Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 28 Jul 2023 11:37:52 +0200
From:   Andrew Jones <ajones@ventanamicro.com>
To:     Haibo Xu <haibo1.xu@intel.com>
Cc:     xiaobo55x@gmail.com, Paul Walmsley <paul.walmsley@sifive.com>,
        Palmer Dabbelt <palmer@dabbelt.com>,
        Albert Ou <aou@eecs.berkeley.edu>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Shuah Khan <shuah@kernel.org>,
        Anup Patel <anup@brainfault.org>,
        Atish Patra <atishp@atishpatra.org>,
        Sean Christopherson <seanjc@google.com>,
        Marc Zyngier <maz@kernel.org>,
        Andrew Jones <andrew.jones@linux.dev>,
        Colton Lewis <coltonlewis@google.com>,
        Vipin Sharma <vipinsh@google.com>,
        Vishal Annapurve <vannapurve@google.com>,
        linux-kernel@vger.kernel.org, linux-riscv@lists.infradead.org,
        kvm@vger.kernel.org, linux-kselftest@vger.kernel.org,
        kvm-riscv@lists.infradead.org
Subject: Re: [PATCH 2/4] KVM: riscv: selftests: Add exception handling support
Message-ID: <20230728-21d0e3ecb0d8916fd9d9bceb@orel>
References: <cover.1690364259.git.haibo1.xu@intel.com>
 <7e9d2d5518375083f10c92a3d1acd98914f69fd8.1690364259.git.haibo1.xu@intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <7e9d2d5518375083f10c92a3d1acd98914f69fd8.1690364259.git.haibo1.xu@intel.com>
Precedence: bulk

On Thu, Jul 27, 2023 at 03:20:06PM +0800, Haibo Xu wrote:
> Add the infrastructure for exception handling in riscv selftests.
> Currently, the guest_unexp_trap handler was used by default, which
> aborts the test. Customized handlers can be enabled by calling
> vm_install_exception_handler(vector) or vm_install_interrupt_handler().
> 
> The code is inspired from that of x86/arm64.
> 
> Signed-off-by: Haibo Xu <haibo1.xu@intel.com>
> ---
>  tools/testing/selftests/kvm/Makefile          |   1 +
>  .../selftests/kvm/include/riscv/processor.h   |  49 +++++++++
>  .../selftests/kvm/lib/riscv/handlers.S        | 101 ++++++++++++++++++
>  .../selftests/kvm/lib/riscv/processor.c       |  57 ++++++++++
>  4 files changed, 208 insertions(+)
>  create mode 100644 tools/testing/selftests/kvm/lib/riscv/handlers.S
> 
> diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
> index c692cc86e7da..70f3a5ba991e 100644
> --- a/tools/testing/selftests/kvm/Makefile
> +++ b/tools/testing/selftests/kvm/Makefile
> @@ -52,6 +52,7 @@ LIBKVM_s390x += lib/s390x/diag318_test_handler.c
>  LIBKVM_s390x += lib/s390x/processor.c
>  LIBKVM_s390x += lib/s390x/ucall.c
>  
> +LIBKVM_riscv += lib/riscv/handlers.S
>  LIBKVM_riscv += lib/riscv/processor.c
>  LIBKVM_riscv += lib/riscv/ucall.c
>  
> diff --git a/tools/testing/selftests/kvm/include/riscv/processor.h b/tools/testing/selftests/kvm/include/riscv/processor.h
> index d00d213c3805..9ea6e7bedc61 100644
> --- a/tools/testing/selftests/kvm/include/riscv/processor.h
> +++ b/tools/testing/selftests/kvm/include/riscv/processor.h
> @@ -9,6 +9,7 @@
>  
>  #include "kvm_util.h"
>  #include <linux/stringify.h>
> +#include <asm/csr.h>
>  
>  static inline uint64_t __kvm_reg_id(uint64_t type, uint64_t idx,
>  				    uint64_t  size)
> @@ -38,6 +39,54 @@ static inline uint64_t __kvm_reg_id(uint64_t type, uint64_t idx,
>  					     KVM_REG_RISCV_TIMER_REG(name), \
>  					     KVM_REG_SIZE_U64)
>  
> +struct ex_regs {
> +	unsigned long ra;
> +	unsigned long sp;
> +	unsigned long gp;
> +	unsigned long tp;
> +	unsigned long t0;
> +	unsigned long t1;
> +	unsigned long t2;
> +	unsigned long s0;
> +	unsigned long s1;
> +	unsigned long a0;
> +	unsigned long a1;
> +	unsigned long a2;
> +	unsigned long a3;
> +	unsigned long a4;
> +	unsigned long a5;
> +	unsigned long a6;
> +	unsigned long a7;
> +	unsigned long s2;
> +	unsigned long s3;
> +	unsigned long s4;
> +	unsigned long s5;
> +	unsigned long s6;
> +	unsigned long s7;
> +	unsigned long s8;
> +	unsigned long s9;
> +	unsigned long s10;
> +	unsigned long s11;
> +	unsigned long t3;
> +	unsigned long t4;
> +	unsigned long t5;
> +	unsigned long t6;
> +	unsigned long epc;
> +	unsigned long status;
> +	unsigned long cause;
> +};
> +
> +#define VECTOR_NUM  2
> +#define EC_NUM  32
> +#define EC_MASK  (EC_NUM - 1)

nit: My personal preference is to use something like NR_VECTORS and
NR_EXCEPTIONS for these, since *_NUM type names are ambiguous with
named indices.

> +
> +void vm_init_trap_vector_tables(struct kvm_vm *vm);
> +void vcpu_init_trap_vector_tables(struct kvm_vcpu *vcpu);

I think we should use a common name for these prototypes that the other
architectures agree to and then put them in a common header. My vote for
the naming is,

  void vm_init_vector_tables(struct kvm_vm *vm);
  void vcpu_init_vector_tables(struct kvm_vcpu *vcpu);

> +
> +typedef void(*handler_fn)(struct ex_regs *);
> +void vm_install_exception_handler(struct kvm_vm *vm, int ec, handler_fn handler);

I'd also put this typedef and prototype in a common header
(with s/ec/vector/ to what you have here)

> +void vm_install_interrupt_handler(struct kvm_vm *vm, handler_fn handler);

I guess this one can stay risc-v specific for now since no other arch is
using it.

> +
>  /* L3 index Bit[47:39] */
>  #define PGTBL_L3_INDEX_MASK			0x0000FF8000000000ULL
>  #define PGTBL_L3_INDEX_SHIFT			39
> diff --git a/tools/testing/selftests/kvm/lib/riscv/handlers.S b/tools/testing/selftests/kvm/lib/riscv/handlers.S
> new file mode 100644
> index 000000000000..ce0b1d5415b9
> --- /dev/null
> +++ b/tools/testing/selftests/kvm/lib/riscv/handlers.S
> @@ -0,0 +1,101 @@
> +// SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (c) 2023 Intel Corporation
> + */
> +
> +#include <asm/csr.h>

General note for all the asm below, please format with the first operand
aligned, so

<tab>op<tab>operand1, operand2, ...

> +
> +.macro save_context
> +	addi sp, sp, (-8*34)
> +
> +	sd x1, 0(sp)
> +	sd x2, 8(sp)
> +	sd x3, 16(sp)
> +	sd x4, 24(sp)
> +	sd x5, 32(sp)
> +	sd x6, 40(sp)
> +	sd x7, 48(sp)
> +	sd x8, 56(sp)
> +	sd x9, 64(sp)
> +	sd x10, 72(sp)
> +	sd x11, 80(sp)
> +	sd x12, 88(sp)
> +	sd x13, 96(sp)
> +	sd x14, 104(sp)
> +	sd x15, 112(sp)
> +	sd x16, 120(sp)
> +	sd x17, 128(sp)
> +	sd x18, 136(sp)
> +	sd x19, 144(sp)
> +	sd x20, 152(sp)
> +	sd x21, 160(sp)
> +	sd x22, 168(sp)
> +	sd x23, 176(sp)
> +	sd x24, 184(sp)
> +	sd x25, 192(sp)
> +	sd x26, 200(sp)
> +	sd x27, 208(sp)
> +	sd x28, 216(sp)
> +	sd x29, 224(sp)
> +	sd x30, 232(sp)
> +	sd x31, 240(sp)
> +
> +	csrr s0, CSR_SEPC
> +	csrr s1, CSR_SSTATUS
> +	csrr s2, CSR_SCAUSE
> +	sd s0, 248(sp)
> +	sd s1, 256(sp)
> +	sd s2, 264(sp)
> +.endm

Let's create a restore_context macro too in order to maintain balance.

> +
> +.balign 4
> +.global exception_vectors
> +exception_vectors:
> +	save_context
> +	move a0, sp
> +	la ra, ret_from_exception
> +	tail route_exception
> +
> +.global ret_from_exception
> +ret_from_exception:
> +	ld s2, 264(sp)
> +	ld s1, 256(sp)
> +	ld s0, 248(sp)
> +	csrw CSR_SCAUSE, s2
> +	csrw CSR_SSTATUS, s1
> +	csrw CSR_SEPC, s0
> +
> +	ld x31, 240(sp)
> +	ld x30, 232(sp)
> +	ld x29, 224(sp)
> +	ld x28, 216(sp)
> +	ld x27, 208(sp)
> +	ld x26, 200(sp)
> +	ld x25, 192(sp)
> +	ld x24, 184(sp)
> +	ld x23, 176(sp)
> +	ld x22, 168(sp)
> +	ld x21, 160(sp)
> +	ld x20, 152(sp)
> +	ld x19, 144(sp)
> +	ld x18, 136(sp)
> +	ld x17, 128(sp)
> +	ld x16, 120(sp)
> +	ld x15, 112(sp)
> +	ld x14, 104(sp)
> +	ld x13, 96(sp)
> +	ld x12, 88(sp)
> +	ld x11, 80(sp)
> +	ld x10, 72(sp)
> +	ld x9, 64(sp)
> +	ld x8, 56(sp)
> +	ld x7, 48(sp)
> +	ld x6, 40(sp)
> +	ld x5, 32(sp)
> +	ld x4, 24(sp)
> +	ld x3, 16(sp)
> +	ld x2, 8(sp)
> +	ld x1, 0(sp)
> +
> +	addi sp, sp, (8*34)
> +	sret
> diff --git a/tools/testing/selftests/kvm/lib/riscv/processor.c b/tools/testing/selftests/kvm/lib/riscv/processor.c
> index d146ca71e0c0..f1b0be58a5dc 100644
> --- a/tools/testing/selftests/kvm/lib/riscv/processor.c
> +++ b/tools/testing/selftests/kvm/lib/riscv/processor.c
> @@ -13,6 +13,8 @@
>  
>  #define DEFAULT_RISCV_GUEST_STACK_VADDR_MIN	0xac0000
>  
> +static vm_vaddr_t exception_handlers;
> +
>  static uint64_t page_align(struct kvm_vm *vm, uint64_t v)
>  {
>  	return (v + vm->page_size) & ~(vm->page_size - 1);
> @@ -367,3 +369,58 @@ void vcpu_args_set(struct kvm_vcpu *vcpu, unsigned int num, ...)
>  void assert_on_unhandled_exception(struct kvm_vcpu *vcpu)
>  {
>  }
> +
> +struct handlers {
> +	handler_fn exception_handlers[VECTOR_NUM][EC_NUM];
> +};
> +
> +void route_exception(struct ex_regs *regs)
> +{
> +	struct handlers *handlers = (struct handlers *)exception_handlers;
> +	int vector = 0, ec;
> +
> +	ec = regs->cause & ~CAUSE_IRQ_FLAG;
> +	if (ec >= EC_NUM)
> +		goto guest_unexpected_trap;
> +
> +	/* Use the same handler for all the interrupts */
> +	if (regs->cause & CAUSE_IRQ_FLAG) {
> +		vector = 1;
> +		ec = 0;
> +	}
> +
> +	if (handlers && handlers->exception_handlers[vector][ec])
> +		return handlers->exception_handlers[vector][ec](regs);
> +
> +guest_unexpected_trap:
> +	return guest_unexp_trap();

I think we want this to have consistent behavior with the other
architectures, so we should be issuing a UCALL_UNHANDLED.

> +}
> +
> +void vcpu_init_trap_vector_tables(struct kvm_vcpu *vcpu)
> +{
> +	extern char exception_vectors;
> +
> +	vcpu_set_reg(vcpu, RISCV_CSR_REG(stvec), (unsigned long)&exception_vectors);
> +}
> +
> +void vm_init_trap_vector_tables(struct kvm_vm *vm)
> +{
> +	vm->handlers = __vm_vaddr_alloc(vm, sizeof(struct handlers),
> +				   vm->page_size, MEM_REGION_DATA);
> +
> +	*(vm_vaddr_t *)addr_gva2hva(vm, (vm_vaddr_t)(&exception_handlers)) = vm->handlers;
> +}
> +
> +void vm_install_exception_handler(struct kvm_vm *vm, int ec, void (*handler)(struct ex_regs *))
> +{
> +	struct handlers *handlers = addr_gva2hva(vm, vm->handlers);
> +

Add assert here that ec is valid.

> +	handlers->exception_handlers[0][ec] = handler;
> +}
> +
> +void vm_install_interrupt_handler(struct kvm_vm *vm, void (*handler)(struct ex_regs *))
> +{
> +	struct handlers *handlers = addr_gva2hva(vm, vm->handlers);
> +
> +	handlers->exception_handlers[1][0] = handler;
> +}
> -- 
> 2.34.1
>

Besides some nits and wanting to get more consistency with the other
architectures, this looks good to me.

Thanks,
drew