Tags

Tags give the ability to mark specific points in history as being important

  • x86-build-2024-01-08

    bcf7ef56 · x86/tools: objdump_reformat.awk: Skip bad instructions from llvm-objdump · 
    x86/build changes for v6.8:

 - Update the objdump & instruction decoder self-test code for
   better LLVM toolchain compatibility

 - Rework CONFIG_X86_PAE dependencies, for better readability
   and higher robustness.

 - Misc cleanups

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • x86-boot-2024-01-08

    257ca14f · x86/boot: Remove redundant initialization of the 'delta' variable in strcmp() · 
    x86/boot changes for v6.8:

 - Ignore NMIs during very early boot, to address kexec crashes

 - Remove redundant initialization in boot/string.c's strcmp()

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • x86-asm-2024-01-08

    bc90aefa · x86/asm: Add DB flag to 32-bit percpu GDT entry · 
    x86/asm changes for v6.8:

 - Replace magic numbers in GDT descriptor definitions & handling:

   - Introduce symbolic names via macros for descriptor types/fields/flags,
     and then use these symbolic names.

   - Clean up definitions a bit, such as GDT_ENTRY_INIT()

   - Fix/clean up details that became visibly inconsistent after the
     symbol-based code was introduced:

      - Unify accessed flag handling

      - Set the D/B size flag consistently & according to the HW specification

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • x86-apic-2024-01-08

    5e1c8a47 · x86/ioapic: Remove unfinished sentence from comment · 
    x86/apic changes for v6.8:

 - Clean up 'struct apic':

    - Drop ::delivery_mode
    - Drop 'enum apic_delivery_modes'
    - Drop 'struct local_apic'

 - Fix comments

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • v5.4.266

    4410df70 · Linux 5.4.266 · 
    This is the 5.4.266 stable release

  • v4.19.304

    88afb447 · Linux 4.19.304 · 
    This is the 4.19.304 stable release

  • v4.14.335

    11fc60f0 · Linux 4.14.335 · 
    This is the 4.14.335 stable release

  • timers-core-2024-01-08

    da65f29d · timers: Fix nextevt calculation when no timers are pending · 
    Timer subsystem changes for v6.8:

 - Various preparatory cleanups & enhancements of the timer-wheel code,
   in preparation for the WIP 'pull timers at expiry' timer migration model
   series (which will replace the current 'push timers at enqueue' migration
   model), by Anna-Maria Behnsen:

      - Update comments and clean up confusing variable names

      - Add debug check to warn about time travel

      - Improve/expand timer-wheel tracepoints

      - Optimize away unnecessary IPIs for deferrable timers

      - Restructure & clean up next_expiry_recalc()

      - Clean up forward_timer_base()

      - Introduce __forward_timer_base() and use it to simplify
        and micro-optimize get_next_timer_interrupt()

 - Restructure the get_next_timer_interrupt()'s idle logic
   for better readability and to enable a minor optimization.

 - Fix the nextevt calculation when no timers are pending

 - Fix the sysfs_get_uname() prototype declaration

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • smp-core-2024-01-08

    fe22944c · cpu/hotplug: Increase the number of dynamic states · 
    CPU hotplug changes for v6.8:

 - Remove unused CPU hotplug states

 - Increase the number of dynamic CPU hotplug states
   from 30 to 40, because existing drivers can exhaust
   the allocation space

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • irq-core-2024-01-08

    69ffab9b · irqchip/irq-xtensa-pic: Clean up · 
    IRQ subsystem changes for v6.8:

  Drivers:

   - Add support for the IA55 interrupt controller on RZ/G3S SoC's

   - Update/fix the Qualcom MPM Interrupt Controller driver's
     register enumeration within the somewhat exotic "RPM Message RAM"
     MMIO-mapped shared memory region that is used for other purposes
     as well.

   - Clean up the Xtensa built-in Programmable Interrupt Controller
     driver (xtensa-pic) a bit.

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • core-entry-2024-01-08

    221a1640 · entry: Move syscall_enter_from_user_mode() to header file · 
    Move various entry functions from kernel/entry/common.c to <linux/entry-common.h>,
and always-inline them, to improve syscall entry performance on s390 by ~11%.

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • core-debugobjects-2024-01-08

    9bb63626 · debugobjects: Stop accessing objects after releasing hash bucket lock · 
    Debugobject changes for v6.8:

 - Make tracking object use more robust: it's not safe to access a
   tracking object after releasing the hashbucket lock. Create a
   persistent copy for debug printouts instead.

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • objtool-core-2024-01-08

    e2e13630 · objtool: Fix calloc call for new -Walloc-size · 
    Address a GCC-14 warning: there's no real bug, but indeed the calloc order doesn't match
the prototype.

(Side note: we should really add zalloc() for such cases.)

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • locking-core-2024-01-08

    2b9d9e0a · locking/mutex: Clarify that mutex_unlock(), and most other sleeping locks, can... · 
    Locking changes for v6.8:

- lock guards:

   - Use lock guards in the ptrace code

   - Introduce conditional guards to extend to conditional lock
     primitives like mutex_trylock()/mutex_lock_interruptible()/etc.

- lockdep:

   - Optimize 'struct lock_class' to be smaller

   - Update file patterns in MAINTAINERS

- mutexes: Document mutex lifetime rules a bit more

Signed-off-by: Ingo Molnar <mingo@kernel.org>

  • rust-6.8

    711cbfc7 · docs: rust: Clarify that 'rustup override' applies to build directory · 
    Rust changes for v6.8

Another routine one in terms of features. In terms of lines, this time
the 'alloc' version upgrade is less prominent, given that it was fairly
small (and we did not have two upgrades).

Toolchain and infrastructure:

 - Upgrade to Rust 1.74.1.

   The patch release includes a fix for an ICE that the Apple AGX GPU
   driver was hitting.

 - Support 'srctree'-relative links in Rust code documentation.

 - Automate part of the manual constants handling (i.e. the ones not
   recognised by 'bindgen').

 - Suppress searching builtin sysroot to avoid confusion with installed
   sysroots, needed for the to-be-merged arm64 support which uses
   a builtin target.

 - Ignore '__preserve_most' functions for 'bindgen'.

 - Reduce header inclusion bloat in exports.

'kernel' crate:

 - Implement 'Debug' for 'CString'.

 - Make 'CondVar::wait()' an uninterruptible wait.

'macros' crate:

 - Update 'paste!' to accept string literals.

 - Improve '#[vtable]' documentation.

Documentation:

 - Add testing section (KUnit and 'rusttest' target).

 - Remove 'CC=clang' mentions.

 - Clarify that 'rustup override' applies to build directory.

  • x86_sev_for_v6.8

    d642ef71 · virt: sev-guest: Convert to platform remove callback returning void · 
    - Convert the sev-guest plaform ->remove callback to return void

- Move the SEV C-bit verification to the BSP as it needs to happen only
  once and not on every AP

  • x86_paravirt_for_v6.8

    7991ed43 · x86/alternative: Correct feature bit debug output · 
    - Replace the paravirt patching functionality using the alternatives
  infrastructure and remove the former

- Misc other improvements

  • v6.7

    0dd3ee31 · Linux 6.7 · 
    Linux 6.7

  • ras_core_for_v6.8

    1f68ce2a · x86/mce: Handle Intel threshold interrupt storms · 
    - Convert the hw error storm handling into a finer-grained, per-bank
  solution which allows for more timely detection and reporting of
  errors

- Start a documentation section which will hold down relevant
  RAS features description and how they should be used

- Add new AMD error bank types

- Slim down and remove error type descriptions from the kernel side of
  error decoding to rasdaemon which can be used from now on to decode
  hw errors on AMD

- Mark pages containing uncorrectable errors as poison so that kdump can
  avoid them and thus not cause another panic

- The usual cleanups and fixlets

  • x86_cpu_for_v6.8

    232afb55 · x86/CPU/AMD: Add X86_FEATURE_ZEN1 · 
    - Add synthetic X86_FEATURE flags for the different AMD Zen generations
  and use them everywhere instead of ad-hoc family/model checks. Drop an
  ancient AMD errata checking facility as a result

- Fix a fragile initcall ordering in intel_epb

- Do not issue the MFENCE+LFENCE barrier for the TSC deadline and X2APIC
  MSRs on AMD as it is not needed there