KVM ARM Archive on lore.kernel.org
 help / color / Atom feed
From: Andrew Jones <drjones@redhat.com>
To: Alexandru Elisei <alexandru.elisei@arm.com>
Cc: kvm@vger.kernel.org, maz@kernel.org, andre.przywara@arm.com,
	pbonzini@redhat.com, kvmarm@lists.cs.columbia.edu
Subject: Re: [kvm-unit-tests RFC PATCH 02/16] arm/arm64: psci: Don't run C code without stack or vectors
Date: Wed, 28 Aug 2019 16:45:22 +0200
Message-ID: <20190828144522.qkmckjcmrdayfq7r@kamzik.brq.redhat.com> (raw)
In-Reply-To: <1566999511-24916-3-git-send-email-alexandru.elisei@arm.com>

On Wed, Aug 28, 2019 at 02:38:17PM +0100, Alexandru Elisei wrote:
> The psci test performs a series of CPU_ON/CPU_OFF cycles for CPU 1. This is
> done by setting the entry point for the CPU_ON call to the physical address
> of the C function cpu_psci_cpu_die.
> 
> The compiler is well within its rights to use the stack when generating
> code for cpu_psci_cpu_die.  However, because no stack initialization has
> been done, the stack pointer is zero, as set by KVM when creating the VCPU.
> This causes a data abort without a change in exception level. The VBAR_EL1
> register is also zero (the KVM reset value for VBAR_EL1), the MMU is off,
> and we end up trying to fetch instructions from address 0x200.
> 
> At this point, a stage 2 instruction abort is generated which is taken to
> KVM. KVM interprets this as an instruction fetch from an I/O region, and
> injects a prefetch abort into the guest. Prefetch abort is a synchronous
> exception, and on guest return the VCPU PC will be set to VBAR_EL1 + 0x200,
> which is...  0x200. The VCPU ends up in an infinite loop causing a prefetch
> abort while fetching the instruction to service the said abort.
> 
> cpu_psci_cpu_die is basically a wrapper over the HVC instruction, so
> provide an assembly implementation for the function which will serve as the
> entry point for CPU_ON.
> 
> Signed-off-by: Alexandru Elisei <alexandru.elisei@arm.com>
> ---
>  arm/cstart.S   | 7 +++++++
>  arm/cstart64.S | 7 +++++++
>  arm/psci.c     | 5 +++--
>  3 files changed, 17 insertions(+), 2 deletions(-)
> 
> diff --git a/arm/cstart.S b/arm/cstart.S
> index 114726feab82..5d4fe4b1570b 100644
> --- a/arm/cstart.S
> +++ b/arm/cstart.S
> @@ -7,6 +7,7 @@
>   */
>  #define __ASSEMBLY__
>  #include <auxinfo.h>
> +#include <linux/psci.h>
>  #include <asm/thread_info.h>
>  #include <asm/asm-offsets.h>
>  #include <asm/ptrace.h>
> @@ -138,6 +139,12 @@ secondary_entry:
>  	blx	r0
>  	b	do_idle
>  
> +.global asm_cpu_psci_cpu_die
> +asm_cpu_psci_cpu_die:
> +	ldr	r0, =PSCI_0_2_FN_CPU_OFF
> +	hvc	#0
> +	b	halt

Shouldn't we load PSCI_POWER_STATE_TYPE_POWER_DOWN into r1 and
zero out r2 and r3, as cpu_psci_cpu_die() does? And maybe we
should just do a 'b .' here instead of 'b halt' in order to
avoid confusion as to how we ended up in halt(), if the psci
invocation were to ever fail.

> +
>  .globl halt
>  halt:
>  1:	wfi
> diff --git a/arm/cstart64.S b/arm/cstart64.S
> index b0e8baa1a23a..20f832fd57f7 100644
> --- a/arm/cstart64.S
> +++ b/arm/cstart64.S
> @@ -7,6 +7,7 @@
>   */
>  #define __ASSEMBLY__
>  #include <auxinfo.h>
> +#include <linux/psci.h>
>  #include <asm/asm-offsets.h>
>  #include <asm/ptrace.h>
>  #include <asm/processor.h>
> @@ -128,6 +129,12 @@ secondary_entry:
>  	blr	x0
>  	b	do_idle
>  
> +.globl asm_cpu_psci_cpu_die
> +asm_cpu_psci_cpu_die:
> +	ldr	x0, =PSCI_0_2_FN_CPU_OFF
> +	hvc	#0
> +	b	halt

Same as above

> +
>  .globl halt
>  halt:
>  1:	wfi
> diff --git a/arm/psci.c b/arm/psci.c
> index 5cb4d5c7c233..0440c4cdbc59 100644
> --- a/arm/psci.c
> +++ b/arm/psci.c
> @@ -72,6 +72,7 @@ static int cpu_on_ret[NR_CPUS];
>  static cpumask_t cpu_on_ready, cpu_on_done;
>  static volatile int cpu_on_start;
>  
> +extern void asm_cpu_psci_cpu_die(void);
>  static void cpu_on_secondary_entry(void)
>  {
>  	int cpu = smp_processor_id();
> @@ -79,7 +80,7 @@ static void cpu_on_secondary_entry(void)
>  	cpumask_set_cpu(cpu, &cpu_on_ready);
>  	while (!cpu_on_start)
>  		cpu_relax();
> -	cpu_on_ret[cpu] = psci_cpu_on(cpus[1], __pa(cpu_psci_cpu_die));
> +	cpu_on_ret[cpu] = psci_cpu_on(cpus[1], __pa(asm_cpu_psci_cpu_die));
>  	cpumask_set_cpu(cpu, &cpu_on_done);
>  }
>  
> @@ -104,7 +105,7 @@ static bool psci_cpu_on_test(void)
>  	cpu_on_start = 1;
>  	smp_mb();
>  
> -	cpu_on_ret[0] = psci_cpu_on(cpus[1], __pa(cpu_psci_cpu_die));
> +	cpu_on_ret[0] = psci_cpu_on(cpus[1], __pa(asm_cpu_psci_cpu_die));
>  	cpumask_set_cpu(0, &cpu_on_done);
>  
>  	while (!cpumask_full(&cpu_on_done))
> -- 
> 2.7.4
>

Thanks,
drew 
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm

  reply index

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-28 13:38 [kvm-unit-tests RFC PATCH 00/16] arm64: Run at EL2 Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 01/16] arm: selftest.c: Remove redundant check for Exception Level Alexandru Elisei
2019-08-28 14:32   ` Andrew Jones
2019-08-28 15:39     ` Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 02/16] arm/arm64: psci: Don't run C code without stack or vectors Alexandru Elisei
2019-08-28 14:45   ` Andrew Jones [this message]
2019-08-28 15:14     ` Alexandru Elisei
2019-09-02 14:55       ` Alexandru Elisei
2019-09-03  6:37         ` Andrew Jones
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 03/16] lib: arm/arm64: Add missing include for alloc_page.h in pgtable.h Alexandru Elisei
2019-08-28 14:47   ` Andrew Jones
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 04/16] arm/arm64: selftest: Add prefetch abort test Alexandru Elisei
2019-08-28 14:09   ` Mark Rutland
2019-08-29  8:18     ` Alexandru Elisei
2019-08-29 10:19       ` Mark Rutland
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 05/16] arm64: timer: Write to ICENABLER to disable timer IRQ Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 06/16] arm64: timer: EOIR the interrupt after masking the timer Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 07/16] arm64: timer: Test behavior when timer disabled or masked Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 08/16] lib: arm/arm64: Refuse to disable the MMU with non-identity stack pointer Alexandru Elisei
2019-08-28 14:55   ` Andrew Jones
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 09/16] lib: arm/arm64: Invalidate TLB before enabling MMU Alexandru Elisei
2019-08-28 14:59   ` Andrew Jones
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 10/16] lib: Add UL and ULL definitions to linux/const.h Alexandru Elisei
2019-08-28 15:10   ` Andrew Jones
2019-08-28 15:46     ` Alexandru Elisei
2019-08-28 16:19       ` Andrew Jones
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 11/16] lib: arm64: Run existing tests at EL2 Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 12/16] arm64: timer: Add test for EL2 timers Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 13/16] arm64: selftest: Add basic test for EL2 Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 14/16] lib: arm64: Add support for disabling and re-enabling VHE Alexandru Elisei
2019-08-28 14:19   ` Mark Rutland
2019-08-29  8:36     ` Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 15/16] arm64: selftest: Expand EL2 test to disable and re-enable VHE Alexandru Elisei
2019-08-28 13:38 ` [kvm-unit-tests RFC PATCH 16/16] arm64: timer: Run tests with VHE disabled Alexandru Elisei

Reply instructions:

You may reply publically to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190828144522.qkmckjcmrdayfq7r@kamzik.brq.redhat.com \
    --to=drjones@redhat.com \
    --cc=alexandru.elisei@arm.com \
    --cc=andre.przywara@arm.com \
    --cc=kvm@vger.kernel.org \
    --cc=kvmarm@lists.cs.columbia.edu \
    --cc=maz@kernel.org \
    --cc=pbonzini@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

KVM ARM Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/kvmarm/0 kvmarm/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 kvmarm kvmarm/ https://lore.kernel.org/kvmarm \
		kvmarm@lists.cs.columbia.edu
	public-inbox-index kvmarm

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/edu.columbia.cs.lists.kvmarm


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git