Re: [PATCH 1/3] KVM: arm64: Stop writing aarch32's CSSELR into ACTLR

[Marc Zyngier <maz@kernel.org>]

Hi James,

On 2020-05-26 17:18, James Morse wrote:
> aarch32 has pairs of registers to access the high and low parts of 
> 64bit
> registers. KVM has a union of 64bit sys_regs[] and 32bit copro[]. The
> 32bit accessors read the high or low part of the 64bit sys_reg[] value
> through the union.
> 
> Both sys_reg_descs[] and cp15_regs[] list access_csselr() as the 
> accessor
> for CSSELR{,_EL1}. access_csselr() is only aware of the 64bit 
> sys_regs[],
> and expects r->reg to be 'CSSELR_EL1' in the enum, index 2 of the 64bit
> array.
> 
> cp15_regs[] uses the 32bit copro[] alias of sys_regs[]. Here CSSELR is
> c0_CSSELR which is the same location in sys_reg[]. r->reg is 
> 'c0_CSSELR',
> index 4 in the 32bit array.
> 
> access_csselr() uses the 32bit r->reg value to access the 64bit array,
> so reads and write the wrong value. sys_regs[4], is ACTLR_EL1, which
> is subsequently save/restored when we enter the guest.

Huhuh... Nice catch.

> 
> ACTLR_EL1 is supposed to be read-only for the guest. This register
> only affects execution at EL1, and the host's value is restored before
> we return to host EL1.
> 
> Rename access_csselr() to access_csselr_el1(), to indicate it expects
> the 64bit register index, and pass it CSSELR_EL1 from cp15_regs[].
> 
> Cc: stable@vger.kernel.org
> Signed-off-by: James Morse <james.morse@arm.com>
> ----
> Providing access_csselr_cp15() wouldn't work as with VHE CSSELR_EL1 is
> loaded on the CPU while this code runs. access_csselr_cp15() would have
> to map it back the 64bit resgister to use vcpu_write_sys_reg(). We may
> as well do it in the table.
> 
>  arch/arm64/kvm/sys_regs.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> index 51db934702b6..2eda539f3281 100644
> --- a/arch/arm64/kvm/sys_regs.c
> +++ b/arch/arm64/kvm/sys_regs.c
> @@ -1302,7 +1302,7 @@ static bool access_clidr(struct kvm_vcpu *vcpu,
> struct sys_reg_params *p,
>  	return true;
>  }
> 
> -static bool access_csselr(struct kvm_vcpu *vcpu, struct sys_reg_params 
> *p,
> +static bool access_csselr_el1(struct kvm_vcpu *vcpu, struct 
> sys_reg_params *p,
>  			  const struct sys_reg_desc *r)
>  {
>  	if (p->is_write)
> @@ -1566,7 +1566,7 @@ static const struct sys_reg_desc sys_reg_descs[] 
> = {
> 
>  	{ SYS_DESC(SYS_CCSIDR_EL1), access_ccsidr },
>  	{ SYS_DESC(SYS_CLIDR_EL1), access_clidr },
> -	{ SYS_DESC(SYS_CSSELR_EL1), access_csselr, reset_unknown, CSSELR_EL1 
> },
> +	{ SYS_DESC(SYS_CSSELR_EL1), access_csselr_el1, reset_unknown, 
> CSSELR_EL1 },
>  	{ SYS_DESC(SYS_CTR_EL0), access_ctr },
> 
>  	{ SYS_DESC(SYS_PMCR_EL0), access_pmcr, reset_pmcr, PMCR_EL0 },
> @@ -2060,7 +2060,7 @@ static const struct sys_reg_desc cp15_regs[] = {
> 
>  	{ Op1(1), CRn( 0), CRm( 0), Op2(0), access_ccsidr },
>  	{ Op1(1), CRn( 0), CRm( 0), Op2(1), access_clidr },
> -	{ Op1(2), CRn( 0), CRm( 0), Op2(0), access_csselr, NULL, c0_CSSELR },
> +	{ Op1(2), CRn( 0), CRm( 0), Op2(0), access_csselr_el1, NULL, 
> CSSELR_EL1 },
>  };
> 
>  static const struct sys_reg_desc cp15_64_regs[] = {

This is a departure from the way we deal with 32bit CP15 registers.
We deal with this exact issue in a very different way for other
CP15 regs, by adjusting the index in the sys_regs array (see the
way we handle the VM regs).

How about something like this (untested):

diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index 12d07e7ced82..515c0c11a668 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -1321,10 +1321,16 @@ static bool access_clidr(struct kvm_vcpu *vcpu, 
struct sys_reg_params *p,
  static bool access_csselr(struct kvm_vcpu *vcpu, struct sys_reg_params 
*p,
  			  const struct sys_reg_desc *r)
  {
+	int reg = r->reg;
+
+	/* See the 32bit mapping in kvm_host.h */
+	if (p->is_aarch32)
+		reg = r->reg / 2;
+
  	if (p->is_write)
-		vcpu_write_sys_reg(vcpu, p->regval, r->reg);
+		vcpu_write_sys_reg(vcpu, p->regval, reg);
  	else
-		p->regval = vcpu_read_sys_reg(vcpu, r->reg);
+		p->regval = vcpu_read_sys_reg(vcpu, reg);
  	return true;
  }

Ideally, I'd like the core sys_reg code to deal with this sort
of funnies, but I'm trying to keep the change minimal...

Thanks,

          M.
-- 
Jazz is not dead. It just smells funny...

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel