linux-kernel - Re: [PATCH 3/4] sh: Add SECCOMP

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20200829004939.GB3265@brightrain.aerifal.cx>
Date:   Fri, 28 Aug 2020 20:49:39 -0400
From:   Rich Felker <dalias@...c.org>
To:     John Paul Adrian Glaubitz <glaubitz@...sik.fu-berlin.de>
Cc:     Michael Karcher <kernel@...rcher.dialup.fu-berlin.de>,
        linux-sh@...r.kernel.org, linux-kernel@...r.kernel.org,
        Yoshinori Sato <ysato@...rs.sourceforge.jp>
Subject: Re: [PATCH 3/4] sh: Add SECCOMP_FILTER

On Fri, Aug 28, 2020 at 01:03:00PM -0400, Rich Felker wrote:
> On Fri, Aug 28, 2020 at 06:38:09PM +0200, John Paul Adrian Glaubitz wrote:
> > Hi!
> > 
> > On 8/28/20 6:30 PM, Rich Felker wrote:
> > > I'm about to test a patch along these lines and will report what I
> > > find.
> > 
> > Let me know when you have something to test and I will test the patch as
> > well, making sure we're not breaking seccomp again.
> 
> If you have a seccomp test setup, please try the following patch. I'm
> not sure if the end result is entirely correct, but I believe it's
> at least much closer to correct than the code was before or after
> adding SECCOMP_FILTER.
> 
> 
> diff --git a/arch/sh/kernel/entry-common.S b/arch/sh/kernel/entry-common.S
> index ad963104d22d..0560a8054215 100644
> --- a/arch/sh/kernel/entry-common.S
> +++ b/arch/sh/kernel/entry-common.S
> @@ -368,9 +368,6 @@ syscall_trace_entry:
>  	mov.l	7f, r11		! Call do_syscall_trace_enter which notifies
>  	jsr	@r11	    	! superior (will chomp R[0-7])
>  	 nop
> -	cmp/eq	#-1, r0
> -	bt	syscall_exit
> -	mov.l	r0, @(OFF_R0,r15)	! Save return value
>  	!			Reload R0-R4 from kernel stack, where the
>  	!   	    	    	parent may have modified them using
>  	!   	    	    	ptrace(POKEUSR).  (Note that R0-R2 are
> @@ -382,7 +379,7 @@ syscall_trace_entry:
>  	mov.l	@(OFF_R5,r15), r5
>  	mov.l	@(OFF_R6,r15), r6
>  	mov.l	@(OFF_R7,r15), r7   ! arg3
> -	mov.l	@(OFF_R3,r15), r3   ! syscall_nr
> +	mov	r0, r3              ! syscall_nr, possibly changed to -1
>  	!
>  	mov.l	6f, r10			! Number of syscalls
>  	cmp/hs	r10, r3
> diff --git a/arch/sh/kernel/ptrace_32.c b/arch/sh/kernel/ptrace_32.c
> index 25ccfbd02bfa..9e86cff041c7 100644
> --- a/arch/sh/kernel/ptrace_32.c
> +++ b/arch/sh/kernel/ptrace_32.c
> @@ -503,7 +503,7 @@ asmlinkage long do_syscall_trace_enter(struct pt_regs *regs)
>  	audit_syscall_entry(regs->regs[3], regs->regs[4], regs->regs[5],
>  			    regs->regs[6], regs->regs[7]);
>  
> -	return ret ?: regs->regs[0];
> +	return ret ?: regs->regs[3];
>  }
>  
>  asmlinkage void do_syscall_trace_leave(struct pt_regs *regs)

This restored my ability to use strace, and I've written and tested a
minimal strace-like hack using SECCOMP_RET_USER_NOTIF that works as
expected on both j2 and qemu-system-sh4, so I think the above is
correct.

Rich