[U-Boot] [PATCH v2 4/7] ARM: switch to non-secure state during bootm execution

Christoffer Dall christoffer.dall at linaro.org
Thu Jun 20 01:13:05 CEST 2013


On Thu, Jun 13, 2013 at 01:01:10PM +0200, Andre Przywara wrote:
> To actually trigger the non-secure switch we just implemented, call
> the switching routine from within the bootm command implementation.
> This way we automatically enable this feature without further user
> intervention.
> 
> The core specific part of the work is done in the assembly routine
> in nonsec_virt.S, introduced with the previous patch, but for the full
> glory we need to setup the GIC distributor interface once for the
> whole system, which is done in C here.
> The routine is placed in arch/arm/lib to allow easy access from
> different boards or CPUs.
> 
> We check the availability of the security extensions first.
> 
> The generic timer base frequency register is only accessible from
> secure state, so we have to program it now. Actually this should be
> done from primary firmware before, but some boards seems to omit
> this, so if needed we do this here with a board specific value.
> The Versatile Express board does not need this, so we remove the
> frequency from the configuration file here.
> 
> Since we need a safe way to access the GIC, we use the PERIPHBASE
> registers on Cortex-A15 and A7 CPUs and do some sanity checks.
> Board not implementing the CBAR can override this value via a
> configuration file variable.
> 
> Then we actually do the GIC enablement:
> a) enable the GIC distributor, both for non-secure and secure state
>    (GICD_CTLR[1:0] = 11b)
> b) allow all interrupts to be handled from non-secure state
>    (GICD_IGROUPRn = 0xFFFFFFFF)
> 
> The core specific GIC setup is then done in the assembly routine.
> 
> The actual bootm trigger is pretty small: calling the routine and
> doing some error reporting.
> 
> Signed-off-by: Andre Przywara <andre.przywara at linaro.org>
> ---
>  arch/arm/include/asm/armv7.h        |   7 ++
>  arch/arm/lib/Makefile               |   2 +
>  arch/arm/lib/bootm.c                |  20 ++++++
>  arch/arm/lib/virt-v7.c              | 137 ++++++++++++++++++++++++++++++++++++
>  include/configs/vexpress_ca15_tc2.h |   2 -
>  5 files changed, 166 insertions(+), 2 deletions(-)
>  create mode 100644 arch/arm/lib/virt-v7.c
> 
> diff --git a/arch/arm/include/asm/armv7.h b/arch/arm/include/asm/armv7.h
> index 989bb72..56d0dd0 100644
> --- a/arch/arm/include/asm/armv7.h
> +++ b/arch/arm/include/asm/armv7.h
> @@ -88,6 +88,13 @@ void v7_outer_cache_flush_range(u32 start, u32 end);
>  void v7_outer_cache_inval_range(u32 start, u32 end);
>  
>  #ifdef CONFIG_ARMV7_VIRT
> +
> +#define HYP_ERR_NO_SEC_EXT		2
> +#define HYP_ERR_NO_GIC_ADDRESS		3
> +#define HYP_ERR_GIC_ADDRESS_ABOVE_4GB	4

enum?

> +
> +int armv7_switch_nonsec(void);
> +
>  /* defined in cpu/armv7/nonsec_virt.S */
>  void _nonsec_init(void);
>  #endif /* CONFIG_ARMV7_VIRT */
> diff --git a/arch/arm/lib/Makefile b/arch/arm/lib/Makefile
> index 8ad9f66..1570ad5 100644
> --- a/arch/arm/lib/Makefile
> +++ b/arch/arm/lib/Makefile
> @@ -60,6 +60,8 @@ COBJS-y	+= reset.o
>  COBJS-y	+= cache.o
>  COBJS-y	+= cache-cp15.o
>  
> +COBJS-$(CONFIG_ARMV7_VIRT) += virt-v7.o
> +
>  SRCS	:= $(GLSOBJS:.o=.S) $(GLCOBJS:.o=.c) \
>  	   $(SOBJS-y:.o=.S) $(COBJS-y:.o=.c)
>  OBJS	:= $(addprefix $(obj),$(SOBJS-y) $(COBJS-y))
> diff --git a/arch/arm/lib/bootm.c b/arch/arm/lib/bootm.c
> index 1b6e0ac..8251a89 100644
> --- a/arch/arm/lib/bootm.c
> +++ b/arch/arm/lib/bootm.c
> @@ -34,6 +34,10 @@
>  #include <asm/bootm.h>
>  #include <linux/compiler.h>
>  
> +#ifdef CONFIG_ARMV7_VIRT
> +#include <asm/armv7.h>
> +#endif
> +
>  DECLARE_GLOBAL_DATA_PTR;
>  
>  static struct tag *params;
> @@ -222,6 +226,22 @@ static void boot_prep_linux(bootm_headers_t *images)
>  		printf("FDT and ATAGS support not compiled in - hanging\n");
>  		hang();
>  	}
> +#ifdef CONFIG_ARMV7_VIRT
> +	switch (armv7_switch_nonsec()) {
> +	case 0:
> +		debug("entered non-secure state\n");
> +		break;

this is weird, why not have a define for the success case?

I still think the debug printing should be done inside
armv7_switch_nonsec instead, and you can just have it be a void();

> +	case HYP_ERR_NO_SEC_EXT:
> +		printf("HYP mode: Security extensions not implemented.\n");
> +		break;
> +	case HYP_ERR_NO_GIC_ADDRESS:
> +		printf("HYP mode: could not determine GIC address.\n");
> +		break;
> +	case HYP_ERR_GIC_ADDRESS_ABOVE_4GB:
> +		printf("HYP mode: PERIPHBASE is above 4 GB, cannot access this.\n");
> +		break;
> +	}
> +#endif
>  }
>  
>  /* Subcommand: GO */
> diff --git a/arch/arm/lib/virt-v7.c b/arch/arm/lib/virt-v7.c
> new file mode 100644
> index 0000000..7876a77
> --- /dev/null
> +++ b/arch/arm/lib/virt-v7.c
> @@ -0,0 +1,137 @@
> +/*
> + * (C) Copyright 2013
> + * Andre Przywara, Linaro
> + *
> + * Routines to transition ARMv7 processors from secure into non-secure state
> + * needed to enable ARMv7 virtualization for current hypervisors
> + *
> + * See file CREDITS for list of people who contributed to this
> + * project.
> + *
> + * This program is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU General Public License as
> + * published by the Free Software Foundation; either version 2 of
> + * the License, or (at your option) any later version.
> + *
> + * 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.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, write to the Free Software
> + * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
> + * MA 02111-1307 USA
> + */
> +
> +#include <common.h>
> +#include <asm/armv7.h>
> +#include <asm/gic.h>
> +#include <asm/io.h>
> +
> +static unsigned int read_id_pfr1(void)
> +{
> +	unsigned int reg;
> +
> +	asm("mrc p15, 0, %0, c0, c1, 1\n" : "=r"(reg));
> +	return reg;
> +}
> +
> +/* The timer frequency for the generic timer needs to be
> + * programmed in secure state. Some primary bootloaders / firmware
> + * omit this, so if the frequency is provided in the configuration,
> + * we do this here instead.
> + * But first check if we have the generic timer.
> + */
> +static void set_generic_timer_frequency(void)
> +{
> +#ifdef CONFIG_SYS_CLK_FREQ
> +	unsigned int reg;
> +
> +	reg = read_id_pfr1();
> +	if ((reg & CPUID_ARM_TIMER_MASK) == 1U << CPUID_ARM_TIMER_SHIFT)
> +		asm("mcr p15, 0, %0, c14, c0, 0\n"
> +		: : "r"(CONFIG_SYS_CLK_FREQ));
> +#endif
> +}
> +
> +static int get_gic_base_address(char **gicdptr)

you could simplify this function and make it an unsigned function and
return (unsigned)-1 on error (and do the debug error print right away in
there).

> +{
> +#ifdef CONFIG_ARM_GIC_BASE_ADDRESS
> +	*gicdptr = (void *)(CONFIG_ARM_GIC_BASE_ADDRESS + GIC_DIST_OFFSET);
> +	return 0;
> +#else
> +	unsigned midr;
> +	unsigned periphbase;
> +
> +	/* check whether we are an Cortex-A15 or A7.
> +	 * The actual HYP switch should work with all CPUs supporting
> +	 * the virtualization extension, but we need the GIC address,
> +	 * which we know only for sure for those two CPUs.
> +	 */
> +	asm("mrc p15, 0, %0, c0, c0, 0\n" : "=r"(midr));
> +	switch (midr & MIDR_PRIMARY_PART_MASK) {
> +	case MIDR_CORTEX_A9_R0P1:
> +	case MIDR_CORTEX_A15_R0P0:
> +	case MIDR_CORTEX_A7_R0P0:
> +		break;
> +	default:
> +		return HYP_ERR_NO_GIC_ADDRESS;
> +	}
> +
> +	/* get the GIC base address from the CBAR register */
> +	asm("mrc p15, 4, %0, c15, c0, 0\n" : "=r" (periphbase));
> +
> +	/* the PERIPHBASE can be mapped above 4 GB (lower 8 bits used to
> +	 * encode this). Bail out here since we cannot access this without
> +	 * enabling paging.
> +	 */
> +	if ((periphbase & 0xff) != 0)
> +		return HYP_ERR_GIC_ADDRESS_ABOVE_4GB;
> +
> +	*gicdptr = (void *)(periphbase + GIC_DIST_OFFSET);

this is weird, the function is called get_gic_base_address, but you're
returning the distributor base address, and the GIC_DIST_OFFSET is
actually an A15/A7 specific thing.

> +
> +	return 0;
> +#endif
> +}
> +
> +int armv7_switch_nonsec(void)
> +{
> +	unsigned int reg, ret;
> +	char *gicdptr;

there's really no need having this be a pointer when you use writel /
readl to access it.

> +	unsigned itlinesnr, i;
> +
> +	/* check whether the CPU supports the security extensions */
> +	reg = read_id_pfr1();
> +	if ((reg & 0xF0) == 0)
> +		return HYP_ERR_NO_SEC_EXT;
> +
> +	set_generic_timer_frequency();
> +
> +	/* the SCR register will be set directly in the monitor mode handler,
> +	 * according to the spec one should not tinker with it in secure state
> +	 * in SVC mode. Do not try to read it once in non-secure state,
> +	 * any access to it will trap.
> +	 */
> +
> +	ret = get_gic_base_address(&gicdptr);
> +	if (ret != 0)
> +		return ret;
> +
> +	/* enable the GIC distributor */
> +	writel(readl(&gicdptr[GICD_CTLR]) | 0x03, &gicdptr[GICD_CTLR]);

I would do the readl first, and then the writel, but it's just a matter
of style.

"gicdptr + GICD_CTLR" seems cleaner

> +
> +	/* TYPER[4:0] contains an encoded number of all interrupts */

nit: s/all/avail./

> +	itlinesnr = readl(&gicdptr[GICD_TYPER]) & 0x1f;
> +
> +	/* set all bits in the GIC group registers to one to allow access
> +	 * from non-secure state
> +	 */
> +	for (i = 0; i <= itlinesnr; i++)
> +		writel((unsigned)-1, &gicdptr[GICD_IGROUPRn + 4 * i]);
> +
> +	/* call the non-sec switching code on this CPU */
> +	_nonsec_init();
> +
> +	return 0;
> +}
> diff --git a/include/configs/vexpress_ca15_tc2.h b/include/configs/vexpress_ca15_tc2.h
> index 9e230ad..4f425ac 100644
> --- a/include/configs/vexpress_ca15_tc2.h
> +++ b/include/configs/vexpress_ca15_tc2.h
> @@ -31,6 +31,4 @@
>  #include "vexpress_common.h"
>  #define CONFIG_BOOTP_VCI_STRING     "U-boot.armv7.vexpress_ca15x2_tc2"
>  
> -#define CONFIG_SYS_CLK_FREQ 24000000
> -
>  #endif
> -- 
> 1.7.12.1
> 

Besides my crazy nit-picking, this looks good to me.

-Christoffer


More information about the U-Boot mailing list