linux-kernel - Re: [PATCH 2/2] riscv: Support CONFIG_STRICT

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <29425dbf7d54bab2733d28480d3adb61@mailhost.ics.forth.gr>
Date:   Tue, 16 Jun 2020 15:27:36 +0300
From:   Nick Kossifidis <mick@....forth.gr>
To:     Zong Li <zong.li@...ive.com>
Cc:     paul.walmsley@...ive.com, palmer@...belt.com,
        linux-riscv@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] riscv: Support CONFIG_STRICT_DEVMEM

Στις 2020-06-16 10:45, Zong Li έγραψε:
> Implement the 'devmem_is_allowed()' interface for RISC-V, like some of
> other architectures have done. It will be called from 
> range_is_allowed()
> when userpsace attempts to access /dev/mem.
> 
> Access to exclusive IOMEM and kernel RAM is denied unless
> CONFIG_STRICT_DEVMEM is set to 'n'.
> 
> Test it by devmem, the result as follows:
> 
>  - CONFIG_STRICT_DEVMEM=y
> 	$ devmem 0x10010000
> 	0x00000000
> 	$ devmem 0x80200000
> 	0x0000106F
> 
>  - CONFIG_STRICT_DEVMEM is not set
> 	$ devmem 0x10010000
> 	devmem: mmap: Operation not permitted
> 	$ devmem 0x80200000
> 	devmem: mmap: Operation not permitted
> 
> Signed-off-by: Zong Li <zong.li@...ive.com>
> ---
>  arch/riscv/Kconfig          |  1 +
>  arch/riscv/include/asm/io.h |  2 ++
>  arch/riscv/mm/init.c        | 19 +++++++++++++++++++
>  3 files changed, 22 insertions(+)
> 
> diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
> index 128192e14ff2..ffd7841ede4c 100644
> --- a/arch/riscv/Kconfig
> +++ b/arch/riscv/Kconfig
> @@ -16,6 +16,7 @@ config RISCV
>  	select ARCH_HAS_BINFMT_FLAT
>  	select ARCH_HAS_DEBUG_VIRTUAL if MMU
>  	select ARCH_HAS_DEBUG_WX
> +	select ARCH_HAS_DEVMEM_IS_ALLOWED
>  	select ARCH_HAS_GCOV_PROFILE_ALL
>  	select ARCH_HAS_GIGANTIC_PAGE
>  	select ARCH_HAS_MMIOWB
> diff --git a/arch/riscv/include/asm/io.h b/arch/riscv/include/asm/io.h
> index 3835c3295dc5..04ac65ab93ce 100644
> --- a/arch/riscv/include/asm/io.h
> +++ b/arch/riscv/include/asm/io.h
> @@ -147,4 +147,6 @@ __io_writes_outs(outs, u64, q, __io_pbr(), 
> __io_paw())
> 
>  #include <asm-generic/io.h>
> 
> +extern int devmem_is_allowed(unsigned long pfn);
> +
>  #endif /* _ASM_RISCV_IO_H */
> diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c
> index bbe816e03b2f..5e7e61519acc 100644
> --- a/arch/riscv/mm/init.c
> +++ b/arch/riscv/mm/init.c
> @@ -517,6 +517,25 @@ void mark_rodata_ro(void)
>  }
>  #endif
> 
> +#ifdef CONFIG_STRICT_DEVMEM
> +#include <linux/ioport.h>
> +/*
> + * devmem_is_allowed() checks to see if /dev/mem access to a certain 
> address
> + * is valid. The argument is a physical page number.
> + *
> + * Disallow access to system RAM as well as device-exclusive MMIO 
> regions.
> + * This effectively disable read()/write() on /dev/mem.
> + */
> +int devmem_is_allowed(unsigned long pfn)
> +{
> +	if (iomem_is_exclusive(pfn << PAGE_SHIFT))
> +		return 0;
> +	if (!page_is_ram(pfn))
> +		return 1;
> +	return 0;
> +}
> +#endif
> +
>  void __init resource_init(void)
>  {
>  	struct memblock_region *region;

This shouldn't be part of /mm/init.c, it has nothing to do with memory 
initialization, I suggest we move it to another file like mmap.c on 
arm/arm64. Also before using iomem_is_exclusive we should probably also 
mark any reserved regions with the no-map attribute as busy|exclusive, 
reserved-memory regions are not necessarily part of the main memory so 
the page_is_ram check may pass and iomem_is_exclusive won't do any good.