[PATCH v5 12/69] KVM: arm64: nv: Handle HCR_EL2.NV system register traps

[Marc Zyngier]

On Tue, 18 Jan 2022 15:51:30 +0000,
"Russell King (Oracle)" <linux at armlinux.org.uk> wrote:
> 
> On Mon, Nov 29, 2021 at 08:00:53PM +0000, Marc Zyngier wrote:
> > From: Jintack Lim <jintack.lim at linaro.org>
> > 
> > ARM v8.3 introduces a new bit in the HCR_EL2, which is the NV bit. When
> > this bit is set, accessing EL2 registers in EL1 traps to EL2. In
> > addition, executing the following instructions in EL1 will trap to EL2:
> > tlbi, at, eret, and msr/mrs instructions to access SP_EL1. Most of the
> > instructions that trap to EL2 with the NV bit were undef at EL1 prior to
> > ARM v8.3. The only instruction that was not undef is eret.
> > 
> > This patch sets up a handler for EL2 registers and SP_EL1 register
> > accesses at EL1. The host hypervisor keeps those register values in
> > memory, and will emulate their behavior.
> > 
> > This patch doesn't set the NV bit yet. It will be set in a later patch
> > once nested virtualization support is completed.
> > 
> > Signed-off-by: Jintack Lim <jintack.lim at linaro.org>
> > [maz: added SCTLR_EL2 RES0/RES1 handling]
> > Signed-off-by: Marc Zyngier <maz at kernel.org>
> > ---
> ...
> > @@ -1825,9 +1882,51 @@ static const struct sys_reg_desc sys_reg_descs[] = {
> >  	{ PMU_SYS_REG(SYS_PMCCFILTR_EL0), .access = access_pmu_evtyper,
> >  	  .reset = reset_val, .reg = PMCCFILTR_EL0, .val = 0 },
> >  
> > +	{ SYS_DESC(SYS_VPIDR_EL2), access_rw, reset_val, VPIDR_EL2, 0 },
> > +	{ SYS_DESC(SYS_VMPIDR_EL2), access_rw, reset_val, VMPIDR_EL2, 0 },
> > +
> > +	{ SYS_DESC(SYS_SCTLR_EL2), access_sctlr_el2, reset_val, SCTLR_EL2, SCTLR_EL2_RES1 },
> > +	{ SYS_DESC(SYS_ACTLR_EL2), access_rw, reset_val, ACTLR_EL2, 0 },
> > +	{ SYS_DESC(SYS_HCR_EL2), access_rw, reset_val, HCR_EL2, 0 },
> > +	{ SYS_DESC(SYS_MDCR_EL2), access_rw, reset_val, MDCR_EL2, 0 },
> > +	{ SYS_DESC(SYS_CPTR_EL2), access_rw, reset_val, CPTR_EL2, CPTR_EL2_DEFAULT },
> > +	{ SYS_DESC(SYS_HSTR_EL2), access_rw, reset_val, HSTR_EL2, 0 },
> > +	{ SYS_DESC(SYS_HACR_EL2), access_rw, reset_val, HACR_EL2, 0 },
> > +
> > +	{ SYS_DESC(SYS_TTBR0_EL2), access_rw, reset_val, TTBR0_EL2, 0 },
> > +	{ SYS_DESC(SYS_TTBR1_EL2), access_rw, reset_val, TTBR1_EL2, 0 },
> > +	{ SYS_DESC(SYS_TCR_EL2), access_rw, reset_val, TCR_EL2, TCR_EL2_RES1 },
> > +	{ SYS_DESC(SYS_VTTBR_EL2), access_rw, reset_val, VTTBR_EL2, 0 },
> > +	{ SYS_DESC(SYS_VTCR_EL2), access_rw, reset_val, VTCR_EL2, 0 },
> > +
> >  	{ SYS_DESC(SYS_DACR32_EL2), NULL, reset_unknown, DACR32_EL2 },
> > +	{ SYS_DESC(SYS_SPSR_EL2), access_rw, reset_val, SPSR_EL2, 0 },
> > +	{ SYS_DESC(SYS_ELR_EL2), access_rw, reset_val, ELR_EL2, 0 },
> > +	{ SYS_DESC(SYS_SP_EL1), access_sp_el1},
> > +
> >  	{ SYS_DESC(SYS_IFSR32_EL2), NULL, reset_unknown, IFSR32_EL2 },
> > +	{ SYS_DESC(SYS_AFSR0_EL2), access_rw, reset_val, AFSR0_EL2, 0 },
> > +	{ SYS_DESC(SYS_AFSR1_EL2), access_rw, reset_val, AFSR1_EL2, 0 },
> > +	{ SYS_DESC(SYS_ESR_EL2), access_rw, reset_val, ESR_EL2, 0 },
> >  	{ SYS_DESC(SYS_FPEXC32_EL2), NULL, reset_val, FPEXC32_EL2, 0x700 },
> > +
> > +	{ SYS_DESC(SYS_FAR_EL2), access_rw, reset_val, FAR_EL2, 0 },
> > +	{ SYS_DESC(SYS_HPFAR_EL2), access_rw, reset_val, HPFAR_EL2, 0 },
> > +
> > +	{ SYS_DESC(SYS_MAIR_EL2), access_rw, reset_val, MAIR_EL2, 0 },
> > +	{ SYS_DESC(SYS_AMAIR_EL2), access_rw, reset_val, AMAIR_EL2, 0 },
> > +
> > +	{ SYS_DESC(SYS_VBAR_EL2), access_rw, reset_val, VBAR_EL2, 0 },
> > +	{ SYS_DESC(SYS_RVBAR_EL2), access_rw, reset_val, RVBAR_EL2, 0 },
> > +	{ SYS_DESC(SYS_RMR_EL2), trap_undef },
> > +
> > +	{ SYS_DESC(SYS_CONTEXTIDR_EL2), access_rw, reset_val, CONTEXTIDR_EL2, 0 },
> > +	{ SYS_DESC(SYS_TPIDR_EL2), access_rw, reset_val, TPIDR_EL2, 0 },
> > +
> > +	{ SYS_DESC(SYS_CNTVOFF_EL2), access_rw, reset_val, CNTVOFF_EL2, 0 },
> > +	{ SYS_DESC(SYS_CNTHCTL_EL2), access_rw, reset_val, CNTHCTL_EL2, 0 },
> > +
> > +	{ SYS_DESC(SYS_SP_EL2), NULL, reset_unknown, SP_EL2 },
> 
> Doesn't this have an effect on the ability to migrate guests between
> identical hardware but running kernels with vs without this patch?
> From what I remember, QEMU fails a migration if the migration target
> has less system registers than the migration source.
> 
> If so, this should at the very least be spelt out in the commit
> message - it's a user experience breaking change. Maybe also preventing
> the exposure of these when NV is disabled would be a good idea?

Yes, that's a known issue. I have taken steps to avoid exposing these
registers (the get-reg-list test now passes on a !NV system, while it
didn't before).

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.