[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [Qemu-arm] [PATCH 6/7] target-arm: Implement NSACR trap
From: |
Alex Bennée |
Subject: |
Re: [Qemu-devel] [Qemu-arm] [PATCH 6/7] target-arm: Implement NSACR trapping behaviour |
Date: |
Fri, 05 Feb 2016 16:07:57 +0000 |
User-agent: |
mu4e 0.9.17; emacs 25.0.90.1 |
Peter Maydell <address@hidden> writes:
> Implement some corner cases of the behaviour of the NSACR
> register on ARMv8:
> * if EL3 is AArch64 then accessing the NSACR from Secure EL1
> with AArch32 should trap to EL3
> * if EL3 is not present or is AArch64 then reads from NS EL1 and
> NS EL2 return constant 0xc00
>
> It would in theory be possible to implement all these with
> a single reginfo definition, but for clarity we use three
> separate definitions for the three cases and install the
> right one based on the CPU feature flags.
>
> Signed-off-by: Peter Maydell <address@hidden>
> ---
> target-arm/helper.c | 62
> +++++++++++++++++++++++++++++++++++++++++++++++++----
> 1 file changed, 58 insertions(+), 4 deletions(-)
>
> diff --git a/target-arm/helper.c b/target-arm/helper.c
> index 8bc3ea3..34dc144 100644
> --- a/target-arm/helper.c
> +++ b/target-arm/helper.c
> @@ -3562,6 +3562,25 @@ static const ARMCPRegInfo el2_cp_reginfo[] = {
> REGINFO_SENTINEL
> };
>
> +static CPAccessResult nsacr_access(CPUARMState *env, const ARMCPRegInfo *ri,
> + bool isread)
> +{
> + /* The NSACR is RW at EL3, and RO for NS EL1 and NS EL2.
> + * At Secure EL1 it traps to EL3.
> + */
> + if (arm_current_el(env) == 3) {
> + return CP_ACCESS_OK;
> + }
> + if (arm_is_secure_below_el3(env)) {
> + return CP_ACCESS_TRAP_EL3;
> + }
> + /* Accesses from EL1 NS and EL2 NS are UNDEF for write but allow reads.
> */
> + if (isread) {
> + return CP_ACCESS_OK;
> + }
> + return CP_ACCESS_TRAP_UNCATEGORIZED;
> +}
> +
> static const ARMCPRegInfo el3_cp_reginfo[] = {
> { .name = "SCR_EL3", .state = ARM_CP_STATE_AA64,
> .opc0 = 3, .opc1 = 6, .crn = 1, .crm = 1, .opc2 = 0,
> @@ -3587,10 +3606,6 @@ static const ARMCPRegInfo el3_cp_reginfo[] = {
> .cp = 15, .opc1 = 0, .crn = 1, .crm = 1, .opc2 = 1,
> .access = PL3_RW, .resetvalue = 0,
> .fieldoffset = offsetoflow32(CPUARMState, cp15.sder) },
> - /* TODO: Implement NSACR trapping of secure EL1 accesses to EL3 */
> - { .name = "NSACR", .cp = 15, .opc1 = 0, .crn = 1, .crm = 1, .opc2 = 2,
> - .access = PL3_W | PL1_R, .resetvalue = 0,
> - .fieldoffset = offsetof(CPUARMState, cp15.nsacr) },
> { .name = "MVBAR", .cp = 15, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 1,
> .access = PL1_RW, .accessfn = access_trap_aa32s_el1,
> .writefn = vbar_write, .resetvalue = 0,
> @@ -4361,6 +4376,45 @@ void register_cp_regs_for_features(ARMCPU *cpu)
> };
> define_one_arm_cp_reg(cpu, &rvbar);
> }
> + /* The behaviour of NSACR is sufficiently various that we don't
> + * try to describe it in a single reginfo:
> + * if EL3 is 64 bit, then trap to EL3 from S EL1,
> + * reads as constant 0xc00 from NS EL1 and NS EL2
> + * if EL3 is 32 bit, then RW at EL3, RO at NS EL1 and NS EL2
> + * if v7 without EL3, register doesn't exist
> + * if v8 without EL3, reads as constant 0xc00 from NS EL1 and NS EL2
> + */
> + if (arm_feature(env, ARM_FEATURE_EL3)) {
> + if (arm_feature(env, ARM_FEATURE_AARCH64)) {
> + ARMCPRegInfo nsacr = {
> + .name = "NSACR", .type = ARM_CP_CONST,
I was thrown by the ARM_CP_CONST here as there is EL dependency. If
nsacr_access says CP_ACCESS_OK where does the write go? The ARM ARM says
RO and the code in translate-a64 says:
if (ri->type & ARM_CP_CONST) {
/* If not forbidden by access permissions, treat as WI */
return;
}
So we might want to make that clear in the comment.
> + .cp = 15, .opc1 = 0, .crn = 1, .crm = 1, .opc2 = 2,
> + .access = PL1_RW, .accessfn = nsacr_access,
PL1_R?
> + .resetvalue = 0xc00
> + };
> + define_one_arm_cp_reg(cpu, &nsacr);
> + } else {
> + ARMCPRegInfo nsacr = {
> + .name = "NSACR",
> + .cp = 15, .opc1 = 0, .crn = 1, .crm = 1, .opc2 = 2,
> + .access = PL3_RW | PL1_R,
> + .resetvalue = 0,
> + .fieldoffset = offsetof(CPUARMState, cp15.nsacr)
> + };
> + define_one_arm_cp_reg(cpu, &nsacr);
> + }
> + } else {
> + if (arm_feature(env, ARM_FEATURE_V8)) {
> + ARMCPRegInfo nsacr = {
> + .name = "NSACR", .type = ARM_CP_CONST,
> + .cp = 15, .opc1 = 0, .crn = 1, .crm = 1, .opc2 = 2,
> + .access = PL1_R,
> + .resetvalue = 0xc00
> + };
> + define_one_arm_cp_reg(cpu, &nsacr);
> + }
> + }
> +
> if (arm_feature(env, ARM_FEATURE_MPU)) {
> if (arm_feature(env, ARM_FEATURE_V6)) {
> /* PMSAv6 not implemented */
I don't know if a more compact definition would make this easier to
follow?
/* The behaviour of NSACR is sufficiently various we tweak the reginfo:
* if EL3 is 64 bit, then trap to EL3 from S EL1,
* reads as constant 0xc00 from NS EL1 and NS EL2
* if EL3 is 32 bit, then RW at EL3, RO at NS EL1 and NS EL2
* if v7 without EL3, register doesn't exist
* if v8 without EL3, reads as constant 0xc00 from NS EL1 and NS EL2
*/
if (arm_feature(env, ARM_FEATURE_V8) || arm_feature(env, ARM_FEATURE_EL3)) {
ARMCPRegInfo nsacr = {
.name = "NSACR",
.cp = 15, .opc1 = 0, .crn = 1, .crm = 1, .opc2 = 2,
.access = PL1_R,
.resetvalue = 0xc00
};
if (arm_feature(env, ARM_FEATURE_EL3)) {
if (arm_feature(env, ARM_FEATURE_AARCH64)) {
nsacr.type = ARM_CP_CONST; /* if no trap RO */
nsacr.accessfn = nsacr_access;
} else {
nsacr.access = PL3_RW | PL1_R;
nsacr.resetvalue = 0;
nsacr.fieldoffset = offsetof(CPUARMState, cp15.nsacr);
};
} else {
nsacr.type = ARM_CP_CONST;
}
define_one_arm_cp_reg(cpu, &nsacr);
}
--
Alex Bennée
- Re: [Qemu-devel] [PATCH 2/7] target-arm: Implement MDCR_EL3 and SDCR, (continued)
Re: [Qemu-devel] [PATCH 5/7] target-arm: Add isread parameter to CPAccessFns, Edgar E. Iglesias, 2016/02/06
Re: [Qemu-devel] [PATCH 5/7] target-arm: Add isread parameter to CPAccessFns, Sergey Fedorov, 2016/02/06
[Qemu-devel] [PATCH 6/7] target-arm: Implement NSACR trapping behaviour, Peter Maydell, 2016/02/03
[Qemu-devel] [PATCH 1/7] target-arm: Fix typo in comment in arm_is_secure_below_el3(), Peter Maydell, 2016/02/03
[Qemu-devel] [PATCH 4/7] target-arm: Update arm_generate_debug_exceptions() to handle EL2/EL3, Peter Maydell, 2016/02/03