Re: [PATCH] arm: Handle starting up in secure mode

From: Dave Martin
Date: Wed Aug 26 2015 - 06:39:52 EST

Next message: Will Deacon: "Re: [PATCH v5 0/8] Add generic support for relaxed atomics"
Previous message: Jagan Teki: "Re: [PATCH 3/3] mtd: spi-nor: sf: Add clear flag status register support"
In reply to: Ard Biesheuvel: "Re: [PATCH] arm: Handle starting up in secure mode"
Next in thread: Russell King - ARM Linux: "Re: [PATCH] arm: Handle starting up in secure mode"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Mon, Aug 24, 2015 at 09:55:26AM -0400, Christopher Covington wrote:
> ARM Linux appears to have never been made aware of the ARMv7 security
> extensions. When CONFIG_ARM_SEC_EXT=y, have it probe for its security
> state by checking whether CNTFRQ is writeable and potentially make
> mode changes based on the information. The most features are available
> from hypervisor (HYP) mode, so switch to it possible. Failing that,
> prefer non-secure supervisor (SVC) mode to secure supervisor mode.

Up to now we've steered clear of this, since it's a bit of a fig leaf
for broken firmware unless Linux actually has some valid use for the
Security Extensions itself.

Shouldn't the bootloader or firmware be doing this stuff, and if not,
why not?

Some other things that would need to be considered in any case:

* SoC-specific setup of the Non-secure view of the system: This has
to happen very early, so making it DT aware is going to be hard --
failing that, we are effectively risking bringing back board files.
The split in responsibility between firmware/bootloader and kernel
needs to be clearly defined and (as far as possible) platform-
independent, otherwise we'll have total chaos.

* Out of reset, generally the CPU state is only fully defined for the
highest exception level. You probably need to be doing more setup
than you're currently doing.

* SMP, secondary boot and suspend/resume -- again involving board-
specific code.

* You need to safely "park" the Secure World before running anything
in Non-Secure. As a minimum, you would need to quiesce any
Secure interrupt sources, disable all interrupt traps to Monitor
mode, and make sure that the Monitor vectors point somewhere
real, so that executing SMC doesn't send the CPU off into the
long grass...

Cheers
---Dave

>
> Signed-off-by: Christopher Covington <cov@xxxxxxxxxxxxxx>
> ---
> arch/arm/include/asm/sec.h | 27 +++++++
> arch/arm/include/uapi/asm/ptrace.h | 1 +
> arch/arm/kernel/Makefile | 1 +
> arch/arm/kernel/head.S | 3 +
> arch/arm/kernel/mon-stub.S | 158 +++++++++++++++++++++++++++++++++++++
> arch/arm/mm/Kconfig | 20 ++++-
> 6 files changed, 206 insertions(+), 4 deletions(-)
> create mode 100644 arch/arm/include/asm/sec.h
> create mode 100644 arch/arm/kernel/mon-stub.S
>
> diff --git a/arch/arm/include/asm/sec.h b/arch/arm/include/asm/sec.h
> new file mode 100644
> index 0000000..4a9a573
> --- /dev/null
> +++ b/arch/arm/include/asm/sec.h
> @@ -0,0 +1,27 @@
> +/*
> + * Copyright (c) 2014, The Linux Foundation. All rights reserved.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 and
> + * only version 2 as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + */
> +
> +#ifndef SEC_H
> +#define SEC_H
> +
> +#ifndef __ASSEMBLY__
> +
> +#ifdef CONFIG_ARM_SEC_EXT
> +extern int __boot_cpu_secure;
> +#else
> +#define __boot_cpu_secure 0
> +#endif
> +
> +#endif /* ! __ASSEMBLY__ */
> +
> +#endif /* SEC_H */
> diff --git a/arch/arm/include/uapi/asm/ptrace.h b/arch/arm/include/uapi/asm/ptrace.h
> index 5af0ed1..70ff6bf 100644
> --- a/arch/arm/include/uapi/asm/ptrace.h
> +++ b/arch/arm/include/uapi/asm/ptrace.h
> @@ -53,6 +53,7 @@
> #endif
> #define FIQ_MODE 0x00000011
> #define IRQ_MODE 0x00000012
> +#define MON_MODE 0x00000016
> #define ABT_MODE 0x00000017
> #define HYP_MODE 0x0000001a
> #define UND_MODE 0x0000001b
> diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
> index e69f7a1..a60a0f7 100644
> --- a/arch/arm/kernel/Makefile
> +++ b/arch/arm/kernel/Makefile
> @@ -87,6 +87,7 @@ head-y := head$(MMUEXT).o
> obj-$(CONFIG_DEBUG_LL) += debug.o
> obj-$(CONFIG_EARLY_PRINTK) += early_printk.o
>
> +obj-$(CONFIG_ARM_SEC_EXT) += mon-stub.o
> obj-$(CONFIG_ARM_VIRT_EXT) += hyp-stub.o
> ifeq ($(CONFIG_ARM_PSCI),y)
> obj-y += psci.o psci-call.o
> diff --git a/arch/arm/kernel/head.S b/arch/arm/kernel/head.S
> index 29e2991..d137ba4 100644
> --- a/arch/arm/kernel/head.S
> +++ b/arch/arm/kernel/head.S
> @@ -85,6 +85,9 @@ ENTRY(stext)
> THUMB( .thumb ) @ switch to Thumb now.
> THUMB(1: )
>
> +#ifdef CONFIG_ARM_SEC_EXT
> + bl __mon_stub_install
> +#endif
> #ifdef CONFIG_ARM_VIRT_EXT
> bl __hyp_stub_install
> #endif
> diff --git a/arch/arm/kernel/mon-stub.S b/arch/arm/kernel/mon-stub.S
> new file mode 100644
> index 0000000..ab1a361
> --- /dev/null
> +++ b/arch/arm/kernel/mon-stub.S
> @@ -0,0 +1,158 @@
> +/*
> + * Copyright (c) 2014, The Linux Foundation. All rights reserved.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 and
> + * only version 2 as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + */
> +
> +#include <linux/linkage.h>
> +#include <asm/assembler.h>
> +#include <asm/sec.h>
> +
> +.data
> +ENTRY(__boot_cpu_secure)
> + .long 0
> +.text
> +
> +/*
> + * ARM Linux has the most features available in hypervisor mode and
> + * running in non-secure mode is recommended. Thus, try to get into
> + * hypervisor mode if we're not already there, or failing that, try
> + * to get into non-secure supervisor mode.
> + */
> +ENTRY(__mon_stub_install)
> + /*
> + * Store the mode field of the CPSR in r4 and return early if we're
> + * already in hypervisor mode.
> + */
> + mrs r4, cpsr
> + and r4, r4, #MODE_MASK
> + cmp r4, #HYP_MODE
> + reteq lr
> +
> + /*
> + * Save the link register in a non-banked register, r5, so that we
> + * still have access to it after mode switches.
> + */
> + mov r5, lr
> +
> + /*
> + * Read ID_PFR1 and store the value in r6. This register indicates
> + * the presence of the security and virtualization extensions. The
> + * former is interesting because we must traverse secure monitor mode
> + * to get to hypervisor mode and it allows easy manipulation of
> + * exception vectors via the Vector Base Address Register (VBAR).
> + *
> + * ID_PFR1 also indicates whether the generic timer is present, which
> + * has a handy register for our purposes, CNTFRQ. Accesses won't trap
> + * even with higher exception levels in AArch64 and writes will only
> + * succeed from the highest exception level on a system (the undefined
> + * exception from a failed write is used as a branch).
> + */
> +
> + mrc p15, 0, r6, c0, c1, 1 @ ID_PFR1
> +
> + /*
> + * If we're in monitor mode then we know the security setting and can
> + * begin the transition to hypervisor or non-secure supervisor mode
> + * immediately.
> + */
> +
> + cmp r4, #MON_MODE
> + beq monitor
> +
> + /*
> + * Check that the security extensions and generic timer are present.
> + */
> + and r7, r6, #0xf0
> + cmp r7, #0x10
> + cmpne r7, #0x20
> + retne lr
> + and r8, r6, #0xf0000
> + cmp r8, #0x10000
> + retne lr
> +
> + /*
> + * Set things up so that if a CNTFRQ access causes an undefined
> + * instruction exception, we return to the address indicated in r5
> + * in the mode indicated in r4.
> + */
> + adr r7, __mon_stub_vectors
> + mcr p15, 0, r7, c12, c0, 0 @ set vector base address (VBAR)
> +
> + mrc p15, 0, r7, c14, c0, 0 @ CNTFRQ
> + mcr p15, 0, r7, c14, c0, 0 @ CNTFRQ
> +
> + /*
> + * If we got this far, switch to monitor mode and prepare for further
> + * switching.
> + */
> + cps #MON_MODE
> +
> +monitor:
> + /*
> + * TODO: Handle SIF and separate secure physical memory in general
> + */
> + mrc p15, 0, r8, c1, c1, 0 @ get secure configuration (SCR)
> + orr r8, r8, #1 @ non-secure (NS)
> +
> + mov r8, #1
> + ldr r7, =__boot_cpu_secure
> + str r8, [r7]
> +
> + /*
> + * See whether the switch will be to hypervisor or supervisor.
> + */
> + and r7, r6, #0xf000
> + cmp r7, #0x1000
> + mcrne p15, 0, r8, c1, c1, 0 @ set secure configuration (SCR)
> + bne supervise
> +
> + orr r8, r8, #0x100 @ hypercall enable (HCE)
> + mcr p15, 0, r8, c1, c1, 0 @ set secure configuration (SCR)
> +
> + mov lr, r5
> + msr spsr, #HYP_MODE
> + __ERET
> +
> +supervise:
> + /*
> + * Switch to supervisor mode and return using a non-banked register,
> + * in case we're not in the mode we started out in.
> + */
> + cps #SVC_MODE
> + ret r5
> +ENDPROC(__mon_stub_install)
> +
> +ENTRY(__mon_stub_do_undef)
> + /*
> + * Return to the address at r5 with the mode in r4. If there is an
> + * illegal or unimplemented mode in r4 the cpsr write will cause an
> + * undefined exception, recursing. To avoid that, put the current
> + * mode in r4 before performing the write.
> + */
> + mrs r6, cpsr
> + bic r7, r6, #MODE_MASK
> + orr r7, r7, r4
> + and r4, r6, #MODE_MASK
> + msr cpsr, r7
> + ret r5
> +ENDPROC(__mon_stub_do_undef)
> +
> +.align 5
> +__mon_stub_vectors:
> +__mon_stub_reset: W(b) .
> +__mon_stub_und: W(b) __mon_stub_do_undef
> +__mon_stub_call: W(b) .
> +__mon_stub_pabort: W(b) .
> +__mon_stub_dabort: W(b) .
> +__mon_stub_trap: W(b) .
> +__mon_stub_irq: W(b) .
> +__mon_stub_fiq: W(b) .
> +ENDPROC(__mon_stub_vectors)
> diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
> index 7c6b976..32fa451 100644
> --- a/arch/arm/mm/Kconfig
> +++ b/arch/arm/mm/Kconfig
> @@ -676,7 +676,7 @@ config ARM_THUMBEE
> make use of it. Say N for code that can run on CPUs without ThumbEE.
>
> config ARM_VIRT_EXT
> - bool
> + bool "Support for Virtualization Extensions"
> depends on MMU
> default y if CPU_V7
> help
> @@ -684,9 +684,21 @@ config ARM_VIRT_EXT
> Extensions to install hypervisors without run-time firmware
> assistance.
>
> - A compliant bootloader is required in order to make maximum
> - use of this feature. Refer to Documentation/arm/Booting for
> - details.
> + A compliant bootloader or enabling ARM_SEC_EXT is required in
> + order to make maximum use of this feature. Refer to
> + Documentation/arm/Booting for details.
> +
> +config ARM_SEC_EXT
> + bool "Support for Security Extensions"
> + depends on MMU
> + default n
> + help
> + Say Y to have the kernel check for the presence of the ARM Security
> + Extensions and where possible, use them to switch to preferred
> + security and mode settings. This decreases the kernel's dependency
> + on bootloaders.
> +
> + If unsure, say N.
>
> config SWP_EMULATE
> bool "Emulate SWP/SWPB instructions" if !SMP
> --
> Qualcomm Innovation Center, Inc.
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
> a Linux Foundation Collaborative Project
>
>
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Next message: Will Deacon: "Re: [PATCH v5 0/8] Add generic support for relaxed atomics"
Previous message: Jagan Teki: "Re: [PATCH 3/3] mtd: spi-nor: sf: Add clear flag status register support"
In reply to: Ard Biesheuvel: "Re: [PATCH] arm: Handle starting up in secure mode"
Next in thread: Russell King - ARM Linux: "Re: [PATCH] arm: Handle starting up in secure mode"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]