x750
[  508.937393]                     tcf_action_init+0x1e8/0x340
[  508.943366]                     tcf_action_add+0xc5/0x240
[  508.949130]                     tc_ctl_action+0x203/0x2a0
[  508.954922]                     rtnetlink_rcv_msg+0x5bd/0x7b0
[  508.961024]                     netlink_rcv_skb+0x184/0x220
[  508.966970]                     netlink_unicast+0x31b/0x460
[  508.972915]                     netlink_sendmsg+0x3fb/0x840
[  508.978859]                     sock_sendmsg+0x7b/0xd0
[  508.984400]                     ___sys_sendmsg+0x4c6/0x610
[  508.990264]                     __sys_sendmsg+0xd7/0x150
[  508.995952]                     do_syscall_64+0x7a/0x3f0
[  509.001643]                     entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  509.008722]    SOFTIRQ-ON-W at:\
[  509.012242]                     _raw_write_lock+0x2c/0x40
[  509.018013]                     register_ife_op+0x118/0x2c0 [act_ife]
[  509.024841]                     do_one_initcall+0xf7/0x4d9
[  509.030720]                     do_init_module+0x18b/0x44e
[  509.036604]                     load_module+0x4167/0x5730
[  509.042397]                     __do_sys_finit_module+0x16d/0x1a0
[  509.048865]                     do_syscall_64+0x7a/0x3f0
[  509.054551]                     entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  509.061636]    SOFTIRQ-ON-R at:
[  509.065145]                     _raw_read_lock+0x2f/0x40
[  509.070854]                     find_ife_oplist+0x1e/0xc0 [act_ife]
[  509.077515]                     tcf_ife_init+0x82f/0xf40 [act_ife]
[  509.084051]                     tcf_action_init_1+0x510/0x750
[  509.090172]                     tcf_action_init+0x1e8/0x340
[  509.096124]                     tcf_action_add+0xc5/0x240
[  509.101891]                     tc_ctl_action+0x203/0x2a0
[  509.107671]                     rtnetlink_rcv_msg+0x5bd/0x7b0
[  509.113811]                     netlink_rcv_skb+0x184/0x220
[  509.119768]                     netlink_unicast+0x31b/0x460
[  509.125716]                     netlink_sendmsg+0x3fb/0x840
[  509.131668]                     sock_sendmsg+0x7b/0xd0
[  509.137167]                     ___sys_sendmsg+0x4c6/0x610
[  509.143010]                     __sys_sendmsg+0xd7/0x150
[  509.148718]                     do_syscall_64+0x7a/0x3f0
[  509.154443]                     entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  509.161533]    INITIAL USE at:
[  509.164956]                    _raw_read_lock+0x2f/0x40
[  509.170574]                    find_ife_oplist+0x1e/0xc0 [act_ife]
[  509.177134]                    tcf_ife_init+0x82f/0xf40 [act_ife]
[  509.183619]                    tcf_action_init_1+0x510/0x750
[  509.189674]                    tcf_action_init+0x1e8/0x340
[  509.195534]                    tcf_action_add+0xc5/0x240
[  509.201229]                    tc_ctl_action+0x203/0x2a0
[  509.206920]                    rtnetlink_rcv_msg+0x5bd/0x7b0
[  509.212936]                    netlink_rcv_skb+0x184/0x220
[  509.218818]                    netlink_unicast+0x31b/0x460
[  509.224699]                    netlink_sendmsg+0x3fb/0x840
[  509.230581]                    sock_sendmsg+0x7b/0xd0
[  509.235984]                    ___sys_sendmsg+0x4c6/0x610
[  509.241791]                    __sys_sendmsg+0xd7/0x150
[  509.247425]                    do_syscall_64+0x7a/0x3f0
[  509.253007]                    entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  509.259975]  }
[  509.261998]  ... key      at: [<ffffffffc1554258>] ife_mod_lock+0x18/0xffffffffffff8dc0 [act_ife]
[  509.271569]  ... acquired at:
[  509.274912]    _raw_read_lock+0x2f/0x40
[  509.279134]    find_ife_oplist+0x1e/0xc0 [act_ife]
[  509.284324]    tcf_ife_init+0x82f/0xf40 [act_ife]
[  509.289425]    tcf_action_init_1+0x510/0x750
[  509.294068]    tcf_action_init+0x1e8/0x340
[  509.298553]    tcf_action_add+0xc5/0x240
[  509.302854]    tc_ctl_action+0x203/0x2a0
[  509.307153]    rtnetlink_rcv_msg+0x5bd/0x7b0
[  509.311805]    netlink_rcv_skb+0x184/0x220
[  509.316282]    netlink_unicast+0x31b/0x460
[  509.320769]    netlink_sendmsg+0x3fb/0x840
[  509.325248]    sock_sendmsg+0x7b/0xd0
[  509.329290]    ___sys_sendmsg+0x4c6/0x610
[  509.333687]    __sys_sendmsg+0xd7/0x150
[  509.337902]    do_syscall_64+0x7a/0x3f0
[  509.342116]    entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  509.349601]
               stack backtrace:
[  509.354663] CPU: 6 PID: 5460 Comm: tc Not tainted 4.18.0-rc8+ #646
[  509.361216] Hardware name: Supermicro SYS-2028TP-DECR/X10DRT-P, BIOS 2.0b 03/30/2017

Fixes: ef6980b6becb ("introduce IFE action")
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetoot...
David S. Miller [Mon, 13 Aug 2018 18:38:54 +0000 (11:38 -0700)]
Merge branch 'for-upstream' of git://git./linux/kernel/git/bluetooth/bluetooth-next

Johan Hedberg says:

====================
pull request: bluetooth-next 2018-08-13

There was one pretty bad bug that slipped into the MediaTek HCI driver
in the last bluetooth-next pull request. Would it be possible to get
this one-liner fix pulled to net-next before you make your first 4.19
pull request for Linus? Thanks.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 Aug 2018 18:25:07 +0000 (11:25 -0700)]
Merge branch 'sched-core-for-linus' of git://git./linux/kernel/git/tip/tip

Pull scheduler updates from Thomas Gleixner:

 - Cleanup and improvement of NUMA balancing

 - Refactoring and improvements to the PELT (Per Entity Load Tracking)
   code

 - Watchdog simplification and related cleanups

 - The usual pile of small incremental fixes and improvements

* 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (41 commits)
  watchdog: Reduce message verbosity
  stop_machine: Reflow cpu_stop_queue_two_works()
  sched/numa: Move task_numa_placement() closer to numa_migrate_preferred()
  sched/numa: Use group_weights to identify if migration degrades locality
  sched/numa: Update the scan period without holding the numa_group lock
  sched/numa: Remove numa_has_capacity()
  sched/numa: Modify migrate_swap() to accept additional parameters
  sched/numa: Remove unused task_capacity from 'struct numa_stats'
  sched/numa: Skip nodes that are at 'hoplimit'
  sched/debug: Reverse the order of printing faults
  sched/numa: Use task faults only if numa_group is not yet set up
  sched/numa: Set preferred_node based on best_cpu
  sched/numa: Simplify load_too_imbalanced()
  sched/numa: Evaluate move once per node
  sched/numa: Remove redundant field
  sched/debug: Show the sum wait time of a task group
  sched/fair: Remove #ifdefs from scale_rt_capacity()
  sched/core: Remove get_cpu() from sched_fork()
  sched/cpufreq: Clarify sugov_get_util()
  sched/sysctl: Remove unused sched_time_avg_ms sysctl
  ...

6 years agoMerge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 Aug 2018 18:21:34 +0000 (11:21 -0700)]
Merge branch 'sched-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull scheduler fix from Thomas Gleixner:
 "A single bugfix to prevent a pinned thread which queues stomp machine
  work to be preempted by the stopper thread on its CPU which causes a
  live lock as it is unable to wake the second CPUs stopper thread"

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  stop_machine: Atomically queue and wake stopper threads

6 years agoMerge branch 'ras-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 18:19:25 +0000 (11:19 -0700)]
Merge branch 'ras-core-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 RAS updates from Thomas Gleixner:
 "A small set of changes to the RAS core:

   - Rework of the MCE bank scanning code

   - Y2038 converion"

* 'ras-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/mce: Cleanup __mc_scan_banks()
  x86/mce: Carve out bank scanning code
  x86/mce: Remove !banks check
  x86/mce: Carve out the crashing_cpu check
  x86/mce: Always use 64-bit timestamps

6 years agoMerge branch 'core-rcu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 17:49:41 +0000 (10:49 -0700)]
Merge branch 'core-rcu-for-linus' of git://git./linux/kernel/git/tip/tip

Pull RCU updates from Thomas Gleixner:
 "A large update to RCU:

  Preparatory work for consolidating the RCU flavors:

   - Introduce grace-period sequence numbers to the RCU-bh, RCU-preempt,
     and RCU-sched flavors, replacing the old ->gpnum and ->completed
     pair of fields.

     This change allows lockless code to obtain the complete
     grace-period state with a single READ_ONCE(), which is needed to
     maintain tolerable lock contention during the upcoming
     consolidation of the three RCU flavors.

     Note that grace-period sequence numbers are already used by
     rcu_barrier(), expedited RCU grace periods, and SRCU, and are thus
     already heavily used and well-tested. Joel Fernandes contributed a
     number of excellent fixes and improvements.

   - Clean up some grace-period-reporting loose ends, including
     improving the handling of quiescent states from offline CPUs and
     fixing some false-positive WARN_ON_ONCE() invocations.

     (Strictly speaking, the WARN_ON_ONCE() invocations were quite
     correct, but their invariants were (harmlessly) violated by the
     earlier sloppy handling of quiescent states from offline CPUs.)

     In addition, improve grace-period forward-progress guarantees so as
     to allow removal of fail-safe checks that required otherwise
     needless lock acquisitions. Finally, add more diagnostics to help
     debug the upcoming consolidation of the RCU-bh, RCU-preempt, and
     RCU-sched flavors.

  The rest:

   - SRCU updates

   - Updates to rcutorture and associated scripting.

   - The usual pile of miscellaneous fixes"

* 'core-rcu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (118 commits)
  rcutorture: Fix rcu_barrier successes counter
  rcutorture: Add support to detect if boost kthread prio is too low
  rcutorture: Use monotonic timestamp for stall detection
  rcutorture: Make boost test more robust
  rcutorture: Disable RT throttling for boost tests
  rcutorture: Emphasize testing of single reader protection type
  rcutorture: Handle extended read-side critical sections
  rcutorture: Make rcu_torture_timer() use rcu_torture_one_read()
  rcutorture: Use per-CPU random state for rcu_torture_timer()
  rcutorture: Use atomic increment for n_rcu_torture_timers
  rcutorture: Extract common code from rcu_torture_reader()
  rcuperf: Remove unused torturing_tasks() function
  rcu: Remove rcutorture test version and sequence number
  rcutorture: Change units of onoff_interval to jiffies
  rcu: Assign higher prio to RCU threads if rcutorture is built-in
  rculist: Improve documentation for list_for_each_entry_from_rcu()
  srcu: Add grace-period number to rcutorture statistics printout
  rcu: Print stall-warning NMI dyntick state in hexadecimal
  MAINTAINERS: Update RCU, SRCU, and TORTURE-TEST entries
  rcu: Make rcu_seq_diff() more exact
  ...

6 years agoMerge branch 'irq-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 17:47:26 +0000 (10:47 -0700)]
Merge branch 'irq-core-for-linus' of git://git./linux/kernel/git/tip/tip

Pull genirq updates from Thomas Gleixner:
 "The irq departement provides:

   - A synchronization fix for free_irq() to synchronize just the
     removed interrupt thread on shared interrupt lines.

   - Consolidate the multi low level interrupt entry handling and mvoe
     it to the generic code instead of adding yet another copy for
     RISC-V

   - Refactoring of the ARM LPI allocator and LPI exposure to the
     hypervisor

   - Yet another interrupt chip driver for the JZ4725B SoC

   - Speed up for /proc/interrupts as people seem to love reading this
     file with high frequency

   - Miscellaneous fixes and updates"

* 'irq-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (23 commits)
  irqchip/gic-v3-its: Make its_lock a raw_spin_lock_t
  genirq/irqchip: Remove MULTI_IRQ_HANDLER as it's now obselete
  openrisc: Use the new GENERIC_IRQ_MULTI_HANDLER
  arm64: Use the new GENERIC_IRQ_MULTI_HANDLER
  ARM: Convert to GENERIC_IRQ_MULTI_HANDLER
  irqchip: Port the ARM IRQ drivers to GENERIC_IRQ_MULTI_HANDLER
  irqchip/gic-v3-its: Reduce minimum LPI allocation to 1 for PCI devices
  dt-bindings: irqchip: renesas-irqc: Document r8a77980 support
  dt-bindings: irqchip: renesas-irqc: Document r8a77470 support
  irqchip/ingenic: Add support for the JZ4725B SoC
  irqchip/stm32: Add exti0 translation for stm32mp1
  genirq: Remove redundant NULL pointer check in __free_irq()
  irqchip/gic-v3-its: Honor hypervisor enforced LPI range
  irqchip/gic-v3: Expose GICD_TYPER in the rdist structure
  irqchip/gic-v3-its: Drop chunk allocation compatibility
  irqchip/gic-v3-its: Move minimum LPI requirements to individual busses
  irqchip/gic-v3-its: Use full range of LPIs
  irqchip/gic-v3-its: Refactor LPI allocator
  genirq: Synchronize only with single thread on free_irq()
  genirq: Update code comments wrt recycled thread_mask
  ...

6 years agoMerge branch 'efi-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 17:25:08 +0000 (10:25 -0700)]
Merge branch 'efi-core-for-linus' of git://git./linux/kernel/git/tip/tip

Pull EFI updates from Thomas Gleixner:
 "The EFI pile:

   - Make mixed mode UEFI runtime service invocations mutually
     exclusive, as mandated by the UEFI spec

   - Perform UEFI runtime services calls from a work queue so the calls
     into the firmware occur from a kernel thread

   - Honor the UEFI memory map attributes for live memory regions
     configured by UEFI as a framebuffer. This works around a coherency
     problem with KVM guests running on ARM.

   - Cleanups, improvements and fixes all over the place"

* 'efi-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  efivars: Call guid_parse() against guid_t type of variable
  efi/cper: Use consistent types for UUIDs
  efi/x86: Replace references to efi_early->is64 with efi_is_64bit()
  efi: Deduplicate efi_open_volume()
  efi/x86: Add missing NULL initialization in UGA draw protocol discovery
  efi/x86: Merge 32-bit and 64-bit UGA draw protocol setup routines
  efi/x86: Align efi_uga_draw_protocol typedef names to convention
  efi/x86: Merge the setup_efi_pci32() and setup_efi_pci64() routines
  efi/x86: Prevent reentrant firmware calls in mixed mode
  efi/esrt: Only call efi_mem_reserve() for boot services memory
  fbdev/efifb: Honour UEFI memory map attributes when mapping the FB
  efi: Drop type and attribute checks in efi_mem_desc_lookup()
  efi/libstub/arm: Add opt-in Kconfig option for the DTB loader
  efi: Remove the declaration of efi_late_init() as the function is unused
  efi/cper: Avoid using get_seconds()
  efi: Use a work queue to invoke EFI Runtime Services
  efi/x86: Use non-blocking SetVariable() for efi_delete_dummy_variable()
  efi/x86: Clean up the eboot code

6 years agoMerge branch 'core-debugobjects-for-linus' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Mon, 13 Aug 2018 17:21:28 +0000 (10:21 -0700)]
Merge branch 'core-debugobjects-for-linus' of git://git./linux/kernel/git/tip/tip

Pull debugobjects update from Thomas Gleixner:
 "Two simple updates for the debug objects code:

   - Make the stack check warning more informative by adding the object
     and the stack page address to the printout

   - Remove a redundant NULL pointer check"

* 'core-debugobjects-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  debugobjects: Remove redundant NULL pointer check
  debugobjects: Make stack check warning more informative

6 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
David S. Miller [Mon, 13 Aug 2018 17:07:23 +0000 (10:07 -0700)]
Merge git://git./linux/kernel/git/bpf/bpf-next

Daniel Borkmann says:

====================
pull-request: bpf-next 2018-08-13

The following pull-request contains BPF updates for your *net-next* tree.

The main changes are:

1) Add driver XDP support for veth. This can be used in conjunction with
   redirect of another XDP program e.g. sitting on NIC so the xdp_frame
   can be forwarded to the peer veth directly without modification,
   from Toshiaki.

2) Add a new BPF map type REUSEPORT_SOCKARRAY and prog type SK_REUSEPORT
   in order to provide more control and visibility on where a SO_REUSEPORT
   sk should be located, and the latter enables to directly select a sk
   from the bpf map. This also enables map-in-map for application migration
   use cases, from Martin.

3) Add a new BPF helper bpf_skb_ancestor_cgroup_id() that returns the id
   of cgroup v2 that is the ancestor of the cgroup associated with the
   skb at the ancestor_level, from Andrey.

4) Implement BPF fs map pretty-print support based on BTF data for regular
   hash table and LRU map, from Yonghong.

5) Decouple the ability to attach BTF for a map from the key and value
   pretty-printer in BPF fs, and enable further support of BTF for maps for
   percpu and LPM trie, from Daniel.

6) Implement a better BPF sample of using XDP's CPU redirect feature for
   load balancing SKB processing to remote CPU. The sample implements the
   same XDP load balancing as Suricata does which is symmetric hash based
   on IP and L4 protocol, from Jesper.

7) Revert adding NULL pointer check with WARN_ON_ONCE() in __xdp_return()'s
   critical path as it is ensured that the allocator is present, from Björn.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge tag 'wireless-drivers-next-for-davem-2018-08-12' of git://git.kernel.org/pub...
David S. Miller [Mon, 13 Aug 2018 17:06:11 +0000 (10:06 -0700)]
Merge tag 'wireless-drivers-next-for-davem-2018-08-12' of git://git./linux/kernel/git/kvalo/wireless-drivers-next

Kalle Valo says:

====================
pull-request: wireless-drivers-next 2018-08-12

wireless-drivers-next patches for 4.19

Last set of new features for 4.19. Most notable is simplifying SSB
debugging code with two Kconfig option removals and fixing mt76 USB
build problems.

Major changes:

ath10k

* add debugfs file warm_hw_reset

wil6210

* add debugfs files tx_latency, link_stats and link_stats_global

* add 3-MSI support

* allow scan on AP interface

* support max aggregation window size 64

ssb

* remove CONFIG_SSB_SILENT and CONFIG_SSB_DEBUG Kconfig options

mt76

* fix build problems with recently added USB support
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge tag 'm68k-for-v4.19-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 13 Aug 2018 16:50:24 +0000 (09:50 -0700)]
Merge tag 'm68k-for-v4.19-tag1' of git://git./linux/kernel/git/geert/linux-m68k

Pull m68k updates from Geert Uytterhoeven:

 - Enable mac_scsi PDMA on PowerBook 500

 - Generic dma_noncoherent_ops conversion

 - Time handling improvements

 - I/O accessor improvements

 - Conversion to MEMBLOCK and NO_BOOTMEM, to bring m68k in line with
   other mainstream architectures

 - Miscellaneous fixes and cleanups

 - Defconfig updates

* tag 'm68k-for-v4.19-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux-m68k:
  m68k/defconfig: Update defconfigs for v4.18-rc6
  m68k: switch to MEMBLOCK + NO_BOOTMEM
  m68k/page_no.h: force __va argument to be unsigned long
  m68k/bitops: convert __ffs to match generic declaration
  m68k/io: Switch mmu variant to <asm-generic/io.h>
  m68k/io: Move mem*io define guards to <asm/kmap.h>
  Input: hilkbd - Add casts to HP9000/300 I/O accessors
  net: mac8390: Use standard memcpy_{from,to}io()
  m68k/io: Add missing ioremap define guards, fix typo
  m68k: Remove unused set_clock_mmss() helpers
  m68k: mac: Use time64_t in RTC handling
  m68k: Use generic dma_noncoherent_ops
  nubus: Set default dma mask for nubus_board devices
  m68k/mac: Enable PDMA for PowerBook 500 series

6 years agoliquidio: remove set but not used variable 'is25G'
YueHaibing [Mon, 13 Aug 2018 09:29:36 +0000 (17:29 +0800)]
liquidio: remove set but not used variable 'is25G'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/net/ethernet/cavium/liquidio/lio_ethtool.c: In function 'lio_set_link_ksettings':
drivers/net/ethernet/cavium/liquidio/lio_ethtool.c:392:6: warning:
 variable 'is25G' set but not used [-Wunused-but-set-variable]

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agocxgb4: remove set but not used variable 'spd'
YueHaibing [Mon, 13 Aug 2018 06:59:02 +0000 (14:59 +0800)]
cxgb4: remove set but not used variable 'spd'

Fixes gcc '-Wunused-but-set-variable' warning:
drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c: In function 'print_port_info':
drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c:5147:14: warning:
 variable 'spd' set but not used [-Wunused-but-set-variable]

variable 'spd' is set but not used since
commit 547fd27241a8 ("cxgb4: Warn if device doesn't have enough PCI bandwidth")

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agolan743x: lan743x: Remove duplicated include from lan743x_ptp.c
Yue Haibing [Mon, 13 Aug 2018 06:39:21 +0000 (06:39 +0000)]
lan743x: lan743x: Remove duplicated include from lan743x_ptp.c

Remove duplicated include.

Signed-off-by: Yue Haibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agovirtio_net: remove duplicated include from virtio_net.c
YueHaibing [Mon, 13 Aug 2018 06:13:15 +0000 (14:13 +0800)]
virtio_net: remove duplicated include from virtio_net.c

Remove duplicated include linux/netdevice.h

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agopacket: switch kvzalloc to allocate memory
Li RongQing [Mon, 13 Aug 2018 02:42:46 +0000 (10:42 +0800)]
packet: switch kvzalloc to allocate memory

The patches includes following change:

*Use modern kvzalloc()/kvfree() instead of custom allocations.

*Remove order argument for alloc_pg_vec, it can get from req.

*Remove order argument for free_pg_vec, free_pg_vec now uses
kvfree which does not need order argument.

*Remove pg_vec_order from struct packet_ring_buffer, no longer
need to save/restore 'order'

*Remove variable 'order' for packet_set_ring, it is now unused

Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: Change the layout of structure trace_event_raw_fib_table_lookup
Zong Li [Mon, 13 Aug 2018 02:26:52 +0000 (10:26 +0800)]
net: Change the layout of structure trace_event_raw_fib_table_lookup

There is an unalignment access about the structure
'trace_event_raw_fib_table_lookup'.

In include/trace/events/fib.h, there is a memory operation which casting
the 'src' data member to a pointer, and then store a value to this
pointer point to.

p32 = (__be32 *) __entry->src;
*p32 = flp->saddr;

The offset of 'src' in structure trace_event_raw_fib_table_lookup is not
four bytes alignment. On some architectures, they don't permit the
unalignment access, it need to pay the price to handle this situation in
exception handler.

Adjust the layout of structure to avoid this case.

Fixes: 9f323973c915 ("net/ipv4: Udate fib_table_lookup tracepoint")
Signed-off-by: Zong Li <zong@andestech.com>
Acked-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'net-sched-actions-rename-for-grep-ability-and-consistency'
David S. Miller [Mon, 13 Aug 2018 16:06:22 +0000 (09:06 -0700)]
Merge branch 'net-sched-actions-rename-for-grep-ability-and-consistency'

Jamal Hadi Salim says:

====================
net: sched: actions rename for grep-ability and consistency

Having a structure (example tcf_mirred) and a function with the same name is
not good for readability or grepability.

This long overdue patchset improves it and make sure there is consistency
across all actions
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_mirred method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:35:01 +0000 (09:35 -0400)]
net: sched: act_mirred method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_vlan method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:35:00 +0000 (09:35 -0400)]
net: sched: act_vlan method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_skbmod method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:59 +0000 (09:34 -0400)]
net: sched: act_skbmod method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_skbedit method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:58 +0000 (09:34 -0400)]
net: sched: act_skbedit method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_simple method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:57 +0000 (09:34 -0400)]
net: sched: act_simple method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_police method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:56 +0000 (09:34 -0400)]
net: sched: act_police method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_pedit method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:55 +0000 (09:34 -0400)]
net: sched: act_pedit method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_nat method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:54 +0000 (09:34 -0400)]
net: sched: act_nat method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_ipt method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:53 +0000 (09:34 -0400)]
net: sched: act_ipt method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_gact method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:52 +0000 (09:34 -0400)]
net: sched: act_gact method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_sum method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:51 +0000 (09:34 -0400)]
net: sched: act_sum method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_bpf method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:50 +0000 (09:34 -0400)]
net: sched: act_bpf method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: act_connmark method rename for grep-ability and consistency
Jamal Hadi Salim [Sun, 12 Aug 2018 13:34:49 +0000 (09:34 -0400)]
net: sched: act_connmark method rename for grep-ability and consistency

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agocpumask: make cpumask_next_wrap available without smp
Willem de Bruijn [Sun, 12 Aug 2018 13:14:03 +0000 (09:14 -0400)]
cpumask: make cpumask_next_wrap available without smp

The kbuild robot shows build failure on machines without CONFIG_SMP:

  drivers/net/virtio_net.c:1916:10: error:
    implicit declaration of function 'cpumask_next_wrap'

cpumask_next_wrap is exported from lib/cpumask.o, which has

    lib-$(CONFIG_SMP) += cpumask.o

same as other functions, also define it as static inline in the
NR_CPUS==1 branch in include/linux/cpumask.h.

If wrap is true and next == start, return nr_cpumask_bits, or 1.
Else wrap across the range of valid cpus, here [0].

Fixes: 2ca653d607ce ("virtio_net: Stripe queue affinities across cores.")
Signed-off-by: Willem de Bruijn <willemb@google.com>
Tested-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agor8169: don't use MSI-X on RTL8168g
Heiner Kallweit [Sun, 12 Aug 2018 11:26:26 +0000 (13:26 +0200)]
r8169: don't use MSI-X on RTL8168g

There have been two reports that network doesn't come back on resume
from suspend when using MSI-X. Both cases affect the same chip version
(RTL8168g - version 40), on different systems. Falling back to MSI
fixes the issue.
Even though we don't really have a proof yet that the network chip
version is to blame, let's disable MSI-X for this version.

Reported-by: Steve Dodd <steved424@gmail.com>
Reported-by: Lou Reed <gogen@disroot.org>
Tested-by: Steve Dodd <steved424@gmail.com>
Tested-by: Lou Reed <gogen@disroot.org>
Fixes: 6c6aa15fdea5 ("r8169: improve interrupt handling")
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'nixge-Minor-cleanups'
David S. Miller [Mon, 13 Aug 2018 15:49:37 +0000 (08:49 -0700)]
Merge branch 'nixge-Minor-cleanups'

Moritz Fischer says:

====================
net: nixge: Minor cleanups

in preparation of my 64-bit support series, here's some
minor cleanup in preparation that gets rid of unneccesary
accesses to the descriptor application fields.

I've confirmed that the hardware does not access the fields
in all our configurations.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: nixge: Don't store skb in app4 field of descriptor
Moritz Fischer [Sat, 11 Aug 2018 01:19:41 +0000 (18:19 -0700)]
net: nixge: Don't store skb in app4 field of descriptor

Don't store skb in app4 field of descriptor since it is
not being used anywhere (including hardware).

Signed-off-by: Moritz Fischer <mdf@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: nixge: Do not zero application specific fields in desc
Moritz Fischer [Sat, 11 Aug 2018 01:19:40 +0000 (18:19 -0700)]
net: nixge: Do not zero application specific fields in desc

Do not zero application specific fields in DMA descriptors.
The hardware does ignore them, so should software.

Signed-off-by: Moritz Fischer <mdf@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agol2tp: use sk_dst_check() to avoid race on sk->sk_dst_cache
Wei Wang [Fri, 10 Aug 2018 18:14:56 +0000 (11:14 -0700)]
l2tp: use sk_dst_check() to avoid race on sk->sk_dst_cache

In l2tp code, if it is a L2TP_UDP_ENCAP tunnel, tunnel->sk points to a
UDP socket. User could call sendmsg() on both this tunnel and the UDP
socket itself concurrently. As l2tp_xmit_skb() holds socket lock and call
__sk_dst_check() to refresh sk->sk_dst_cache, while udpv6_sendmsg() is
lockless and call sk_dst_check() to refresh sk->sk_dst_cache, there
could be a race and cause the dst cache to be freed multiple times.
So we fix l2tp side code to always call sk_dst_check() to garantee
xchg() is called when refreshing sk->sk_dst_cache to avoid race
conditions.

Syzkaller reported stack trace:
BUG: KASAN: use-after-free in atomic_read include/asm-generic/atomic-instrumented.h:21 [inline]
BUG: KASAN: use-after-free in atomic_fetch_add_unless include/linux/atomic.h:575 [inline]
BUG: KASAN: use-after-free in atomic_add_unless include/linux/atomic.h:597 [inline]
BUG: KASAN: use-after-free in dst_hold_safe include/net/dst.h:308 [inline]
BUG: KASAN: use-after-free in ip6_hold_safe+0xe6/0x670 net/ipv6/route.c:1029
Read of size 4 at addr ffff8801aea9a880 by task syz-executor129/4829

CPU: 0 PID: 4829 Comm: syz-executor129 Not tainted 4.18.0-rc7-next-20180802+ #30
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0x1c9/0x2b4 lib/dump_stack.c:113
 print_address_description+0x6c/0x20b mm/kasan/report.c:256
 kasan_report_error mm/kasan/report.c:354 [inline]
 kasan_report.cold.7+0x242/0x30d mm/kasan/report.c:412
 check_memory_region_inline mm/kasan/kasan.c:260 [inline]
 check_memory_region+0x13e/0x1b0 mm/kasan/kasan.c:267
 kasan_check_read+0x11/0x20 mm/kasan/kasan.c:272
 atomic_read include/asm-generic/atomic-instrumented.h:21 [inline]
 atomic_fetch_add_unless include/linux/atomic.h:575 [inline]
 atomic_add_unless include/linux/atomic.h:597 [inline]
 dst_hold_safe include/net/dst.h:308 [inline]
 ip6_hold_safe+0xe6/0x670 net/ipv6/route.c:1029
 rt6_get_pcpu_route net/ipv6/route.c:1249 [inline]
 ip6_pol_route+0x354/0xd20 net/ipv6/route.c:1922
 ip6_pol_route_output+0x54/0x70 net/ipv6/route.c:2098
 fib6_rule_lookup+0x283/0x890 net/ipv6/fib6_rules.c:122
 ip6_route_output_flags+0x2c5/0x350 net/ipv6/route.c:2126
 ip6_dst_lookup_tail+0x1278/0x1da0 net/ipv6/ip6_output.c:978
 ip6_dst_lookup_flow+0xc8/0x270 net/ipv6/ip6_output.c:1079
 ip6_sk_dst_lookup_flow+0x5ed/0xc50 net/ipv6/ip6_output.c:1117
 udpv6_sendmsg+0x2163/0x36b0 net/ipv6/udp.c:1354
 inet_sendmsg+0x1a1/0x690 net/ipv4/af_inet.c:798
 sock_sendmsg_nosec net/socket.c:622 [inline]
 sock_sendmsg+0xd5/0x120 net/socket.c:632
 ___sys_sendmsg+0x51d/0x930 net/socket.c:2115
 __sys_sendmmsg+0x240/0x6f0 net/socket.c:2210
 __do_sys_sendmmsg net/socket.c:2239 [inline]
 __se_sys_sendmmsg net/socket.c:2236 [inline]
 __x64_sys_sendmmsg+0x9d/0x100 net/socket.c:2236
 do_syscall_64+0x1b9/0x820 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x446a29
Code: e8 ac b8 02 00 48 83 c4 18 c3 0f 1f 80 00 00 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 0f 83 eb 08 fc ff c3 66 2e 0f 1f 84 00 00 00 00
RSP: 002b:00007f4de5532db8 EFLAGS: 00000246 ORIG_RAX: 0000000000000133
RAX: ffffffffffffffda RBX: 00000000006dcc38 RCX: 0000000000446a29
RDX: 00000000000000b8 RSI: 0000000020001b00 RDI: 0000000000000003
RBP: 00000000006dcc30 R08: 00007f4de5533700 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 00000000006dcc3c
R13: 00007ffe2b830fdf R14: 00007f4de55339c0 R15: 0000000000000001

Fixes: 71b1391a4128 ("l2tp: ensure sk->dst is still valid")
Reported-by: syzbot+05f840f3b04f211bad55@syzkaller.appspotmail.com
Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Cc: Guillaume Nault <g.nault@alphalink.fr>
Cc: David Ahern <dsahern@gmail.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoipv6: Add icmp_echo_ignore_all support for ICMPv6
Virgile Jarry [Fri, 10 Aug 2018 15:48:15 +0000 (17:48 +0200)]
ipv6: Add icmp_echo_ignore_all support for ICMPv6

Preventing the kernel from responding to ICMP Echo Requests messages
can be useful in several ways. The sysctl parameter
'icmp_echo_ignore_all' can be used to prevent the kernel from
responding to IPv4 ICMP echo requests. For IPv6 pings, such
a sysctl kernel parameter did not exist.

Add the ability to prevent the kernel from responding to IPv6
ICMP echo requests through the use of the following sysctl
parameter : /proc/sys/net/ipv6/icmp/echo_ignore_all.
Update the documentation to reflect this change.

Signed-off-by: Virgile Jarry <virgile@acceis.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'net-tls-Combined-memory-allocation-for-decryption-request'
David S. Miller [Mon, 13 Aug 2018 15:41:09 +0000 (08:41 -0700)]
Merge branch 'net-tls-Combined-memory-allocation-for-decryption-request'

Vakul Garg says:

====================
net/tls: Combined memory allocation for decryption request

This patch does a combined memory allocation from heap for scatterlists,
aead_request, aad and iv for the tls record decryption path. In present
code, aead_request is allocated from heap, scatterlists on a conditional
basis are allocated on heap or on stack. This is inefficient as it may
requires multiple kmalloc/kfree.

The initialization vector passed in cryption request is allocated on
stack. This is a problem since the stack memory is not dma-able from
crypto accelerators.

Doing one combined memory allocation for each decryption request fixes
both the above issues. It also paves a way to be able to submit multiple
async decryption requests while the previous one is pending i.e. being
processed or queued.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/tls: Combined memory allocation for decryption request
Vakul Garg [Fri, 10 Aug 2018 15:16:41 +0000 (20:46 +0530)]
net/tls: Combined memory allocation for decryption request

For preparing decryption request, several memory chunks are required
(aead_req, sgin, sgout, iv, aad). For submitting the decrypt request to
an accelerator, it is required that the buffers which are read by the
accelerator must be dma-able and not come from stack. The buffers for
aad and iv can be separately kmalloced each, but it is inefficient.
This patch does a combined allocation for preparing decryption request
and then segments into aead_req || sgin || sgout || iv || aad.

Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branches 'fixes', 'misc' and 'spectre' into for-linus
Russell King [Mon, 13 Aug 2018 15:28:50 +0000 (16:28 +0100)]
Merge branches 'fixes', 'misc' and 'spectre' into for-linus

Conflicts:
arch/arm/include/asm/uaccess.h

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
6 years agoARM: 8778/1: clkdev: don't call __of_clk_get_by_name() unnecessarily from clk_get()
Bartosz Golaszewski [Thu, 28 Jun 2018 14:42:20 +0000 (15:42 +0100)]
ARM: 8778/1: clkdev: don't call __of_clk_get_by_name() unnecessarily from clk_get()

The way this function is implemented caused some confusion when
converting the TI DaVinci platform to using the common clock framework.

Current kernel supports booting DaVinci boards both in device tree as
well as legacy, board-file mode. In the latter, we always end up
calling clk_get_sys() as of_node is NULL and __of_clk_get_by_name()
returns -ENOENT.

It was not obvious at first glance how clk_get(dev, NULL) will work in
board-file mode since we always call __of_clk_get_by_name(). Let's make
it clearer by checking if of_node is NULL and skipping right to
clk_get_sys().

Cc: Sekhar Nori <nsekhar@ti.com>
Cc: Kevin Hilman <khilman@baylibre.com>
Cc: David Lechner <david@lechnology.com>
Reviewed-by: David Lechner <david@lechnology.com>
Reviewed-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
6 years agoDocumentation: remove dynamic-resolution-notes reference to non-existent file
Harish Jenny K N [Thu, 19 Jul 2018 11:54:04 +0000 (17:24 +0530)]
Documentation: remove dynamic-resolution-notes reference to non-existent file

File dt-object-internal.txt does not exist. This patch removes
a reference to it.

Signed-off-by: Harish Jenny K N <harish_kandiga@mentor.com>
Reviewed-by: Frank Rowand <frank.rowand@sony.com>
Signed-off-by: Rob Herring <robh@kernel.org>
6 years agoBluetooth: mediatek: pass correct size to h4_recv_buf()
Dan Carpenter [Mon, 13 Aug 2018 09:32:35 +0000 (12:32 +0300)]
Bluetooth: mediatek: pass correct size to h4_recv_buf()

We're supposed to pass the number of elements in the mtk_recv_pkts, not
the number of bytes.

Fixes: 7237c4c9ec92 ("Bluetooth: mediatek: Add protocol support for MediaTek serial devices")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
6 years agoMerge tag 'asoc-v4.19' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie...
Takashi Iwai [Mon, 13 Aug 2018 10:12:31 +0000 (12:12 +0200)]
Merge tag 'asoc-v4.19' of https://git./linux/kernel/git/broonie/sound into for-linus

ASoC: Updates for v4.19

A fairly big update, including quite a bit of core activity this time
around (which is good to see) along with a fairly large set of new
drivers.

 - A new snd_pcm_stop_xrun() helper which is now used in several
   drivers.
 - Support for providing name prefixes to generic component nodes.
 - Quite a few fixes for DPCM as it gains a bit wider use and more
   robust testing.
 - Generalization of the DIO2125 support to a simple amplifier driver.
 - Accessory detection support for the audio graph card.
 - DT support for PXA AC'97 devices.
 - Quirks for a number of new x86 systems.
 - Support for AM Logic Meson, Everest ES7154, Intel systems with
   RT5682, Qualcomm QDSP6 and WCD9335, Realtek RT5682 and TI TAS5707.

6 years agoparisc: Fix and improve kernel stack unwinding
Helge Deller [Sat, 4 Aug 2018 22:03:29 +0000 (00:03 +0200)]
parisc: Fix and improve kernel stack unwinding

This patchset fixes and improves stack unwinding a lot:
1. Show backward stack traces with up to 30 callsites
2. Add callinfo to ENTRY_CFI() such that every assembler function will get an
   entry in the unwind table
3. Use constants instead of numbers in call_on_stack()
4. Do not depend on CONFIG_KALLSYMS to generate backtraces.
5. Speed up backtrace generation

Make sure you have this patch to GNU as installed:
https://sourceware.org/ml/binutils/2018-07/msg00474.html
Without this patch, unwind info in the kernel is often wrong for various
functions.

Signed-off-by: Helge Deller <deller@gmx.de>
6 years agoparisc: Remove unnecessary barriers from spinlock.h
John David Anglin [Sun, 12 Aug 2018 20:31:17 +0000 (16:31 -0400)]
parisc: Remove unnecessary barriers from spinlock.h

Now that mb() is an instruction barrier, it will slow performance if we issue
unnecessary barriers.

The spinlock defines have a number of unnecessary barriers.  The __ldcw()
define is both a hardware and compiler barrier.  The mb() barriers in the
routines using __ldcw() serve no purpose.

The only barrier needed is the one in arch_spin_unlock().  We need to ensure
all accesses are complete prior to releasing the lock.

Signed-off-by: John David Anglin <dave.anglin@bell.net>
Cc: stable@vger.kernel.org # 4.0+
Signed-off-by: Helge Deller <deller@gmx.de>
6 years agoparisc: Remove ordered stores from syscall.S
John David Anglin [Sun, 12 Aug 2018 20:38:03 +0000 (16:38 -0400)]
parisc: Remove ordered stores from syscall.S

Now that we use a sync prior to releasing the locks in syscall.S, we don't need
the PA 2.0 ordered stores used to release some locks.  Using an ordered store,
potentially slows the release and subsequent code.

There are a number of other ordered stores and loads that serve no purpose.  I
have converted these to normal stores.

Signed-off-by: John David Anglin <dave.anglin@bell.net>
Cc: stable@vger.kernel.org # 4.0+
Signed-off-by: Helge Deller <deller@gmx.de>
6 years agoparisc: prefer _THIS_IP_ and _RET_IP_ statement expressions
Nick Desaulniers [Wed, 1 Aug 2018 18:22:56 +0000 (11:22 -0700)]
parisc: prefer _THIS_IP_ and _RET_IP_ statement expressions

As part of the effort to reduce the code duplication between _THIS_IP_
and current_text_addr(), let's consolidate callers of
current_text_addr() to use _THIS_IP_.

Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Helge Deller <deller@gmx.de>
6 years agoparisc: Add HAVE_REGS_AND_STACK_ACCESS_API feature
Helge Deller [Thu, 28 Jun 2018 20:47:11 +0000 (22:47 +0200)]
parisc: Add HAVE_REGS_AND_STACK_ACCESS_API feature

Some parts of the HAVE_REGS_AND_STACK_ACCESS_API feature is needed for
the rseq syscall. This patch adds the most important parts, and as long
as we don't support kprobes, we should be fine.

Signed-off-by: Helge Deller <deller@gmx.de>
6 years agoparisc: Drop architecture-specific ENOTSUP define
Helge Deller [Sun, 1 Jul 2018 16:18:15 +0000 (18:18 +0200)]
parisc: Drop architecture-specific ENOTSUP define

parisc is the only Linux architecture which has defined a value for ENOTSUP.
All other architectures #define ENOTSUP as EOPNOTSUPP in their libc headers.

Having an own value for ENOTSUP which is different than EOPNOTSUPP often gives
problems with userspace programs which expect both to be the same.  One such
example is a build error in the libuv package, as can be seen in
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=900237.

Since we dropped HP-UX support, there is no real benefit in keeping an own
value for ENOTSUP. This patch drops the parisc value for ENOTSUP from the
kernel sources. glibc needs no patch, it reuses the exported headers.

Signed-off-by: Helge Deller <deller@gmx.de>
6 years agoparisc: use generic dma_noncoherent_ops
Christoph Hellwig [Tue, 19 Jun 2018 07:04:55 +0000 (09:04 +0200)]
parisc: use generic dma_noncoherent_ops

Switch to the generic noncoherent direct mapping implementation.

Fix sync_single_for_cpu to do skip the cache flush unless the transfer
is to the device to match the more tested unmap_single path which should
have the same cache coherency implications.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Helge Deller <deller@gmx.de>
6 years agoparisc: always use flush_kernel_dcache_range for DMA cache maintainance
Christoph Hellwig [Tue, 19 Jun 2018 07:04:54 +0000 (09:04 +0200)]
parisc: always use flush_kernel_dcache_range for DMA cache maintainance

Current the S/G list based DMA ops use flush_kernel_vmap_range which
contains a few UP optimizations, while the rest of the DMA operations
uses flush_kernel_dcache_range.  The single vs sg operations are supposed
to have the same effect, so they should use the same routines.  Use
the more conservation version for now, but if people more familiar with
parisc think the vmap version is generally fine for DMA we should switch
all interfaces over to it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Helge Deller <deller@gmx.de>
6 years agoparisc: merge pcx_dma_ops and pcxl_dma_ops
Christoph Hellwig [Tue, 19 Jun 2018 07:04:53 +0000 (09:04 +0200)]
parisc: merge pcx_dma_ops and pcxl_dma_ops

The only difference is that pcxl supports dma coherent allocations, while
pcx only supports non-consistent allocations and otherwise fails.

But dma_alloc* is not in the fast path, and merging these two allows an
easy migration path to the generic dma-noncoherent implementation, so
do it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Helge Deller <deller@gmx.de>
6 years agokconfig: fix the rule of mainmenu_stmt symbol
Masahiro Yamada [Thu, 9 Aug 2018 06:47:06 +0000 (15:47 +0900)]
kconfig: fix the rule of mainmenu_stmt symbol

The rule of mainmenu_stmt does not have debug print of zconf_lineno(),
but if it had, it would print a wrong line number for the same reason
as commit b2d00d7c61c8 ("kconfig: fix line numbers for if-entries in
menu tree").

The mainmenu_stmt does not need to eat following empty lines because
they are reduced to common_stmt.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
6 years agoMerge branch 'bpf-ancestor-cgroup-id'
Daniel Borkmann [Sun, 12 Aug 2018 23:02:40 +0000 (01:02 +0200)]
Merge branch 'bpf-ancestor-cgroup-id'

Andrey Ignatov says:

====================
This patch set adds new BPF helper bpf_skb_ancestor_cgroup_id that returns
id of cgroup v2 that is ancestor of cgroup associated with the skb at the
ancestor_level.

The helper is useful to implement policies in TC based on cgroups that are
upper in hierarchy than immediate cgroup associated with skb.

v1->v2:
- more reliable check for testing IPv6 to become ready in selftest.
====================

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
6 years agoselftests/bpf: Selftest for bpf_skb_ancestor_cgroup_id
Andrey Ignatov [Sun, 12 Aug 2018 17:49:30 +0000 (10:49 -0700)]
selftests/bpf: Selftest for bpf_skb_ancestor_cgroup_id

Add selftests for bpf_skb_ancestor_cgroup_id helper.

test_skb_cgroup_id.sh prepares testing interface and adds tc qdisc and
filter for it using BPF object compiled from test_skb_cgroup_id_kern.c
program.

BPF program in test_skb_cgroup_id_kern.c gets ancestor cgroup id using
the new helper at different levels of cgroup hierarchy that skb belongs
to, including root level and non-existing level, and saves it to the map
where the key is the level of corresponding cgroup and the value is its
id.

To trigger BPF program, user space program test_skb_cgroup_id_user is
run. It adds itself into testing cgroup and sends UDP datagram to
link-local multicast address of testing interface. Then it reads cgroup
ids saved in kernel for different levels from the BPF map and compares
them with those in user space. They must be equal for every level of
ancestry.

Example of run:
  # ./test_skb_cgroup_id.sh
  Wait for testing link-local IP to become available ... OK
  Note: 8 bytes struct bpf_elf_map fixup performed due to size mismatch!
  [PASS]

Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
6 years agoselftests/bpf: Add cgroup id helpers to bpf_helpers.h
Andrey Ignatov [Sun, 12 Aug 2018 17:49:29 +0000 (10:49 -0700)]
selftests/bpf: Add cgroup id helpers to bpf_helpers.h

Add bpf_skb_cgroup_id and bpf_skb_ancestor_cgroup_id helpers to
bpf_helpers.h to use them in tests and samples.

Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
6 years agobpf: Sync bpf.h to tools/
Andrey Ignatov [Sun, 12 Aug 2018 17:49:28 +0000 (10:49 -0700)]
bpf: Sync bpf.h to tools/

Sync skb_ancestor_cgroup_id() related bpf UAPI changes to tools/.

Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
6 years agobpf: Introduce bpf_skb_ancestor_cgroup_id helper
Andrey Ignatov [Sun, 12 Aug 2018 17:49:27 +0000 (10:49 -0700)]
bpf: Introduce bpf_skb_ancestor_cgroup_id helper

== Problem description ==

It's useful to be able to identify cgroup associated with skb in TC so
that a policy can be applied to this skb, and existing bpf_skb_cgroup_id
helper can help with this.

Though in real life cgroup hierarchy and hierarchy to apply a policy to
don't map 1:1.

It's often the case that there is a container and corresponding cgroup,
but there are many more sub-cgroups inside container, e.g. because it's
delegated to containerized application to control resources for its
subsystems, or to separate application inside container from infra that
belongs to containerization system (e.g. sshd).

At the same time it may be useful to apply a policy to container as a
whole.

If multiple containers like this are run on a host (what is often the
case) and many of them have sub-cgroups, it may not be possible to apply
per-container policy in TC with existing helpers such as
bpf_skb_under_cgroup or bpf_skb_cgroup_id:

* bpf_skb_cgroup_id will return id of immediate cgroup associated with
  skb, i.e. if it's a sub-cgroup inside container, it can't be used to
  identify container's cgroup;

* bpf_skb_under_cgroup can work only with one cgroup and doesn't scale,
  i.e. if there are N containers on a host and a policy has to be
  applied to M of them (0 <= M <= N), it'd require M calls to
  bpf_skb_under_cgroup, and, if M changes, it'd require to rebuild &
  load new BPF program.

== Solution ==

The patch introduces new helper bpf_skb_ancestor_cgroup_id that can be
used to get id of cgroup v2 that is an ancestor of cgroup associated
with skb at specified level of cgroup hierarchy.

That way admin can place all containers on one level of cgroup hierarchy
(what is a good practice in general and already used in many
configurations) and identify specific cgroup on this level no matter
what sub-cgroup skb is associated with.

E.g. if there is a cgroup hierarchy:
  root/
  root/container1/
  root/container1/app11/
  root/container1/app11/sub-app-a/
  root/container1/app12/
  root/container2/
  root/container2/app21/
  root/container2/app22/
  root/container2/app22/sub-app-b/

, then having skb associated with root/container1/app11/sub-app-a/ it's
possible to get ancestor at level 1, what is container1 and apply policy
for this container, or apply another policy if it's container2.

Policies can be kept e.g. in a hash map where key is a container cgroup
id and value is an action.

Levels where container cgroups are created are usually known in advance
whether cgroup hierarchy inside container may be hard to predict
especially in case when its creation is delegated to containerized
application.

== Implementation details ==

The helper gets ancestor by walking parents up to specified level.

Another option would be to get different kind of "id" from
cgroup->ancestor_ids[level] and use it with idr_find() to get struct
cgroup for ancestor. But that would require radix lookup what doesn't
seem to be better (at least it's not obviously better).

Format of return value of the new helper is same as that of
bpf_skb_cgroup_id.

Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
6 years agobpf: decouple btf from seq bpf fs dump and enable more maps
Daniel Borkmann [Sat, 11 Aug 2018 23:59:17 +0000 (01:59 +0200)]
bpf: decouple btf from seq bpf fs dump and enable more maps

Commit a26ca7c982cb ("bpf: btf: Add pretty print support to
the basic arraymap") and 699c86d6ec21 ("bpf: btf: add pretty
print for hash/lru_hash maps") enabled support for BTF and
dumping via BPF fs for array and hash/lru map. However, both
can be decoupled from each other such that regular BPF maps
can be supported for attaching BTF key/value information,
while not all maps necessarily need to dump via map_seq_show_elem()
callback.

The basic sanity check which is a prerequisite for all maps
is that key/value size has to match in any case, and some maps
can have extra checks via map_check_btf() callback, e.g.
probing certain types or indicating no support in general. With
that we can also enable retrieving BTF info for per-cpu map
types and lpm.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Yonghong Song <yhs@fb.com>
6 years agoLinux 4.18
Linus Torvalds [Sun, 12 Aug 2018 20:41:04 +0000 (13:41 -0700)]
Linux 4.18

6 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sun, 12 Aug 2018 19:52:05 +0000 (12:52 -0700)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "Eight fixes.

  The most important one is the mpt3sas fix which makes the driver work
  again on big endian systems. The rest are mostly minor error path or
  checker issues and the vmw_scsi one fixes a performance problem"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: vmw_pvscsi: Return DID_RESET for status SAM_STAT_COMMAND_TERMINATED
  scsi: sr: Avoid that opening a CD-ROM hangs with runtime power management enabled
  scsi: mpt3sas: Swap I/O memory read value back to cpu endianness
  scsi: fcoe: clear FC_RP_STARTED flags when receiving a LOGO
  scsi: fcoe: drop frames in ELS LOGO error path
  scsi: fcoe: fix use-after-free in fcoe_ctlr_els_send
  scsi: qedi: Fix a potential buffer overflow
  scsi: qla2xxx: Fix memory leak for allocating abort IOCB

6 years agoinit: rename and re-order boot_cpu_state_init()
Linus Torvalds [Sun, 12 Aug 2018 19:19:42 +0000 (12:19 -0700)]
init: rename and re-order boot_cpu_state_init()

This is purely a preparatory patch for upcoming changes during the 4.19
merge window.

We have a function called "boot_cpu_state_init()" that isn't really
about the bootup cpu state: that is done much earlier by the similarly
named "boot_cpu_init()" (note lack of "state" in name).

This function initializes some hotplug CPU state, and needs to run after
the percpu data has been properly initialized.  It even has a comment to
that effect.

Except it _doesn't_ actually run after the percpu data has been properly
initialized.  On x86 it happens to do that, but on at least arm and
arm64, the percpu base pointers are initialized by the arch-specific
'smp_prepare_boot_cpu()' hook, which ran _after_ boot_cpu_state_init().

This had some unexpected results, and in particular we have a patch
pending for the merge window that did the obvious cleanup of using
'this_cpu_write()' in the cpu hotplug init code:

  -       per_cpu_ptr(&cpuhp_state, smp_processor_id())->state = CPUHP_ONLINE;
  +       this_cpu_write(cpuhp_state.state, CPUHP_ONLINE);

which is obviously the right thing to do.  Except because of the
ordering issue, it actually failed miserably and unexpectedly on arm64.

So this just fixes the ordering, and changes the name of the function to
be 'boot_cpu_hotplug_init()' to make it obvious that it's about cpu
hotplug state, because the core CPU state was supposed to have already
been done earlier.

Marked for stable, since the (not yet merged) patch that will show this
problem is marked for stable.

Reported-by: Vlastimil Babka <vbabka@suse.cz>
Reported-by: Mian Yousaf Kaukab <yousaf.kaukab@suse.com>
Suggested-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
6 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sun, 12 Aug 2018 18:21:17 +0000 (11:21 -0700)]
Merge branch 'fixes' of git://git./linux/kernel/git/viro/vfs

Pull vfs fixes from Al Viro:
 "A bunch of race fixes, mostly around lazy pathwalk.

  All of it is -stable fodder, a large part going back to 2013"

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  make sure that __dentry_kill() always invalidates d_seq, unhashed or not
  fix __legitimize_mnt()/mntput() race
  fix mntput/mntput race
  root dentries need RCU-delayed freeing

6 years agoxfs: fix a null pointer dereference in xfs_bmap_extents_to_btree
Shan Hai [Sat, 11 Aug 2018 00:55:55 +0000 (17:55 -0700)]
xfs: fix a null pointer dereference in xfs_bmap_extents_to_btree

Fuzzing tool reports a write to null pointer error in the
xfs_bmap_extents_to_btree, fix it by bailing out on encountering
a null pointer.

Signed-off-by: Shan Hai <shan.hai@oracle.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
6 years agoxfs: remove b_last_holder & associated macros
Eric Sandeen [Fri, 10 Aug 2018 20:56:25 +0000 (13:56 -0700)]
xfs: remove b_last_holder & associated macros

The old lock tracking infrastructure in xfs using the b_last_holder
field seems to only be useful if you can get into the system with a
debugger; it seems that the existing tracepoints would be the way to
go these days, and this old infrastructure can be removed.

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>