[RFT 1/2] RISC-V: handle missing "no-map" properties for OpenSBI's PMP protected regions

[Alexandre Ghiti]

Hi Conor,

On 02/08/2023 13:12, Conor Dooley wrote:
> Add an erratum for versions [v0.8 to v1.3) of OpenSBI which fail to add
> the "no-map" property to the reserved memory nodes for the regions it
> has protected using PMPs.
>
> Our existing fix sweeping hibernation under the carpet by marking it
> NONPORTABLE is insufficient as there are other ways to generate
> accesses to these reserved memory regions, as Petr discovered [1]
> while testing crash kernels & kdump.
>
> Intercede during the boot process when the afflicted versions of OpenSBI
> are present & set the "no-map" property in all "mmode_resv" nodes before
> the kernel does its reserved memory region initialisation.
>
> Reported-by: Song Shuai <suagrfillet at gmail.com>
> Link: https://lore.kernel.org/all/CAAYs2=gQvkhTeioMmqRDVGjdtNF_vhB+vm_1dHJxPNi75YDQ_Q@mail.gmail.com/
> Reported-by: JeeHeng Sia <jeeheng.sia at starfivetech.com>
> Link: https://groups.google.com/a/groups.riscv.org/g/sw-dev/c/ITXwaKfA6z8
> Reported-by: Petr Tesarik <petrtesarik at huaweicloud.com>
> Closes: https://lore.kernel.org/linux-riscv/76ff0f51-d6c1-580d-f943-061e93073306@huaweicloud.com/ [1]
> CC: stable at vger.kernel.org
> Signed-off-by: Conor Dooley <conor.dooley at microchip.com>
> ---
>   arch/riscv/include/asm/sbi.h |  5 +++++
>   arch/riscv/kernel/sbi.c      | 42 +++++++++++++++++++++++++++++++++++-
>   arch/riscv/mm/init.c         |  3 +++
>   3 files changed, 49 insertions(+), 1 deletion(-)
>
> diff --git a/arch/riscv/include/asm/sbi.h b/arch/riscv/include/asm/sbi.h
> index 5b4a1bf5f439..5360f3476278 100644
> --- a/arch/riscv/include/asm/sbi.h
> +++ b/arch/riscv/include/asm/sbi.h
> @@ -252,6 +252,9 @@ enum sbi_pmu_ctr_type {
>   #define SBI_ERR_ALREADY_STARTED -7
>   #define SBI_ERR_ALREADY_STOPPED -8
>   
> +/* SBI implementation IDs */
> +#define SBI_IMP_OPENSBI	1
> +
>   extern unsigned long sbi_spec_version;
>   struct sbiret {
>   	long error;
> @@ -259,6 +262,8 @@ struct sbiret {
>   };
>   
>   void sbi_init(void);
> +void sbi_apply_reserved_mem_erratum(void *dtb_va);
> +
>   struct sbiret sbi_ecall(int ext, int fid, unsigned long arg0,
>   			unsigned long arg1, unsigned long arg2,
>   			unsigned long arg3, unsigned long arg4,
> diff --git a/arch/riscv/kernel/sbi.c b/arch/riscv/kernel/sbi.c
> index c672c8ba9a2a..aeb27263fa53 100644
> --- a/arch/riscv/kernel/sbi.c
> +++ b/arch/riscv/kernel/sbi.c
> @@ -5,8 +5,10 @@
>    * Copyright (c) 2020 Western Digital Corporation or its affiliates.
>    */
>   
> +#include <linux/acpi.h>
>   #include <linux/bits.h>
>   #include <linux/init.h>
> +#include <linux/libfdt.h>
>   #include <linux/pm.h>
>   #include <linux/reboot.h>
>   #include <asm/sbi.h>
> @@ -583,6 +585,40 @@ long sbi_get_mimpid(void)
>   }
>   EXPORT_SYMBOL_GPL(sbi_get_mimpid);
>   
> +static long sbi_firmware_id;
> +static long sbi_firmware_version;
> +
> +/*
> + * For devicetrees patched by OpenSBI a "mmode_resv" node is added to cover
> + * the region OpenSBI has protected by means of a PMP. Some versions of OpenSBI,
> + * [v0.8 to v1.3), omitted the "no-map" property, but this trips up hibernation
> + * among other things.
> + */
> +void __init sbi_apply_reserved_mem_erratum(void *dtb_pa)
> +{
> +	int child, reserved_mem;
> +
> +	if (sbi_firmware_id != SBI_IMP_OPENSBI)
> +		return;
> +
> +	if (!acpi_disabled)
> +		return;
> +
> +	if (sbi_firmware_version >= 0x10003 || sbi_firmware_version < 0x8)
> +		return;
> +
> +	reserved_mem = fdt_path_offset((void *)dtb_pa, "/reserved-memory");
> +	if (reserved_mem < 0)
> +		return;
> +
> +	fdt_for_each_subnode(child, (void *)dtb_pa, reserved_mem) {
> +		const char *name = fdt_get_name((void *)dtb_pa, child, NULL);
> +
> +		if (!strncmp(name, "mmode_resv", 10))


I would check that name != NULL before strncmp.


> +			fdt_setprop((void *)dtb_pa, child, "no-map", NULL, 0);
> +	}
> +};
> +
>   void __init sbi_init(void)
>   {
>   	int ret;
> @@ -596,8 +632,12 @@ void __init sbi_init(void)
>   		sbi_major_version(), sbi_minor_version());
>   
>   	if (!sbi_spec_is_0_1()) {
> +		sbi_firmware_id = sbi_get_firmware_id();
> +		sbi_firmware_version = sbi_get_firmware_version();
> +
>   		pr_info("SBI implementation ID=0x%lx Version=0x%lx\n",
> -			sbi_get_firmware_id(), sbi_get_firmware_version());
> +			sbi_firmware_id, sbi_firmware_version);
> +
>   		if (sbi_probe_extension(SBI_EXT_TIME)) {
>   			__sbi_set_timer = __sbi_set_timer_v02;
>   			pr_info("SBI TIME extension detected\n");
> diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c
> index 70fb31960b63..cb16bfdeacdb 100644
> --- a/arch/riscv/mm/init.c
> +++ b/arch/riscv/mm/init.c
> @@ -29,6 +29,7 @@
>   #include <asm/tlbflush.h>
>   #include <asm/sections.h>
>   #include <asm/soc.h>
> +#include <asm/sbi.h>
>   #include <asm/io.h>
>   #include <asm/ptdump.h>
>   #include <asm/numa.h>
> @@ -253,6 +254,8 @@ static void __init setup_bootmem(void)
>   	 * in the device tree, otherwise the allocation could end up in a
>   	 * reserved region.
>   	 */
> +
> +	sbi_apply_reserved_mem_erratum(dtb_early_va);
>   	early_init_fdt_scan_reserved_mem();
>   
>   	/*


Otherwise the patch looks good to me:

You can add:

Reviewed-by: Alexandre Ghiti <alexghiti at rivosinc.com>

I just have one question though (that we discussed privately already): 
should we fix openSBI in the kernel? If yes, what makes a bug worth 
being fixed in the kernel?

Thanks,