linux-kernel - Re: [PATCH] sh: use generic uaccess

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: <1ab7594855718e24ddc629ebbab1edc8bdcd799a.camel@physik.fu-berlin.de>
Date: Tue, 23 Jan 2024 14:55:35 +0100
From: John Paul Adrian Glaubitz <glaubitz@...sik.fu-berlin.de>
To: Arnd Bergmann <arnd@...nel.org>, Yoshinori Sato
 <ysato@...rs.sourceforge.jp>,  Rich Felker <dalias@...c.org>
Cc: Arnd Bergmann <arnd@...db.de>, linux-sh@...r.kernel.org, 
	linux-kernel@...r.kernel.org, linux-riscv@...ts.infradead.org
Subject: Re: [PATCH] sh: use generic uaccess

Hi Arnd,

On Tue, 2024-01-23 at 14:23 +0100, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@...db.de>
> 
> As reported by many people, the nommu SH code runs into a compiler error
> with a newly added syscall:
> 
>   + {standard input}: Error: displacement to undefined symbol .L105 overflows 8-bit field :  => 590, 593
>   + {standard input}: Error: displacement to undefined symbol .L135 overflows 8-bit field :  => 603
>   + {standard input}: Error: displacement to undefined symbol .L140 overflows 8-bit field :  => 606
>   + {standard input}: Error: displacement to undefined symbol .L76 overflows 12-bit field:  => 591, 594
>   + {standard input}: Error: displacement to undefined symbol .L77 overflows 8-bit field : 607 => 607, 582, 585
>   + {standard input}: Error: displacement to undefined symbol .L97 overflows 12-bit field:  => 607
>   + {standard input}: Error: pcrel too far: 604, 590, 577, 593, 572, 569, 598, 599, 596, 610 => 610, 574, 599, 569, 598, 596, 601, 590, 604, 595, 572, 577, 593
> 
> Avoid the code that triggers this entirely by using the same generic
> uaccess code that m68k and riscv have on nommu.
> 
> Link: https://lore.kernel.org/all/07d8877b-d933-46f4-8ca4-c10ed602f37e@app.fastmail.com/
> Signed-off-by: Arnd Bergmann <arnd@...db.de>
> ---
>  arch/sh/include/asm/uaccess.h    |  5 +++++
>  arch/sh/include/asm/uaccess_32.h | 23 -----------------------
>  2 files changed, 5 insertions(+), 23 deletions(-)
> 
> diff --git a/arch/sh/include/asm/uaccess.h b/arch/sh/include/asm/uaccess.h
> index a79609eb14be..b42764d55901 100644
> --- a/arch/sh/include/asm/uaccess.h
> +++ b/arch/sh/include/asm/uaccess.h
> @@ -2,6 +2,7 @@
>  #ifndef __ASM_SH_UACCESS_H
>  #define __ASM_SH_UACCESS_H
>  
> +#ifdef CONFIG_MMU
>  #include <asm/extable.h>
>  #include <asm-generic/access_ok.h>
>  
> @@ -130,4 +131,8 @@ struct mem_access {
>  int handle_unaligned_access(insn_size_t instruction, struct pt_regs *regs,
>  			    struct mem_access *ma, int, unsigned long address);
>  
> +#else
> +#include <asm-generic/uaccess.h>
> +#endif
> +
>  #endif /* __ASM_SH_UACCESS_H */
> diff --git a/arch/sh/include/asm/uaccess_32.h b/arch/sh/include/asm/uaccess_32.h
> index 5d7ddc092afd..e053f2fd245c 100644
> --- a/arch/sh/include/asm/uaccess_32.h
> +++ b/arch/sh/include/asm/uaccess_32.h
> @@ -35,7 +35,6 @@ do {								\
>  	}							\
>  } while (0)
>  
> -#ifdef CONFIG_MMU
>  #define __get_user_asm(x, addr, err, insn) \
>  ({ \
>  __asm__ __volatile__( \
> @@ -56,16 +55,6 @@ __asm__ __volatile__( \
>  	".previous" \
>  	:"=&r" (err), "=&r" (x) \
>  	:"m" (__m(addr)), "i" (-EFAULT), "0" (err)); })
> -#else
> -#define __get_user_asm(x, addr, err, insn)		\
> -do {							\
> -	__asm__ __volatile__ (				\
> -		"mov." insn "	%1, %0\n\t"		\
> -		: "=&r" (x)				\
> -		: "m" (__m(addr))			\
> -	);						\
> -} while (0)
> -#endif /* CONFIG_MMU */
>  
>  extern void __get_user_unknown(void);
>  
> @@ -140,7 +129,6 @@ do {							\
>  	}						\
>  } while (0)
>  
> -#ifdef CONFIG_MMU
>  #define __put_user_asm(x, addr, err, insn)			\
>  do {								\
>  	__asm__ __volatile__ (					\
> @@ -164,17 +152,6 @@ do {								\
>  		: "memory"					\
>  	);							\
>  } while (0)
> -#else
> -#define __put_user_asm(x, addr, err, insn)		\
> -do {							\
> -	__asm__ __volatile__ (				\
> -		"mov." insn "	%0, %1\n\t"		\
> -		: /* no outputs */			\
> -		: "r" (x), "m" (__m(addr))		\
> -		: "memory"				\
> -	);						\
> -} while (0)
> -#endif /* CONFIG_MMU */
>  
>  #if defined(CONFIG_CPU_LITTLE_ENDIAN)
>  #define __put_user_u64(val,addr,retval) \

Wouldn't that make these operations slower or do you think that GCC is able
to optimize this well enough?

Also, this is something that should definitely be boot-tested to make sure
this doesn't introduce any regressions.

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer
`. `'   Physicist
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913