[v2,0/3] KVM/arm64: timer fixes for 6.14

Message

Marc Zyngier Feb. 4, 2025, 11 a.m. UTC

It's been recently pointed out that the NV timer code that is
(hopefully) on its way upstream is marginally sub-optimal (read:
terminally broken). This short series attempts to fix things:

- Correctly arm a background timer in all situations

- Avoid corrupting the EL2 state by shoving the EL1 state into it, and
  simplify the handling of emulated timers

- Drop the weird HV timer offset dance dance on E2H switch that never
  happens and make HV timer correctly UNDEF on E2H==0.

Thanks to Volodymyr Babchuk and Wei-Lin Chang for their help.

Patches on top of kvmarm/fixes, and merged back on top of my
kvm-arm64/nv-next branch.

Marc Zyngier (3):
  KVM: arm64: timer: Always evaluate the need for a soft timer
  KVM: arm64: timer: Correctly handle EL1 timer emulation when !FEAT_ECV
  KVM: arm64: timer: Don't adjust the EL2 virtual timer offset

 arch/arm64/kvm/arch_timer.c | 49 +++++++++----------------------------
 arch/arm64/kvm/sys_regs.c   | 16 +++++++++---
 2 files changed, 24 insertions(+), 41 deletions(-)

Comments

Oliver Upton Feb. 4, 2025, 3:08 p.m. UTC | #1

On Tue, Feb 04, 2025 at 11:00:47AM +0000, Marc Zyngier wrote:
> It's been recently pointed out that the NV timer code that is
> (hopefully) on its way upstream is marginally sub-optimal (read:
> terminally broken). This short series attempts to fix things:
> 
> - Correctly arm a background timer in all situations
> 
> - Avoid corrupting the EL2 state by shoving the EL1 state into it, and
>   simplify the handling of emulated timers
> 
> - Drop the weird HV timer offset dance dance on E2H switch that never
>   happens and make HV timer correctly UNDEF on E2H==0.
> 
> Thanks to Volodymyr Babchuk and Wei-Lin Chang for their help.
> 
> Patches on top of kvmarm/fixes, and merged back on top of my
> kvm-arm64/nv-next branch.

Reviewed-by: Oliver Upton <[e-mail]>

Marc Zyngier Feb. 4, 2025, 3:12 p.m. UTC | #2

On Tue, 04 Feb 2025 11:00:47 +0000, Marc Zyngier wrote:
> It's been recently pointed out that the NV timer code that is
> (hopefully) on its way upstream is marginally sub-optimal (read:
> terminally broken). This short series attempts to fix things:
> 
> - Correctly arm a background timer in all situations
> 
> - Avoid corrupting the EL2 state by shoving the EL1 state into it, and
>   simplify the handling of emulated timers
> 
> [...]

Applied to fixes, thanks!

[1/3] KVM: arm64: timer: Always evaluate the need for a soft timer
      commit: b450dcce93bc2cf6d2bfaf5a0de88a94ebad8f89
[2/3] KVM: arm64: timer: Correctly handle EL1 timer emulation when !FEAT_ECV
      commit: 1b8705ad5365b5333240b46d5cd24e88ef2ddb14
[3/3] KVM: arm64: timer: Don't adjust the EL2 virtual timer offset
      commit: 0e459810285503fb354537e84049e212c5917c33

Cheers,

	M.