mbox series

[v3,0/4] jump_label: Fix __ro_after_init keys for modules & annotate some keys

Message ID 20240313180106.2917308-1-vschneid@redhat.com (mailing list archive)
Headers show
Series jump_label: Fix __ro_after_init keys for modules & annotate some keys | expand

Message

Valentin Schneider March 13, 2024, 6:01 p.m. UTC
Hi folks,

This series fixes __ro_after_init keys used in modules (courtesy of PeterZ) and
flags more keys as __ro_after_init.

Compile & boot tested for x86_64_defconfig and i386_defconfig.

@Peter, regarding making __use_tsc x86_32-only, I hit a few snags:

Currently, for the static key to be enabled, we (mostly) need:
o X86_FEATURE_TSC is in CPUID
o determine_cpu_tsc_frequencies passes

All X86_64 systems have a TSC, so the CPUID feature is a given there.

Calibrating the TSC can end up depending on different things:
o CPUID accepting 0x16 as eax input (cf. cpu_khz_from_cpuid())
o MSR_FSB_FREQ being available (cf. cpu_khz_from_msr())
o pit_hpet_ptimer_calibrate_cpu() doesn't mess up

I couldn't find any guarantees for X86_64 on having the processor frequency
information CPUID leaf, nor for the FSB_FREQ MSR (both tsc_msr_cpu_ids and
the SDM seem to point at only a handful of models).

pit_hpet_ptimer_calibrate_cpu() relies on having either HPET or the ACPI PM
timer, the latter being widely available, though X86_PM_TIMER can be
disabled via EXPERT.

The question here is: are there any guarantees that at least one of these
can be relied upon for x86_64?

And with all of that, there is still the "apicpmtimer" cmdline option which
currently invokes notsc_setup() on x86_64. The justification I found for it was
in 0c3749c41f5e ("[PATCH] x86_64: Calibrate APIC timer using PM timer"):

  """
  On some broken motherboards (at least one NForce3 based AMD64 laptop)
  the PIT timer runs at a incorrect frequency.  This patch adds a new
  option "apicpmtimer" that allows to use the APIC timer and calibrate it
  using the PMTimer.
  """


Revisions
=========

v2 -> v3
++++++++

o Rebased against latest upstream
o Fixed up jump_label_del_module() error in first patch reported by kernel test
  robot
  http://lore.kernel.org/r/202402191313.3276317d-oliver.sang@intel.com
o Removed mds_user_clear patch; key was switched to an ALTERNATIVE by
  6613d82e617d ("x86/bugs: Use ALTERNATIVE() instead of mds_user_clear static key")


v1 -> v2
++++++++

o Collected tags (Josh, Sean)
o Fixed CONFIG_JUMP_LABEL=n compile fail (lkp)

Cheers,
Valentin

Peter Zijlstra (1):
  jump_label,module: Don't alloc static_key_mod for __ro_after_init keys

Valentin Schneider (3):
  context_tracking: Make context_tracking_key __ro_after_init
  x86/kvm: Make kvm_async_pf_enabled __ro_after_init
  x86/tsc: Make __use_tsc __ro_after_init

 arch/x86/kernel/kvm.c          |  2 +-
 arch/x86/kernel/tsc.c          |  2 +-
 include/asm-generic/sections.h |  5 ++++
 include/linux/jump_label.h     |  3 ++
 init/main.c                    |  1 +
 kernel/context_tracking.c      |  2 +-
 kernel/jump_label.c            | 53 ++++++++++++++++++++++++++++++++++
 7 files changed, 65 insertions(+), 3 deletions(-)

--
2.43.0