diff mbox series

[RFC,04/25] KVM: arm64: Introduce struct id_reg_info

Message ID 20211012043535.500493-5-reijiw@google.com (mailing list archive)
State New, archived
Headers show
Series KVM: arm64: Make CPU ID registers writable by userspace | expand

Commit Message

Reiji Watanabe Oct. 12, 2021, 4:35 a.m. UTC
This patch lays the groundwork to make ID registers writable.

Introduce struct id_reg_info for an ID register to manage the
register specific control of its value for the guest.
It is used to do register specific initialization, validation
of the ID register and etc.  Not all ID registers must have
the id_reg_info. ID registers that don't have the id_reg_info
are handled in a common way that is applied to all other
ID registers.

At present, changing an ID register from userspace is allowed only
if the ID register has the id_reg_info, but that will be changed
by the following patches.

No ID register has the structure yet and the following patches
will add the id_reg_info for some ID registers.

Signed-off-by: Reiji Watanabe <reijiw@google.com>
---
 arch/arm64/kvm/sys_regs.c | 120 +++++++++++++++++++++++++++++++++++---
 1 file changed, 111 insertions(+), 9 deletions(-)

Comments

Andrew Jones Oct. 15, 2021, 1:47 p.m. UTC | #1
On Mon, Oct 11, 2021 at 09:35:14PM -0700, Reiji Watanabe wrote:
> This patch lays the groundwork to make ID registers writable.
> 
> Introduce struct id_reg_info for an ID register to manage the
> register specific control of its value for the guest.
> It is used to do register specific initialization, validation
> of the ID register and etc.  Not all ID registers must have
> the id_reg_info. ID registers that don't have the id_reg_info
> are handled in a common way that is applied to all other
> ID registers.
> 
> At present, changing an ID register from userspace is allowed only
> if the ID register has the id_reg_info, but that will be changed
> by the following patches.
> 
> No ID register has the structure yet and the following patches
> will add the id_reg_info for some ID registers.
> 
> Signed-off-by: Reiji Watanabe <reijiw@google.com>
> ---
>  arch/arm64/kvm/sys_regs.c | 120 +++++++++++++++++++++++++++++++++++---
>  1 file changed, 111 insertions(+), 9 deletions(-)
> 
> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> index 72ca518e7944..8a0b88f9a975 100644
> --- a/arch/arm64/kvm/sys_regs.c
> +++ b/arch/arm64/kvm/sys_regs.c
> @@ -263,6 +263,76 @@ static bool trap_raz_wi(struct kvm_vcpu *vcpu,
>  		return read_zero(vcpu, p);
>  }
>  
> +struct id_reg_info {
> +	u32	sys_reg;	/* Register ID */
> +	u64	sys_val;	/* Sanitized system value */
> +
> +	/*
> +	 * Limit value of the register for a vcpu. The value is sys_val
> +	 * with bits cleared for unsupported features for the guest.
> +	 */
> +	u64	vcpu_limit_val;

Maybe I'll see a need for both later, but at the moment I'd think we only
need sys_val with the bits cleared for disabled features.

> +
> +	/* Bits that are not validated when userspace sets the register. */
> +	u64	ignore_mask;
> +
> +	/* Initialization function of the id_reg_info */
> +	void (*init)(struct id_reg_info *id_reg);
> +
> +	/* Register specific validation function */
> +	int (*validate)(struct kvm_vcpu *vcpu, u64 val);
> +
> +	/* Return the reset value of the register for the vCPU */
> +	u64 (*get_reset_val)(struct kvm_vcpu *vcpu, struct id_reg_info *id_reg);
> +};
> +
> +static void id_reg_info_init(struct id_reg_info *id_reg)
> +{
> +	id_reg->sys_val = read_sanitised_ftr_reg(id_reg->sys_reg);
> +	id_reg->vcpu_limit_val = id_reg->sys_val;
> +}
> +
> +/*
> + * An ID register that needs special handling to control the value for the
> + * guest must have its own id_reg_info in id_reg_info_table.
> + * (i.e. the reset value is different from the host's sanitized value,
> + * the value is affected by opt-in features, some fields needs specific
> + * validation, etc.)
> + */
> +#define	GET_ID_REG_INFO(id)	(id_reg_info_table[IDREG_IDX(id)])
> +static struct id_reg_info *id_reg_info_table[KVM_ARM_ID_REG_MAX_NUM] = {};
> +
> +static int validate_id_reg(struct kvm_vcpu *vcpu,
> +			   const struct sys_reg_desc *rd, u64 val)
> +{
> +	u32 id = reg_to_encoding(rd);
> +	const struct id_reg_info *id_reg = GET_ID_REG_INFO(id);
> +	u64 limit;
> +	u64 check_val = val;
> +	int err;
> +
> +	if (id_reg) {
> +		/*
> +		 * Clear bits with ignore_mask, which we don't want
> +		 * arm64_check_features to check.
> +		 */
> +		check_val &= ~id_reg->ignore_mask;
> +		limit = id_reg->vcpu_limit_val;
> +	} else
> +		limit = read_sanitised_ftr_reg(id);
> +
> +	/* Check if the value indicates any feature that is not in the limit. */
> +	err = arm64_check_features(id, check_val, limit);
> +	if (err)
> +		return err;
> +
> +	if (id_reg && id_reg->validate)
> +		/* Run the ID register specific validity check. */
> +		err = id_reg->validate(vcpu, val);
> +
> +	return err;
> +}
> +
>  /*
>   * ARMv8.1 mandates at least a trivial LORegion implementation, where all the
>   * RW registers are RES0 (which we can implement as RAZ/WI). On an ARMv8.0
> @@ -1176,11 +1246,19 @@ static unsigned int sve_visibility(const struct kvm_vcpu *vcpu,
>  static void reset_id_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd)
>  {
>  	u32 id = reg_to_encoding(rd);
> +	struct id_reg_info *id_reg;
> +	u64 val;
>  
>  	if (vcpu_has_reset_once(vcpu))
>  		return;
>  
> -	__vcpu_sys_reg(vcpu, IDREG_SYS_IDX(id)) = read_sanitised_ftr_reg(id);
> +	id_reg = GET_ID_REG_INFO(id);
> +	if (id_reg && id_reg->get_reset_val)
> +		val = id_reg->get_reset_val(vcpu, id_reg);
> +	else
> +		val = read_sanitised_ftr_reg(id);
> +
> +	__vcpu_sys_reg(vcpu, IDREG_SYS_IDX(id)) = val;
>  }
>  
>  static int set_id_aa64pfr0_el1(struct kvm_vcpu *vcpu,
> @@ -1225,11 +1303,7 @@ static int set_id_aa64pfr0_el1(struct kvm_vcpu *vcpu,
>  	return 0;
>  }
>  
> -/*
> - * cpufeature ID register user accessors
> - *
> - * We don't allow the effective value to be changed.
> - */
> +/* cpufeature ID register user accessors */
>  static int __get_id_reg(const struct kvm_vcpu *vcpu,
>  			const struct sys_reg_desc *rd, void __user *uaddr,
>  			bool raz)
> @@ -1240,11 +1314,12 @@ static int __get_id_reg(const struct kvm_vcpu *vcpu,
>  	return reg_to_user(uaddr, &val, id);
>  }
>  
> -static int __set_id_reg(const struct kvm_vcpu *vcpu,
> +static int __set_id_reg(struct kvm_vcpu *vcpu,
>  			const struct sys_reg_desc *rd, void __user *uaddr,
>  			bool raz)
>  {
>  	const u64 id = sys_reg_to_index(rd);
> +	u32 encoding = reg_to_encoding(rd);
>  	int err;
>  	u64 val;
>  
> @@ -1252,10 +1327,18 @@ static int __set_id_reg(const struct kvm_vcpu *vcpu,
>  	if (err)
>  		return err;
>  
> -	/* This is what we mean by invariant: you can't change it. */
> -	if (val != read_id_reg(vcpu, rd, raz))
> +	/* Don't allow to change the reg unless the reg has id_reg_info */
> +	if (val != read_id_reg(vcpu, rd, raz) && !GET_ID_REG_INFO(encoding))
>  		return -EINVAL;
>  
> +	if (raz)
> +		return (val == 0) ? 0 : -EINVAL;

This is already covered by the val != read_id_reg(vcpu, rd, raz) check.

> +
> +	err = validate_id_reg(vcpu, rd, val);
> +	if (err)
> +		return err;
> +
> +	__vcpu_sys_reg(vcpu, IDREG_SYS_IDX(encoding)) = val;
>  	return 0;
>  }
>  
> @@ -2818,6 +2901,23 @@ int kvm_arm_copy_sys_reg_indices(struct kvm_vcpu *vcpu, u64 __user *uindices)
>  	return write_demux_regids(uindices);
>  }
>  
> +static void id_reg_info_init_all(void)
> +{
> +	int i;
> +	struct id_reg_info *id_reg;
> +
> +	for (i = 0; i < ARRAY_SIZE(id_reg_info_table); i++) {
> +		id_reg = (struct id_reg_info *)id_reg_info_table[i];
> +		if (!id_reg)
> +			continue;
> +
> +		if (id_reg->init)
> +			id_reg->init(id_reg);
> +		else
> +			id_reg_info_init(id_reg);

Maybe call id_reg->init(id_reg) from within id_reg_info_init() in case we
wanted to apply some common id register initialization at some point?

> +	}
> +}
> +
>  void kvm_sys_reg_table_init(void)
>  {
>  	unsigned int i;
> @@ -2852,4 +2952,6 @@ void kvm_sys_reg_table_init(void)
>  			break;
>  	/* Clear all higher bits. */
>  	cache_levels &= (1 << (i*3))-1;
> +
> +	id_reg_info_init_all();
>  }
> -- 
> 2.33.0.882.g93a45727a2-goog
>

Thanks,
drew
Reiji Watanabe Oct. 17, 2021, 4:43 a.m. UTC | #2
> > --- a/arch/arm64/kvm/sys_regs.c
> > +++ b/arch/arm64/kvm/sys_regs.c
> > @@ -263,6 +263,76 @@ static bool trap_raz_wi(struct kvm_vcpu *vcpu,
> >               return read_zero(vcpu, p);
> >  }
> >
> > +struct id_reg_info {
> > +     u32     sys_reg;        /* Register ID */
> > +     u64     sys_val;        /* Sanitized system value */
> > +
> > +     /*
> > +      * Limit value of the register for a vcpu. The value is sys_val
> > +      * with bits cleared for unsupported features for the guest.
> > +      */
> > +     u64     vcpu_limit_val;
>
> Maybe I'll see a need for both later, but at the moment I'd think we only
> need sys_val with the bits cleared for disabled features.

Uh, yes, sys_val is used in patch-15 and I should have introduced
the field in the patch.  I will fix it in v2.


> > -static int __set_id_reg(const struct kvm_vcpu *vcpu,
> > +static int __set_id_reg(struct kvm_vcpu *vcpu,
> >                       const struct sys_reg_desc *rd, void __user *uaddr,
> >                       bool raz)
> >  {
> >       const u64 id = sys_reg_to_index(rd);
> > +     u32 encoding = reg_to_encoding(rd);
> >       int err;
> >       u64 val;
> >
> > @@ -1252,10 +1327,18 @@ static int __set_id_reg(const struct kvm_vcpu *vcpu,
> >       if (err)
> >               return err;
> >
> > -     /* This is what we mean by invariant: you can't change it. */
> > -     if (val != read_id_reg(vcpu, rd, raz))
> > +     /* Don't allow to change the reg unless the reg has id_reg_info */
> > +     if (val != read_id_reg(vcpu, rd, raz) && !GET_ID_REG_INFO(encoding))
> >               return -EINVAL;
> >
> > +     if (raz)
> > +             return (val == 0) ? 0 : -EINVAL;
>
> This is already covered by the val != read_id_reg(vcpu, rd, raz) check.

Yes, it can simply return 0 for raz case in this patch.
I will fix this in v2.


> > +     err = validate_id_reg(vcpu, rd, val);
> > +     if (err)
> > +             return err;
> > +
> > +     __vcpu_sys_reg(vcpu, IDREG_SYS_IDX(encoding)) = val;
> >       return 0;
> >  }
> >
> > @@ -2818,6 +2901,23 @@ int kvm_arm_copy_sys_reg_indices(struct kvm_vcpu *vcpu, u64 __user *uindices)
> >       return write_demux_regids(uindices);
> >  }
> >
> > +static void id_reg_info_init_all(void)
> > +{
> > +     int i;
> > +     struct id_reg_info *id_reg;
> > +
> > +     for (i = 0; i < ARRAY_SIZE(id_reg_info_table); i++) {
> > +             id_reg = (struct id_reg_info *)id_reg_info_table[i];
> > +             if (!id_reg)
> > +                     continue;
> > +
> > +             if (id_reg->init)
> > +                     id_reg->init(id_reg);
> > +             else
> > +                     id_reg_info_init(id_reg);
>
> Maybe call id_reg->init(id_reg) from within id_reg_info_init() in case we
> wanted to apply some common id register initialization at some point?

Thank you for the nice suggestion.
That sounds like a better idea. I'll look into fixing it in v2.

Thanks,
Reiji
diff mbox series

Patch

diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index 72ca518e7944..8a0b88f9a975 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -263,6 +263,76 @@  static bool trap_raz_wi(struct kvm_vcpu *vcpu,
 		return read_zero(vcpu, p);
 }
 
+struct id_reg_info {
+	u32	sys_reg;	/* Register ID */
+	u64	sys_val;	/* Sanitized system value */
+
+	/*
+	 * Limit value of the register for a vcpu. The value is sys_val
+	 * with bits cleared for unsupported features for the guest.
+	 */
+	u64	vcpu_limit_val;
+
+	/* Bits that are not validated when userspace sets the register. */
+	u64	ignore_mask;
+
+	/* Initialization function of the id_reg_info */
+	void (*init)(struct id_reg_info *id_reg);
+
+	/* Register specific validation function */
+	int (*validate)(struct kvm_vcpu *vcpu, u64 val);
+
+	/* Return the reset value of the register for the vCPU */
+	u64 (*get_reset_val)(struct kvm_vcpu *vcpu, struct id_reg_info *id_reg);
+};
+
+static void id_reg_info_init(struct id_reg_info *id_reg)
+{
+	id_reg->sys_val = read_sanitised_ftr_reg(id_reg->sys_reg);
+	id_reg->vcpu_limit_val = id_reg->sys_val;
+}
+
+/*
+ * An ID register that needs special handling to control the value for the
+ * guest must have its own id_reg_info in id_reg_info_table.
+ * (i.e. the reset value is different from the host's sanitized value,
+ * the value is affected by opt-in features, some fields needs specific
+ * validation, etc.)
+ */
+#define	GET_ID_REG_INFO(id)	(id_reg_info_table[IDREG_IDX(id)])
+static struct id_reg_info *id_reg_info_table[KVM_ARM_ID_REG_MAX_NUM] = {};
+
+static int validate_id_reg(struct kvm_vcpu *vcpu,
+			   const struct sys_reg_desc *rd, u64 val)
+{
+	u32 id = reg_to_encoding(rd);
+	const struct id_reg_info *id_reg = GET_ID_REG_INFO(id);
+	u64 limit;
+	u64 check_val = val;
+	int err;
+
+	if (id_reg) {
+		/*
+		 * Clear bits with ignore_mask, which we don't want
+		 * arm64_check_features to check.
+		 */
+		check_val &= ~id_reg->ignore_mask;
+		limit = id_reg->vcpu_limit_val;
+	} else
+		limit = read_sanitised_ftr_reg(id);
+
+	/* Check if the value indicates any feature that is not in the limit. */
+	err = arm64_check_features(id, check_val, limit);
+	if (err)
+		return err;
+
+	if (id_reg && id_reg->validate)
+		/* Run the ID register specific validity check. */
+		err = id_reg->validate(vcpu, val);
+
+	return err;
+}
+
 /*
  * ARMv8.1 mandates at least a trivial LORegion implementation, where all the
  * RW registers are RES0 (which we can implement as RAZ/WI). On an ARMv8.0
@@ -1176,11 +1246,19 @@  static unsigned int sve_visibility(const struct kvm_vcpu *vcpu,
 static void reset_id_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd)
 {
 	u32 id = reg_to_encoding(rd);
+	struct id_reg_info *id_reg;
+	u64 val;
 
 	if (vcpu_has_reset_once(vcpu))
 		return;
 
-	__vcpu_sys_reg(vcpu, IDREG_SYS_IDX(id)) = read_sanitised_ftr_reg(id);
+	id_reg = GET_ID_REG_INFO(id);
+	if (id_reg && id_reg->get_reset_val)
+		val = id_reg->get_reset_val(vcpu, id_reg);
+	else
+		val = read_sanitised_ftr_reg(id);
+
+	__vcpu_sys_reg(vcpu, IDREG_SYS_IDX(id)) = val;
 }
 
 static int set_id_aa64pfr0_el1(struct kvm_vcpu *vcpu,
@@ -1225,11 +1303,7 @@  static int set_id_aa64pfr0_el1(struct kvm_vcpu *vcpu,
 	return 0;
 }
 
-/*
- * cpufeature ID register user accessors
- *
- * We don't allow the effective value to be changed.
- */
+/* cpufeature ID register user accessors */
 static int __get_id_reg(const struct kvm_vcpu *vcpu,
 			const struct sys_reg_desc *rd, void __user *uaddr,
 			bool raz)
@@ -1240,11 +1314,12 @@  static int __get_id_reg(const struct kvm_vcpu *vcpu,
 	return reg_to_user(uaddr, &val, id);
 }
 
-static int __set_id_reg(const struct kvm_vcpu *vcpu,
+static int __set_id_reg(struct kvm_vcpu *vcpu,
 			const struct sys_reg_desc *rd, void __user *uaddr,
 			bool raz)
 {
 	const u64 id = sys_reg_to_index(rd);
+	u32 encoding = reg_to_encoding(rd);
 	int err;
 	u64 val;
 
@@ -1252,10 +1327,18 @@  static int __set_id_reg(const struct kvm_vcpu *vcpu,
 	if (err)
 		return err;
 
-	/* This is what we mean by invariant: you can't change it. */
-	if (val != read_id_reg(vcpu, rd, raz))
+	/* Don't allow to change the reg unless the reg has id_reg_info */
+	if (val != read_id_reg(vcpu, rd, raz) && !GET_ID_REG_INFO(encoding))
 		return -EINVAL;
 
+	if (raz)
+		return (val == 0) ? 0 : -EINVAL;
+
+	err = validate_id_reg(vcpu, rd, val);
+	if (err)
+		return err;
+
+	__vcpu_sys_reg(vcpu, IDREG_SYS_IDX(encoding)) = val;
 	return 0;
 }
 
@@ -2818,6 +2901,23 @@  int kvm_arm_copy_sys_reg_indices(struct kvm_vcpu *vcpu, u64 __user *uindices)
 	return write_demux_regids(uindices);
 }
 
+static void id_reg_info_init_all(void)
+{
+	int i;
+	struct id_reg_info *id_reg;
+
+	for (i = 0; i < ARRAY_SIZE(id_reg_info_table); i++) {
+		id_reg = (struct id_reg_info *)id_reg_info_table[i];
+		if (!id_reg)
+			continue;
+
+		if (id_reg->init)
+			id_reg->init(id_reg);
+		else
+			id_reg_info_init(id_reg);
+	}
+}
+
 void kvm_sys_reg_table_init(void)
 {
 	unsigned int i;
@@ -2852,4 +2952,6 @@  void kvm_sys_reg_table_init(void)
 			break;
 	/* Clear all higher bits. */
 	cache_levels &= (1 << (i*3))-1;
+
+	id_reg_info_init_all();
 }