Re: [PATCH 7/9] KVM: arm64: selftests: Add a test case for a linked breakpoint

[Ricardo Koller <ricarkol@google.com>]

On Wed, Aug 24, 2022 at 10:08:44PM -0700, Reiji Watanabe wrote:
> Currently, the debug-exceptions test doesn't have a test case for
> a linked breakpoint. Add a test case for the linked breakpoint to
> the test.

I would add some more detail, like the fact that this is a pair of
breakpoints: one is a context-aware breakpoint, and the other one
is an address breakpoint linked to the first one.

> 
> Signed-off-by: Reiji Watanabe <reijiw@google.com>
> 
> ---
>  .../selftests/kvm/aarch64/debug-exceptions.c  | 59 +++++++++++++++++--
>  1 file changed, 55 insertions(+), 4 deletions(-)
> 
> diff --git a/tools/testing/selftests/kvm/aarch64/debug-exceptions.c b/tools/testing/selftests/kvm/aarch64/debug-exceptions.c
> index ab8860e3a9fa..9fccfeebccd3 100644
> --- a/tools/testing/selftests/kvm/aarch64/debug-exceptions.c
> +++ b/tools/testing/selftests/kvm/aarch64/debug-exceptions.c
> @@ -11,6 +11,10 @@
>  #define DBGBCR_EXEC	(0x0 << 3)
>  #define DBGBCR_EL1	(0x1 << 1)
>  #define DBGBCR_E	(0x1 << 0)
> +#define DBGBCR_LBN_SHIFT	16
> +#define DBGBCR_BT_SHIFT		20
> +#define DBGBCR_BT_ADDR_LINK_CTX	(0x1 << DBGBCR_BT_SHIFT)
> +#define DBGBCR_BT_CTX_LINK	(0x3 << DBGBCR_BT_SHIFT)
>  
>  #define DBGWCR_LEN8	(0xff << 5)
>  #define DBGWCR_RD	(0x1 << 3)
> @@ -21,7 +25,7 @@
>  #define SPSR_D		(1 << 9)
>  #define SPSR_SS		(1 << 21)
>  
> -extern unsigned char sw_bp, sw_bp2, hw_bp, hw_bp2, bp_svc, bp_brk, hw_wp, ss_start;
> +extern unsigned char sw_bp, sw_bp2, hw_bp, hw_bp2, bp_svc, bp_brk, hw_wp, ss_start, hw_bp_ctx;
>  static volatile uint64_t sw_bp_addr, hw_bp_addr;
>  static volatile uint64_t wp_addr, wp_data_addr;
>  static volatile uint64_t svc_addr;
> @@ -103,6 +107,7 @@ static void reset_debug_state(void)
>  	isb();
>  
>  	write_sysreg(0, mdscr_el1);
> +	write_sysreg(0, contextidr_el1);
>  
>  	/* Reset all bcr/bvr/wcr/wvr registers */
>  	dfr0 = read_sysreg(id_aa64dfr0_el1);
> @@ -164,6 +169,28 @@ static void install_hw_bp(uint8_t bpn, uint64_t addr)
>  	enable_debug_bwp_exception();
>  }
>  
> +void install_hw_bp_ctx(uint8_t addr_bp, uint8_t ctx_bp, uint64_t addr,
> +		       uint64_t ctx)
> +{
> +	uint32_t addr_bcr, ctx_bcr;
> +
> +	/* Setup a context-aware breakpoint */
> +	ctx_bcr = DBGBCR_LEN8 | DBGBCR_EXEC | DBGBCR_EL1 | DBGBCR_E |
> +		  DBGBCR_BT_CTX_LINK;
                               ^^^^^
                          isn't this a regular context-aware breakpoint?
			  the other one is the linked one.

> +	write_dbgbcr(ctx_bp, ctx_bcr);
> +	write_dbgbvr(ctx_bp, ctx);
> +
> +	/* Setup a linked breakpoint (linked to the context-aware breakpoint) */
> +	addr_bcr = DBGBCR_LEN8 | DBGBCR_EXEC | DBGBCR_EL1 | DBGBCR_E |
> +		   DBGBCR_BT_ADDR_LINK_CTX |
> +		   ((uint32_t)ctx_bp << DBGBCR_LBN_SHIFT);

Just a curiosity, can the context-aware one link to this one?

> +	write_dbgbcr(addr_bp, addr_bcr);
> +	write_dbgbvr(addr_bp, addr);
> +	isb();
> +
> +	enable_debug_bwp_exception();
> +}
> +
>  static void install_ss(void)
>  {
>  	uint32_t mdscr;
> @@ -177,8 +204,10 @@ static void install_ss(void)
>  
>  static volatile char write_data;
>  
> -static void guest_code(uint8_t bpn, uint8_t wpn)
> +static void guest_code(uint8_t bpn, uint8_t wpn, uint8_t ctx_bpn)
>  {
> +	uint64_t ctx = 0x1;	/* a random context number */

nit: make this number a bit more unlikely to happen by mistake.
I guess you could use all available 32 bits.

> +
>  	GUEST_SYNC(0);
>  
>  	/* Software-breakpoint */
> @@ -281,6 +310,19 @@ static void guest_code(uint8_t bpn, uint8_t wpn)
>  		     : : : "x0");
>  	GUEST_ASSERT_EQ(ss_addr[0], 0);
>  
> +	/* Linked hardware-breakpoint */
> +	hw_bp_addr = 0;
> +	reset_debug_state();
> +	install_hw_bp_ctx(bpn, ctx_bpn, PC(hw_bp_ctx), ctx);
> +	/* Set context id */
> +	write_sysreg(ctx, contextidr_el1);
> +	isb();
> +	asm volatile("hw_bp_ctx: nop");
> +	write_sysreg(0, contextidr_el1);
> +	GUEST_ASSERT_EQ(hw_bp_addr, PC(hw_bp_ctx));
> +
> +	GUEST_SYNC(10);
> +
>  	GUEST_DONE();
>  }
>  
> @@ -327,6 +369,7 @@ int main(int argc, char *argv[])
>  	struct ucall uc;
>  	int stage;
>  	uint64_t aa64dfr0;
> +	uint8_t brps;
>  
>  	vm = vm_create_with_one_vcpu(&vcpu, guest_code);
>  	ucall_init(vm, NULL);
> @@ -349,8 +392,16 @@ int main(int argc, char *argv[])
>  	vm_install_sync_handler(vm, VECTOR_SYNC_CURRENT,
>  				ESR_EC_SVC64, guest_svc_handler);
>  
> -	/* Run tests with breakpoint#0 and watchpoint#0. */
> -	vcpu_args_set(vcpu, 2, 0, 0);
> +	/* Number of breakpoints, minus 1 */
> +	brps = cpuid_get_ufield(aa64dfr0, ID_AA64DFR0_BRPS_SHIFT);

If brps is "number of breakpoints", then there should be a "+ 1" above.
Otherwise brps is really "last breakpoint" (last_brp).

> +	__TEST_REQUIRE(brps > 0, "At least two breakpoints are required");

Yes, based on this test, brps is really "last breakpoint". I would
suggest changing the name to "last_brp" (or something similar).

> +
> +	/*
> +	 * Run tests with breakpoint#0 and watchpoint#0, and the higiest

	 * Run tests with breakpoint#0, watchpoint#0, and the highest
	
> +	 * numbered (context-aware) breakpoint.
> +	 */
> +	vcpu_args_set(vcpu, 3, 0, 0, brps);
> +
>  	for (stage = 0; stage < 11; stage++) {
>  		vcpu_run(vcpu);
>  
> -- 
> 2.37.1.595.g718a3a8f04-goog
> 
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm